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
a @xmath0-expansion with deleted digits of a number @xmath1 is an expression of the form @xmath2 where @xmath3 is a real number and the @xmath4 are all elements of a set of real numbers @xmath5 . in @xcite , pedicini showed that if @xmath3 and if the set of real numbers @xmath6 satisfies the following two conditions : * @xmath7 , * @xmath8 , then every @xmath9 $ ] has an expansion of the form ( [ q : exp ] ) , with @xmath10 for all @xmath11 . he also gave an iterative algorithm that generates greedy expansions of the form ( [ q : exp ] ) , in the sense that at each step of the algorithm , @xmath4 is chosen to be the largest element of @xmath12 such that @xmath13 in @xcite a dynamical system is given that generates all possible expansions of the form ( [ q : exp ] ) for points @xmath9 $ ] , if @xmath12 satisfies ( i ) and ( ii ) . we call a set @xmath12 , satisfying ( i ) and ( ii ) an _ allowable digit set_. if @xmath12 is not allowable , then not every point has an expansion . the size of the set of real numbers that can be represented by expression ( [ q : exp ] ) , for different choices of @xmath0 and for @xmath14 , was already studied by keane , smorodinsky and solomyak in @xcite . pollicott and simon studied the hausdorff dimension of this set of reals in the case @xmath12 is a subset of the non - negative integers in @xcite . + the @xmath0-expansions with deleted digits are a generalization of the @xmath0-expansions with what we could call a complete digit set , i.e. with @xmath15 , where @xmath16 denotes the largest integer less than or equal to @xmath1 . a lot of research has been done on this topic and we mention some results here . for a given @xmath0 , all @xmath17 $ ] have an expansion with a complete digit set and almost all @xmath1 in this interval have continuum expansions of this form . ( see @xcite by erds , jo and komornik and @xcite by sidorov for more information . ) one way to generate @xmath0-expansions with a complete digit set is by iterating the transformation @xmath18 \to \left [ 0 , \frac{\lfloor \beta \rfloor}{\beta -1 } \right ] : x \to \left\ { \begin{array}{ll } \beta x \ ; ( \text{mod } 1 ) , & \text{if } 0 \leq x < \displaystyle \frac{\lfloor \beta \rfloor}{\beta},\\ \\ \beta x - \lfloor \beta \rfloor , & \text{if } \displaystyle \frac{\lfloor \beta \rfloor}{\beta } \leq x \leq \frac{\lfloor \beta \rfloor}{\beta-1}. \end{array } \right.\ ] ] if we set @xmath19 , \end{array } \right.\ ] ] and for @xmath11 , @xmath20 , then @xmath21 , and for any @xmath22 , @xmath23 letting @xmath24 , it is easily seen that @xmath25 we call the @xmath0-expansion generated by the transformation @xmath26 the greedy @xmath0-expansion with a complete digit set of @xmath1 , since it gives us at each step in the expansion the largest digit possible . more precisely , for each @xmath11 , if @xmath27 are already known , then @xmath28 is the largest element of the complete digit set @xmath29 , such that @xmath30 throughout the rest of the paper , @xmath31 will denote the 1-dimensional lebesgue measure . the transformation @xmath26 has a unique invariant measure , absolutely continuous with respect to @xmath31 . rnyi proved the existence of this measure ( @xcite ) and gelfond and parry independently gave an explicit formula for the density function of this measure in @xcite and @xcite respectively . the invariant measure has the unit interval @xmath32 as its support and the density function @xmath33 is given as follows . @xmath34 where @xmath35 is a normalizing constant . from now on we will refer to an invariant measure , absolutely continuous with respect to @xmath31 as an _ acim_. + for the @xmath0-expansions with deleted digits we also have a transformation that generates the greedy expansions by iteration . this transformation is called the greedy @xmath0-transformation with deleted digits . if @xmath3 is a real number and @xmath36 is an allowable digit set of which the first digit equals zero , i.e. @xmath37 , then this transformation @xmath38 \to \left[0 , \frac{a_m}{\beta -1 } \right]$ ] is given by @xmath39 . \end{array } \right.\ ] ] in @xcite it is shown that a greedy @xmath0-transformation with deleted digits for a given @xmath3 and an allowable digit set @xmath40 is isomorphic to a greedy @xmath0-transformation with deleted digits for the same @xmath3 , but with allowable digit set @xmath41 , i.e. a digit set of which the first digit equals zero . therefore we can assume without loss of generality that @xmath37 for all allowable digit sets . the sequence of digits @xmath42 can be defined in a way similar to ( [ q : digitset ] ) as follows . put @xmath43 if @xmath44 and @xmath45 if @xmath46 $ ] and for @xmath11 , put @xmath47 . the definition of the transformation @xmath48 is based on the greedy algorithm with deleted digits that was defined by pedicini in @xcite and the transformation therefore generates greedy @xmath0-expansions with deleted digits satisfying ( [ q : greedy ] ) . ( for more information , see @xcite . ) if @xmath49 is the greedy @xmath0-expansion with deleted digits for @xmath1 , we also write @xmath50 which is understood to mean the same as ( [ q : exp ] ) . + from @xcite we know that the transformation @xmath48 has an acim that is unique and ergodic . the support of this measure is given by the interval @xmath51 , where @xmath52 an explicit expression for the density function of this measure however , is given only for special cases . one of these special cases is , when @xmath53 before we can give this density , however , we need some more notation . in @xcite it is proven that the minimal amount of digits in an allowable digit set is @xmath54 . in other words , the amount of digits in @xmath12 is at least equal to the smallest integer larger than or equal to @xmath0 . + let @xmath55 be the largest element of the set @xmath56 such that @xmath57 . define a partition @xmath58 of the support of the acim of @xmath48 , where for @xmath59 , we have @xmath60 and @xmath61 note that @xmath62 and for @xmath63 , @xmath64 . using @xmath65 and @xmath48 , we can make a sequence of partitions @xmath66 : for @xmath67 , @xmath68 the elements of @xmath69 are intervals and are called the _ fundamental intervals of rank _ @xmath70 . if @xmath71 is an element of @xmath69 , denote it by @xmath72 . we call a fundamental interval @xmath72 _ full _ of rank @xmath70 if @xmath73 and _ non - full _ otherwise . this means that for a full fundamental interval , @xmath74 we have @xmath75 and if @xmath74 is non - full , then @xmath76 let @xmath77 be the collection of all non - full fundamental intervals of rank @xmath70 , that are not subsets of any full fundamental interval of lower rank . for @xmath78 , define @xmath79 and for @xmath80 , let @xmath81 put @xmath82 . then the function @xmath83 is the density function of the acim of @xmath48 . ( this density is a special case of the density found by wilkinson in @xcite . ) notice that for the classical greedy @xmath0-transformation , @xmath77 contains at most one element @xmath84 and for this element we have @xmath85 so the density function from ( [ q : parrymeasure ] ) is a special case of the density function from ( [ q : wilkinson ] ) . + in this article we will give an expression for the density function of the acim of the greedy @xmath0-transformation with three deleted digits . this means that we will only be looking at @xmath86 and allowable digit sets @xmath12 of the form @xmath87 . since @xmath88 fits the framework above , we already know that the density is given by ( [ q : wilkinson ] ) . however , ( [ q : wilkinson ] ) will also turn out to be valid for @xmath89 . in the first section we will introduce some notation and discuss when we can directly relate the density function of @xmath48 to the density from equation ( [ q : wilkinson ] ) . for the case in which this is not immediately clear , we will derive a formula for the density by defining a version of the natural extension of the dynamical system @xmath90 here @xmath91 is the borel @xmath92-algebra on the support of the acim of @xmath48 and @xmath93 is the measure obtained by `` pulling back '' the appropriate measure , defined on the natural extension . we will define this version of the natural extension in the second section . the density of the measure @xmath93 will again be the density from ( [ q : wilkinson ] ) . for the greedy @xmath0-transformation with a complete digit set , a version of the natural extension is given in @xcite and also in @xcite by brown and yin . our definition is based on the version given in the first paper . using the measure @xmath93 , we obtain that the transformation @xmath48 is exact and weakly bernoulli . to illustrate the construction of our version of the natural extension , an example of a specific greedy @xmath0-transformation with three deleted digits can be found in the last section . here @xmath0 is the golden mean and @xmath94 . the aim of this paper is to give an explicit expression for the acim of the greedy @xmath0-transformation with three deleted digits . for @xmath88 , we already know that the density function of this measure is given by ( [ q : wilkinson ] ) . we will first give some conditions under which we can directly see that the density from ( [ q : wilkinson ] ) is the density of the acim . + let @xmath48 be the greedy @xmath0-transformation with deleted digits for which @xmath86 and for which @xmath87 is an allowable digit set . so , we set @xmath37 . the transformation then becomes : @xmath95 . \end{array } \right.\ ] ] the sequence of digits @xmath96 is defined for any @xmath97 $ ] as in the introduction . before we can give the density function of the acim , we need to find its support . from ( [ q : i0 ] ) we know that this support is either the interval @xmath98 or the interval @xmath99 . the following situations can occur . + suppose first that @xmath100 . either @xmath101 or @xmath102 . + if @xmath101 , then the support is @xmath98 and since @xmath103 , we get that @xmath104 . the transformation @xmath48 on the interval @xmath98 is then isomorphic to the classical greedy @xmath0-transformation , @xmath26 , for the same @xmath0 . figure [ f : attractorcases](a ) is an example of this . + if @xmath102 , then the support of the invariant measure is the other interval , @xmath99 , and we can deduce that @xmath105 . so , criterion ( [ q : critwilkinson ] ) applies and the density for the invariant measure is given by ( [ q : wilkinson ] ) . in figure [ f : attractorcases](b ) we see an example . + suppose that @xmath106 . either @xmath107 or @xmath108 . + if @xmath109 , then the support of the acim is @xmath99 and we have that @xmath110 . so again @xmath105 and the density from equation ( [ q : wilkinson ] ) is the density of the invariant measure we are looking for . see figure [ f : attractorcases](c ) for an example . + if @xmath111 , then the support is @xmath112 and @xmath113 . the transformation in this case is isomorphic to the classical greedy transformation for the same @xmath0 . similarly , if @xmath114 and @xmath115 , then @xmath104 and the support of the acim is @xmath98 . the transformation @xmath48 is again isomorphic to the classical greedy @xmath0-transformation for the same @xmath0 . + lastly , suppose that @xmath116 . if @xmath104 , then this would imply @xmath117 which gives a contradiction . so this means that @xmath105 and thus that the density from ( [ q : wilkinson ] ) is the density for the acim . + the only situation we did not yet consider , is when @xmath118 and @xmath119 . to this , the rest of the paper is dedicated . figure [ f : attractorcases](d ) gives an example of a transformation that satisfies these conditions . from now on , suppose that @xmath48 is a greedy @xmath0-transformation with an allowable digit set @xmath120 , that satisfies the following equations : @xmath121 and @xmath122 . of course , since @xmath12 is allowable and @xmath123 , we also have @xmath124 . this is all captured in the following condition . @xmath125 notice that we do not assume that @xmath104 , although we already know that the density from ( [ q : wilkinson ] ) is the density of the acim in this case . the reason is that the construction of the natural extension that will be given in what follows , is also valid for @xmath126 . [ r:1beta2 ] observe that , if @xmath89 , then @xmath127 implies condition ( [ q : conditions ] ) . so , if @xmath89 and a digit set @xmath128 satisfies @xmath129 , then @xmath12 is an allowable digit set . the support of the acim of the transformation @xmath48 is the interval @xmath98 . the partition @xmath130 of this interval is given in the following way : @xmath131 as explained in the introduction , we can construct the sequence of partitions @xmath132 , with @xmath69 as defined in ( [ q : deltan ] ) . the elements of @xmath69 are denoted by @xmath84 and are either full or non - full fundamental intervals of rank @xmath70 . we have the following obvious lemma . [ l : fullfi ] let @xmath133 and @xmath134 be two full fundamental intervals of rank @xmath135 and @xmath136 respectively . then @xmath137 is a full fundamental interval of rank @xmath138 . recall that for @xmath11 , @xmath77 is the collection of all non - full fundamental intervals of rank @xmath70 that are not contained in any full fundamental interval of lower rank . let @xmath139 be the number of elements in @xmath77 . so @xmath140 , since this is the number of non - full fundamental intervals of rank 1 and for all @xmath11 , @xmath141 . the version of the natural extension that we will define in the next section , uses all the elements of @xmath77 for all @xmath11 . to make sure that the total measure of the underlying space of this version is finite , we need upper bounds for the values of @xmath139 . to obtain these , we will first describe the structure of the elements of @xmath77 . + notice that by ( [ q : gennonfull ] ) we have that for all elements @xmath142 , @xmath143 for an @xmath144 , the set @xmath145 is called the _ orbit _ of @xmath1 under the transformation @xmath48 . let @xmath146 be the greedy @xmath0-expansions with deleted digit set @xmath12 of the points @xmath147 and @xmath148 , i.e. the expansions generated by iterations of @xmath48 . the number @xmath148 would be the image of @xmath149 under @xmath48 if @xmath48 were not restricted to the interval @xmath150 . the values of the numbers @xmath139 depend on the orbits of the points @xmath151 and @xmath152 under @xmath48 and whether or not @xmath153 and @xmath154 are elements of @xmath155 for @xmath156 . to see this , notice that for any @xmath142 one has @xmath157 and the set @xmath158 has the form @xmath159 for some @xmath160 . suppose @xmath161 . + if @xmath162 , then @xmath84 contains exactly one element of @xmath163 , namely @xmath164 in case @xmath165 or @xmath166 in case @xmath167 . furthermore , in the first case , @xmath168 and also in the second case , @xmath169 on the other hand , if @xmath170 , then @xmath84 contains exactly two elements of @xmath163 , namely the sets @xmath171 and @xmath172 . now , @xmath173 and @xmath174 similar arguments hold in case @xmath175 , except that @xmath153 is replaced by @xmath154 . + for @xmath11 , let @xmath176 be the number of elements from @xmath77 that contain two elements from @xmath163 . then clearly for all @xmath11 , @xmath177 from the above we see that in order to get an upper bound on @xmath139 , we need to study the behavior of the orbits of @xmath178 and @xmath151 . the following three lemmas say something about the first few elements of the orbits of these points . these lemmas are needed to guarantee the total measure of the underlying space of the natural extension will be finite . [ l : gbeta2 ] if @xmath89 and @xmath127 , then @xmath179 . since @xmath104 , we have @xmath180 . thus @xmath181 and hence @xmath182 . this proves the lemma . observe that @xmath77 only contains fundamental intervals of which the first digit is either @xmath149 or @xmath183 . let @xmath184 denote the number of elements @xmath74 in @xmath77 such that @xmath185 and @xmath186 the number of elements in @xmath77 that have @xmath183 as their first digit . then of course for all @xmath11 , @xmath187 let @xmath188 denote the fibonacci sequence , i.e. let @xmath189 , @xmath190 and for @xmath191 , let @xmath192 . lemma [ l : gbeta2 ] implies that the number of elements of @xmath163 would be maximal if the only elements of @xmath77 that do not contain two elements from @xmath163 are the elements @xmath84 for which @xmath193 . in this maximal situation we would have @xmath194 and for @xmath195 , @xmath196 for @xmath197 we would have that @xmath198 . this means that under the conditions from lemma [ l : gbeta2 ] , we have for all @xmath11 that @xmath199 and @xmath200 .3 cm let @xmath201 be the golden mean , i.e. the positive solution of the equation @xmath202 . [ l : lessthang ] let @xmath203 and @xmath122 . then @xmath204 . since @xmath205 , we have @xmath206 , so by equation ( [ q : conditions ] ) , @xmath207 . thus @xmath208 and hence @xmath209 . on the other hand , since @xmath210 , we have @xmath211 . thus @xmath212 and we have that @xmath213 . [ r : wortel2 ] this lemma implies that for @xmath203 and for digit sets satisfying condition ( [ q : conditions ] ) , we have @xmath214 . the largest amount of elements for @xmath77 would be obtained if @xmath215 and @xmath216 for all odd values of @xmath217 and thus @xmath218 for all even values of @xmath217 . in this case , @xmath219 this would imply that for @xmath11 , @xmath220 . so , in general we have that for all @xmath11 , @xmath221 . [ l:2to1overm ] let @xmath222 and @xmath122 . if @xmath223 , then @xmath224 for all @xmath225 . the proof is by induction on @xmath226 . note that from lemma [ l : lessthang ] we know that @xmath227 . assume first that @xmath228 and thus @xmath229 . then @xmath230 if @xmath231 , then @xmath232 and @xmath233 . so @xmath234 . on the other hand , since @xmath235 , we have @xmath236 if @xmath231 , then @xmath237 . so @xmath238 and thus @xmath239 . + now , assume that the result is true for some @xmath240 . let @xmath241 and @xmath242 . then certainly @xmath243 , so by induction @xmath244 for all @xmath245 . we only need to show that @xmath246 . first consider @xmath247 . we have @xmath248 since @xmath242 , then @xmath249 thus @xmath250 . we now consider @xmath251 and see that @xmath252 since @xmath242 , then @xmath253 thus @xmath254 and this proves the lemma . [ r : smallbeta ] suppose @xmath255 and @xmath256 . lemma [ l:2to1overm ] implies that @xmath257 for @xmath258 . by the same reasoning as in remark [ r : wortel2 ] , @xmath139 would obtain the largest possible value if @xmath259 for all @xmath260 , @xmath261 , and @xmath218 for all other values of @xmath217 . this would imply @xmath262 thus , to get the maximal number of elements for @xmath77 , we would have that if @xmath263 for some @xmath264 , then @xmath265 . so , in general we have that @xmath266 . for all @xmath11 , let @xmath267 be the union of all full fundamental intervals of rank @xmath70 , that are not a subset of any full fundamental interval of lower rank . from the next lemma it follows that the full fundamental intervals generate the borel @xmath92-algebra on @xmath98 . [ l : generate ] @xmath268 notice that all of the sets @xmath267 are disjoint . by ( [ q : nonfull ] ) we have for each @xmath11 , that @xmath269 so it is enough to prove that @xmath270 . if @xmath126 , then since @xmath141 , we immediately have the result . for @xmath203 , it follows from remark [ r : wortel2 ] and remark [ r : smallbeta ] . now , suppose @xmath271 . then by ( [ q : gbeta2 ] ) , we have that @xmath272 , where @xmath273 is the @xmath274-th element of the fibonacci sequence . for the elements of this sequence , there is a closed formula which gives @xmath275 so @xmath276.\ ] ] since @xmath271 , also in this case @xmath270 and this proves the lemma . [ r : generate ] the fact that @xmath277 is a full fundamental interval of rank 1 allows us to construct full fundamental intervals of arbitrary small lebesgue measure . this , together with the previous lemma , guarantees that we can write each subinterval of @xmath150 as a countable union of full fundamental intervals . thus , the full fundamental intervals generate the borel @xmath92-algebra on @xmath150 . notice that for the cases illustrated by figure [ f : attractorcases](b ) and [ f : attractorcases](c ) , we can define the partitions @xmath69 , the sets @xmath77 and the numbers @xmath139 in a similar way . the only differences are that the support of the acim is given by the interval @xmath99 and that @xmath278 is the only full fundamental interval of rank 1 . in that sense , @xmath278 plays the role of @xmath277 above . since in these cases we have @xmath126 and since @xmath141 for all @xmath11 , we can prove a lemma similar to lemma [ l : generate ] , i.e. we can prove that the full fundamental intervals generate the borel @xmath92-algebra on the support of the acim . for the version of the natural extension , we will define a space @xmath279 , using the element of @xmath77 . for @xmath11 , define the collections @xmath280 so to each element of @xmath77 , there corresponds an element of @xmath281 and thus the number of elements in @xmath281 is given by @xmath139 . we enumerate the elements of @xmath281 and write @xmath282 . thus , for each @xmath142 there exists a unique @xmath283 such that @xmath284 . let @xmath285 and let @xmath279 be the disjoint union of all these sets , i.e. @xmath286 the @xmath92-algebra @xmath287 on @xmath279 is the disjoint union of the borel @xmath92-algebras on all the rectangles @xmath288 and @xmath289 . let @xmath290 be the measure on @xmath279 , given by the two dimensional lebesgue measure on each of these rectangles . define the probability measure @xmath291 on @xmath279 by setting @xmath292 for all @xmath293 . the next lemma says that this measure is well defined and finite . [ l : finitemeasure ] @xmath294 set @xmath295 . then , @xmath296 using the same arguments as in the proof of lemma [ l : generate ] , we can show that the sum on the right hand side converges for all @xmath297 . the transformation @xmath298 is defined piecewise on each rectangle . if @xmath299 , then @xmath300 for each @xmath11 , @xmath301 , if @xmath302 then @xmath303 maps this rectangle to the rectangles corresponding to the fundamental intervals contained in @xmath84 in the following way . if @xmath304 is full and @xmath305 with @xmath306 , then @xmath307 if @xmath308 and @xmath309 is the index of the corresponding set in @xmath310 , then for @xmath305 with @xmath311 , we set @xmath312 in figure [ f : mathcalt ] we show the different situations in case @xmath313 for some @xmath314 . if @xmath315 for some @xmath316 , the pictures look exactly the same with @xmath317 in place of @xmath318 . notice that if a rectangle @xmath288 corresponds to a fundamental interval @xmath74 such that @xmath319 is non - full , then this is the only fundamental interval contained in @xmath84 . @xmath303 then maps the rectangle @xmath288 bijectively to the rectangle @xmath320 , corresponding to @xmath304 . otherwise , @xmath304 is a full fundamental interval contained in @xmath84 and the rectangle @xmath288 is partly mapped surjectively onto some rectangle @xmath321 and partly into @xmath289 . from lemma [ l : generate ] it now follows that @xmath303 is bijective . + .3 cm let @xmath322 be the projection onto the first coordinate . define the measure @xmath93 on @xmath323 by pulling back the measure @xmath291 , i.e. for all measurable sets @xmath324 , let @xmath325 . in order to show that @xmath326 is a version of the natural extension of the system @xmath327 with @xmath328 as the factor map , we will prove all of the following . * @xmath328 is a surjective , measurable and measure preserving map from @xmath279 to @xmath150 . * for all @xmath329 , we have @xmath330 . * @xmath331 is an invertible transformation . * @xmath332 , where @xmath333 is the smallest @xmath92-algebra containing the @xmath92-algebras @xmath334 for all @xmath11 . it is clear that @xmath328 is surjective and measurable and that @xmath335 . since @xmath303 expands by a factor @xmath0 in the first coordinate and contracts by a factor @xmath0 in the second coordinate , it is also clear that @xmath303 is invariant with respect to the measure @xmath291 . then @xmath336 defines a @xmath48-invariant probability measure on @xmath337 , that is equivalent to the normalized lebesgue measure on @xmath150 and @xmath328 is a measure preserving map . this shows ( i ) and ( ii ) . the invertibility of @xmath303 follows from remark [ r : generate ] , so that leaves only ( iv ) . to prove ( iv ) we will have a look at the structure of the fundamental intervals and we will introduce some more notation . + let @xmath74 be a fundamental interval . we can divide the block of digits @xmath338 into @xmath339 subblocks , @xmath340 , for some @xmath341 , where each subblock @xmath342 , @xmath343 , corresponds to a full fundamental interval . the last subblock , @xmath344 , corresponds to a full fundamental interval exactly when @xmath74 is full . we can make this precise , using the notion of return times to @xmath289 . for points @xmath299 define the _ first return time to _ @xmath289 by @xmath345 and for @xmath346 , let the @xmath347_-th return time to _ @xmath289 be given recursively by @xmath348 by the poincar recurrence theorem , we have @xmath349 for almost all @xmath299 . notice that this notion of return time depends only on @xmath1 , i.e. for all @xmath350 and all @xmath346 , @xmath351 . so we can write @xmath352 instead of @xmath353 . if @xmath354 , then for all @xmath355 , @xmath356 maps the whole rectangle @xmath357 to the same rectangle in @xmath279 . so up to a certain @xmath358 , the @xmath217-th return time to @xmath289 is equal for all elements in @xmath84 . now suppose that @xmath359 is a full fundamental interval , then there is an @xmath341 and there are numbers @xmath360 , @xmath361 such that @xmath362 for all @xmath363 and @xmath364 . put @xmath365 . we can also obtain the numbers @xmath360 inductively as follows . let @xmath366 and if @xmath367 are already known , let @xmath368 take for @xmath361 , @xmath369 let @xmath370 denote the number of digits of the block @xmath342 . the blocks have the following properties . * for @xmath361 , @xmath371 . * if @xmath372 , then @xmath373 . this means that if a subblock begins with the digit 0 , then 0 is the only digit in this subblock . so , @xmath374 consists just of the digit 0 . * for all @xmath375 , @xmath376 is a full fundamental interval of rank @xmath377 . the next lemma is the last step in the proving that @xmath326 is the natural extension of the space @xmath327 . [ l : bigvee ] let @xmath326 and @xmath327 be the dynamical systems defined above . then @xmath378 it is clear that @xmath379 by lemma [ l : generate ] we know that the direct products of the full fundamental intervals contained in the rectangle @xmath289 generate the borel @xmath92-algebra on this rectangle . the same holds for all the rectangles @xmath288 . first , let @xmath380 be a generating rectangle in @xmath289 , where @xmath133 and @xmath134 are full fundamental intervals . for the set @xmath134 construct the subblocks @xmath381 as in ( [ q : subblocks ] ) . by property ( p3 ) and lemma [ l : fullfi ] , @xmath382 is a full fundamental interval of rank @xmath138 . then @xmath383 @xmath384 since @xmath134 is a full fundamental interval , it can be proven by induction that for all @xmath385 , @xmath386 . this , together with the definitions of the blocks @xmath387 and the transformation @xmath303 leads to @xmath388 @xmath389 so @xmath390 now , for @xmath11 and @xmath391 , let @xmath288 be a rectangle in @xmath281 and suppose that it corresponds to the fundamental interval @xmath392 . hence , @xmath393 let @xmath394 be a generating rectangle for the borel @xmath92-algebra on the rectangle @xmath288 . so @xmath133 and @xmath134 are again full fundamental intervals . notice that @xmath395 which means that @xmath396 . also , for all @xmath397 , @xmath398 and thus @xmath399 . so , if we divide @xmath400 into subblocks @xmath387 as before , we get that @xmath401 , that @xmath402 and that @xmath403 . consider the set @xmath404 we will show the following . + claim : the set @xmath405 is a fundamental interval of rank @xmath138 and @xmath406 . .2 cm first notice that @xmath407 so obviously , @xmath408 by lemma [ l : fullfi ] , @xmath409 is a full fundamental interval of rank @xmath410 , so @xmath411 . now , by the definition of @xmath288 we have that @xmath412 and thus @xmath413 . + for the other inclusion , let @xmath414 . by ( [ q : deltaa ] ) , there is an element @xmath415 in @xmath84 , such that @xmath416 . and since @xmath417 , there is an @xmath418 with @xmath419 , so @xmath420 . this means that @xmath421 so @xmath422 and this proves the claim . .2 cm consider the set @xmath423 . then as before , we have @xmath424 and after @xmath70 more steps , @xmath425 so , @xmath426 and thus we see that @xmath427 this gives the following theorem . the dynamical system @xmath326 is a version of the natural extension of the dynamical system @xmath428 . here , the measure @xmath93 , given by @xmath429 for all measurable sets @xmath430 , is the acim of @xmath48 and the density of @xmath93 is equal to the density from equation ( [ q : wilkinson ] ) . the fact that @xmath326 is a version of the natural extension of the system @xmath428 follows from remark [ r : generate ] , the properties of the map @xmath328 and lemma [ l : bigvee ] . now for each measurable set @xmath431 , we have @xmath432\\ & = & \frac{a_1}{\lambda_r(r ) } \int_e \left [ 1 + \sum_{n=1}^{\infty } \sum_{\delta(b_0 \ldots b_{n-1 } ) \in b_n } \frac{a_1}{\beta^n } \ ; 1_{t^n \delta(b_0 \ldots b_{n-1 } ) } \right ] d\lambda.\end{aligned}\ ] ] here @xmath158 has the form @xmath159 for some @xmath160 . so , the density of the measure @xmath93 equals the density from equation ( [ q : wilkinson ] ) . \(i ) the above definitions of the space @xmath279 and the transformation @xmath303 can be adapted quite easily for the cases illustrated by figure [ f : attractorcases](b ) and [ f : attractorcases](c ) . we let @xmath278 take the role of @xmath277 and consider the orbits of the points @xmath149 and @xmath433 . in general , the sets @xmath77 will contain more elements , but since @xmath126 , it is immediate that @xmath434 . this shows that we can construct a version of the natural extension of @xmath48 , also for these two cases . + ( ii ) let @xmath435 be the set obtained from @xmath289 by removing the set of measure zero of elements which do not return to @xmath289 , i.e. we remove those @xmath436 for which @xmath437 . let @xmath438 be the transformation induced by @xmath303 , i.e. for all @xmath439 , let @xmath440 then the dynamical system @xmath441 , where @xmath442 is the borel @xmath92-algebra on @xmath435 , is isomorphic to the natural extension of a gls - transformation as defined in @xcite . this implies that the system @xmath441 is bernoulli . using this invariant measure , we will show that @xmath48 is an exact transformation . since the full fundamental intervals generate the borel @xmath92-algebra on the support of the acim , by a result of rohlin ( @xcite ) , it is enough to show that there exists a universal constant @xmath443 , such that for any full fundamental interval @xmath84 and any measurable subset @xmath444 , we have @xmath445 to this end , define two constants , @xmath446 , by @xmath447 then for all measurable sets @xmath430 , we have @xmath448 now , let @xmath74 be a full fundamental interval of rank @xmath70 . then by ( [ q : genfull ] ) , @xmath449 let @xmath450 be a measurable set . then @xmath451 now , @xmath452 if we take @xmath453 , then @xmath443 and @xmath445 thus , @xmath48 is exact and hence mixing of all orders . furthermore , the natural extension @xmath303 is a @xmath454-automorphism . by a result of rychlik ( see @xcite ) , it follows immediately that @xmath48 is weakly bernoulli . in this section we will consider one specific example of a @xmath0-transformation with three deleted digits . let @xmath455 be the golden mean as before and take @xmath456 . consider the allowable digit set @xmath457 . the acim of the greedy @xmath0-transformation with deleted digits with this @xmath0 and @xmath12 , has the interval @xmath458 as its support . the partition @xmath459 is given by @xmath460 and the transformation is then @xmath461 , if @xmath462 . the points @xmath463 and @xmath464 are of special interest and their orbits under @xmath48 are as follows . @xmath465 thus their greedy @xmath0-expansions are given by @xmath466 where the bar on the right hand side of these equations indicates a repeating block in the expansions . notice that @xmath467 would be the image of 3 under @xmath48 , if @xmath48 were not restricted to the half open interval @xmath458 . figure [ f : goud034 ] shows the orbits of both points under @xmath48 . for each non - full fundamental interval , @xmath84 , the set @xmath158 is one of the following , @xmath468 this means that we could give all the elements of @xmath281 explicitly . we will not do this , since figure [ f : tree ] speaks for itself . the space @xmath279 contains all the rectangles shown in the figure . + consists of all these rectangles . the arrows indicate where the rectangles are mapped under @xmath303 . ] to determine the invariant measure of @xmath48 , we need to project the lebesgue measure on the rectangles of @xmath279 onto the first coordinate . therefore , we need to add up the heights of all the rectangles which have the same interval in the first coordinate . to do this , we only need to determine the total height of the rectangles of the form @xmath469 , since the total height of all the other rectangles in @xmath279 can be deduced from this . these heights can be found by using the fibonacci numbers @xmath470 , as defined before . then @xmath471 and observe that for @xmath191 , @xmath472 . for @xmath473 , @xmath474 , the number of rectangles in @xmath279 of the form @xmath469 is equal to @xmath475 and for @xmath476 , @xmath346 , this number is equal to @xmath477 . formula ( [ q : fib ] ) gives that the total lebesgue measure of all these rectangle is equal to @xmath478\\ & = & \frac{3}{\beta}\left[1+\frac{1}{\sqrt 5}\left(\beta \sum_{k=1}^{\infty } \frac{1}{\beta^{2k } } + \frac{1}{\beta}\sum_{k=1}^{\infty } \frac{1}{\beta^{2k } } \right)\right]\\ & = & \frac{3}{\beta}\left[1+\frac{1}{\sqrt 5}\left ( ( \beta + \frac{1}{\beta } ) ( \frac{\beta^2}{\beta^2 - 1 } -1 ) \right)\right]\\ & = & \frac{3}{\beta}\left[1+\frac{1}{\sqrt 5}(3-\beta)\right]=3.\end{aligned}\ ] ] the total height of the rectangles of the form @xmath479 is now equal to @xmath480 , that of the rectangles of the form @xmath481 is @xmath482 , etcetera . the total height of the rectangles @xmath483 is given by @xmath484 . then the density function of the invariant probability measure of @xmath48 , @xmath485 , equivalent to the normalized lebesgue measure on @xmath458 , is given by @xmath486 .\end{aligned}\ ] ] we have seen that it is possible to find an explicit expression for the density function of the acim of the greedy @xmath0-transformation with three deleted digits . this density function is equal to the density found by wilkinson in @xcite . we have constructed a version of the natural extension of the dynamical system of the greedy @xmath0-transformation with deleted digits to find this density in case the real number @xmath487 and the digit set @xmath12 satisfy condition ( [ q : conditions ] ) . a very similar construction can be used to give a version of the natural extension for the transformation in case @xmath488 does not satisfy this condition . the transformation @xmath48 is exact and weakly bermoulli . + of course , it still remains to find an explicit expression for the acim of the greedy @xmath0-transformation with deleted digits , of which the digit set contains more than three digits . if the number of digits , @xmath226 , satisfies @xmath489 , the density is given by the density from ( [ q : wilkinson ] ) . for this case , one could try to generalize the definition of the version of the natural extension that we gave in this paper . the space @xmath279 would then become an @xmath226-ary tree and the finiteness of the measure of @xmath279 is guaranteed , since @xmath490 . + recently , gra found an invariant density for piecewise linear and increasing maps of constant slope , for certain values of @xmath0 . see @xcite .	we give an explicit expression for the invariant measure , absolutely continuous with respect to the lebesgue measure , of the greedy @xmath0-transformation with three deleted digits . we define a version of the natural extension of the transformation to obtain this expression . we get that the transformation is exact and weakly bernoulli . [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] [ section ]
in a previous paper , we studied an extension of the standard model in which a real scalar @xmath2 triplet with zero hypercharge is added to the usual scalar @xmath2 doublet @xcite . we showed that such an extension is allowed by the precision data and that the mass of the lightest higgs boson can be as big as 500 gev . to recap , the lagrangian of the model in terms of the usual standard model higgs , @xmath3 , and the new triplet , @xmath4 , reads @xmath5 with a scalar potential @xmath6 @xmath7 are the pauli matrices . the expansion of the field components is @xmath8 where @xmath9 and @xmath10 is the goldstone boson which is eaten by the @xmath11 . the model violates custodial symmetry at tree level giving a prediction for the @xmath12-parameter of @xmath13 as discussed in @xcite , it is precisely this violation of custodial symmetry which allows the lightest higgs to be much heavier than in the standard model . by giving the triplet a non - zero vacuum expectation value , one is in effect making a positive tree - level contribution to the @xmath14-parameter , and this is enough to allow a heavier higgs . in the neutral higgs sector we have two cp - even states which mix with angle @xmath15 . the mass eigenstates @xmath16 are defined by @xmath17 there is also mixing in the charged higgs sector . we define the mass eigenstates @xmath18 by @xmath19 the @xmath20 are the goldstone bosons corresponding to @xmath21 and , at tree level , the mixing angle is @xmath22 the precision electroweak data constrain @xmath23 to be smaller than about @xmath24 @xcite . in this paper we wish to examine the renormalization group flow of the couplings and hence establish bounds on the scalar masses under the assumption that the triplet model remain valid up to some scale @xmath25 . we take @xmath26 tev and make no statements about physics at higher scales . for the lagrangian of ( [ eqn : lag ] ) to remain appropriate up to @xmath25 , we demand that the scalar couplings @xmath27 remain perturbative and that the vacuum remain stable ( i.e. is a local minimum ) up to @xmath25 . we begin in the next section with the calculation of the beta - functions . in section 3 we present our results away from the decoupling limit of the model and in section 4 we discuss decoupling . the effective potential @xcite has the following one - loop expansion in the @xmath28 renormalization scheme and t hooft - landau gauge : @xmath29 @xmath30 is the renormalization scale and @xmath31 . we have included the contributions from all the relevant physical states including the heaviest fermion , the top quark . the terms with @xmath32 correspond to the counterterms of the theory and the tree - level masses are @xmath33 it is understood that we should substitute explicitly for the mixing angles , which are solutions to the equations @xmath34 the expressions for the counterterms are thus @xmath35 where @xmath36 is the counterterm for the vacuum energy . the fact that the theory should be independent of the unphysical mass @xmath30 implies that the couplings and masses acquire a @xmath30 dependence governed by the renormalization group ( rg ) equation for the one - loop effective potential , i.e. @xmath37 in terms of the tree level masses this equation is equivalent to @xmath38 and , matching powers of fields , we can derive the beta functions : @xmath39 we can now make use of the anomalous dimensions for the two neutral higgs fields @xmath40 to write down our final expressions for the one - loop beta functions : @xmath41 in the gauge and top quark sector the beta functions for the @xmath42 , @xmath43 and yukawa couplings are the same as in the standard model , i.e. @xmath44 the @xmath2 coupling is modified due to the extra higgs triplet in the adjoint representation , i.e. @xmath45 working with the tree - level effective potential with couplings evolved using the one - loop @xmath23 and @xmath15 functions we are able to resum the leading logarithms to all orders in the effective potential . it would be possible to include the next - to - leading logarithmic contributions by using the two - loop @xmath23 and @xmath15 functions and including the one - loop part of the effective potential , see @xcite . let us now turn to the rg analysis . we first introduce the parameter @xmath46 , related to the scale @xmath30 through @xmath47 . we shall perform evolution starting at @xmath48 . the rg equations are coupled differential equations in the set @xmath49 we choose rather to use the following set to define the input to the rg equations : @xmath50 within the accuracy to which we are working , the values of the couplings at @xmath48 can be obtained from the input set using the appropriate tree - level expressions . the vacuum conditions , @xmath51 allow us to write ( defining @xmath52 and @xmath53 ) @xmath54 inverting these relations we can thus fix the @xmath48 boundary conditions for the subsequent evolution : @xmath55 to ensure that the system remains in a local minimum we impose the condition that the squared masses should remain positive , i.e. @xmath56 we impose the further requirement that the couplings remain perturbative . in particular we insist that @xmath57 for @xmath58 and @xmath59 . we run the evolution from @xmath60 to @xmath61 , with @xmath26 tev . in this section we present our results of the higgs mass bounds in the regime where the triplet higgs can not be arbitrarily heavy . as we shall see in the next section , decoupling of the triplet occurs when both mixing angles and their sum @xmath62 tend to zero and in this case , the triplet decouples from the doublet and can be arbitrarily heavy . we are free to choose the 3 scalar masses and the 2 mixing angles at @xmath48 . in figure [ b004g0 ] we show the range of higgs masses allowed when there is no mixing in the neutral higgs sector , @xmath63 , for a value of @xmath64 . such a value is towards the upper end of the range allowed by the precision data and is interesting because it allows a rather heavy lightest higgs ( e.g. for @xmath65 gev and for @xmath66 gev ) @xcite . the strong correlation between the @xmath67 and @xmath68 masses arises in order that @xmath69 remain perturbative ( @xmath70 for masses @xmath71 ) . the upper bound on the triplet higgs masses ( @xmath72 gev ) comes about from the perturbativity of @xmath73 whilst that on @xmath74 ( @xmath75 gev ) comes from the perturbativity of @xmath76 . these latter two bounds can be estimated crudely by ignoring the evolution of the couplings directly from equations ( [ lam1 ] ) and ( [ lam3 ] ) . evolution tightens the bounds due to the positivity of the beta functions , especially for the @xmath74 since @xmath77 . the hole at low masses is due to vacuum stability . in figure [ b004g01 ] we show the allowed regions for @xmath78 . the correlation of the mainly triplet higgses is as in figure [ b004g0 ] . for large @xmath79 ( @xmath80 gev ) , the upper limit on the triplet higgs mass arises because @xmath76 becomes too large ( in this region @xmath81 . for smaller @xmath79 , @xmath76 is much smaller than @xmath73 and the upper bound comes from the largeness of @xmath73 with the tree - level estimate being @xmath82 . the upper limit on @xmath79 is again a consequence of the perturbativity of @xmath76 , except at low @xmath67 masses , where it is due to the negativity of @xmath73 driving the vacuum unstable . for very low @xmath83 , @xmath69 becoming too large is the problem . in figure [ b004gpi4 ] we show the allowed regions for @xmath84 . in this maximal mixing scenario one loses the distinction between doublet and triplet higgses and the bounds are correspondingly more democratic . the largeness of @xmath85 can be arranged either by tuning @xmath86 or by having small enough @xmath76 and @xmath87 . in the former case , all masses are approximately degenerate , as can be seen in the plot . in the latter case , which corresponds to light masses , the degeneracy is lifted . the bounds for @xmath88 are very similar to those for @xmath89 on interchanging the neutral higgses @xmath68 and @xmath74 . for @xmath90 and small @xmath15 ( but still away from the decoupling regime ) the allowed regions are very similar to those for @xmath91 , i.e. as in figure [ b004g0 ] . for larger @xmath15 , the mass bounds are again as for larger @xmath23 but with the correlation between the neutral and charged higgs masses becoming even stronger than for larger @xmath23 . we should stress that all of the previous discussion is valid for strictly non - zero @xmath23 . the situation is quite different for @xmath92 . if the neutral mixing is not zero ( which is required if we are to avoid decoupling ) then the vacuum conditions dictate that @xmath93 and @xmath94 and this renders equation ( [ eq : beta ] ) redundant . equation ( [ eq : gamma ] ) then yields @xmath95 and we have complete degeneracy , i.e. @xmath96 . so far we have worked in a regime where the triplet does not decouple from the doublet . clearly for @xmath97 there is no mixing between the doublet and triplet and there is no bound on the triplet mass . this is a special case of the more general decoupling scenario , which occurs when @xmath98 , which we now discuss . for small mixing angles , the ( mainly ) triplet higgs has mass squared @xmath99 . one possible solution to the mixing angle equations ( [ eq : beta ] ) and ( [ eq : gamma ] ) is that @xmath100 and any @xmath15 . in this case the triplet higgs has mass @xmath71 . this is the regime of the previous section . however , it is also possible to solve the mixing angle equations with @xmath101 by keeping @xmath102 large , i.e. ( [ eq : beta ] ) gives @xmath103 . in this case , equation ( [ eq : gamma ] ) forces @xmath104 . this is the decoupling limit in which the triplet mass lies far above the mass of the doublet and the low energy model looks identical to the standard model . tree level arguments on the perturbativity of @xmath73 allow us to quantify the approach to decoupling from the point of view of the triplet higgs mass . in particular ( [ lam3 ] ) dictates that , for small @xmath23 and @xmath15 , @xmath105 by virtue of the smallness of @xmath106 this relation picks up relatively small loop corrections . this bound clearly demonstrates decoupling . it also re - iterates the results of the previous section , i.e. for small @xmath107 the limit is as in figure 1 and for small @xmath108 the limit is as in figure 2 . we remark that the pseudo - decoupling regime , where @xmath23 is not too small , is of particular interest in that it again allows one to relax the mass bound on the lightest higgs coming from the precision data without otherwise changing the physics of the standard model @xcite . we have computed the one - loop beta functions for the scalar couplings in an extension to the standard model which contains an additional real triplet higgs . through considerations of perturbativity of the couplings and vacuum stability we have been able to identify the allowed masses of the higgs bosons in the non - decoupling regime . in the decoupling regime , the model tends to resemble the standard model . we note that the theoretical mass bounds presented here will of course be tightened after considering the precision electroweak and direct search data . such a study requires that the impact of the quantum corrections ( to the @xmath14 parameter ) for non - zero @xmath15 be computed ( they were not explored in @xcite ) . as a final remark , we wish to emphasise that the near degeneracy of the triplet higgs masses ( the mass splitting is naturally @xmath109 ) ensures that , at least for small @xmath15 , the quantum corrections to the @xmath14 parameter are negligible ( the @xmath110 parameter vanishing since the triplet has zero hypercharge ) @xcite . as shown in @xcite , this means that the lightest higgs boson can be heavy as a result of the compensation arising from the explicit tree - level violation of custodial symmetry which the real triplet induces . thus it is quite possible to be in a regime where all the higgs bosons are heavy without any dramatic deviation from the physics of the standard model . * acknowledgements * : we would like to thank ben allanach , arthur hebecker , apostolos pilaftsis and douglas ross for discussions . asv acknowledges the support of pparc ( postdoctoral fellowship : ppa / p / s/1999/00446 ) . j. r. forshaw , d. a. ross and b. e. white , jhep * 0110 * ( 2001 ) 007 . s. r. coleman and e. weinberg , phys . d * 7 * ( 1973 ) 1888 . r. jackiw , phys . d * 9 * ( 1974 ) 1686 . j. iliopoulos , c. itzykson and a. martin , rev . * 47 * ( 1975 ) 165 . b. kastening , phys . b * 283 * ( 1992 ) 287 . c. ford , d. r. jones , p. w. stephenson and m. b. einhorn , nucl . b * 395 * ( 1993 ) 17 . m. quiros , helv . acta * 67 * ( 1994 ) 451 .	we perform an analysis of the renormalization group evolution of the couplings in an extension to the standard model which contains a real triplet in the higgs sector . insisting that the model remain valid up to 1 tev allow us to map out the region of allowed mass for the higgs bosons . we conclude that it is possible for there to be no light higgs bosons without any otherwise dramatic deviation from the physics of the standard model . plus 2 mm minus 2 mm 23.0 cm 17.0 cm -1.0 in -65pt mc - th-2002/14 + cavendish - hep-2003/01 + february 2003 + + j.r . forshaw@xmath0 , a. sabio vera@xmath1 and b.e . white@xmath0 @xmath0_department of physics & astronomy _ , + _ university of manchester _ , + _ oxford road _ , + _ manchester m13 9pl , u.k . _ + @xmath1 _ cavendish laboratory _ , + _ university of cambridge _ , + _ madingley road _ , + _ cambridge cb3 ohe , u.k . _
let @xmath0 be a polynomial map such that its jacobian determinant , @xmath9 , is nowhere zero in @xmath4 . by the inverse function theorem , @xmath3 is locally injective . the _ real jacobian conjecture _ asserts that @xmath3 is globally injective . this is closely related to the famous _ jacobian conjecture _ , which claims that given @xmath10 a field of characteristic zero , any polynomial map from @xmath11 to @xmath11 such that its jacobian determinant is equal to @xmath12 is injective . jacobian conjecture was stated at the first time in 1939 by keller ( @xcite ) and up to now is open if @xmath13 . we refer to @xcite for further informations on jacobian conjecture . the real jacobian conjecture is nevertheless not true : in 1994 , pinchuk constructed , in @xcite , a non injective polynomial map @xmath8 such that @xmath14 in @xmath4 . in this example , the polynomials @xmath1 and @xmath6 have high degrees : @xmath1 has degree @xmath15 and @xmath6 has degree @xmath16 . since the injectivity of @xmath3 is clear when the degree of @xmath3 is one ( in this case @xmath3 is an affine change of coordinates ) , it is natural to ask what would be the highest degree of @xmath1 or @xmath6 guaranteeing the global injectivity of @xmath3 . in this direction , gwodziewicz proved , in @xcite , that if the degrees of @xmath1 and @xmath6 are less than or equal to @xmath17 then @xmath3 is injective . in @xcite , braun and santos generalized gwodziewicz result proving that the injectivity of @xmath3 is true provided just the degree of @xmath1 being less than or equal to @xmath17 . thus a natural question is : what is the maximum degree of @xmath1 , between @xmath17 and @xmath18 , in order that the injectivity of @xmath3 is necessary independently of the degree of @xmath6 ? as a partial answer to this question , we prove in this note the following result . [ main ] let @xmath0 be a polynomial map such that @xmath14 in @xmath4 . if the degree of @xmath1 is less than or equal to @xmath2 , then @xmath3 is injective . since the assumption `` @xmath14 in @xmath4 '' gives that @xmath6 is strictly monotone along each connected component of a level set of @xmath1 ( as the level sets of @xmath6 are transversal to the level sets of @xmath1 ) , one strategy to show the injectivity of @xmath3 is to prove that the level sets of @xmath1 are all connected . actually , since injective polynomial maps are bijections ( see @xcite ) , this is also necessary to the injectivity of @xmath3 . thus to prove theorem [ main ] it is enough to consider all the polynomial submersions @xmath1 with disconnected level sets and to show that for these there is not a polynomial @xmath6 such that @xmath14 in @xmath4 . before continuing , we need the following definition . [ def : equi ] we say that two functions @xmath19 are _ equivalent _ if there exist an affine change of coordinates , where @xmath20 , with @xmath21 . ] @xmath22 and constants @xmath23 , with @xmath24 , such that @xmath25 . let @xmath1 be a polynomial submersion . it is quite simple to show that all its level sets are connected if the degree of @xmath1 is less than or equal to @xmath26 . if @xmath1 has degree @xmath17 and has at least one disconnected level set , one of the results of the above cited paper @xcite shows that @xmath1 is equivalent to @xmath27 . in our next theorem , we give the classification in the case of degree @xmath2 . [ classification ] if @xmath28 is a polynomial submersion of degree @xmath2 which has at least one disconnected level set , then @xmath1 is equivalent to one of the following 1 . [ 1 ] @xmath29 , 2 . [ 2 ] @xmath30 , with @xmath31 or @xmath12 , 3 . [ 3 ] @xmath32 , 4 . [ 4 ] @xmath33 , with @xmath34 . for each case of theorem [ classification ] , we will prove there is no polynomial @xmath6 such that @xmath35 is nowhere zero . as a consequence , we will obtain theorem [ main ] , since when the degree of @xmath1 is less than or equal to @xmath17 , the injectivity was already proven in @xcite ( in that paper , it was also shown there is no @xmath6 such that @xmath36 is nowhere zero ) . the paper is organized as follows . in section [ section_common_zeros ] , we use subresultants of two polynomials to develop results to decide when two special polynomials in @xmath37 $ ] have common zeros . these special polynomials have the form of @xmath38 and @xmath39 , where @xmath1 is a polynomials of degree @xmath2 . in section [ section_level_sets ] , we construct results to decide when a special polynomial submersion has all its level sets connected . then in section [ section_classification ] , we use the results of sections [ section_common_zeros ] and [ section_level_sets ] to prove theorem [ classification ] . in section [ quase_final ] , we will consider the polynomials of theorem [ classification ] and prove that for each of them there is not a polynomial @xmath40 such that @xmath41 , for all @xmath42 . for the arguments we shall use the concept of half - reeb component of a foliation . then , in section [ fim ] , we depicted the proof of theorem [ main ] . the arguments in section [ quase_final ] are divided in two groups . in the first one , we study the polynomials [ 1 ] and [ 2 ] of theorem [ classification ] and use techniques analogous to @xcite to conclude the non existence of a polynomial @xmath6 . now in the second group , when we analyse the polynomial [ 3 ] and [ 4 ] , these techniques no longer work . thus we transform part of the problem in being able to decide when a special polynomial of one variable is positive . namely , lemma [ bruna ] asserts that @xmath43 can not be a positive polynomial , if not identically zero . the proof of this lemma is in the appendix . let us recall the concept of subresultants of polynomials . consider two polynomials @xmath44 $ ] , @xmath45 and take their sylvester matrix written in the following form @xmath46 for each @xmath47\right\}$ ] , we define the _ @xmath48-subresultant _ of @xmath1 and @xmath6 , @xmath49 , by the determinant of the @xmath50 matrix obtained when we delete the first and the latest @xmath48 columns and rows of @xmath51 . the following result is classical but we include a proof of it for the sake of completeness . our proof was based on @xcite . [ resultant ] consider @xmath52 and @xmath53 as above , with @xmath54 . then @xmath1 and @xmath6 have exactly @xmath48 common roots ( counting multiplicity ) if and only if @xmath55 let us denote by @xmath56 the @xmath57th column of @xmath51 , for @xmath58 . by substituting the last column @xmath59 by @xmath60 , and calculating the determinant using the laplace expansion on the last column , we get that @xmath61 , where @xmath62 and @xmath63 are polynomials of degree less than or equal to @xmath64 and @xmath65 , respectively . since @xmath66 is a complex number with does not depend on @xmath67 , it follows that @xmath1 and @xmath6 have a common root if and only if @xmath68 . now the lemma follows readily from the following assertion : _ if @xmath69 and @xmath70 , @xmath71 , then @xmath72 , for @xmath73_. to prove the assertion , we make the following operations on the columns of @xmath74 : we change column @xmath75 by @xmath76 for @xmath77 . then we observe this is exactly the matrix @xmath51 without the first and the last rows and columns . [ calculo_res ] let @xmath78 be functions . for @xmath79 we have the following formulas @xmath80 whereas for @xmath81 we have @xmath82 the proof is straightforward . [ derivadapar ] let @xmath83 be smooth functions , and @xmath84 be such that @xmath85 and @xmath86 . define @xmath87 if there is a non negative integer @xmath88 such that @xmath89 , then @xmath90 . stands for the @xmath57th derivative of @xmath91 . ] in order to simplify the notation , we denote @xmath92 , @xmath93 , @xmath94 , @xmath95 and @xmath96 . we will first make the proof supposing @xmath97 . if @xmath98 , we have @xmath99 which only can be true if @xmath100 . since by hypothesis @xmath101 , we obtain @xmath102 and thus @xmath103 . this implies by leibniz rule that @xmath104 showing the result for @xmath105 . let us make the following induction hypothesis : @xmath106 we will show this is true for @xmath88 . then the first part of the proof will be completed . leibniz rule gives @xmath107 by , if @xmath108 , @xmath109 , hence @xmath110 and @xmath111 ( apply leibniz rule ) . now if @xmath112 , @xmath113 , since @xmath114 . then last sum simplifies to @xmath115 if @xmath116 , we have to have @xmath117 , which by hypothesis gives @xmath118 , hence @xmath119 . now it remains to prove that @xmath90 . apply again leibniz rule to obtain @xmath120 if @xmath121 , @xmath122 and @xmath123 . if @xmath124 , @xmath125 , and it follows that @xmath126 this gives us @xmath90 . in the general case , we define @xmath127 . leibniz rule gives for @xmath128 @xmath129 then the hypothesis @xmath130 shows @xmath131 . thus by the first part of the proof we obtain @xmath132 . hence , putting @xmath133 above , we conclude @xmath134 . now we apply the preceding results to produce criteria to decide when two special polynomials have common zeros . [ raiz_23 ] let @xmath135 and @xmath40 in @xmath37 $ ] as follows @xmath136 write @xmath137 . if 1 . @xmath138 has no common zeros with @xmath139 , 2 . [ raiz_23_2 ] there exists @xmath140 such that @xmath141 , then there exists @xmath142 such that @xmath143 . let @xmath144 be the distinct real zeros of @xmath138 ( by [ raiz_23_2 ] we have at least one ) . if @xmath145 , for some @xmath146 , we have two possibilities : @xmath147 in the first one , define @xmath148 and @xmath149 , and observe that @xmath150 , and @xmath151 . thus by lemma [ resultant ] there is exactly one @xmath152 such that @xmath153 ( since the coefficients of @xmath154 and @xmath155 are real ) . on the other hand , if @xmath156 , it is simple to see that @xmath157 . thus @xmath158 , hence @xmath159 . moreover , there exists @xmath152 such that @xmath160 , since @xmath161 is a polynomial of degree @xmath17 in @xmath67 . from now on , we will suppose @xmath162 , for all @xmath146 . denoting @xmath163 , if @xmath164 for some @xmath146 , lemma [ resultant ] shows the existence of @xmath152 such that @xmath153 as we wanted . thus we suppose @xmath165 for each @xmath146 . _ we assert there is @xmath146 such that @xmath166_. in this case , since by lemma [ resultant ] @xmath167 and @xmath161 have exactly two zeros in common ( observe @xmath168 ) , they must be real zeros and we are done . let us then prove the assertion . suppose by contradiction that for all @xmath146 , @xmath169 . by definition of subresultant and by lemma [ calculo_res ] , we have in a neighborhood of each @xmath170 where @xmath171 , @xmath172 where we have written @xmath173 , @xmath174 , @xmath175 and @xmath176 . then by lemma [ derivadapar ] , since @xmath177 if and only if @xmath101 , we obtain that each @xmath170 is a zero of multiplicity even of @xmath138 , say @xmath178 . then @xmath179 where @xmath180 is a monic polynomial without real zeros . thus @xmath181 , for all @xmath182 , a contradiction with assumption [ raiz_23_2 ] . [ raiz_22 ] let @xmath135 and @xmath40 in @xmath37 $ ] be defined by @xmath183 write @xmath137 . if 1 . [ raiz_22_1 ] @xmath138 has no common zeros with @xmath184 and @xmath139 , 2 . [ raiz_22_2 ] there exists @xmath140 such that @xmath141 , then there exists @xmath142 such that @xmath143 . moreover , if we keep other hypotheses and change [ raiz_22_1 ] by * [ 78 ] @xmath138 has no common zeros with @xmath139 and there is @xmath84 with @xmath185 , then there exists @xmath152 such that @xmath143 if and only if @xmath186 or @xmath187 and @xmath188 . as in the proof of theorem [ raiz_23 ] , let @xmath144 be the distinct real zeros of @xmath138 . if @xmath145 for some @xmath88 , and @xmath189 , take @xmath148 and @xmath149 , and observe that @xmath150 . since @xmath190 , we have by lemma [ resultant ] that there is @xmath152 such that @xmath153 . if now @xmath156 , @xmath191 , hence @xmath158 . thus @xmath192 , and @xmath193 for some @xmath194 if and only if @xmath195 . this proves the second part of the theorem . we suppose now that @xmath162 for each @xmath146 . the proof from now on is similar to the proof of theorem [ raiz_23 ] . if @xmath164 for some @xmath88 , we are done . if @xmath165 for each @xmath196 , _ we assert there is @xmath88 such that @xmath197_. if this is in force , since @xmath167 and @xmath161 have two zeros in common , they must be real . thus let us prove the assertion . suppose @xmath198 , for each @xmath146 . by lemma [ calculo_res ] and definition of resultant , we get in a small neighbourhood of each @xmath170 @xmath199 where @xmath173 , @xmath174 , @xmath200 and @xmath176 . then by lemma [ derivadapar ] , we have that all the zeros of @xmath138 have multiplicity even @xmath178 . thus @xmath201 where @xmath180 is a monic polynomial without real zeros . this gives @xmath202 for all @xmath182 , a contradiction with hypothesis [ raiz_22_2 ] . the following two corollaries use the @xmath12-subresultant to analyse common zeros of polynomials when hypotheses [ raiz_23_2 ] of the preceding theorems are difficult to be verified . [ raiz_23_cor ] let @xmath135 and @xmath40 as in theorem [ raiz_23 ] . if 1 . @xmath138 has no common zeros with @xmath139 , 2 . [ iiiii ] there exists @xmath140 such that @xmath203 , @xmath204 and @xmath205 , then there exists @xmath142 such that @xmath143 . since @xmath206 , in a neighborhood of @xmath207 we have , as in , @xmath208 if @xmath209 , then @xmath210 ( recall that @xmath176 ) , and thus the result follows from theorem [ raiz_23 ] . if , on the other hand , @xmath211 , we have @xmath212 , which guarantees two common zeros of @xmath213 and @xmath214 . if these zeros are real we are done . thus let us suppose @xmath215 . if @xmath216 , since the zeros of @xmath62 are isolated ( it is a rational function not identically zero , since if @xmath217 , @xmath218 by , hence @xmath219 , a contradiction with [ iiiii ] ) , we have by for @xmath220 near @xmath207 @xmath221 and we are under the hypotheses of theorem [ raiz_23 ] . now if @xmath222 , we have @xmath223 and @xmath224 for @xmath220 near @xmath207 . this gives @xmath225 which guarantees @xmath226 for @xmath220 near @xmath207 . this together with and assumption gives @xmath227 for @xmath220 near @xmath207 , and we are again under the hypotheses of theorem [ raiz_23 ] . [ raiz_22_cor ] let @xmath135 and @xmath40 as in theorem [ raiz_22 ] . if 1 . [ 7897 ] @xmath138 has no common zeros with @xmath184 and @xmath139 , 2 . [ 7898 ] @xmath228 and there exists @xmath140 such that @xmath203 and @xmath206 , then there exists @xmath142 such that @xmath143 . moreover , if we keep other hypotheses and change [ raiz_22_1 ] by * @xmath138 has no common zeros with @xmath139 and there is @xmath84 with @xmath185 , then there exists @xmath152 such that @xmath143 if and only if @xmath186 or @xmath187 and @xmath188 . the second part of the corollary is clearly similar to the second part of theorem [ raiz_22 ] . thus suppose we are under hypothesis [ 7897 ] and [ 7898 ] . since @xmath206 , in a neighborhood of @xmath207 we have , as in equation , @xmath229 if @xmath209 , we have @xmath230 , hence we are under the hypotheses of theorem [ raiz_22 ] . if , on the other hand , @xmath211 , we have @xmath212 , which guarantees two common zeros of @xmath213 and @xmath214 . if these zeros are real we are done . thus let us suppose @xmath215 . if @xmath216 , since the zeros of @xmath62 are isolated ( as in the proof of corollary [ raiz_23_cor ] ) , we have @xmath231 for @xmath220 near @xmath207 , and we are again under the hypotheses of theorem [ raiz_22 ] . now if @xmath222 , we have @xmath223 and @xmath224 for @xmath220 near @xmath207 . this gives @xmath225 and thus @xmath226 for @xmath220 near @xmath207 . this together with and assumption [ 7898 ] gives @xmath232 for @xmath233 near @xmath207 , and we are also under hypotheses of theorem [ raiz_22 ] . [ unbounded ] let @xmath234 be an open set . if @xmath235 is a smooth submersion then the connected components of the level sets of @xmath1 induce a smooth foliation of dimension @xmath12 of @xmath236 . see , for example , @xcite . in particular , when @xmath237 , any connected component of a level set of @xmath62 is an unbounded curve in both directions . [ zero_interval ] let @xmath238 be a smooth function . then for each interval @xmath239\subset\{y\in{\mathbb{r}}\ \|\ a_n(y)\neq 0\}$ ] , there exists an interval @xmath240 $ ] such that @xmath241,\ p(x , y)=0\}\subset [ a , b]\times [ c , d].\ ] ] define @xmath242}\left\|\frac{a_i(y)}{a_n(y)}\right\|,\ i=0,1,\ldots , n-1,\ \ \ \ \ \ \ \ \ \ \ b=1+\sum_{k=1}^n\left(na_{n - k}\right)^{1/k}.\ ] ] if @xmath243 $ ] and @xmath244 , then @xmath245 taking @xmath240=[-b , b]$ ] , the lemma follows . [ level_sets_2ttt ] let @xmath246 be a smooth submersion with the expression @xmath247 if 1 . @xmath248 , @xmath249 , 2 . @xmath250 is a _ polynomial _ with _ odd _ degree , then @xmath251 is connected . we suppose that the leader coefficient of @xmath252 is positive . the proof in the other case is analogous . therefore there exists @xmath253 such that @xmath254 and @xmath255 , @xmath256 . in particular , @xmath257 and there exists exactly one @xmath258 such that @xmath259 . let @xmath260 be the connected component of @xmath251 which contains @xmath261 and consider @xmath262 such that @xmath263 . it is clearly enough to prove that @xmath264 . by lemma [ zero_interval ] , there exists an interval @xmath240 $ ] such that @xmath265\right)\subset(a , b)\times[c , q_2].\ ] ] now by lemma [ unbounded ] , both ends of @xmath260 must escape the compact @xmath240\times[c , q_2]$ ] . therefore , by , @xmath260 will cut the line @xmath266 in two points , say @xmath267 and @xmath268 . since @xmath260 does not have self intersections , @xmath269 . in particular @xmath270 and @xmath271 , hence @xmath264 . [ level_sets_22tt ] let @xmath246 be the smooth submersion @xmath247 if 1 . @xmath272 for exactly one @xmath194 , 2 . [ segundo ] when @xmath272 , there exists exactly one @xmath182 such that @xmath273 , 3 . [ terceiro ] @xmath250 is a _ polynomial _ with _ even _ degree and its leader coefficient is _ negative _ , then @xmath251 is connected . let @xmath274 such that @xmath275 . by item [ terceiro ] , there exist @xmath276 , with @xmath277 , such that if @xmath278 or if @xmath279 , @xmath255 . hence @xmath280 by item [ segundo ] , there exists exactly one point @xmath281 in the line @xmath282 such that @xmath283 . let @xmath260 be the connected component of @xmath251 which contains @xmath281 . _ we assert that @xmath260 can not be entirely contained in @xmath284 $ ] nor in @xmath285_. indeed , if @xmath286 , then there exists @xmath287 such that ( using lemmas [ zero_interval ] and [ unbounded ] ) there exist four different solutions of @xmath288 , a contradiction since this is a quadratic equation in @xmath289 . similar contradiction can be obtained if we suppose @xmath290 $ ] . now take @xmath291 $ ] such that @xmath263 . we will prove that @xmath264 . in fact , by the assertion above , there exists @xmath292 with @xmath293 . thus by lemma [ zero_interval ] , there exists an interval @xmath240 $ ] such that @xmath294\right)\subset[a , b]\times[y_2,y_3]$ ] . then by and lemma [ unbounded ] , the connected component of @xmath251 which contains @xmath295 must cut the line @xmath296 in two different points . since @xmath288 is a quadratic equation in @xmath289 , one of these points must be @xmath297 , and then @xmath264 . the same can be done if we take @xmath298 $ ] such that @xmath263 . thus @xmath251 is connected . to prove propositions [ level_sets_2ttt ] and [ level_sets_22tt ] , we used the properties of the discriminant of a quadratic equation . for the next proposition , which already appeared in @xcite , let us recall the properties of the discriminant of a cubic equation @xmath299 take @xmath300 and @xmath301 , and define the _ discriminant _ of the above equation by @xmath302 . we have that @xmath303 as a consequence , we obtain [ level_sets_3 ] let @xmath246 be the smooth submersion @xmath304 if the discriminant @xmath305 of the equation @xmath273 is a _ polynomial _ with _ even _ degree and with _ positive _ leader coefficient , then @xmath251 is connected . the proof is similar to the proof of propositions [ level_sets_2ttt ] , and [ level_sets_22tt ] , thus we give just the outline . take @xmath239 $ ] such that @xmath306 , @xmath307 $ ] . prove first that @xmath308 and @xmath309\big)$ ] are non empty connected sets . then conclude these two sets connect each other by a curve in @xmath240\times[c , d]$ ] , where @xmath240 $ ] is given by lemma [ zero_interval ] . in order to prove theorem [ classification ] , we shall use next corollary , which is an immediate consequence of the classification of homogeneous polynomials of degree @xmath2 given in theorem 2.6 of @xcite . [ second_classif ] any polynomial of degree exactly @xmath2 can be transformed by means of a _ linear change of variables _ in one of the following . @xmath310 where @xmath311 is a polynomial of degree less than or equal to @xmath17 . we will work with the polynomials of corollary [ second_classif ] showing that each of them is not a submersion , or has all its level sets connected , or is equivalent ( in the sense of definition [ def : equi ] ) to one of the cases of theorem [ classification ] . we divide the 9 cases in 5 groups , each of them in one of the subsections bellow . theorem [ classification ] will be a direct consequence of the propositions [ i_iii_viii_ix ] , [ pro_case_v ] , [ pro_case_vii ] , [ pro_case_iv ] and [ pro_case_ii_vi ] contained in the subsections [ primeira ] , ... , [ ultima ] , respectively . the arguments use the results of sections [ section_common_zeros ] and [ section_level_sets ] . the calculations of the subresultants and of the discriminants were made using maple . to stablish notation , we write the polynomial @xmath311 in the following form @xmath312 [ i_iii_viii_ix ] the polynomials of cases ( i ) , ( ii ) , ( iii ) and ( vi ) are not submersions . next lemma will help us in the proof of proposition [ i_iii_viii_ix ] . [ lemma_i_iii_viii_ix ] the polynomials of cases ( i ) , ( ii ) , ( iii ) and ( vi ) are equivalent to @xmath313 moreover , if @xmath314 , then @xmath315 . we remark that if @xmath314 and @xmath316 , then the change of coordinates @xmath317 transforms @xmath135 in case ( vii ) . polynomial ( i ) is counted above taking @xmath318 and @xmath314 . dividing polynomial ( ii ) by @xmath319 and taking @xmath318 , we have that it is also counted in the lemma , with @xmath314 . now case ( iii ) is counted above with @xmath318 and @xmath320 . finally , dividing polynomial ( vi ) by @xmath319 , it is in the form of the lemma by taking @xmath321 and @xmath314 . we will analyse the polynomial of lemma [ lemma_i_iii_viii_ix ] . observe @xmath39 and @xmath38 have the forms @xmath322 and @xmath323 , respectively , with @xmath324 . with the notations of theorem [ raiz_23 ] , observe @xmath325 with @xmath326 . if @xmath327 , this theorem gives that there exists a common zero of @xmath39 and @xmath38 . therefore @xmath1 is not a submersion . we suppose thus @xmath328 , i.e. @xmath314 and @xmath321 . in this case we get @xmath329 and @xmath330 . then by theorem [ raiz_23 ] , if @xmath331 or @xmath332 , @xmath1 is not a submersion . on the other hand , if @xmath333 and @xmath334 , we have @xmath335 and @xmath336 . if @xmath337 , the same theorem guarantees that @xmath1 is not a submersion . now if @xmath338 , i.e. @xmath339 and @xmath340 , then @xmath341 and @xmath342 then , as above , if @xmath343 , @xmath1 is not a submersion , whereas if @xmath344 and @xmath345 , we get @xmath346 and @xmath347 , which clearly have a common zero and so @xmath1 is not a submersion . [ pro_case_v ] if a polynomial of case ( viii ) is a submersion , it is equivalent to @xmath348 in particular , the level set @xmath349 is not connected . to prove this proposition , we will first prove the following lemma . [ v ] the polynomials of case ( viii ) are equivalent to one of the following . @xmath350 by multiplying the polynomial in case ( viii ) by @xmath351 , and after composing it with the change of coordinates @xmath352 , we obtain it is equivalent to ( keeping the notations of the coefficients ) @xmath353 if @xmath354 , we obtain case , whereas if these coefficients are zero , we could have @xmath355 zero or not . in the first case we get ( if @xmath356 , simply divide @xmath1 by @xmath357 and compose it with the change of coordinates @xmath358 ) and in the second case , if @xmath359 , we get case . on the other hand , if @xmath360 , then @xmath361 and the change @xmath362{2/a_{01}}x , a_{01}y/2\big)$ ] gives case . for case and of lemma [ v ] , @xmath38 and @xmath39 have the forms @xmath322 and @xmath323 , respectively , with @xmath363 and @xmath364 . thus we are under the hypotheses of theorem [ raiz_23 ] . calculating the resultant we have @xmath365 in case , @xmath366 . thus if @xmath367 , @xmath1 is not a submersion . on the other hand , if @xmath368 , we have ( @xmath369 ) and @xmath370 . then observe that @xmath371 therefore if @xmath372 and @xmath356 , we have a zero @xmath207 of @xmath373 with opposite signal with @xmath374 . then @xmath375 , and corollary [ raiz_23_cor ] guarantees that @xmath1 is not a submersion . on the other hand , if @xmath376 and @xmath377 , we have that @xmath378 . suppose first @xmath379 . if @xmath380 and @xmath374 have opposite signals , then @xmath381 and @xmath382 have opposite signals and theorem [ raiz_23 ] gives that @xmath1 is not a submersion . now if @xmath380 and @xmath374 have the same signal , we observe @xmath373 have a negative and a positive zero , which gives @xmath383 , for one of these zeros . this proves that @xmath1 is not a submersion by corollary [ raiz_23_cor ] . now if @xmath384 , we observe that @xmath385 , and @xmath386 is a polynomial of degree @xmath17 . thus , again , @xmath1 is not a submersion . if now @xmath387 , then @xmath388 and hence @xmath389 . then theorem [ raiz_23 ] gives that @xmath1 is not a submersion . in the case , @xmath390 and @xmath391 , which by theorem [ raiz_23 ] proves that @xmath1 is not a submersion . now for the case , it is simple to observe that @xmath392 , and @xmath393 is a polynomial of degree @xmath12 , hence @xmath1 is not a submersion . finally , in the case , it is clear that if @xmath384 then @xmath1 is not a submersion , whereas if @xmath379 , we multiply @xmath1 by @xmath394 and apply the change @xmath395 to get the polynomial map of the proposition . it is clear this is a submersion . [ pro_case_vii ] if the polynomial of the case ( ix ) is a submersion and has at least one disconnected level set , then it is equivalent to @xmath396 [ vii ] the polynomials in case ( ix ) are equivalent to one of the following . @xmath397 first we divide the polynomial of case ( ix ) by @xmath319 . then we get the case above if @xmath367 by composing @xmath1 with @xmath398{a_{03}}y\big)$ ] . if @xmath368 and @xmath376 , we get the first part of case above , whereas if @xmath399 , we obtain the second one . now if @xmath400 and @xmath401 , we take the transformation @xmath402 to obtain @xmath403 then , if @xmath388 , we apply @xmath404 to obtain case . on the other hand , if @xmath31 we obtain case if @xmath361 and , if @xmath405 , the transformation @xmath406 gives case . the equation @xmath407 , where @xmath1 is the polynomial of case of lemma [ vii ] , has the form @xmath408 where @xmath409 , @xmath410 and @xmath411 . calculating the discriminant of the equation as in proposition [ level_sets_3 ] , we obtain @xmath412 thus by this proposition , we conclude that if @xmath1 is a submersion , it has all its level sets connected . now we observe that in case of lemma [ vii ] , @xmath413 with @xmath414 , and @xmath415 ( which _ can _ be zero , for the firs part of this case and _ is _ zero for the second part ) . then we calculate @xmath416 , and observe that @xmath417 if @xmath372 , which by theorem [ raiz_23 ] gives , we can yet apply this theorem . ] that @xmath1 is not a submersion . now if @xmath418 , we have that @xmath419 and @xmath420 . therefore , if @xmath421 , the same theorem guarantees @xmath1 is not a submersion . finally , if @xmath31 , we have @xmath422 and @xmath423 . then if @xmath424 , we also have @xmath1 is not a submersion . if @xmath425 and @xmath361 , it is simple to see the level sets of @xmath1 are all connected , while if @xmath426 , it follows that @xmath427 , and @xmath428 annihilates for some @xmath289 , since it has degree @xmath17 in @xmath289 . consider now case of lemma [ vii ] and observe that @xmath39 and @xmath38 have again the form of . we have @xmath429 , with @xmath430 . if this is not zero , we obtain by theorem [ raiz_23 ] that @xmath1 is not a submersion . if this is zero , i.e. @xmath431 ( since @xmath421 ) , we obtain @xmath432 moreover , equation @xmath407 has the form of proposition [ level_sets_2ttt ] , with discriminant @xmath433 thus if @xmath434 , proposition [ level_sets_2ttt ] gives us that if @xmath1 is a submersion , then all its level sets are connected . on the other hand , supposing @xmath435 , if @xmath436 , and theorem [ raiz_23 ] give us @xmath1 is not a submersion . if @xmath377 and @xmath379 , and proposition [ level_sets_2ttt ] give us again that if @xmath1 is a submersion , all its level sets are connected . finally , if @xmath437 , it is clear that @xmath438 . now we consider case . we have @xmath439 and @xmath440 . if @xmath441 , it is clear that @xmath1 is not a submersion . on the other hand , if @xmath442 , we have that @xmath1 is a submersion , but in this case @xmath443 gives us @xmath444 , i.e. all the level sets of @xmath1 are connected . finally , case with @xmath384 is not a submersion , whereas if @xmath445 , we divide @xmath1 by @xmath446{a_{10}^4}$ ] and then the change @xmath447{a_{10}^2},x/\sqrt[3]{a_{10 } } \big)$ ] , gives the polynomial of the proposition . [ pro_case_iv ] if the polynomial of case ( vii ) is a submersion and has no all its level sets connected , it is equivalent to one of the following . @xmath448 [ iv ] the polynomials of case ( vii ) are equivalent to one of the following . we first divide @xmath1 by @xmath450 . then we compose it with the transformation @xmath451 to get the following form @xmath452 if @xmath453 , we compose the polynomial with the transformation @xmath454{a_{30}a_{03}^2},y/\sqrt[3]{a_{30}^2a_{03}}\big)$ ] and multiply by @xmath455 to get . if @xmath435 and @xmath456 , the transformation @xmath457{a_{03}},\sqrt[3]{a_{03}}y\big)$ ] takes @xmath1 to . if @xmath458 and @xmath368 , we change @xmath289 by @xmath67 to get the case just studied . if now @xmath459 , and if @xmath421 , the change of coordinates @xmath460 gives . if @xmath436 , change @xmath289 by @xmath67 to get the case just studied . thus we suppose @xmath461 , to obtain case . we will analyse each case of lemma [ iv ] . for case , we notice that @xmath38 and @xmath39 have the forms @xmath462 and @xmath463 , respectively , with @xmath464 and @xmath465 . the subresultant is @xmath466 with @xmath467 and @xmath468 . thus by theorem [ raiz_22 ] , @xmath469 has a zero if @xmath470 . moreover , by the same theorem , if @xmath471 and @xmath472 , we have again that @xmath469 has a zero . now if @xmath471 and @xmath425 , we obtain @xmath384 and then @xmath473 . taking @xmath474 , we have @xmath475 , and @xmath39 is a polynomial of degree @xmath476 in @xmath67 . hence @xmath1 is not a submersion . now consider case and observe that the equation @xmath407 is of the form @xmath477 , with the discriminant @xmath478 being a polynomial of degree @xmath479 with leader coefficient @xmath480 . if @xmath481 , proposition [ level_sets_3 ] gives us that if @xmath1 is a submersion , all its level sets are connected . if @xmath482 , observe that @xmath38 and @xmath39 have the forms @xmath462 and @xmath463 , respectively , with @xmath464 and @xmath465 . then we have @xmath483 with @xmath484 and @xmath468 , and @xmath485 we first suppose @xmath486 . then if @xmath470 and @xmath424 , we have by corollary [ raiz_22_cor ] , that @xmath469 has a zero . if @xmath470 and @xmath425 , we get @xmath487 and @xmath488 , which by theorem [ raiz_22 ] gives a zero of @xmath489 . now if @xmath471 and @xmath472 , we have by the same theorem , that @xmath1 is not a submersion , whereas if @xmath425 , then @xmath384 . therefore , again by theorem [ raiz_22 ] , if @xmath490 , we have a zero of @xmath489 . thus let us suppose @xmath491 . we observe that @xmath492 annihilates @xmath428 , and there exists @xmath289 such that @xmath493 , by our last hypothesis . now if @xmath377 , then @xmath335 and @xmath494 . thus if @xmath379 and @xmath470 , theorem [ raiz_22 ] gives a zero of @xmath489 . if yet @xmath379 and @xmath471 , the same theorem guarantees @xmath1 is not a submersion , since in this case @xmath495 has to be non zero . now if @xmath384 , we see that @xmath419 and @xmath496 and @xmath497 . therefore , if @xmath498 , theorem [ raiz_22 ] guarantees that @xmath1 is not a submersion . if @xmath405 , it is clear that @xmath438 , whereas if @xmath499 and @xmath425 , we obtain that @xmath1 is a submersion if and only if @xmath491 . then @xmath442 and multiplying @xmath1 by @xmath500 and composing with the change @xmath501{a_{01}},y/\sqrt{a_{01 } } \big)$ ] , we get the first case of this proposition . for case , @xmath502 gives @xmath503 , and it follows that @xmath504 where @xmath505 , with @xmath506 . if @xmath436 and @xmath507 , we get a zero of @xmath489 . if @xmath508 , we observe that @xmath509 and @xmath510 . if @xmath511 and @xmath512 are not zero , we have a zero of @xmath513 different from @xmath514 , and thus a zero of @xmath489 . if @xmath515 , i.e. @xmath516 , we have @xmath517 then if @xmath425 it is simple to see that @xmath518 . on the other hand , if @xmath424 , gives that @xmath519 annihilates @xmath39 , and @xmath520 which clearly has a zero @xmath521 . analogously , if @xmath522 , we obtain a zero of @xmath489 . thus we can suppose @xmath377 . solving @xmath289 in @xmath475 , we get @xmath523 , which substituting in @xmath39 gives @xmath524 if @xmath379 and @xmath508 , we have a zero of @xmath489 . if @xmath379 and @xmath507 , we consider the discriminant of @xmath525 as in proposition [ level_sets_2ttt ] , and notice that it is a polynomial of degree @xmath17 in @xmath289 . therefore , if @xmath1 is a submersion , all its level sets are connected . now to finish this case , we suppose @xmath384 and observe that @xmath526 . then if @xmath527 , we have that @xmath528 . on the other hand if @xmath425 , we see that @xmath529 as well . finally , for case , we first observe that if @xmath1 is a submersion , then changing @xmath289 to @xmath67 if necessary , we can suppose @xmath361 . then @xmath475 gives that @xmath530 and hence @xmath531 . thus if @xmath379 , we get that @xmath489 has a zero . on the other hand , if @xmath384 , we get that @xmath532 annihilates @xmath38 and if @xmath424 , we have a zero of @xmath533 . finally , if @xmath425 , the change of variables @xmath534 gives the second case of this proposition . [ pro_case_ii_vi ] if the polynomials of cases ( iv ) and ( v ) are submersions , then all its level sets are connected . the polynomials of cases ( iv ) and ( v ) are equivalent to one of the following @xmath535 with @xmath536 . dividing polynomials of cases ( iv ) and ( v ) by @xmath319 and @xmath537 , respectively , and applying the transformation @xmath538 , we obtain they are equivalent to @xmath539 with @xmath536 . then we compose this with the transformation @xmath540 to eliminate @xmath374 and @xmath541 . if @xmath456 , we compose the result with @xmath542 and by multiplying it by @xmath543 we obtain case . on the other hand , if @xmath544 , we get case . take case of the lemma . observe that for each @xmath253 , @xmath545 has the form @xmath477 as in proposition [ level_sets_3 ] . calculating the discriminant @xmath546 we observe it is a polynomial of degree @xmath15 with leader coefficient @xmath547 . thus if @xmath548 , we have that if @xmath1 is a submersion , all its level sets are connected . in case @xmath549 , we observe that @xmath39 and @xmath38 have the forms @xmath322 and @xmath323 , respectively , with @xmath550 and @xmath414 . with the notations of theorem [ raiz_23 ] , @xmath551 with @xmath552 , and @xmath553 if @xmath424 , there exist @xmath554 such that @xmath203 . thus @xmath555 , and corollary [ raiz_23_cor ] , gives us @xmath1 is not a submersion . now if @xmath425 , we observe that @xmath556 . if @xmath441 , there exists @xmath140 such that @xmath557 , which by theorem [ raiz_23 ] , guarantees @xmath1 is not a submersion . on the other hand , if @xmath442 , there exists @xmath554 such that @xmath203 and @xmath558 , which again by corollary [ raiz_23_cor ] guarantees that @xmath1 is not a submersion . finally , if @xmath405 , we see that @xmath559 and @xmath560 is a polynomial of degree @xmath17 , which always has a zero . now in the case , @xmath39 and @xmath38 also have their forms as in theorem [ raiz_23 ] , with @xmath561 and @xmath414 . here @xmath562 , with @xmath563 . thus by theorem [ raiz_23 ] , @xmath1 is not a submersion if @xmath388 . on the other hand , if @xmath31 , we divide the analysis in two cases : @xmath548 and @xmath549 . in the first case , if @xmath405 , @xmath564 annihilates @xmath39 and transforms @xmath475 in an equation of degree @xmath17 in @xmath289 with constant term @xmath379 ( otherwise @xmath565 ) . thus we get a zero of @xmath489 . if @xmath361 , we observe that @xmath545 has the form @xmath566 with @xmath567 , @xmath568 and @xmath569 . if @xmath570 , there is exactly one @xmath67 such that @xmath571 . moreover , the discriminant @xmath478 is a polynomial of degree @xmath572 with leader coefficient @xmath573 . therefore , we are under the hypotheses of proposition [ level_sets_22tt ] , and hence if @xmath1 is a submersion , all its level sets are connected . finally , in the second case , i.e. @xmath549 , we observe that @xmath39 and @xmath38 are as in theorem [ raiz_23 ] , with @xmath550 and @xmath414 , and @xmath574 with @xmath575 , and @xmath576 then , if @xmath361 and @xmath424 , we have by corollary [ raiz_23_cor ] that @xmath1 is not a submersion . if @xmath361 and @xmath425 , we also get by this corollary that @xmath1 is not a submersion . on the other hand , if @xmath577 , it is simple to conclude that @xmath489 has a zero . let @xmath234 be an open set and @xmath578 be a smooth submersion . we denote by @xmath579 the foliation of @xmath236 given by the connected components of the level sets of @xmath62 ( lemma [ unbounded ] ) . we recall the concept of half - reeb component . let @xmath580 be a smooth submersion , @xmath581 be defined by @xmath582 , and @xmath583\times[0,2]\ \|\ 0<x+y\leq 2\right\}.\ ] ] we say that @xmath584 is a _ half - reeb component _ , or simply a _ hrc _ , of @xmath579 if there is a homeomorphism @xmath585 which is a topological equivalence between @xmath586 and @xmath587 with the following properties : 1 . the segment @xmath588 is sent by @xmath589 onto a transversal section to the leaves of @xmath579 in the complement of @xmath590 . this section is called the _ compact edge _ of @xmath591 . both the segments @xmath592 and @xmath593 are sent by @xmath589 onto full half leaves of @xmath579 , called the _ non - compact edges _ of @xmath591 . ( 10,20 ) ( -295,12)@xmath594 ( -200,12)@xmath26 ( -295,110)@xmath26 ( -265,45)@xmath595 ( -75,70)@xmath591 ( -170,80)@xmath589 the existence of hrc is equivalent to the existence of inseparable leaves on the foliation @xmath579 ( see @xcite for details ) . moreover , the following proposition of @xcite relates this definition to connectedness of level sets . let @xmath580 be a smooth submersion . then @xmath579 has a hrc if and only if there exists @xmath253 such that @xmath596 is not connected . a particular version of the following result is already contained in @xcite . the proof in the general case is a simple extension of that and we add it here for completeness . [ integral ] let @xmath580 be a smooth submersion , @xmath597 be a hrc of @xmath579 and @xmath598 be a neighbourhood of @xmath591 . if @xmath599 is a measurable function such that @xmath600 then there is not a differentiable @xmath601 such that @xmath602 at @xmath598 . let @xmath603\to\mathcal{a}$ ] be an injective curve that parametrize the compact edge of @xmath591 . for @xmath604 big enough , the leaf of @xmath579 through @xmath605 cuts @xmath606 in @xmath607 , with @xmath608 . we denote by @xmath609 the interval of this leaf between @xmath605 and @xmath607 , and by @xmath610 the interval of @xmath606 between this two points . we denote yet by @xmath611 the compact region bounded by @xmath610 and @xmath609 . by the monotone convergence theorem , @xmath612 if there is @xmath63 such that @xmath613 in @xmath598 , green s theorem gives @xmath614 since @xmath609 is orthogonal to @xmath615 . thus @xmath616 is uniformly bounded , a contradiction with and @xmath617 . we shall also need the following simple lemma ( a reduced version of it was already used in @xcite ) . [ calculus ] let @xmath618 , and @xmath619 and @xmath620 be defined by @xmath621 , where @xmath622 and @xmath623 . if @xmath624 , for all @xmath625 in the domain of @xmath626 , then 1 . [ 1045 ] @xmath627 , for @xmath628 . [ 1046 ] if @xmath629 , then @xmath630 . [ 1047 ] if @xmath631 , then @xmath632 . if we multiply @xmath633 by @xmath634 , hypothesis in [ 1045 ] gives that @xmath635 , for all @xmath625 in the domain of @xmath626 . taking @xmath636 , if @xmath637 or @xmath638 , if @xmath639 , it follows that @xmath627 , proving [ 1045 ] . the proof of [ 1046 ] and [ 1047 ] follows by hlder s inequality : defining @xmath640 and @xmath641 , we get that @xmath642 , since for both @xmath628 , the last integral is @xmath643 plus the integral of a @xmath644 function . we now apply these results to analyse each polynomial of theorem [ classification ] in each of the subsections bellow . as a consequence , we will obtain the following theorem . [ naoq ] if @xmath135 is one of the polynomials of theorem [ classification ] , then there is not a polynomial @xmath40 such that @xmath41 , @xmath645 . we consider the polynomial @xmath646 . it is simple to observe that the following set is a hrc of @xmath647 : @xmath648 _ we claim that given a polynomial @xmath649 such that @xmath650 , @xmath645 , then @xmath617_. thus by proposition [ integral ] , there is not a polynomial @xmath40 such that @xmath41 . to prove the claim , we define @xmath651 applying the change of variables @xmath652 in the interior of @xmath591 , we obtain @xmath653 where @xmath654 and @xmath655 are suitable polynomials in @xmath289 . by , the polynomial @xmath656 is not identically zero , hence there exists a @xmath657 such that @xmath658 , @xmath659 . since @xmath650 , for all @xmath42 , it follows by lemma [ calculus ] that @xmath660 for all @xmath661 . moreover , as @xmath662 , we have that @xmath663 and thus , by applying lemma [ calculus ] once more , it follows that , for each @xmath664 , @xmath665 then @xmath666 which by gives that @xmath617 , and the claim is proven . we consider the quite analogous case of the polynomial @xmath30 , with @xmath31 or @xmath12 . we observe that the closure of the following set is a hrc of @xmath647 . @xmath667 _ we claim that given a positive polynomial @xmath668 , then @xmath617_. as above , this will show there is not a polynomial @xmath6 such that @xmath669 . the proof of this claim is similar to the one made above : by defining @xmath670 and taking the bounded set @xmath671 , we get . ] @xmath672 then since @xmath663 , we get as before that @xmath673 , hence @xmath617 , proving the claim . we consider now the polynomial @xmath32 . this is quite different from the former ones . we first observe that the following set is a hrc of @xmath647 . @xmath674 then @xmath675 , and thus proposition [ integral ] can not be used to prove that there is not a polynomial @xmath6 such that @xmath676 , for example . but we will use this proposition to eliminate candidates to be the polynomial @xmath6 . then we will use a different argument to show that there is not such a polynomial . let us suppose there exist @xmath677 such that @xmath678 for all @xmath42 . we define @xmath679 and _ we claim that @xmath680_. indeed , calculating @xmath681 by applying the change of variables @xmath682 , we obtain @xmath683 where @xmath684 and @xmath685 are suitable polynomials in @xmath67 . if @xmath686 , we have that @xmath687 is not identically zero and thus there exists @xmath688 such that @xmath689 in @xmath690 . moreover , since @xmath668 is a positive polynomial , it follows by lemma [ calculus ] that @xmath691 in @xmath690 . by the same lemma , it follows that the above integral is infinite . but this contradicts proposition [ integral ] ( since we are supposing there exists @xmath6 such that @xmath692 ) . hence the claim is proven . now since we are supposing @xmath650 , for all @xmath42 , considering @xmath693 we have that @xmath694 for all @xmath695 , with @xmath696 . since @xmath697 for @xmath698 ( as @xmath699 ) , we have that @xmath700 for all @xmath701 . then by lemma [ bruna ] bellow , it follows that @xmath702 is identically zero , which guarantees in particular that @xmath703 . but this is a contradiction with @xmath704 . hence we conclude there is not a polynomial @xmath40 such that @xmath41 , for all @xmath42 . finally , we consider @xmath33 , with @xmath705 . we shall use the notations of subsection [ 33333 ] . we first observe that the closure of @xmath706 is a half - reeb component of @xmath647 . given @xmath707 , since the inequality @xmath708 is valid for all @xmath709 , for some @xmath710 , we have that @xmath711 we suppose also that @xmath712 and define @xmath713 now if @xmath714 , then @xmath715 and @xmath716 . hence by , we have that @xmath717 , where @xmath718 is the bounded set @xmath719 , if @xmath720 , or @xmath721 . thus @xmath722 , for all positive function @xmath723 defined in @xmath4 , and hence @xmath617 if @xmath724 . as in the preceding subsection , we suppose there is a polynomial @xmath725 such that @xmath726 , @xmath645 . then we define @xmath727 and _ we claim that @xmath680_. indeed calculating @xmath728 by applying the change of variables @xmath729 , we get that @xmath730 where @xmath731 . as in subsection [ 33333 ] , it follows that if @xmath686 , this integral is infinite , which is a contradiction with proposition [ integral ] , and the claim is proven . by considering again @xmath693 , we have that @xmath732 for all @xmath733 , with @xmath696 . since @xmath680 , it follows in particular that the polynomial @xmath702 defined in is such that @xmath734 for all @xmath701 . therefore , by lemma [ bruna ] bellow , we get that @xmath702 is identically zero , hence @xmath703 , which is a contradiction with @xmath704 . thus we conclude that there is not a polynomial @xmath6 such that @xmath669 . [ bruna ] given @xmath735 and @xmath736 , let @xmath702 be the polynomial @xmath737 then @xmath702 is the zero polynomial or there exist @xmath738 such that @xmath739 . see the appendix for the proof of lemma [ bruna ] . suppose we are under the hypotheses of theorem [ main ] . if all the level sets of @xmath1 are connected or the degree of @xmath1 is less than or equal to @xmath17 , then @xmath3 is injective by the argument given in introduction or by the main result of @xcite , respectively . thus we can suppose that @xmath1 has least one disconnected level set and that the degree of @xmath1 is @xmath2 . by theorem [ classification ] , there exist @xmath23 , @xmath24 , and an affine change of coordinates @xmath589 such that @xmath740 is one of the polynomials of this theorem . if we take @xmath741 and @xmath742 , we have that @xmath743 , a contradiction with theorem [ naoq ] . from now on , we consider the following convention : given @xmath744 , @xmath745 if @xmath746 . we also denote by @xmath747 the determinant of a quadratic matrix @xmath748 . for each integer @xmath755 , denote @xmath756 denote also @xmath757 the @xmath758 matrix whose element in the @xmath759th row and @xmath760th column is @xmath761 . then @xmath762 , where @xmath763 for each @xmath759 , divide the row @xmath759 by @xmath764 and for each @xmath760 , divide the column @xmath760 by the factors above depending only on @xmath760 . after that , we obtain that @xmath765 , where @xmath766 and @xmath767 since @xmath768 , we have that @xmath769 and it follows that @xmath236 is positive . thus it is enough to prove that @xmath770 is positive . we first operate the rows . we denote @xmath771 for each @xmath772 . then for each @xmath773 , we consider @xmath774 and replace the row @xmath775 by this row minus two times the row @xmath776 and denote by @xmath777 the elements of this new matrix . that is , for each @xmath625 , @xmath778 it is not difficult to prove by induction on @xmath625 that @xmath779 in the last step @xmath780 , we obtain the matrix @xmath781 , with @xmath782 since the products of the @xmath783 are equal to @xmath12 when @xmath784 . we now operate the columns of this modified matrix . denoting again @xmath785 , for each @xmath786 and for each @xmath787 , replace the column @xmath775 by the column @xmath775 minus the column @xmath788 . denoting by @xmath777 the elements of this new matrix , we have @xmath789 it is not difficult to show by induction on @xmath625 that @xmath790 in the last step @xmath780 , we have the matrix @xmath791 , with @xmath792 if @xmath793 , @xmath794 will annihilate the product on the first factor of @xmath795 and hence @xmath781 is a triangular matrix . moreover , if @xmath796 , we have @xmath797 which is the product of @xmath57 _ negative _ factors . thus @xmath798 is the product of @xmath799 negative factors . hence @xmath800 is positive . given @xmath801 and @xmath802 , it is simple to see that @xmath803 since @xmath804 , for all @xmath805 , we have that the leading principal minors of @xmath754 are positive multiples of @xmath806 , with @xmath807 , which are positive by lemma [ l1 ] . if @xmath702 is a positive polynomial , i.e. @xmath734 , for all @xmath701 , then the degree of @xmath808 is even , say @xmath809 . besides it is well known that there are polynomials @xmath810 such that @xmath811 that is , @xmath812 _ we assert that if we use only and , then for each @xmath816 , @xmath817 _ indeed , shows for @xmath818 . we assume [ e2 ] is true for @xmath57 . using to write @xmath819 in terms of @xmath820 , after elementary and long calculations we can prove that is true for @xmath821 , and the assertion is proven . now considering @xmath827 as variables in @xmath828 , last equation shows that @xmath829 is a quadratic form in @xmath828 . _ we assert that @xmath829 is a positive definite quadratic form_. this guarantees that @xmath830 provided that @xmath825 . thus @xmath63 and @xmath723 are identically zero , and by , @xmath808 is identically zero . therefore if @xmath831 , we have shown that there exists @xmath832 such that @xmath833 . to conclude the proof , we just apply this result to the polynomial @xmath834 . in order to show the assertion , we first observe that it is enough to prove that @xmath835 is positive definite , since @xmath836 and @xmath837 . now to verify that @xmath835 is positive definite , we shall show that @xmath838 , where @xmath839 is the hessian matrix of @xmath10 , has all its leading principal minors positive . it is simple to show that for each @xmath840 , @xmath841 if we denote @xmath842 with @xmath843 , then @xmath844 , where @xmath845 is the matrix of lemma [ l1 ] with @xmath846 . thus by corollary [ c1 ] , all the leading principal minors of @xmath847 are positive .	the main result of this paper is the following version of the real jacobian conjecture : `` let @xmath0 be a polynomial map with nowhere zero jacobian determinant . if the degree of @xmath1 is less than or equal to @xmath2 , then @xmath3 is injective '' . assume that two polynomial maps from @xmath4 to @xmath5 are equivalent when they are the same up to affine changes of coordinates in the source and in the target . we completely classify the polynomial submersions of degree @xmath2 with at least one disconnected level set up to this equivalence , obtaining four classes . then , analysing the half - reeb components of the foliation induced by a representative @xmath1 of each of these classes , we prove there is not a polynomial @xmath6 such that the jacobian determinant of the map @xmath7 is nowhere zero . recalling that the real jacobian conjecture is true for maps @xmath8 when all the level sets of @xmath1 are connected , we conclude the proof of the main result .
networks have significant benefits in terms of diversity and robustness over non - cooperative networks . consequently , they have been presented as a topology for the next generation of mobile networks @xcite . antenna selection , relay selection ( rs ) and diversity maximization are central themes in mimo relaying literature @xcite . however , current approaches are often limited to stationary , single relay systems and channels which assume the direct path from the source to the destination is negligible @xcite . in this letter , the problems of transmit diversity selection ( tds ) and rs are formulated as joint discrete optimization problems , where rs refines the set from which tds is made ; leading to improved convergence , performance and complexity . complexity discrete stochastic algorithms ( dsa ) with mean square error ( mse ) cost functions are employed to arrive at a solution . continuous recursive least squares ( rls ) channel estimation ( ce ) is introduced to form a combined framework , where adaptive rs and tds are performed jointly with no forward channel state information ( csi ) . the proposed algorithms are implemented , and bit error - rate ( ber ) and diversity comparisons given against the exhaustive search solution and the unmodified cooperative system . we consider a qpsk , two - phase , decode - and - forward ( df ) , multi - relay mimo system with half - duplex relays . linear minimum mean square error ( mmse ) receivers are used at all nodes and an error - free control channel is assumed @xcite . all channels between antenna pairs are flat fading , have a coherence time equal to the period of an @xmath0 symbol packet and are represented by a complex gain . the direct path is non - negligible and has an expected gain of a fraction of that of the indirect paths ; reflecting the increased distance and shadowing involved . an outline system model is given by fig . [ fig : system_model ] . the system comprises @xmath1 intermediate relay nodes which lie between single source and destination nodes which have @xmath2 and @xmath3 antennas , respectively . the relay nodes have @xmath4 antennas , where @xmath4 is an integer multiple of @xmath2 in order to reduce feedback requirements . the transmitted data consists of @xmath2 independent , . the source node transmits to the relay and destination nodes during the first phase , and the second phase involves the relay nodes decoding and forwarding their received signals to the destination . the maximum spatial multiplexing gain and diversity advantage simultaneously available in the system @xmath5 and @xmath6 , respectively @xcite . the @xmath7 and @xmath8 first phase received signals at the destination and the @xmath9 relay are given by @xmath10 = \mathbf{h}_{\mathrm{sd}}[i]a_{s}\mathbf{t}_{\mathrm{s}}\mathbf{s}[i ] + \eta_{\mathrm{sd}}[i],\ ] ] @xmath11 = \mathbf{h}_{\mathrm{sr}_{n}}[i]a_{\mathrm{s}}\mathbf{t}_{\mathrm{s}}\mathbf{s}[i ] + \eta_{\mathrm{sr}_{n}}[i ] , \label{eq : source_relay}\ ] ] respectively . the matrices @xmath12 and @xmath13 are the @xmath14 source - destination and @xmath15 source - @xmath9 relay channel matrices , respectively . the subscripts s , d and @xmath16 refer to the source , destination and @xmath9 relay nodes , respectively . the quantity @xmath17 is a vector of zero mean additive white gaussian noise , @xmath18 is the @xmath19 data vector , and @xmath20 is the scalar transmit power allocation . the tds matrix , @xmath21 , is a @xmath22 diagonal matrix where each element on the main diagonal specifies whether the correspondingly numbered antenna is active . the received signal of the second phase at the destination is the sum of the forwarded signals from the @xmath1 relays and is expressed as @xmath23 = \boldsymbol{\boldsymbol{\mathcal{h}}}_{\mathrm{rd}}[i]a_{\mathrm{r}}\boldsymbol{\mathcal{t}}_{\mathrm{r}}[i]\mathbf{\hat{\bar{s}}}[i ] + \eta_{\mathrm{rd}}[i ] , \label{eq : rd_compound}\ ] ] where @xmath24 $ ] is the @xmath25 relay tds matrix , @xmath26 = \big[\mathbf{\hat{s}}_{\mathrm{r}_{1}}^{t}[i ] $ ] @xmath27 @xmath28 $ ] @xmath29 @xmath30 $ ] @xmath27 @xmath31\big]$ ] is the @xmath32 estimated data vector and @xmath33 = \big[\mathbf{h}_{\mathrm{r}_{1}\mathrm{d}}[i ] \ \mathbf{h}_{\mathrm{r}_{2}\mathrm{d}}[i ] ... \mathbf{h}_{\mathrm{r}_{n_{\mathrm{r}}}\mathrm{d}}[i]\big]$ ] is the @xmath34 channel matrix . the linear mmse receiver at each relay is given by @xmath35=\underset{\mathbf{w}_{\mathrm{sr}_{n}}}{\mbox{arg\,min}}\;e\big[\big\vert \mathbf{s}[i]-\mathbf{w}_{\mathrm{sr}_{n}}^{h}[i]\mathbf{r}_{\mathrm{sr}_{n}}[i]\big\vert^{2}\big ] , \label{eq : relay_wiener_filter}\ ] ] resulting in the following wiener filter , @xmath36 , where @xmath37\mathbf{r}_{\mathrm{sr}_{n}}^{h}[i]\big]$ ] and @xmath38\mathbf{s}^{h}[i]\big]$ ] are the autocorrelation and cross - correlation matrices , respectively . at the destination , the received signals are stacked to give @xmath39 = \big[\mathbf{r}^{t}_{\mathrm{sd}}[i ] \mathbf{r}^{t}_{\mathrm{rd}}[i]\big]^{t}$ ] . the mmse filter which operates upon @xmath39 $ ] is given by @xmath40=\underset{\mathbf{w}_{\mathrm{d}}}{\mbox{arg\,min}}\;e\big[\big\vert \mathbf{s}[i]-\mathbf{w}_{\mathrm{d}}^{h}[i]\mathbf{r}_{\mathrm{d}}[i]\big\vert^{2}\big ] \label{eq : dest_wiener_filter}\ ] ] and the resulting wiener filter is @xmath41 where @xmath42\mathbf{r}_{\mathrm{d}}^{h}[i]\big]$ ] , @xmath43\mathbf{s}^{h}[i]\big]$ ] . a qpsk slicer follows mmse reception at all nodes ; the output of which is taken as the symbol estimate @xcite . using ( [ eq : relay_wiener_filter ] ) and ( [ eq : dest_wiener_filter ] ) , the mse at the @xmath9 relay and destination are given by @xmath44 and @xmath45 , respectively , where @xmath46\mathbf{s}[i]\big]$ ] . in this section , we formulate the joint tds and rs task as a discrete combinatorial mse problem . the tds optimization problem is given by @xmath47\\ = \underset{\boldsymbol{\mathcal{t}}_{\mathrm{r}}\in \omega_{\mathrm{t}}}{\arg\ \min}\ e \big [ \big\vert\mathbf{s}[i]-\mathbf{w}_{\mathrm{d}}[i,\boldsymbol{\mathcal{t}}_{\mathrm{r}},\hat{\boldsymbol{\mathcal{h}}}_{\mathrm{rd}},\hat{\boldsymbol{\mathcal{h}}}_{\mathrm{sd } } ] \mathbf{r}_{\mathrm{d}}[i]\big\vert^{2}\big ] , \label{eq : mmse_opt_function}\end{gathered}\ ] ] where @xmath48 is the tds matrix set of cardinality @xmath49 and @xmath50 is the number of active relay antennas . the performance and complexity of solutions to ( [ eq : mmse_opt_function ] ) @xmath51 . we decrease @xmath51 whilst ensuring a minimum level of diversity by fixing the number of active antennas at @xmath52 . however , @xmath51 is significant at modest levels of antennas and relays , e.g. @xmath53 and @xmath54 . further improvements can be achieved by a process we term rs . by removing one or more relays from consideration based on their mse performance , the cardinality and quality @xmath48 is improved restricting the phase channels available to the tds process . tds using this refined set then leads to the optimization of both phases . the selection of the single highest mse relay can be expressed as a discrete maximization problem given by @xmath55\\ = \underset{r_{n } \in \omega_{\mathrm{r}}}{\arg\ \max}\ e\big[\big\vert\mathbf{s}[i]-\mathbf{w}_{\mathrm{sr}_{n}}^{h}[i , r_{n},\hat{\boldsymbol{\mathcal{h}}}_{\mathrm{sr}_{n } } ] \mathbf{r}_{\mathrm{sr}_{n}}[i]\big\vert^{2}\big ] , \label{eq : discrete_opt_rs}\end{gathered}\ ] ] where @xmath56 is the set of candidate relays . extension to the selection of multiple relays involves summing the mse from candidate relays and populating @xmath57 with sets of these relays . this results in @xmath58 where @xmath59 is the number of relays to be removed . once rs optimization is complete , a refined subset , @xmath60 , is generated by removing members of @xmath48 which involve transmission from @xmath61 . tds then operates with this subset , where @xmath62 . we propose a complexity dsa which jointly optimizes rs and tds in accordance with ( [ eq : mmse_opt_function ] ) and ( [ eq : discrete_opt_rs ] ) , and converges to the optimal exhaustive solution . 1 .proposed discrete stochastic algorithm [ cols= " < " , ] [ tab : rs_algorithm ] the rs portion of the dsa is given by the algorithm of table [ tab : rs_algorithm ] . at each iteration the mse of a randomly chosen candidate relay ( @xmath63 ) ( step 2 ) and that of the worst performing relay currently known ( @xmath64 ) are calculated ( step 3 ) . via a comparison , the higher mse relay is designated @xmath64 for the next iteration ( step 3 ) . the current solution and the relay chosen for removal ( @xmath65 ) is denoted as the current optimum and is the relay which has occupied @xmath66 most frequently over the course of the packet up to the @xmath67 time instant ; effectively an average of the occupiers of @xmath66 . this averaging / selection process is performed by allocating each member of @xmath56 a @xmath68 unit vector , @xmath69 , which has a one in its corresponding position in @xmath56 , i.e. , @xmath70 $ ] is the label of the worst performing relay at the @xmath67 iteration . the current optimum is then chosen and tracked by means of a @xmath71 state occupation probability ( sop ) vector , @xmath72 . this vector is updated at each iteration by adding @xmath73 $ ] and subtracting the previous value of @xmath72 ( step 4 ) . the current optimum is then determined by selecting the largest element in @xmath72 and its corresponding entry in @xmath56 ( step 5 ) . through this process , the current optimum converges towards and tracks the exhaustive solution @xcite . an alternative interpretation of the proposed algorithm is to view the transitions , @xmath74\rightarrow r^{\mathrm{w}}[i+1]$ ] , as a markov chain and the members of @xmath56 as the possible transition states . the current optimum can then be defined as the most visited state . once rs is complete at each time instant , set reduction ( @xmath75 , step 6 ) and tds can take place . to perform tds , modified versions of steps 1 - 5 are used . the considered set is replaced , @xmath76 ; the structure of interest is replaced , @xmath77 ; the best performing matrix is sought @xmath78 ; the sop vector is replaced @xmath79 and @xmath80 from ( [ eq : mmse_opt_function ] ) . finally , the inequality of step 3 is reversed to enable convergence to the lowest mse tds matrix which is the feedback to the relays . convergence of the proposed algorithm to the optimal exhaustive solution is dependent on the independence of the cost function observations and the satisfaction of @xmath81\big]>\mathcal{f}\big[r[i]\big]\big\ } > \mathrm{pr}\big\{\mathcal{f}\big[r[i]\big]>\mathcal{f}\big[r^{\mathrm{opt}}[i]\big]\big\ } $ ] and @xmath82\big]>\mathcal{f}\big[r^{\mathrm{c}}[i]\big]\big\ } > \mathrm{pr}\big\{\mathcal{f}\big[r[i]\big]>\mathcal{f}\big[r^{\mathrm{c}}[i]\big]\big\ } $ ] for rs and tds ( with the afore mentioned modifications ) . in this work , to minimize complexity , independent observations are not used , therefore the proof of convergence is intractable . however , excellent convergence has been observed under these conditions in @xcite and throughout the simulations conducted for this work . significant complexity savings result from the proposed algorithm ; savings which increase with @xmath2 , @xmath4 , @xmath3 , @xmath1 and @xmath59 . when @xmath83 , @xmath84 , @xmath59 and @xmath85 , the number of complex multiplications for mmse reception and exhaustive tds , exhaustive tds with rs , iterative tds and iterative tds with rs are @xmath86 , @xmath87 , @xmath88 and @xmath89 , respectively , for each time instant . in this section , simulations of the proposed algorithms ( iterative tds with rs ) are presented and comparisons drawn against the optimal exhaustive solutions ( exhaustive tds with rs ) , the unmodified system ( no tds ) , and the direct transmission ( non - cooperative ) . plots of the schemes with tds only ( exhaustive tds , iterative tds ) are also included to illustrate the performance improvement obtained by rs . equal power allocation is maintained in each phase , where @xmath90 when tds is employed and @xmath91 for the unmodified system . for the rls ce , @xmath92 , @xmath93 and @xmath94 are initialized as identity matrices and the exponential forgetting factor is 0.9 . the initial values of @xmath95 , @xmath96 and @xmath97 are zeros matrices . each simulation is averaged over 1000 packets ( @xmath98 ) ; each made up @xmath0 of pilot symbols . 1#21 fig . [ fig : ber_ce ] gives the ber convergence performance of the proposed algorithms . the iterative tds with rs algorithm converges to the optimal ber as does tds with rs and ce , albeit in a delayed fashion due to the ce . the tds with rs scheme exhibits quicker convergence and lower steady state ber . these results and the interdependence between elements of the algorithm confirm that both the rs and tds portions of the algorithm converge to their exhaustive solutions but also the satisfaction of the probability conditions of section [ sec : proposed_algorithms ] . # 1#21 fig . [ fig : snr ] shows the ber versus snr performance of the proposed and conventional algorithms . increased diversity has been achieved whilst maintaining @xmath99 , illustrating that although the maximum available diversity advantage decreases with rs with tds to @xmath100 , the actual diversity achieved has increased . these diversity effects can be attributed to the removal of poor paths which bring little benefit in terms of diversity , but also the increase in transmit power over the remaining paths . the largest gains in diversity are present in region and begin to diminish above this region because relay decoding becomes increasingly reliable and lower power paths become more viable for transmission . this work presented a joint dsa which combines tds and rs along with continuous ce for multi - relay cooperative mimo systems.the scheme exceeds the performance of systems which lack tds and matches that of the optimal exhaustive solution whilst saving considerable computational expense , making it ideal for realtime mobile use . p. clarke and r. c. de lamare , `` joint transmit diversity optimization and relay selection for multi - relay cooperative mimo systems using discrete stochastic algorithms , '' _ ieee communications letters _ , vol.15 , no.10 , pp.1035 - 1037 , october 2011 . p. clarke and r. c. de lamare , `` transmit diversity and relay selection algorithms for multirelay cooperative mimo systems '' _ ieee transactions on vehicular technology _ , vol.61 , no . 3 , pp . 1084 - 1098 , october 2011 . r. c. de lamare and a. alcaim , `` strategies to improve the performance of very low bit rate speech coders and application to a 1.2 kb / s codec , '' _ iee proceedings- vision , image and signal processing . 1 , feb . 2005 . r. c. de lamare and r. sampaio - neto , `` minimum mean squared error iterative successive parallel arbitrated decision feedback detectors for ds - cdma systems , '' _ ieee transactions on communications . 5 , may , 2008 . de lamare and r. sampaio - neto , adaptive reduced - rank equalization algorithms based on alternating optimization design techniques for mimo systems , " _ ieee trans . vehicular technology _ , vol . 60 , no . 6 , pp.2482 - 2494 , july 2011 . p. li , r. c. de lamare and r. fa , multiple feedback successive interference cancellation detection for multiuser mimo systems , " _ ieee transactions on wireless communications _ , vol . 10 , no . 8 , pp . 2434 - 2439 , august 2011 .	we propose a joint discrete stochastic optimization based transmit diversity selection ( tds ) and relay selection ( rs ) algorithm for decode - and - forward ( df ) , cooperative mimo systems with a non - negligible direct path . tds and rs are performed jointly with continuous least squares channel estimation ( ce ) , linear minimum mean square error ( mmse ) receivers are used at all nodes and no inter - relay communication is required . the performance of the proposed scheme is evaluated via bit - error rate ( ber ) comparisons and diversity analysis , and is shown to converge to the optimum exhaustive solution . mimo relaying , transmit diversity , cooperative systems , relay selection
recurrence intervals , defined as the time periods between consecutive extreme events , have been a topic of extensive research across many fields , financial markets in particular . the primary contribution of the published research is an understanding of the statistical regularities in recurrence intervals . the memory behavior in the underlying process strongly affects the distribution form of recurrence intervals @xcite . the interval distribution is exponential if the process has no memory . incorporating a long memory into the underlying process greatly alters the recurrence interval distribution . for example , the stretched exponential and weibull recurrence interval distribution are analytically and numerically confirmed in a process with a long linear memory @xcite . when a process has a long nonlinear memory ( a multifractual process ) , the recurrence intervals are power - law distributed @xcite . there is extensive literature that examines the empirical distribution of recurrence intervals in financial markets . the distribution form is found to be dependent on data source , data type , and data resolution . for example , recurrence interval distributions with a power - law tail are found in the daily volatilities in the japanese market @xcite , in the minute volatilities in the korean @xcite and italian markets @xcite , in the daily returns in the us stock markets @xcite , in the minute returns in the chinese markets @xcite , and in the minute volume in the us @xcite and chinese markets @xcite . in addition , stretched recurrence interval distributions are also observed in the financial volatility at different resolutions in a range of different markets @xcite . the @xmath0-exponential distribution has also been observed in the recurrence intervals between losses in financial returns @xcite , and the corresponding distribution in the chinese stock index future market is a stretched exponential @xcite . in addition to the inconsistent findings on the distribution of empirical recurrence intervals , the existence of scaling behaviors in the recurrence interval distribution for the extremes filtered by different thresholds is under debate . analyzing the distribution of recurrence intervals has indicated that the extreme event filtering threshold should influence the recurrence interval distribution @xcite . this indication was supported when the estimated distributional parameters were found to be strongly dependent on the thresholds when the recurrence intervals are fitted by such distribution functions as the stretched exponential distribution @xcite and the @xmath0-exponential distribution @xcite . @xcite and @xcite propose that the distribution of recurrence intervals depends only on the mean recurrence interval @xmath2 , and not on a specific asset or on the time resolution of the data . only a limited amount of research has used recurrence interval analysis to assess and manage risks in financial markets . an improved method for estimating the value at risk ( var ) based on the recurrence interval is significantly more accurate than traditional estimates based on the overall or local return distributions @xcite . another way of predicting extremes using statistics of recurrence intervals is also superior to the precursory pattern recognition technique when the underlying process is multifractal @xcite . defining a conditional loss probability as the inverse of the expected waiting time before observing another extreme determined by the latest recurrence interval , @xcite finds that the risk of extreme loss events is high if the latest recurrence interval is long or short . in all of these studies , however , only in - sample tests are conducted , and a good performance in in - sample tests can not ensure good results in out - of - sample tests . in contrast , @xcite recently found that the extreme predicting method using recurrence interval analysis does provide good predictions in out - of - sample tests . such events as market crashes , currency crises , and bank failures are financial crisis in which the value of assets or the equity of financial institutions shrinks rapidly . financial crises shock the real - world economy and can cause recessions or depressions if left unchecked . to reduce investor losses and shocks to the economy and to reduce financial turbulence , much effort has gone into predicting financial extremes . there is a plethora of literature on forecasting financial crises , especially currency crises and bank failures , and most of the research relies on the early warning model ( ewm ) @xcite . the ewm identifies the leading indicators of emerging financial problems and uses such techniques as logit ( or probit ) regressions and intelligence approaches to translate them into the hazard probability of crises occurring in the future , which is used as an early warning signal that indicates whether a crisis is imminent . compared to the vast ewm research predicting bank failures and currency crisis , early warning models to monitor stock markets and provide warning signals of market extremes have received little attention . the contributions of the existing literature are as follows . a number of indicators are able to warn of incoming financial extremes . @xcite show that risk aversion indicators are useful in predicting stock market crises , but not currency crises . @xcite finds that such macroeconomic indicators as yield curve spreads and inflation rates can be used to predict stock market recessions . @xcite show that a global measure of liquidity can predict asset price booms . @xcite show that the price - to - book ratio can predict emerging price bubbles . @xcite show that such variables of index futures and options as the vix , open interest , dollar volume , put option price , and put option effective spread can predict equity market crises . @xcite define the average value at risk ( avars ) based on the arma - garch model with standard infinitely divisible innovations as an early warning indicator and find that avars can predict both extreme events and highly volatile markets . by constructing two investment networks based on the cross - border equity and a long - term debt securities portfolio , @xcite identify two network - based indicators ( algebraic connectivity and edge density ) that could have predicted the 2008 global financial crisis . @xcite show that the interconnectedness in the global network of financial linkages could have predicted the financial crises that occurred during the 19782010 period . composite indices averaged from crisis - related variables have been proposed to predict financial crises . @xcite propose a daily financial condition indicator , market volatility , to determine whether a stock market is unstable or not . @xcite define and propose a stock market instability index based on the difference between the current market condition and the past conditions when the market was stable . @xcite propose a model to predict stock market collapse that signals when a massive selling by global institutional investors occurs . @xcite integrate all crisis - related variables into a monthly financial market condition indicator and find that by using a support vector machine the indicator can detect market crises . @xcite use a market instability index to capture risk warning levels , quantify the instability level of the current market , and predict its future behavior there is a pattern of price trajectories that signals near - future market crashes . @xcite develops a log - periodic power law singularity ( lppls ) model for detecting bubbles by combining ( i ) the economic theory of rational expectation bubbles , ( ii ) the effect on the market of imitation and herding behaviors among investors and traders , and ( iii ) the mathematical and statistical physics of bifurcations and phase transitions . the faster - than - exponential ( power law with finite - time singularity ) increase in asset prices accompanied by accelerating oscillations is the main diagnostic that indicates bubbles @xcite . @xcite also corroborate that the lppls pattern can be used as an early warning signal for market crashes . in addition , @xcite convert the price series into networks using a visible graph alogorithm and use the degree - of - price network to measure the magnitude of the faster - than - exponential growth of stock prices , and to predict imminent financial extreme events . on average this indicator performs better than the lppls pattern - recognition indicator . the patterns of financial crises are modeled to predict financial extreme events . @xcite uncover the distribution pattern of waiting time between consecutive market extremes and use it to define a hazard probability that subsequent extremes will occur within a certain time period . they find that this hazard probability performs well in out - of - sample predictions . as an analogue to the seismic activity around earthquakes , @xcite adopt an epidemic - type aftershock sequence model ( a type of mutually self - exciting hawkes point process ) to capture the occurring dynamics of stock market crashes , which can serve as an early warning model for predicting the probability of medium - term crashes . we analyze the daily dow jones industrial average ( djia ) index from 16 february 1885 to 31 december 2015 . the logarithmic return of the djia index over a time scale of one day is defined @xmath3 figures [ fig : index : return](a ) and [ fig : index : return](b ) show plots of the logarithmic djia and its return , respectively . the djia index grows from 30.92 on 16 february 1885 to 17425.03 on 31 december 2015 with a total logarithmic return greater than 6 . although the index exhibits a rising trend throughout sample period , there are falling trends and range - bounds in different subperiods . figure [ fig : index : return ] shows six turbulent periods ( highlighted in shadow ) , the wall street crash of 19291932 , the oil crisis of 19731975 , the black monday crash of 19871989 , the dot - com bubble of 20002003 , the subprime crisis 20072009 , the 2008 financial crisis , and the european sovereign debt crisis 20112015 . ( color online ) . plots of the logarithmic djia index @xmath4 and it s difference , return @xmath5 . ( a ) @xmath4 . ( b ) @xmath5 . , title="fig:",width=272 ] ( color online ) . plots of the logarithmic djia index @xmath4 and it s difference , return @xmath5 . ( a ) @xmath4 . ( b ) @xmath5 . , title="fig:",width=283 ] an extreme value is usually defined as a peak above a threshold ( pot ) @xcite that is @xmath6 times the sample standard deviation . the parameter @xmath6 is a predefined value ( see a summary in table 1 of @xcite ) . although identifying extreme events in terms of pot is widely applied in empirical analysis , the pot has drawbacks . a small @xmath6 value will produce many `` extreme values , '' not all of which are truly extreme , and a large @xmath6 value will indicate genuine extremes but not necessarily include all of them . ( color online ) . determining the extreme value threshold @xmath13 for negative , positive , and absolute returns . ( a ) plots of the tail exponents @xmath14 as a function of the sorted returns . ( b ) plots of the ks statistics @xmath15 with respect to the sorted returns . the ks statistics is defined as the maximum absolute difference between the empirical and fitting tail distributions.,title="fig:",width=275 ] ( color online ) . determining the extreme value threshold @xmath13 for negative , positive , and absolute returns . ( a ) plots of the tail exponents @xmath14 as a function of the sorted returns . ( b ) plots of the ks statistics @xmath15 with respect to the sorted returns . the ks statistics is defined as the maximum absolute difference between the empirical and fitting tail distributions.,title="fig:",width=283 ] we estimate the shape parameter @xmath12 using the hill estimator @xcite , which is a non - parametric method . for a given sample @xmath16 , we sort the data in ascending order , @xmath17 the @xmath12 value given by the hill estimator is @xmath18,\ ] ] where @xmath19 corresponds to the extreme value threshold @xmath13 that will be determined . one way to find threshold @xmath13 is by ( i ) estimating the value of @xmath12 with respect to all possible values of @xmath13 , and ( ii ) plotting @xmath14 against @xmath13 to find a range of @xmath13 values within which the estimated @xmath14 values are stable @xcite . in practice , this `` stable behavior '' between @xmath14 and @xmath19 is difficult to quantify . for example , fig . [ fig : evt : threshold](a ) uses djia returns to illustrate the estimated @xmath14 as a function of the sorted djia ( negative , positive , and absolute ) returns . the @xmath14 values strongly fluctuate and there is no stable range . an alternative approach is to use ks statistics to measure the agreement between the empirical and fitting tail distributions . ks statistics quantify the maximum absolute difference between both distributions . the most suitable threshold @xmath13 is associated with the best fits to the tail distribution , which has the smallest ks statistical values @xcite . figure [ fig : evt : threshold](b ) shows the plots of the ks statistics @xmath15 with respect to the sorted ( negative , positive , and absolute ) returns . the significant low point in each curve allows us to more easily determine the extreme value threshold @xmath13 . for sake of comparison , we also use the quantiles of 95% , 97.5% , and 99% to define the extremes . definitions based on the quantile are common in the analysis of value - at - risk ( var ) . @xcite also define the 95% quantile of returns and the 95% quantile of negative returns as extremes and crashes . by taking into consideration only the time in which extremes occur , we base our prediction of extreme returns on the hazard probability @xmath20 , which measures the probability that following an extreme return occurring at @xmath21 time in the past there is an additional waiting time @xmath22 before another extreme return occurs . @xcite and @xcite theoretically derived the hazard probability @xmath23 using the distribution of recurrence intervals between extreme events , @xmath24 where @xmath25 is the probability distribution of the recurring intervals . once we have the distribution form of @xmath25 , the formula for @xmath26 can be derived from eq . ( [ eq : wq ] ) . although the recurrence intervals of poisson processes are exponentially distributed @xcite , which generates a constant hazard probability when @xmath22 is given , financial processes always exhibit such non - poissonian characteristics as long - term dependence and multifractality in volatilities @xcite , medium - term dependence ( e.g. , momentum and contrarian behaviors @xcite ) , and multiscaling behaviors in returns @xcite , which leads to that the recurrence intervals are no longer exponentially distributed , and that the derivation of the close distribution form for the recurrence intervals is obstructed @xcite . the non - poissonian features also result in a controversial situation in the empirical analysis of the distribution formula of recurrence intervals . for example , the reported distributions range from a power - law distribution with an exponential cutoff @xcite to a stretched exponential distribution @xcite , from a @xmath0-exponential distribution @xcite to a @xmath0-weibull distribution @xcite . here we employ three common functions to fit the recurrence interval distributions . the three formulas are the stretched exponential distribution , @xmath27,\ ] ] the @xmath0-exponential distribution , @xmath28^{-\frac{1}{q-1}},\ ] ] and the weibull distribution , @xmath29,\ ] ] by putting the three probability distributions eqs . ( [ eq : pdf : sexp])([eq : pdf : wbl ] ) into eq . ( [ eq : wq ] ) , we obtain the hazard probability @xmath30 for the stretched exponential distribution , @xmath31^\mu\right)}{\gamma_u\left ( \frac{1}{\mu } , ( bt)^{\mu}\right)},\ ] ] the hazard probability @xmath32 for @xmath0-exponential distribution , @xmath33 ^{1-\frac{1}{q-1}},\ ] ] and the hazard probability @xmath34 for weibull distribution , @xmath35,\ ] ] where @xmath36 and @xmath37 are lower and upper incomplete gamma functions . for fixed @xmath22 , all three hazard probabilities decrease as @xmath21 increases , which explains the clustering of extremes in financial returns and volatilities . to use the hazard probability @xmath38 to predict the extremes we must set a hazard threshold @xmath39 to trigger the early warning indicator of an approaching extreme event . if the hazard probability @xmath40 is greater than the hazard threshold @xmath39 , an alarm that an extreme return will occur during the next @xmath22 time is activated . the hazard threshold @xmath39 is not an arbitrary given value but depending on the risk level preferences of investors is optimized to balance between false alarms and not detecting events . the hazard probability @xmath20 becomes a binary extreme forecast that equals one when @xmath20 exceeds the hazard threshold @xmath39 and equals zero otherwise . when comparing the forecasted extremes with the actual events we see ( i ) correct predictions of an extreme return occurring , ( ii ) correct predictions of a non - extreme return occurring , ( iii ) missed events , and ( iv ) false alarms . by counting how many times each outcome occurs we can compute a range of evaluation measurements including the correct prediction rate , the false alarm rate , and the accuracy . our primary interest here is correct prediction rate @xmath41 and false alarm rate @xmath42 , which are defined as @xmath43 where @xmath44 is the number of extreme returns that are correctly predicted , @xmath45 the number of non - extreme returns that are correctly predicted , @xmath46 the number of missed events , and @xmath47 the number of false alarms . following @xcite , we use the hanssen - kuiper skill score ( kss ) to assess the validity of extreme forecasts . the kss is the difference @xmath48 between the correct prediction rate and the false alarm rate . the kss encompasses both missing occurrence errors and false alarms errors . decreasing these two errors increases the value of kss . our goal is to find a balanced signal for investors when they prefer either type 1 and type 2 errors and to take into account whether they use or discard the predictive signals . following @xcite we define a loss function when a hazard probability threshold is added issue extreme forecasts , @xmath49 where @xmath50 is the ratio of missing events ( type 1 errors ) and @xmath42 is the ratio of false alarms ( type 2 errors ) . the parameter @xmath51 is the investor preference for avoiding either type 1 or type 2 errors @xcite . we further define the usefulness of extreme forecasts as @xmath52 where @xmath53 is the loss faced by investors when they ignore the predictive signals , and @xmath54 is the extent to which the extreme forecasting model offers better performance than no model at all @xcite . extreme forecasts are useful when @xmath55 , which means that losses using the forecasts are lower than when the forecasts are ignored . the usefulness definition here ignores any influence from the data imbalance , i.e. , that non - extreme events occur much more frequently than extreme events @xcite . given hazard probability @xmath56 , we need a hazard threshold @xmath39 that maximizes usefulness @xmath54 @xcite . @xcite optimizes the threshold by minimizing the noise - to - signal ratio @xmath57 . when we optimize the usefulness there is a marginal rate of substitution between type 1 and type 2 errors , but this marginal rate is not clear in the optimization of the noise - to - signal ratio , and this can result in an unacceptable level of type 1 and type 2 errors @xcite . by introducing the stretched exponential function of eq . ( [ eq : pdf : sexp ] ) into the probability density function @xmath58 , we obtain @xmath59 where @xmath60 is the gamma function . @xcite and @xcite describe the one - to - one correspondence between the average recurrence interval @xmath2 and the percentage of extremes , @xmath61 where @xmath62 is the quantile that is used to define the extreme values . for this equation to be valid , the extremes must be positive . when extremes are negative , we convert them into positives by multiplying by @xmath63 . @xcite find that the average recurrence interval is universal irrespective of the dependence structure of the underlying process . from the definition of expectation , the average recurrence interval can also be written @xmath64 . for the stretched exponential distribution , we have @xmath65 by solving eqs . ( [ eq : se:1 ] ) and ( [ eq : se:2 ] ) and using @xmath10 and @xmath2 for the stretched exponential distribution , parameters @xmath66 and @xmath67 are @xmath68 this strategy reduces the number of estimated parameters from three to one . the mean of the @xmath0-exponential distribution is @xmath69 $ ] . for there to be a mean , @xmath0 must be less than 3/2 . then the parameter @xmath70 can be found using @xmath0 and @xmath2 , @xmath71 the expectation for the weibull distribution is @xmath72 . similarly , the @xmath73 can be expressed in terms of @xmath74 and @xmath2 , @xmath75 we need to estimate only one parameter for the three distributions when they are used to fit the recurrence intervals . we adopt the maximum likelihood estimation ( mle ) to estimate the distributional parameters . the logarithmic likelihood functions are the stretched exponential distribution @xmath76 the @xmath0-exponential distribution @xmath77 -\frac{1}{q-1 } \sum_{i=1}^{n } \ln [ 1+(q-1)\lambda \tau_i],\ ] ] and the weibull distribution @xmath78,\ ] ] in which @xmath79 is the number of recurrence intervals . taking the stretched exponential distribution as an example , the logarithmic likelihood function @xmath80 is a one - variable function of @xmath10 . although usually we can solve the equation by taking the first order derivative of @xmath80 with respect to @xmath10 to find the solution that maximizes the likelihood , here the analytical expression of the derivative of @xmath80 with respect to @xmath10 is more difficult to obtain . we thus discretize @xmath10 in the @xmath81 range with a step of @xmath82 and calculate the logarithmic likelihood function for each discrete value of @xmath10 . the @xmath10 associated with the maximum value of @xmath80 is the maximum likelihood estimation . in the same way we estimate the parameters @xmath83 and @xmath84 for the @xmath0-exponential and weibull distributions , respectively . in order to test the validity of our extreme - return - prediction model , we use the data before each turbulent period to calibrate the model and each turbulent period that follows for out - of - sample forecasting . we obtain six in - sample calibrating periods : 18851928 , 18851972 , 18851986 , 18851999 , 18852006 , and 18852010 their out - of - sample predicting periods are 19291932 , 19731975 , 19871989 , 20002003 , 20072009 , and 20112015 , respectively . in each in - sample calibrating period , we identify the extreme value threshold @xmath13 and extract the extreme values associated with @xmath13 . we also locate the extreme values based on the quantile thresholds of 95% , 97.5% , and 99% . for each group of extreme values , we estimate the waiting time between consecutive extreme values , i.e. , the recurrence interval . we take positive , negative , and absolute returns into account in our analysis because they may have connections with specific trading strategies . the investors holding long positions in the market are more sensitive to extreme negative returns and those holding short positions less sensitive . table [ tb : ri : statistics ] lists the recurrence intervals for different thresholds from different calibrating periods . note that unlike the observations from the quantile threshold , the observations from the extreme value threshold do not increase monotonically as the calibration periods are expanded . the sharp decease in the number of recurrence intervals , for example from panel b to panel c for positive returns and from panel c to panel d for absolute returns , indicates that there is a dramatic increase in the extreme value threshold , suggesting that the market after 1973 became more volatile [ see fig . [ fig : index : return](b ) ] . the mean values of the recurrence interval are strongly influenced by the extreme values , as indicated by the large gap between the means and medians . because the skewness is positive and the kurtosis is much greater than 3 , the recurrence intervals also exhibit a right - skewed and fat - tailed distribution . this affirms the finding that the recurrence intervals obey a stretched exponential @xcite or a @xmath0-exponential distribution @xcite . we see a significantly positive autocorrelation of lag 1 in the 95% column in panel a and significant ljung - box q statistics at the 0.01 level for the three types of returns . in addition the autocorrelation of lag 5 is also positive when returns are positive . this indicates that there are autocorrelations at short and long lags in the recurrence intervals at the 95% quantile threshold . in the 99% column the autocorrelations are close to 0 and the ljung - box q statistics are insignificant for positive , negative , and absolute returns , suggesting that there are no correlations in the recurrence intervals . the results in both columns also show that the autocorrelation of recurrence intervals gradually decreases to insignificance when the quantile threshold increases from 95% to 99% , which is also seen in the columns of @xmath13 and 97.5% quantile . in panels b through f , the autocorrelation coefficients at lags 1 and 5 are all positive and statistically significant , indicating the presence of strong autocorrelations in the recurrence intervals . in addition , the ljung - box q statistics of lag 30 are statistically significant at the 1% level , implying that significant autocorrelations are also prevalent when recurrence interval lags are longer . these results are supported by the long - memory behavior results of a detrended fluctuation analysis ( dfa ) of the recurrence intervals @xcite . [email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@.l & & & & & & + & & & & & & & & & & & & & & & + + obsv & & 278&0 & 652&0 & 326&0 & 130&0 & & 477&0 & 652&0 & 326&0 & 130&0 & & 145&0 & 652&0 & 326&0 & 130&0 + mean & & 46&950 & 20&0 & 40&0 & 100&0 & & 27&363 & 20&0 & 40&0 & 100&0 & & 90&014 & 20&0 & 40&0 & 100&0 + median & & 12&0 & 6&500 & 11&0 & 21&0 & & 12&0 & 9&0 & 16&0 & 23&500 & & 12&0 & 5&0 & 6&500 & 12&0 + stdev & & 91&524 & 38&122 & 78&407 & 173&978 & & 44&790 & 32&100 & 78&043 & 184&695 & & 182&677 & 46&844 & 94&819 & 201&517 + skew & & 4&713 & 6&703 & 5&131 & 3&859 & & 3&782 & 4&286 & 6&067 & 3&677 & & 3&704 & 9&852 & 6&000 & 3&402 + kurt & & 33&706 & 80&856 & 43&300 & 24&150 & & 20&607 & 27&888 & 55&805 & 20&586 & & 20&540 & 157&674 & 57&471 & 16&804 + rho(1 ) & & 0&120@xmath85 & 0&121@xmath86 & 0&140@xmath85 & 0&070 & & 0&257@xmath86 & 0&218@xmath86 & 0&081 & @xmath870&028 & & 0&105 & 0&143@xmath86 & 0&090 & 0&082 + rho(5 ) & & @xmath870&064 & 0&062 & @xmath870&053 & @xmath870&054 & & 0&083@xmath88 & 0&112@xmath86 & 0&142@xmath85 & @xmath870&086 & & @xmath870&040 & 0&005 & 0&041 & @xmath870&058 + q(30 ) & & 48&876@xmath85 & 66&026@xmath86 & 36&545 & 18&890 & & 131&947@xmath86 & 143&181@xmath86 & 41&158@xmath88 & 14&910 & & 18&024 & 53&899@xmath86 & 30&233 & 10&549 + + obsv & & 2032&0 & 1254&0 & 627&0 & 250&0 & & 1171&0 & 1254&0 & 627&0 & 250&0 & & 2546&0 & 1254&0 & 627&0 & 250&0 + mean & & 12&349 & 20&0 & 40&0 & 100&0 & & 21&430 & 20&0 & 40&0 & 100&0 & & 9&856 & 20&0 & 40&0 & 100&0 + median & & 4&0 & 5&0 & 6&0 & 8&0 & & 7&0 & 7&0 & 8&0 & 12&0 & & 3&0 & 3&0 & 4&0 & 6&0 + stdev & & 22&546 & 43&451 & 110&804 & 251&118 & & 48&351 & 43&793 & 105&520 & 281&113 & & 23&928 & 59&488 & 138&074 & 276&742 + skew & & 5&010 & 5&606 & 7&684 & 4&409 & & 8&208 & 7&289 & 6&213 & 6&072 & & 7&690 & 7&306 & 7&189 & 4&007 + kurt & & 41&635 & 47&837 & 84&779 & 26&219 & & 110&246 & 83&273 & 50&569 & 50&613 & & 89&933 & 73&524 & 65&717 & 19&536 + rho(1 ) & & 0&193@xmath86 & 0&295@xmath86 & 0&140@xmath86 & 0&532@xmath86 & & 0&310@xmath86 & 0&303@xmath86 & 0&193@xmath86 & 0&215@xmath86 & & 0&219@xmath86 & 0&194@xmath86 & 0&075@xmath88 & 0&311@xmath86 + rho(5 ) & & 0&112@xmath86 & 0&098@xmath86 & 0&245@xmath86 & 0&090 & & 0&117@xmath86 & 0&106@xmath86 & 0&131@xmath86 & 0&309@xmath86 & & 0&158@xmath86 & 0&085@xmath86 & 0&110@xmath86 & 0&407@xmath86 + q(30 ) & & 843&689@xmath86 & 755&150@xmath86 & 206&777@xmath86 & 111&717@xmath86 & & 753&825@xmath86 & 733&710@xmath86 & 432&781@xmath86 & 82&437@xmath86 & & 1153&606@xmath86 & 585&612@xmath86 & 225&441@xmath86 & 110&987@xmath86 + + obsv & & 716&0 & 1431&0 & 715&0 & 286&0 & & 1360&0 & 1431&0 & 715&0 & 286&0 & & 2822&0 & 1431&0 & 715&0 & 286&0 + mean & & 39&989 & 20&0 & 40&0 & 100&0 & & 21&053 & 20&0 & 40&0 & 100&0 & & 10&146 & 20&0 & 40&0 & 100&0 + median & & 7&0 & 5&0 & 7&0 & 8&0 & & 8&0 & 7&0 & 9&0 & 12&0 & & 3&0 & 4&0 & 4&0 & 6&0 + stdev & & 99&845 & 43&138 & 99&905 & 253&139 & & 46&590 & 44&787 & 102&571 & 272&215 & & 24&320 & 58&219 & 128&684 & 270&823 + skew & & 5&139 & 5&507 & 5&135 & 4&222 & & 8&144 & 8&664 & 6&099 & 5&977 & & 7&815 & 7&182 & 6&986 & 3&832 + kurt & & 34&807 & 45&983 & 34&762 & 24&000 & & 111&272 & 124&207 & 50&108 & 50&643 & & 93&261 & 72&012 & 66&050 & 18&315 + rho(1 ) & & 0&242@xmath86 & 0&284@xmath86 & 0&242@xmath86 & 0&565@xmath86 & & 0&296@xmath86 & 0&259@xmath86 & 0&196@xmath86 & 0&221@xmath86 & & 0&233@xmath86 & 0&205@xmath86 & 0&098@xmath86 & 0&213@xmath86 + rho(5 ) & & 0&176@xmath86 & 0&117@xmath86 & 0&174@xmath86 & 0&307@xmath86 & & 0&114@xmath86 & 0&130@xmath86 & 0&136@xmath86 & 0&271@xmath86 & & 0&148@xmath86 & 0&075@xmath86 & 0&080@xmath85 & 0&237@xmath86 + q(30 ) & & 276&588@xmath86 & 667&824@xmath86 & 277&421@xmath86 & 249&183@xmath86 & & 791&779@xmath86 & 736&976@xmath86 & 450&736@xmath86 & 101&007@xmath86 & & 1277&446@xmath86 & 597&888@xmath86 & 238&276@xmath86 & 182&886@xmath86 + + obsv & & 782&0 & 1595&0 & 797&0 & 319&0 & & 1448&0 & 1595&0 & 797&0 & 319&0 & & 1421&0 & 1595&0 & 797&0 & 319&0 + mean & & 40&816 & 20&0 & 40&0 & 100&0 & & 22&043 & 20&0 & 40&0 & 100&0 & & 22&462 & 20&0 & 40&0 & 100&0 + median & & 8&0 & 6&0 & 7&0 & 8&0 & & 8&0 & 8&0 & 9&0 & 13&0 & & 4&0 & 4&0 & 4&0 & 5&0 + stdev & & 97&920 & 41&851 & 96&353 & 248&548 & & 47&351 & 43&650 & 99&684 & 270&301 & & 62&523 & 56&713 & 127&547 & 270&274 + skew & & 5&055 & 5&511 & 5&155 & 4&210 & & 7&660 & 8&550 & 6&076 & 5&746 & & 6&589 & 7&148 & 7&004 & 3&903 + kurt & & 34&510 & 46&938 & 35&971 & 23&986 & & 99&818 & 124&073 & 50&734 & 46&911 & & 60&862 & 72&525 & 65&444 & 18&877 + rho(1 ) & & 0&256@xmath86 & 0&278@xmath86 & 0&252@xmath86 & 0&548@xmath86 & & 0&327@xmath86 & 0&259@xmath86 & 0&228@xmath86 & 0&199@xmath86 & & 0&203@xmath86 & 0&214@xmath86 & 0&089@xmath85 & 0&206@xmath86 + rho(5 ) & & 0&176@xmath86 & 0&131@xmath86 & 0&189@xmath86 & 0&273@xmath86 & & 0&142@xmath86 & 0&139@xmath86 & 0&131@xmath86 & 0&326@xmath86 & & 0&081@xmath86 & 0&082@xmath86 & 0&148@xmath86 & 0&203@xmath86 + q(30 ) & & 277&454@xmath86 & 675&904@xmath86 & 303&462@xmath86 & 254&647@xmath86 & & 849&279@xmath86 & 790&724@xmath86 & 467&774@xmath86 & 102&416@xmath86 & & 677&953@xmath86 & 657&786@xmath86 & 195&112@xmath86 & 154&916@xmath86 + + obsv & & 834&0 & 1683&0 & 841&0 & 336&0 & & 1539&0 & 1683&0 & 841&0 & 336&0 & & 1518&0 & 1683&0 & 841&0 & 336&0 + mean & & 40&380 & 20&0 & 40&0 & 100&0 & & 21&882 & 20&0 & 40&0 & 100&0 & & 22&185 & 20&0 & 40&0 & 100&0 + median & & 8&0 & 6&0 & 8&0 & 8&0 & & 8&0 & 8&0 & 9&0 & 13&0 & & 4&0 & 4&0 & 4&0 & 6&0 + stdev & & 95&125 & 41&361 & 94&757 & 241&773 & & 46&746 & 43&235 & 102&416 & 265&009 & & 60&659 & 56&179 & 124&604 & 264&232 + skew & & 5&213 & 5&472 & 5&240 & 4&322 & & 7&580 & 8&417 & 6&467 & 5&925 & & 6&795 & 7&162 & 7&148 & 3&985 + kurt & & 36&645 & 46&813 & 36&980 & 25&403 & & 99&236 & 122&461 & 56&734 & 50&015 & & 64&680 & 72&654 & 68&333 & 19&714 + rho(1 ) & & 0&258@xmath86 & 0&295@xmath86 & 0&254@xmath86 & 0&536@xmath86 & & 0&337@xmath86 & 0&286@xmath86 & 0&282@xmath86 & 0&199@xmath86 & & 0&206@xmath86 & 0&221@xmath86 & 0&102@xmath86 & 0&204@xmath86 + rho(5 ) & & 0&178@xmath86 & 0&105@xmath86 & 0&193@xmath86 & 0&278@xmath86 & & 0&151@xmath86 & 0&141@xmath86 & 0&081@xmath85 & 0&251@xmath86 & & 0&083@xmath86 & 0&091@xmath86 & 0&058@xmath88 & 0&204@xmath86 + q(30 ) & & 304&082@xmath86 & 873&194@xmath86 & 304&969@xmath86 & 260&485@xmath86 & & 980&303@xmath86 & 875&671@xmath86 & 491&718@xmath86 & 106&786@xmath86 & & 743&033@xmath86 & 673&330@xmath86 & 200&311@xmath86 & 158&391@xmath86 + + obsv & & 858&0 & 1734&0 & 867&0 & 346&0 & & 924&0 & 1734&0 & 867&0 & 346&0 & & 1540&0 & 1734&0 & 867&0 & 346&0 + mean & & 40&425 & 20&0 & 40&0 & 100&0 & & 37&538 & 20&0 & 40&0 & 100&0 & & 22&523 & 20&0 & 40&0 & 100&0 + median & & 7&0 & 6&0 & 7&0 & 8&0 & & 9&0 & 8&0 & 9&0 & 12&0 & & 4&0 & 4&0 & 4&0 & 5&0 + stdev & & 111&346 & 42&218 & 109&226 & 259&435 & & 97&671 & 43&424 & 104&937 & 269&598 & & 68&579 & 59&451 & 132&133 & 275&164 + skew & & 7&095 & 5&598 & 7&081 & 4&464 & & 6&397 & 8&211 & 6&287 & 5&680 & & 8&097 & 7&761 & 6&888 & 3&898 + kurt & & 71&630 & 49&644 & 72&329 & 26&510 & & 56&115 & 117&442 & 52&915 & 46&183 & & 91&999 & 83&058 & 61&436 & 18&701 + rho(1 ) & & 0&146@xmath86 & 0&296@xmath86 & 0&161@xmath86 & 0&495@xmath86 & & 0&308@xmath86 & 0&297@xmath86 & 0&302@xmath86 & 0&184@xmath86 & & 0&218@xmath86 & 0&201@xmath86 & 0&112@xmath86 & 0&165@xmath86 + rho(5 ) & & 0&174@xmath86 & 0&088@xmath86 & 0&132@xmath86 & 0&234@xmath86 & & 0&084@xmath85 & 0&138@xmath86 & 0&191@xmath86 & 0&218@xmath86 & & 0&113@xmath86 & 0&099@xmath86 & 0&057@xmath88 & 0&249@xmath86 + q(30 ) & & 227&668@xmath86 & 877&559@xmath86 & 215&488@xmath86 & 223&093@xmath86 & & 552&569@xmath86 & 994&993@xmath86 & 465&599@xmath86 & 99&923@xmath86 & & 565&484@xmath86 & 613&619@xmath86 & 218&575@xmath86 & 146&530@xmath86 + the recurrence intervals are fitted by the stretched exponential distribution , @xmath0-exponential distribution , and weibull distribution in each in - sample calibrating period . figure [ fig : distribution : ri : nr : qt99 ] shows the probability distribution of the recurrence intervals between the negative extreme events in the 99% quantile during the 1928 to 1985 in - sample calibrating period . the best fits to the three fitting distributions are also plotted as solid curves for comparison . note that the stretched exponential distribution gives the best fit . note also that the stretch exponential fits are most likely , which also agrees with the distribution of recurrence intervals between the negative and absolute extreme returns in the chinese markets and the us markets @xcite . because all the distribution curves are very similar , we do not show the recurrence intervals in other calibrating periods . ( color online ) . distribution of the recurrence intervals . this figure presents the empirical distributions of the recurrence intervals between negative extreme returns in the quantile of 99% and the best fits to the three distributions , stretched exponential distribution , @xmath0-exponential distribution , and weibull distribution . the analysis is performed in the period from 1885 to 1928.,width=302 ] table [ tb : ri : fit : par ] shows the estimated parameters of three fitting distributions of the recurrence intervals obtained from different return types and different thresholds . to assess the validity of the distributional fits , the logarithmic likelihoods are also listed . the likelihoods of the distribution with the maximum value are show in bold to indicate that the corresponding distribution gives the best fit . panel a shows that the best recurrence interval fits transition from the @xmath0-exponential distribution to the stretched exponential distribution as the threshold is increased . this distribution transition behavior is also seen in the recurrence intervals in the minute volatilities in the chinese stock markets @xcite . note that in panels b f the maximum likelihood comes from the @xmath0-exponential distribution for all the recurrence interval fits . @xcite and @xcite also found that the recurrence intervals between extreme loses are captured by the @xmath0-exponential distribution . the possible explanation for the lack of maximum likelihoods for the stretched exponential distribution in panels b f is that the 99% quantile threshold is not sufficiently large for a distribution transition from a @xmath0-exponential to a stretched exponential to occur . table [ tb : ri : fit : par ] shows a monotonic trend between the estimated parameters and the quantile thresholds for each type of returns , such that @xmath10 and @xmath74 decrease and @xmath0 increases as the quantile threshold increases . our results support the dependence of the recurrence interval distribution on the quantile threshold @xcite . for the same quantile threshold and the same type of return , we also find that the estimated distributional parameters are close to each other in panels b f , which indicates that the recurrence interval distribution depends solely on the quantile @xcite . [email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@.l & & & & & & + & & & & & & & & & & & & & & & + + @xmath10 & & 0&350 & 0&461 & 0&360 & 0&280 & & 0&513 & 0&573 & 0&451 & 0&307 & & 0&227 & 0&383 & 0&284 & 0&215 + @xmath80 & & * @xmath871264&269 * & @xmath872496&914 & * @xmath871434&225 * & * @xmath87674&465 * & & @xmath872000&951 & @xmath872553&728 & @xmath871473&376 & * @xmath87680&297 * & & * @xmath87700&468 * & @xmath872424&402 & @xmath871365&008 & * @xmath87634&993 * + @xmath0 & & 1&408 & 1&357 & 1&405 & 1&435 & & 1&316 & 1&286 & 1&345 & 1&423 & & 1&461 & 1&399 & 1&443 & 1&465 + @xmath89 & & @xmath871266&953 & * @xmath872485&817 * & @xmath871434&396 & @xmath87684&927 & & * @xmath871996&519 * & * @xmath872544&451 * & * @xmath871471&303 * & @xmath87685&261 & & @xmath87711&080 & * @xmath872411&504 * & * @xmath871360&696 * & @xmath87647&154 + @xmath74 & & 0&625 & 0&718 & 0&634 & 0&563 & & 0&757 & 0&802 & 0&708 & 0&587 & & 0&498 & 0&651 & 0&556 & 0&483 + @xmath90 & & @xmath871275&718 & @xmath872523&094 & @xmath871448&428 & @xmath87678&027 & & @xmath872014&244 & @xmath872571&295 & @xmath871483&428 & @xmath87684&853 & & @xmath87707&857 & @xmath872458&188 & @xmath871386&102 & @xmath87641&552 + + @xmath10 & & 0&512 & 0&398 & 0&294 & 0&207 & & 0&433 & 0&451 & 0&315 & 0&229 & & 0&452 & 0&312 & 0&230 & 0&177 + @xmath80 & & @xmath876870&875 & @xmath874662&711 & @xmath872602&562 & @xmath871173&240 & & @xmath874501&362 & @xmath874759&720 & @xmath872654&990 & @xmath871205&945 & & @xmath877787&724 & @xmath874359&017 & @xmath872384&191 & @xmath871098&368 + @xmath0 & & 1&336 & 1&396 & 1&442 & 1&473 & & 1&373 & 1&363 & 1&429 & 1&464 & & 1&373 & 1&438 & 1&467 & 1&483 + @xmath89 & & @xmath87 * 6805 * & * 955 * & @xmath87 * 4610 * & * 407 * & @xmath87 * 2571 * & * 429 * & @xmath87 * 1165 * & * 118 * & & @xmath87 * 4463 * & * 297 * & @xmath87 * 4718 * & * 507 * & @xmath87 * 2632 * & * 444 * & @xmath87 * 1200 * & * 464 * & & @xmath87 * 7624 * & * 907 * & @xmath87 * 4248 * & * 552 * & @xmath87 * 2312 * & * 401 * & @xmath87 * 1081 * & * 346 * + @xmath74 & & 0&758 & 0&663 & 0&563 & 0&465 & & 0&692 & 0&706 & 0&585 & 0&492 & & 0&704 & 0&577 & 0&487 & 0&422 + @xmath90 & & @xmath876957&774 & @xmath874735&734 & @xmath872650&054 & @xmath871195&844 & & @xmath874559&812 & @xmath874820&370 & @xmath872696&831 & @xmath871226&002 & & @xmath877951&129 & @xmath874469&435 & @xmath872453&110 & @xmath871126&514 + + @xmath10 & & 0&302 & 0&409 & 0&302 & 0&208 & & 0&449 & 0&463 & 0&325 & 0&229 & & 0&464 & 0&322 & 0&237 & 0&179 + @xmath80 & & @xmath873011&043 & @xmath875359&481 & @xmath873008&233 & @xmath871348&487 & & @xmath875236&811 & @xmath875455&453 & @xmath873056&744 & @xmath871382&459 & & @xmath878773&031 & @xmath875036&183 & @xmath872770&080 & @xmath871272&260 + @xmath0 & & 1&437 & 1&388 & 1&437 & 1&473 & & 1&362 & 1&354 & 1&424 & 1&464 & & 1&365 & 1&433 & 1&465 & 1&482 + @xmath89 & & @xmath87 * 2980 * & * 534 * & @xmath87 * 5305 * & * 811 * & @xmath87 * 2978 * & * 056 * & @xmath87 * 1342 * & * 692 * & & @xmath87 * 5195 * & * 029 * & @xmath87 * 5410 * & * 298 * & @xmath87 * 3035 * & * 525 * & @xmath87 * 1377 * & * 135 * & & @xmath87 * 8608 * & * 555 * & @xmath87 * 4924 * & * 461 * & @xmath87 * 2701 * & * 469 * & @xmath87 * 1258 * & * 164 * + @xmath74 & & 0&571 & 0&672 & 0&571 & 0&465 & & 0&705 & 0&715 & 0&594 & 0&493 & & 0&714 & 0&587 & 0&496 & 0&426 + @xmath90 & & @xmath873062&519 & @xmath875438&722 & @xmath873059&523 & @xmath871373&600 & & @xmath875300&912 & @xmath875521&812 & @xmath873101&494 & @xmath871405&021 & & @xmath878944&445 & @xmath875155&682 & @xmath872843&627 & @xmath871302&926 + + @xmath10 & & 0&307 & 0&419 & 0&308 & 0&208 & & 0&457 & 0&477 & 0&335 & 0&233 & & 0&314 & 0&329 & 0&241 & 0&178 + @xmath80 & & @xmath873325&185 & @xmath876004&925 & @xmath873375&885 & @xmath871506&977 & & @xmath875657&201 & @xmath876104&959 & @xmath873430&370 & @xmath871552&323 & & @xmath875155&298 & @xmath875655&759 & @xmath873115&198 & @xmath871416&369 + @xmath0 & & 1&435 & 1&382 & 1&434 & 1&472 & & 1&357 & 1&345 & 1&418 & 1&462 & & 1&435 & 1&429 & 1&463 & 1&482 + @xmath89 & & @xmath87 * 3298 * & * 423 * & @xmath87 * 5953 * & * 468 * & @xmath87 * 3348 * & * 336 * & @xmath87 * 1501 * & * 836 * & & @xmath87 * 5615 * & * 392 * & @xmath87 * 6055 * & * 245 * & @xmath87 * 3408 * & * 686 * & @xmath87 * 1548 * & * 247 * & & @xmath87 * 5054 * & * 298 * & @xmath87 * 5540 * & * 644 * & @xmath87 * 3047 * & * 366 * & @xmath87 * 1403 * & * 017 * + @xmath74 & & 0&577 & 0&681 & 0&578 & 0&467 & & 0&711 & 0&726 & 0&604 & 0&498 & & 0&581 & 0&595 & 0&502 & 0&424 + @xmath90 & & @xmath873378&117 & @xmath876087&515 & @xmath873429&832 & @xmath871534&442 & & @xmath875722&565 & @xmath876175&819 & @xmath873477&890 & @xmath871576&590 & & @xmath875271&119 & @xmath875783&800 & @xmath873193&905 & @xmath871450&209 + + @xmath10 & & 0&310 & 0&422 & 0&311 & 0&214 & & 0&455 & 0&472 & 0&334 & 0&235 & & 0&318 & 0&331 & 0&241 & 0&181 + @xmath80 & & @xmath873529&464 & @xmath876339&332 & @xmath873552&077 & @xmath871598&990 & & @xmath875994&090 & @xmath876429&892 & @xmath873604&772 & @xmath871636&240 & & @xmath875476&112 & @xmath875959&563 & @xmath873276&784 & @xmath871502&338 + @xmath0 & & 1&434 & 1&381 & 1&433 & 1&470 & & 1&359 & 1&350 & 1&419 & 1&461 & & 1&434 & 1&429 & 1&463 & 1&481 + @xmath89 & & @xmath87 * 3498 * & * 388 * & @xmath87 * 6285 * & * 001 * & @xmath87 * 3519 * & * 699 * & @xmath87 * 1595 * & * 926 * & & @xmath87 * 5947 * & * 393 * & @xmath87 * 6375 * & * 880 * & @xmath87 * 3577 * & * 750 * & @xmath87 * 1632 * & * 522 * & & @xmath87 * 5364 * & * 948 * & @xmath87 * 5835 * & * 504 * & @xmath87 * 3203 * & * 838 * & @xmath87 * 1491 * & * 093 * + @xmath74 & & 0&582 & 0&684 & 0&582 & 0&475 & & 0&710 & 0&723 & 0&603 & 0&501 & & 0&585 & 0&597 & 0&503 & 0&430 + @xmath90 & & @xmath873585&778 & @xmath876425&731 & @xmath873609&157 & @xmath871626&494 & & @xmath876064&641 & @xmath876506&061 & @xmath873656&471 & @xmath871661&245 & & @xmath875599&175 & @xmath876094&643 & @xmath873359&717 & @xmath871536&739 + + @xmath10 & & 0&305 & 0&421 & 0&308 & 0&208 & & 0&338 & 0&469 & 0&333 & 0&232 & & 0&312 & 0&330 & 0&237 & 0&177 + @xmath80 & & @xmath873616&004 & @xmath876527&378 & @xmath873651&908 & @xmath871632&635 & & @xmath873910&386 & @xmath876620&877 & @xmath873712&984 & @xmath871682&142 & & @xmath875558&880 & @xmath876135&844 & @xmath873356&883 & @xmath871529&646 + @xmath0 & & 1&436 & 1&382 & 1&435 & 1&472 & & 1&418 & 1&352 & 1&421 & 1&462 & & 1&437 & 1&429 & 1&465 & 1&483 + @xmath89 & & @xmath87 * 3577 * & * 687 * & @xmath87 * 6465 * & * 676 * & @xmath87 * 3613 * & * 537 * & @xmath87 * 1625 * & * 505 * & & @xmath87 * 3875 * & * 292 * & @xmath87 * 6562 * & * 953 * & @xmath87 * 3680 * & * 043 * & @xmath87 * 1677 * & * 062 * & & @xmath87 * 5441 * & * 337 * & @xmath87 * 6002 * & * 342 * & @xmath87 * 3273 * & * 580 * & @xmath87 * 1512 * & * 333 * + @xmath74 & & 0&574 & 0&682 & 0&577 & 0&466 & & 0&606 & 0&720 & 0&601 & 0&497 & & 0&577 & 0&595 & 0&497 & 0&422 + @xmath90 & & @xmath873678&097 & @xmath876619&216 & @xmath873714&149 & @xmath871663&022 & & @xmath873969&370 & @xmath876701&078 & @xmath873768&877 & @xmath871708&772 & & @xmath875688&562 & @xmath876278&192 & @xmath873446&235 & @xmath871567&267 + when we obtain the distribution parameters , we can then find the theoretical curve of the hazard function for the @xmath30 , @xmath91 , and @xmath92 by putting the parameters into the theoretical formula for hazard probability @xmath20 given by eqs . ( [ eq : sexp : hazard ] ) , ( [ eq : qexp : hazard ] ) , and ( [ eq : wbl : hazard ] ) for the stretched exponential distribution , @xmath0-exponential distribution , and weibull distribution , respectively . on the other hand , using eq . ( [ eq : wq ] ) we can evaluate the empirical hazard function @xmath93 , @xmath94 where the denominator @xmath95 is the number of recurrence intervals with values greater than @xmath21 , and the numerator @xmath96 the number of recurrence intervals within the range of @xmath97 $ ] . ( color online ) . plots of hazard probability @xmath38 with @xmath98 . the hazard events correspond to the extreme negative returns obtained from the quantile threshold of 99% . the analysis is performed in the period from 1885 to 1928.,width=302 ] figure [ fig : hp : ri : w ] shows a plot of the hazard probability @xmath56 as a function of the elapsing time @xmath21 for the extreme negative returns obtained from the 99% quantile threshold when @xmath99 . it shows the empirical hazard probability estimated from the real data ( filled markers ) and the analytical hazard probabilities obtained from the theoretical equations ( solid curves ) . note that although all the theoretical lines do not overlap on the same curve they all decrease with respect to the elapsing time @xmath21 , as does the empirical hazard probability . the statistics are poor and the empirical hazard probability strongly oscillates , but for a given value of @xmath21 the analytical hazard probability values are comparable to those of the empirical hazard probability , suggesting that the analytical hazard probabilities agree with the empirical hazard probability . these decreasing patterns in the hazard probability are also seen in energy futures @xcite , spot index and index futures @xcite , and stock returns @xcite , indicating that the probability of observing a follow - up extreme return decreases as time @xmath21 elapses . this reveals the existence of extreme return clustering and a potential dependent structure in the triggering processes of the extreme returns , which supports the argument that `` many extreme price movements are triggered by previous extreme movements '' and that `` larger extremes occur more often after big events or frequent events than after tranquil periods '' @xcite . this is caused by the positive herding behavior of investors and the endogenous growth of instability in financial markets @xcite . because the results are all similar , we do not show the hazard probabilities @xmath56 for different thresholds and other types of return . using hazard probabilities and an optimized hazard threshold , we build a model to predict the occurrence of positive , negative , and absolute extreme returns in financial markets within a given time period . the hazard probabilities are specified by the distribution parameter of the recurrence intervals between extreme events in the return history . the indicators of incoming extreme events are generated when the hazard probability exceeds the optimized hazard threshold , and this maximizes the usefulness of these extreme forecasts . we perform out - of - sample tests to evaluate the predictive power of this extreme - return - prediction model as follows . 1 . we mark extreme events according to a specified extreme value or quantile threshold during a given in - sample calibrating period . 2 . fitting the recurrence intervals between the marked extreme events , we estimate the stretched exponential distribution , @xmath0-exponential distribution , or weibull distribution parameters . 3 . using the estimated distribution parameters in the in - sample calibrating period , we determine the hazard probability @xmath100 and find the optimized hazard threshold @xmath39 by maximizing the usefulness @xmath54 . using the distribution parameters and optimized hazard threshold from the in - sample calibrating period , we forecast the indicators of incoming extreme events within time period @xmath22 and evaluate the forecasting signals . to find the optimized hazard threshold , we vary the hazard threshold in @xmath101 $ ] to obtain all possible pairs of @xmath102 . plotting @xmath42 with respect to @xmath41 , we obtain the well - known `` receiver operator characteristic '' ( roc ) curve @xcite . using the roc curve we measure the validity of the predicting power of early warning models . figure [ fig : roc : insample ] shows the roc predictive curves of extreme negative returns for in - sample tests and out - of - sample tests . the in - sample ( out - of - sample ) period is from 1885 to 1928 ( from 1929 to 1932 ) . the diagonal line is a random guess . note that the roc curves of the three fitting distributions are overlap exactly on the same curve for in - sample and out - of - sample tests , suggesting that the results do not depend on the distribution formula used to fit the recurrence intervals . all roc curves are above the random guess line , indicating that both in - sample and out - of - sample tests have a better predictive power than a random guess . note also that the out - of - sample curves are lower than the in - sample curves , which confirms the observation that out - of - sample predictions are usually worse than in - sample tests @xcite . because they all exhibit very similar patterns , we do not show the roc curves obtained from different thresholds and different types of extreme returns . ( color online ) . roc curves of in - sample tests and out - of - sample predictions . the extreme returns correspond to the negative returns in the quantile of 99% . the in - sample period covers from 1885 to 1928 . the out - of - sample period spans from 1929 to 1932.,width=302 ] because all three fitting distributions give the same roc curve , we evaluate only the in - sample and out - of - sample performance of the extreme return prediction model for the @xmath0-exponential distribution . we find the optimized hazard threshold , which maximizes the usefulness in the in - sample calibrating period , and estimate such performance measurements as the rate of correct predictions , the false alarm rate , the usefulness , and the kss score during in - sample and out - of - sample periods . the results are shown in table [ tb : pe : performance ] . first , we observe that all usefulness @xmath103 values are positive except in the positive returns in the 97.5% quantile in panel a and in the 95% quantile in panel b , indicating that when missing - event and false - alarm errors are weighted equally our model provides more accurate results than the benchmark of ignoring the forecasting signals . second , excluding the above two exceptions all kss scores are greater than 0 , which corresponds to random guessing , indicating that the rate of correct predictions exceeds that of false alarms . third , note that in most of the results , the @xmath103 and kss scores of in - sample performance are larger than those of out - of - sample performance , which is in consistent with the observation that out - of - sample predictions are inferior to the in - sample tests . we do find one exception in panel e and nine exceptions in panel f in which the out - of - sample predictions surpass the in - sample tests , and this indicates the predictive power of the testing model . the results also imply that the more data available for the in - sample tests , the better the performance of out - of - sample predictions , and this is further supported by the predictions during two recent turbulent periods , which were better than predictions during other periods . fourth , note that the predictions of the extreme returns in the 99% quantile produce a lower false alarm rate and a higher correct prediction rate than those in the 95% and 97.5% quantiles , and this produces high usefulness and kss scores . the results imply that the extreme events with a high quantile can be predicted more accurately . table [ tb : ri : statistics ] shows the statistics of ljung - box q tests that exhibit a decreasing pattern as quantile thresholds increase in all panels , indicating that increasing the quantile threshold could decrease the memory strength in the extremes . we neglect the potential dependence structure in the extreme series in our model because the larger the quantile threshold , the weaker the memory in extremes and the better the forecasting performance . compared to the model based on the hawkes processes @xcite our model has the advantage of fewer model parameters , easier estimating methods , and a faster prediction implementation . [email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@.l & & & & & & + & & & & & & & & & & & & & & & + + in : @xmath42 & & 0&256 & 0&319 & 0&286 & 0&228 & & 0&317 & 0&382 & 0&494 & 0&263 & & 0&159 & 0&207 & 0&242 & 0&142 + out : @xmath42 & & 0&807 & 0&814 & 0&834 & 0&895 & & 0&821 & 0&862 & 0&967 & 0&773 & & 0&719 & 0&699 & 0&805 & 0&721 + in : @xmath41 & & 0&616 & 0&623 & 0&657 & 0&626 & & 0&543 & 0&602 & 0&755 & 0&656 & & 0&637 & 0&582 & 0&713 & 0&626 + out : @xmath41 & & 0&953 & 0&949 & 0&967 & 0&980 & & 0&891 & 0&898 & 0&963 & 0&963 & & 0&960 & 0&935 & 0&975 & 0&958 + in : @xmath103 & & 0&180 & 0&152 & 0&186 & 0&199 & & 0&113 & 0&110 & 0&131 & 0&197 & & 0&239 & 0&187 & 0&235 & 0&242 + out : @xmath103 & & 0&073 & 0&068 & 0&066 & 0&042 & & 0&035 & 0&018 & @xmath870&002 & 0&095 & & 0&121 & 0&118 & 0&085 & 0&119 + in : kss & & 0&361 & 0&304 & 0&372 & 0&398 & & 0&226 & 0&220 & 0&261 & 0&393 & & 0&478 & 0&375 & 0&470 & 0&484 + out : kss & & 0&146 & 0&135 & 0&133 & 0&084 & & 0&069 & 0&036 & @xmath870&005 & 0&190 & & 0&241 & 0&236 & 0&170 & 0&238 + + in : @xmath42 & & 0&413 & 0&332 & 0&263 & 0&120 & & 0&316 & 0&337 & 0&257 & 0&191 & & 0&303 & 0&170 & 0&159 & 0&075 + out : @xmath42 & & 0&760 & 0&634 & 0&377 & 0&111 & & 0&677 & 0&676 & 0&608 & 0&361 & & 0&641 & 0&360 & 0&245 & 0&042 + in : @xmath41 & & 0&727 & 0&729 & 0&760 & 0&725 & & 0&657 & 0&664 & 0&702 & 0&757 & & 0&723 & 0&700 & 0&799 & 0&737 + out : @xmath41 & & 0&824 & 0&750 & 0&769 & 0&250 & & 0&776 & 0&782 & 0&714 & 0&692 & & 0&780 & 0&671 & 0&789 & 0&667 + in : @xmath103 & & 0&157 & 0&199 & 0&248 & 0&302 & & 0&170 & 0&163 & 0&222 & 0&283 & & 0&210 & 0&265 & 0&320 & 0&331 + out : @xmath103 & & 0&032 & 0&058 & 0&196 & 0&069 & & 0&049 & 0&053 & 0&053 & 0&166 & & 0&069 & 0&155 & 0&272 & 0&312 + in : kss & & 0&314 & 0&397 & 0&497 & 0&605 & & 0&341 & 0&327 & 0&445 & 0&566 & & 0&420 & 0&531 & 0&640 & 0&663 + out : kss & & 0&064 & 0&116 & 0&392 & 0&139 & & 0&099 & 0&106 & 0&107 & 0&331 & & 0&139 & 0&311 & 0&545 & 0&624 + + in : @xmath42 & & 0&284 & 0&342 & 0&284 & 0&193 & & 0&405 & 0&362 & 0&309 & 0&191 & & 0&309 & 0&177 & 0&167 & 0&099 + out : @xmath42 & & 0&370 & 0&512 & 0&370 & 0&291 & & 0&721 & 0&688 & 0&676 & 0&394 & & 0&546 & 0&336 & 0&229 & 0&205 + in : @xmath41 & & 0&763 & 0&717 & 0&763 & 0&780 & & 0&726 & 0&675 & 0&732 & 0&753 & & 0&703 & 0&680 & 0&785 & 0&749 + out : @xmath41 & & 0&786 & 0&667 & 0&786 & 0&750 & & 0&763 & 0&683 & 0&789 & 0&733 & & 0&645 & 0&662 & 0&828 & 0&733 + in : @xmath103 & & 0&239 & 0&187 & 0&239 & 0&294 & & 0&161 & 0&157 & 0&211 & 0&281 & & 0&197 & 0&252 & 0&309 & 0&325 + out : @xmath103 & & 0&208 & 0&077 & 0&208 & 0&230 & & 0&021 & @xmath870&003 & 0&057 & 0&169 & & 0&050 & 0&163 & 0&299 & 0&264 + in : kss & & 0&479 & 0&375 & 0&478 & 0&587 & & 0&321 & 0&313 & 0&423 & 0&562 & & 0&394 & 0&504 & 0&618 & 0&650 + out : kss & & 0&416 & 0&155 & 0&416 & 0&459 & & 0&042 & @xmath870&005 & 0&113 & 0&339 & & 0&099 & 0&325 & 0&599 & 0&529 + + in : @xmath42 & & 0&245 & 0&350 & 0&248 & 0&140 & & 0&414 & 0&370 & 0&317 & 0&185 & & 0&233 & 0&181 & 0&172 & 0&105 + out : @xmath42 & & 0&522 & 0&690 & 0&523 & 0&330 & & 0&768 & 0&702 & 0&725 & 0&461 & & 0&539 & 0&463 & 0&418 & 0&278 + in : @xmath41 & & 0&709 & 0&707 & 0&712 & 0&725 & & 0&725 & 0&669 & 0&726 & 0&734 & & 0&733 & 0&669 & 0&773 & 0&747 + out : @xmath41 & & 0&731 & 0&821 & 0&745 & 0&533 & & 0&883 & 0&820 & 0&855 & 0&769 & & 0&784 & 0&664 & 0&764 & 0&700 + in : @xmath103 & & 0&232 & 0&178 & 0&232 & 0&292 & & 0&156 & 0&149 & 0&204 & 0&275 & & 0&250 & 0&244 & 0&301 & 0&321 + out : @xmath103 & & 0&104 & 0&065 & 0&111 & 0&102 & & 0&058 & 0&059 & 0&065 & 0&154 & & 0&122 & 0&100 & 0&173 & 0&211 + in : kss & & 0&463 & 0&357 & 0&463 & 0&585 & & 0&311 & 0&298 & 0&409 & 0&549 & & 0&500 & 0&487 & 0&601 & 0&642 + out : kss & & 0&209 & 0&130 & 0&223 & 0&204 & & 0&115 & 0&118 & 0&130 & 0&308 & & 0&244 & 0&201 & 0&345 & 0&422 + + in : @xmath42 & & 0&248 & 0&351 & 0&249 & 0&144 & & 0&414 & 0&367 & 0&315 & 0&186 & & 0&236 & 0&258 & 0&172 & 0&107 + out : @xmath42 & & 0&660 & 0&711 & 0&666 & 0&333 & & 0&633 & 0&603 & 0&640 & 0&423 & & 0&557 & 0&616 & 0&471 & 0&175 + in : @xmath41 & & 0&710 & 0&708 & 0&713 & 0&715 & & 0&733 & 0&678 & 0&734 & 0&730 & & 0&736 & 0&746 & 0&774 & 0&742 + out : @xmath41 & & 0&855 & 0&902 & 0&875 & 0&900 & & 0&913 & 0&869 & 0&887 & 0&933 & & 0&892 & 0&911 & 0&863 & 0&969 + in : @xmath103 & & 0&231 & 0&179 & 0&232 & 0&286 & & 0&160 & 0&155 & 0&209 & 0&272 & & 0&250 & 0&244 & 0&301 & 0&317 + out : @xmath103 & & 0&097 & 0&095 & 0&104 & 0&283 & & 0&140 & 0&133 & 0&124 & 0&255 & & 0&168 & 0&147 & 0&196 & * 0 * & * 397 * + in : kss & & 0&462 & 0&357 & 0&463 & 0&571 & & 0&319 & 0&311 & 0&419 & 0&544 & & 0&500 & 0&489 & 0&603 & 0&634 + out : kss & & 0&195 & 0&191 & 0&209 & 0&567 & & 0&279 & 0&266 & 0&247 & 0&510 & & 0&335 & 0&294 & 0&392 & * 0 * & * 793 * + + in : @xmath42 & & 0&258 & 0&350 & 0&254 & 0&140 & & 0&326 & 0&439 & 0&312 & 0&183 & & 0&230 & 0&256 & 0&169 & 0&100 + out : @xmath42 & & 0&180 & 0&352 & 0&174 & 0&128 & & 0&229 & 0&358 & 0&229 & 0&136 & & 0&155 & 0&195 & 0&112 & 0&066 + in : @xmath41 & & 0&729 & 0&712 & 0&725 & 0&726 & & 0&747 & 0&752 & 0&737 & 0&738 & & 0&740 & 0&751 & 0&780 & 0&758 + out : @xmath41 & & 0&667 & 0&636 & 0&682 & 0&667 & & 0&682 & 0&696 & 0&682 & 0&714 & & 0&737 & 0&720 & 0&842 & 0&571 + in : @xmath103 & & 0&235 & 0&181 & 0&235 & 0&293 & & 0&211 & 0&157 & 0&213 & 0&277 & & 0&255 & 0&248 & 0&306 & 0&329 + out : @xmath103 & & * 0&243 * & 0&142 & * 0&254 * & 0&269 & & * 0&227 * & * 0&169 * & * 0&227 * & * 0&289 * & & * 0&291 * & * 0&262 * & * 0&365 * & 0&253 + in : kss & & 0&470 & 0&362 & 0&471 & 0&587 & & 0&421 & 0&313 & 0&426 & 0&554 & & 0&510 & 0&495 & 0&611 & 0&658 + out : kss & & * 0&486 * & 0&285 & * 0&508 * & 0&538 & & * 0&453 * & * 0&338 * & * 0&453 * & * 0&578 * & & * 0&582 * & * 0&525 * & * 0&730 * & 0&505 + we have performed a recurrence interval analysis of financial extremes in the djia index during the period from 1885 to 2015 . we determine the extreme returns according to a newly proposed extreme identifying approach , as well as quantile thresholds . with the extreme identifying approach we are able to locate the optimal extreme threshold associated with the minimum ks statistics of tail distributions . we find that the recurrence intervals , which are the periods of time between the successive extremes of different types of returns and thresholds , follow a @xmath0-exponential distribution . this allows us to analytically derive the hazard probability @xmath104 that within the time interval @xmath22 since the last extreme event that occurred at time @xmath21 we will observe the next extreme event . the analytical @xmath20 value agrees well with the empirical hazard probability estimated from real data . using the hazard probability , we develop an extreme - return - prediction model for forecasting imminent financial extreme events . when the hazard probability is greater than the hazard threshold , this model can warn when an extreme event is about to occur . the hazard threshold is obtained by maximizing the usefulness of extreme forecasts . both in - sample tests and out - of - sample predictions reveal that the signals generated by our prediction model are better statistically than the benchmark of neglecting these signals and that the input distribution formula used to fit the recurrence intervals has no influence on the final outcome of our early warning model . although in most cases the predictive performance of in - sample tests are better than that of out - of - sample predictions , expanding the in - sample calibrating period could yield out - of - sample predictions that are better than in - sample tests . in addition , increasing the extreme - extracting threshold could improve the predictive power of our model in both in - sample tests and out - of - sample predictions . our results may shed new light the occurrence of extremes in financial markets and on the application of recurrence interval analysis to forecasting financial extremes . z .- q.j . and acknowledge support from the national natural science foundation of china ( 71131007 and 71532009 ) , shanghai `` chen guang '' project ( 2012cg34 ) , program for changjiang scholars and innovative research team in university ( irt1028 ) , china scholarship council ( 201406745014 ) and the fundamental research funds for the central universities . g .- j.w . and c.x . acknowledge support from the national natural science foundation of china ( 71501066 , 71373072 , and 71521061 ) . a.c . acknowledges the support from brazilian agencies fapeal ( ppp 20110902 - 011 - 0025 - 0069/60030 - 733/2011 ) and cnpq ( pde 20736012014 - 6 ) . was supported by nsf ( grants cmmi 1125290 , phy 1505000 , and che- 1213217 ) and by doe contract ( de - ac07 - 05id14517 ) . 72 natexlab#1#1url # 1`#1`urlprefix babecky , j. , havrnek , t. , matj , j. , rusnk , m. , mdkov , k. , vaek , b. , 2014 . banking , debt , and currency crises in developed countries : stylized facts and early warning indicators . j. financ . stab . 15 , 117 . jiang , z .- q . , canabarro , a. a. , podobnik , b. , stanley , h. e. , zhou , w .- x . , 2016 . early warning of large volatilities based on recurrence interval analysis in chinese stock markets . to be appeared in quant . finance . jiang , z .- q . , zhou , w .- x . , sornette , d. , woodard , r. , bastiaensen , k. , cauwels , p. , 2010 . bubble diagnosis and prediction of the 2005 - 2007 and 2008 - 2009 chinese stock market bubbles . 74 , 149162 . petersen , a. m. , wang , f .- z . , havlin , s. , stanley , h. e. , 2010 . market dynamics immediately before and after financial shocks : quantifying the omori , productivity , and bath laws . e 82 , 036114 . son , i. s. , oh , k. j. , kim , t. y. , kim , d. h. , 2009 . an early warning system for global institutional investors at emerging stock markets based on machine learning forecasting . expert sys . appl . 36 , 49514957 . sornette , d. , demos , g. , zhang , q. , cauwels , p. , filimonov , v. , zhang , q .- z . , real - time prediction and post - mortem analysis of the shanghai 2015 stock market bubble and crash . j. of invest . 4 ( 7795 ) .	being able to predict the occurrence of extreme returns is important in financial risk management . using the distribution of recurrence intervals the waiting time between consecutive extremes we show that these extreme returns are predictable on the short term . examining a range of different types of returns and thresholds we find that recurrence intervals follow a @xmath0-exponential distribution , which we then use to theoretically derive the hazard probability @xmath1 introduction ------------ predicting such extreme financial events as market crashes , bank failures , and currency crises is of great importance to investors and policy markers because they destabilize the financial system and can greatly shrink asset value . much research has been carried out in an attempt to detect the underlying vulnerabilities and the common precursors to financial extremes . a number of different models have been developed to predict the occurrence of financial distresses including those using probability @xcite , signal approaches @xcite and intelligence @xcite . a faster - than - exponential increase in price accompanied by accelerating price oscillations indicates the presence of bubbles @xcite . the behavior of these bubbles can be characterized using the log - period power - law singularity ( lppls ) model , which is capable of accurately forecasting a bubble s tipping point @xcite . recent research on the occurrence of financial extremes and on the market dynamics around financial crashes has enabled us to better forecast emerging financial crises . we can understand the occurrence pattern of extremes by determining the distribution of waiting times between consecutive financial extremes ( the `` recurrence intervals '' ) and charting the memory behavior within the occurring extremes . @xcite built an early warning model of this waiting time distribution to predict the probability that extremes will occur within a given time period . following a financial crisis the financial system gradually transitions back to a stasis @xcite . this relaxation behavior following a financial market crash is similar to the aftershocks following an earthquake @xcite . @xcite indicates that a possible theoretical explanation for bursts of speculating bubbles is a positive herding behavior of traders that causes local self - excited crashes @xcite . this is in accordance with the phenomenon that extremes cluster and are interdependent . @xcite show that approximately 7685% of occurring extremes are triggered by other extremes , and they develop an early warning model that treats financial crashes as earthquakes and compute the probability that an extreme event will occur within a certain time period . here we extend the probabilistic framework for extreme returns presented in @xcite to predict extremes by using the conditional probability of an future extreme event within a fixed time frame in which type 1 and type 2 errors are balanced in current market state . the contributions of our works are in four ways . * we identify extremes by locating the threshold at the minimum ks value between the empirical and fitting distributions of the extreme values . * we classify the returns as either extreme or non - extreme by quantifying the extreme threshold , and we assume that the extremes are independent . this simplifies the modeling and reduces the computational complexity when estimating parameters but provides an adequate performance when doing out - of - sample prediction . * we define a hazard probability that is dependent on the distribution formula of recurrence intervals between extremes , and this translates the problem into finding a suitable distribution form for recurrence intervals . unlike the hawkes point process , our modeling framework is easy to implement . * instead of using a predefined threshold of hazard probability , we predict extremes when the hazard probability exceeds an optimized hazard threshold , obtained by maximizing a usefulness function that takes into account an investor s preference for either type 1 or type 2 errors . we organize the paper as follows . in section 2 we present a brief review of recurrence interval analysis and early warning models . in section 3 we provide the dataset . in section 4 we describe the model and methods . in section 5 we present the results of our recurrence interval analysis for different subperiods . in section 6 we document and discuss the performance of our out - of - sample predictions . in section 7 we present our conclusions .
the ultrarelativistic heavy ion collisions provide a means to create a new state of matter at high temperature and density as the quantum chromodynamics ( qcd ) predicts a phase transition from normal hadronic matter to quark gluon plasma ( qgp ) , a state of unconfined quarks and gluons . in recent years a considerable amount of theoretical and experimental activity is going on in this field . after the collision of two lorentz contracted nuclei , qgp is assumed to be formed and equilibrated in a very small time of the order of @xmath0 1 fm . in the usual description of the evolution of the plasma , it cools by expanding hydrodynamically till the critical temperature @xmath1 is reached at which a transition from qgp to normal hadronic matter takes place . the temperature remains fixed at @xmath1 until hadronization gets completed . this hadronization scenario corresponds to the ideal maxwell construction . however , in reality hadronization does not begin at @xmath2 due to the large nucleation barrier . recently csernai and kapusta have proposed a model for nucleation for the relativistic first order phase transition @xcite . supercooling , through nucleation of hadronic bubbles in qgp has been studied by several authors @xcite using the csernai - kapusta model of nucleation . a general outcome of these studies is that the plasma will cool according to the law @xmath3 till @xmath1 . the matter continues to cool below @xmath1 until the temperature goes down to about @xmath0 .8 @xmath1 , where bubble formation and growth becomes sufficient to reheat the system due to the release of latent heat . compared to the idealized maxwell construction the supercooling delays the transition and leads to an extra entropy production as the nucleation process allows dissipation around the hadronic bubbles . the dynamical prefactor @xcite in the nucleation rate includes quark viscosity coefficients which bring dissipative effect in the medium . in fact , the nucleation rate is limited by the ability of the dissipative processes to carry latent heat away from the bubbles s surface . however , for the dynamical evolution of the plasma the ideal hydrodynamics is used . this is not consistent as the viscosity dependent terms in hydrodynamics would also contribute to the entropy production . therefore , in this work we study the supercooling in the viscous hydrodynamics . the role of viscosity on supercooling and entropy production has been investigated in detail . the nucleation model computes the probability that a bubble of the hadronic matter appears in a system , initially in qgp phase near the critical temperature . further in the model a small baryon chemical potential is assumed . the bubble formation is assumed to take place in a homogeneous qgp phase consisting of @xmath4 and @xmath5 quarks and gluons , ignoring the role of inhomogenities such as strange quarks . it is further assumed that there is not substantial supercooling . langer s theory of nucleation gives the nucleation rate per unit volume at temperature @xmath6 as @xmath7 where @xmath8 is the change in the free energy of the system due to the formation of a critical hadronic droplet , @xmath9 is a statistical prefactor which measures the available phase volume and @xmath10 is the volume of the system . the dynamical prefactor , @xmath11 determines the exponential growth rate of critical droplets which are perturbed from their equilibrium radius @xmath12 . the coarse - grained effective field theory approximation to qcd is utilised to obtain @xmath11 and @xmath9 . the factor @xmath9 is obtained as @xmath13 where @xmath14 is the surface free energy and is determined as 50 mev/@xmath15 by lattice guage theory simulations without dynamical quarks . the correlation length @xmath16 is estimated as 0.7 fm . the dynamical prefactor @xmath11 is obtained as @xmath17 where @xmath18 is the difference in the enthalpy densities of the two phases . @xmath19 and @xmath20 are respectively , the shear and bulk viscosity coefficients . inserting @xmath9 from eq . ( [ omega ] ) and @xmath11 from eq . ( [ kappa ] ) in eq . ( [ basic ] ) we get the nucleation rate per unit volume as @xmath21 the critical radius @xmath12 is given by the laplace formula as @xmath22 where @xmath23 is the pressure of the quark / hadron phase at temperature @xmath6 and the @xmath8 as @xmath24 from the nucleation rate eq . ( [ rate ] ) , one can calculate the fraction of volume @xmath25 which has been converted from qcd plasma to hadronic gas at proper time @xmath26 . if the system cools to @xmath1 at time @xmath27 , then at some later time @xmath26 the fraction @xmath28 of space which has been converted to hadronic gas is @xmath29 v(\tau',\tau).\end{aligned}\ ] ] here @xmath30 is the volume of a bubble at time @xmath26 which had been nucleated at an earlier time @xmath31 ; this takes into account the bubble growth . the factor @xmath32 accounts for the fact that new bubbles can only be nucleated in the fraction of space not already occupied by the hadronic gas . the model for bubble growth is simply taken as @xcite @xmath33 where @xmath34 is the velocity of the bubble growth at temperature @xmath6 . by definition a critical size bubble is metastable and will not grow without a perturbation . the growth of bubbles has been studied numerically with relativistic hydrodynamics by miller and pantano @xcite . their results are consistent with the growth law @xmath35^{3/2}.\end{aligned}\ ] ] this expression is intended to apply only when @xmath36 so that the growth velocity stays below the speed of sound of a massless gas , @xmath37 . at the critical temperature , @xmath38 , @xmath39 and the rate of nucleation vanishes . the system must supercool at least @xmath0 5 % to attain a finite rate . in the evolution of the matter from qgp to hadron phase , the temperature varies with @xmath26 and its description by scaling hydrodynamics provides another equation so that @xmath25 and @xmath40 are determined from these equations at any proper time @xmath26 . the scaling viscous hydrodynamics is discussed by danielewicz and gyulassy @xcite and others . the form of the dissipative terms depends on the choice of the definition of what constitutes the local rest frame of the fluid . the landau - lifshitz definition is appropriate for describing systems with small ( or zero ) chemical potential . here we give a simple derivation of the viscous hydrodynamics . in scaling hydrodynamics the expansion takes place only along the direction of collision which we chose as z axis . the proper time @xmath26 and space - time rapidity @xmath41 are used in place of @xmath42 and @xmath43 which are defined by @xmath44 and @xmath45 in the rest frame of the fluid we take a volume element as @xmath46 , where @xmath47 is the area of the element transverse to z direction . the expansion takes place with the velocity @xmath48 , in the rest frame of the element . due to the viscous effects in longitudinal direction the heat density per unit time is given by @xcite @xmath49 after time increment @xmath50 , the volume expands by @xmath51 . the amount of work done in expansion is @xmath52 . if the energy density at @xmath26 is @xmath53 and at @xmath54 is @xmath55 then the energy conservation implies @xmath56 which leads to the scaling navier - stokes equation in the limit @xmath57 @xmath58 equation ( [ visc ] ) has been solved earlier @xcite for the qgp to hadron transition with the maxwell construction , i.e. , assuming @xmath2 for the mixed phase . to solve eq . ( [ visc ] ) we require knowledge of the equation of state and the temperature dependence of @xmath19 and @xmath20 . in this work we use the bag equation of state for qgp . the energy density , pressure and entropy densities in pure qgp and hadron phases are taken as @xmath59 @xmath60 here , @xmath61 and @xmath62 are related to the degrees of freedom operating in two phases and @xmath63 is the bag pressure . for ultrarelativistic gases , the bulk viscosity @xmath20 is usually much smaller than the shear viscosity @xmath19 @xcite . danielewicz and gyulassy @xcite give the acceptable range of @xmath19 for the applicability of the navier - stokes equation to the expansion of the plasma as @xmath64 we define @xmath65 and assume the temperature dependence of viscosity coefficient as @xmath66 with @xmath67 being replaced by @xmath28 for the hadronic phase . solving eq . ( [ visc ] ) for temperature in pure quark phase , we get @xcite @xmath68\end{aligned}\ ] ] for qgp formed at time @xmath69 while in pure hadron phase @xmath70,\end{aligned}\ ] ] where @xmath71 is the time where hadronization gets completed . the energy density in the mixed phase at a time @xmath26 can be written in terms of hadronic fraction @xmath25 as @xmath72 t^4 + b [ 1 - h(\tau)],\end{aligned}\ ] ] while the enthalpy density is @xmath73 t^4.\end{aligned}\ ] ] the temperature @xmath6 can be deduced as @xmath74 } \right)^{1/4}.\end{aligned}\ ] ] the shear and bulk viscosities are also taken to be the functions of time according to @xmath75 ( t(\tau))^3.\end{aligned}\ ] ] for the maxwell construction , i.e. , for @xmath2 in the mixed phase , solution of eq . ( [ visc ] ) is @xmath76 e^{-b/\tau}/\tau + ( a+1 ) [ 1- \tau_c/\tau e^{(b/\tau_c - b/\tau)}].\end{aligned}\ ] ] here , @xmath77 is exponential integral and @xmath78 the constants a , b , and c are evaluated for @xmath2 . the solution given by eq . ( [ consh ] ) does not account for supercooling . due to supercooling , i.e. , @xmath79 in the mixed phase , the temperature is not constant and depends on @xmath26 . equation ( [ visc ] ) then becomes equation ( [ etau ] ) and eq . ( [ frac ] ) are coupled equations , finally to be solved for @xmath25 and @xmath81 [ or @xmath40 ] . once we get @xmath25 and @xmath40 we can calculate the entropy density . by solving together eq . ( [ etau ] ) and eq . ( [ frac ] ) , we have studied the plasma evolution and calculated @xmath85the entropy production as a function of @xmath26 . the initial conditions are taken at @xmath86 as @xmath2 , @xmath87 and @xmath88 . now , @xmath25 is calculated using eq . ( [ consh ] ) with step @xmath89 in @xmath26 . with this value of @xmath28 , eq . ( [ etau ] ) is solved for @xmath81 [ or @xmath40 ] . then eq . ( [ frac ] ) is evaluated by the trapezoidal rule , thereby yielding new value of @xmath28 . using new value of @xmath28 , we solve again eq . ( [ etau ] ) to improve the value of @xmath53 . this is repeated till an accuracy of @xmath90 is obtained . then we proceed to the next step @xmath91 . we take @xmath92 and @xmath93 . the value 4.6 is taken instead of 3 to account for @xmath94 , @xmath95 and @xmath19 mesons apart from pions . we chose @xmath1 = 160 mev in this work implying @xmath96 = 219 mev . for the hadronic matter @xmath97 , @xmath98 while for the quark matter @xmath99 @xcite . the coeficient @xmath100 has been varied from 2.5 to 20 . the nucleation involves the surface free energy , correlation length , and the velocity of bubble growth , which have already been described in sec . the volume of a fluid element is @xmath101 . as @xmath102 does not change with @xmath26 in the scaling hydrodynamics , @xmath85 provides a measure of total entropy . so as to understand the role of supercooling and viscous heat generation , we have compared various scenarious ; ideal hydrodynamics ( ihd ) , ihd with supercooling , viscous hydrodynamics ( vhd ) and vhd with supercooling . we assumed that at initial time @xmath103 = 1 fm/@xmath104 , the temperature @xmath105=268 mev marked the beginning of the evolution . this corresponds to energy density @xmath53=8.51 gev/@xmath106 and entropy @xmath107=40.65 @xmath108 . these are appropriate for the relativistic heavy ion collider ( rhic ) energies . for further discussions , it is instructive to examine @xmath26 versus @xmath6 plot . figure 1 shows such a plot for vhd calculations with supercooling included . from a initial point ( @xmath109 ) , one moves on to @xmath110 where critical temperature is reached . however , the nucleation rate is zero at this point and system does not hadronize . the system keeps cooling to @xmath111 where significant nucleation rate and hadronization is reached . the system can not cool further as entropy can not decrease according to the second law of thermodynamics . at this point , the hadron fraction is around 11 - 18 @xmath112 for various cases . as the temperature increases towards @xmath1 , the hadron fraction @xmath28 increases and slowly approaches a value of one at the point @xmath113 . after this , temperature starts decreasing finally reaching the point @xmath114 at the freeze - out temperature @xmath115 . this is the general character of @xmath116 curve . if one uses the maxwell construction , one reaches directly from ( @xmath117 to @xmath113 without any change in the temperature . in figs . 2 and 3 , we show the results for @xmath6 , @xmath28 and @xmath85 as a function of @xmath26 for @xmath100=2.5 and 5 . for ihd , the critical temperature @xmath1 is reached earlier than vhd . similarly with supercooling included , @xmath118 is approached earlier in ihd than vhd . the supercooling leads to an abrupt change in entropy starting from @xmath119 onwards , i.e. , in the reheating region . we find in our calculations that @xmath85 does not increase much beyond @xmath12020 fm/@xmath104 . therefore , @xmath121 represents reasonably well the total entropy produced . as @xmath122 is evaluated at @xmath1 for all cases , the total entropy is also a measure of the life time of the system upto the hadronization stage . figure 4 shows @xmath121 as a function of @xmath100 . with vhd , entropy production increases with viscosity . supercooling leads almost to a constant shift of the curve for the vhd . in fig . 5 , the excess entropy production due to supercooling is shown as a function of viscosity . for all values of @xmath100 , excess entropy does not change much . it is almost constant for the vhd calculation . we also found that the values of @xmath123 do not lead to significant hadronization and the system continues to cool in the qgp phase . sufficient value of @xmath100 is required for the transition to the hadron phase . the bubble growth velocity coefficient ( 3@xmath104 in our calculation ) also plays an important role in the transition as setting it to zero also blocks the transition . the viscosity plays very crucial role in the transition through nucleation . however , once viscosity enters into picture , the viscous hydrodynamics is to be used for consistency . in the presence of viscosity , the initial energy density estimates from the experimental rapidity density distribution would be lower than the ones calculated with the bjorken formula . though supercooling does not lead to significant increase in entropy production , it is very much needed . for the description of the phenomena in relativistic nuclear collisions , the viscous hydrodynamics with supercooling is a more appropriate framework than the ideal hydrodynamics with or without supercooling . further theoretical work is needed to compute the viscosity with narrower bounds for the analysis of the experimental data . we have studied the entropy production both in ideal and viscous scaling hydrodynamics with and without supercooling . excess entropy produced due to supercooling in viscous hydrodynamics , weakly depends on viscosity of the plasma phase . though , in general entropy produced due to supercooling is much less than that due to viscous heat generation , the phenomenon of supercooling provides an important physical mechanism for the quark - hadron transition to occur . the viscous hydrodynamics with supercooling also leads to an increase in the lifetime of the plasma . * @xmath124 as a function of proper time @xmath26 in viscous hydrodynamics with supercooling . * fig . the @xmath124 , @xmath25 and @xmath85 as a function of @xmath26 in ihd ( dotted ) , ihd with supercooling ( short dashed ) , vhd ( long dashed ) and vhd with supercooling ( solid ) at @xmath100=2.5 . * 3 . same as fig . 2 , but with @xmath100=5.0 . 4 . @xmath121 versus @xmath100 in ihd , ihd with supercooling , vhd and vhd with supercooling . * 5 . excess entropy produced due to supercooling versus @xmath100 both with ihd and vhd .	first order quark hadron phase transition is considered in scaling hydrodynamics including the csernai - kapusta model of nucleation of hadronic bubbles leading to supercooling . the effect of viscosity on the entropy production is studied in ideal as well as in viscous hydrodynamics with and without supercooling . it is found that the excess entropy produced due to supercooling depends weakly on viscosity . * supercooling in viscous hydrodynamics for qcd phase transition * + p. shukla , s. k. gupta , a. k. mohanty + _ nuclear physics division , bhabha atomic research centre , trombay , mumbai 400 085 , india _
we begin with the basic background on the ionospheric operator , followed by that of the ionospheric correction operator . these sections expand upon the mathematical framework created and briefly outlined in morales , et . al . , @xcite . the ionospheric operator * a(@xmath2 ) is the operator which takes an unperturbed map of the sky @xmath3 and maps it to a perturbed map @xmath4 which has been distorted by the ionosphere , @xmath5 put simply , a(@xmath6 ) is a generalized coordinate change from @xmath7 to @xmath8 . ( the order of the arguments of a(@xmath6 ) here , and with other operators later , is indicative of the direction of change . ) in the regime of mwa , it is a very good approximation that the mapping of angles is approximately linear with only a small deviation @xmath9 , @xmath10 the perturbed intensity @xmath4 at @xmath8 is the summed contribution from the intensities @xmath11 at all @xmath12 where this relation holds ; ie , @xmath13 where @xmath14 represents the dirac delta function . for example , if ( only ) @xmath15 and @xmath16 are mapped to @xmath17 according to equation [ delthetadef ] , then @xmath18 . in the limit that @xmath7 is a continuous variable , this discrete sum for @xmath4 is altered to an integral , @xmath19 where @xmath20 ( notice that we ve used that the magnitude of the delta - function s argument s gradient is approximately 1 here ) . from this equation it is evident that in the limit of continuous @xmath7 , the ionospheric operator equation becomes @xmath21 so far our calculations have been confined to real space , using the variables \ { @xmath2 } . however , as is common for antenna arrays in radio astronomy , the output of the correlator for mwa will actually be the fourier transform of the real space sky intensity , @xmath22 . in addition , the final power spectrum of the sky will also be measured in this fourier transfrom space ( also known as _ visibility space _ , or _ @xmath0-space _ ) . therefore , correcting for the ionosphere in real space requires inverse fourier transforming to real space , making the correction , and then fourier transforming back to visibility space . the problem with this is that fourier transforming is computationally expensive , especially for a system operating in real time , such as the mwa s _ real time system _ @xcite . as such , correcting for the ionosphere in the @xmath0-plane would greatly reduce computation . we now study the nature of this @xmath0-plane correction . let @xmath22 represent the fourier transform of @xmath3 . define this fourier transform by @xmath23 and the corresponding inverse fourier transform by @xmath24 now define a@xmath25 to be the ionospheric operator in the @xmath26plane ; that is , the operator that maps the unperturbed map @xmath22 to the perturbed map @xmath27 , @xmath28 one way to obtain @xmath27 from @xmath22 is to inverse fourier transform @xmath22 to @xmath3 using f@xmath29 , apply the ionospheric operator a@xmath30 in real space to obtain @xmath4 , and then fourier transform @xmath4 to @xmath27 using f@xmath31 . in all , @xmath32 comparison of this to the definition of a@xmath25 shows that @xmath33 this expression may be thought of as simply a basis change of a * from \{@xmath34 } to \{@xmath35}. the three operators on the right here have all been previously given . plugging in these predetermined expressions ( equations [ aop ] , [ fourierdef ] , and [ inversefourierdef ] ) and simplifying as much as possible , we obtain @xmath36 notice that the integral over @xmath8 has been evaluated by using the delta function from the expression for @xmath37 ( see equations [ a2 ] and [ aop ] ; here again we use that the magnitude of the delta - function s argument s gradient is approximately 1 ) . one interesting characteristic of this expression is the non - local nature , by which finding the value of the perturbed intensity @xmath27 at one particular @xmath38 requires knowing the value of the pure intensity @xmath22 at other @xmath39 . this property will also appear in the ionospheric correction operator , found below . the previous section corresponds to the operator which distorts the pure data into the perturbed data , but the reverse process is what actually interests us we want to correct for the effect of the ionosphere to obtain the pure data from the perturbed data . define * a@xmath40@xmath41 to be this ionospheric correction operator which corrects for the influence of the ionosphere by mapping the perturbed map of the sky @xmath4 back to the unperturbed map @xmath3 , @xmath42 the mwa will not run during periods of scintillation ( at which times multiple values of @xmath7 are perturbed to the same @xmath8 ) , but will instead run during times when it is a very good approximation that the mapping from @xmath8 to @xmath7 is one - to - one and approximately linear with only a small correction , @xmath43 the derivation of an expression for the ionospheric correction operator in the @xmath0-plane @xmath44 follows analogously to the derivation of @xmath45 , so i ll merely quote the result : @xmath46 this is the most general expression for obtaining @xmath22 from @xmath27 ; proceeding further requires knowledge of the ionospheric perturbation @xmath47 . numerically solving for @xmath22 using this equation is an incredibly daunting task for an arbitrary choice of @xmath47 , as it involves a double integral over all space for every value of @xmath1 . therefore , unless we find a choice of @xmath47 which offers an analytic solution for @xmath22 , the correction for the ionosphere in the @xmath0-plane will actually be more computationally expensive than the two fourier transforms necessary to correct for the ionosphere in real space . unfortunately , choices of @xmath47 which lend themselves to analytic solutions are hard to come by . there are a couple , however , and they will be discussed in the following sections . the above equation for @xmath22 contains an exponential with @xmath48 in the exponent . by expanding this exponential , we obtain a form for @xmath22 which may be solved analytically for a couple of choices of @xmath48 . to be explicit , expanding the exponential in @xmath48 , @xmath49 leads to @xmath50 ( interchanging an infinite sum and an integral requires that the sum be uniformly convergent , which will be true for all choices of @xmath48 that we choose . ) with the above expansion , @xmath22 may be solved analytically if we choose @xmath51 here , the @xmath52 are chosen to be purely real , but the @xmath53 are allowed to assume complex values . physically , this choice of @xmath48 corresponds to modeling the integral along the line of sight of the ionosphere s electron density @xmath54 ( @xmath55 is the distance along the line of sight ) as a sum over sinusoidal modes , @xmath56 the reflection by the ionosphere is then related to this by @xmath57 where @xmath58 represents the two - dimensional gradient with respect to @xmath59 . ( the actual shift , of course , is a hermitian observable , so only the real part of this is included in @xmath48 . ) notice that this choice ultimately stems from our decision to model density fluctuations from the ionosphere as sinusoidal modes . actually , any orthonormal basis would have sufficed here . once again , this particular choice was made because it allows an analytic solution for the unperturbed intensity @xmath22 . ( an alternate choice which likewise offers an analytic solution will be briefly discussed later on in section [ choice2 ] . ) with this choice of @xmath48 , the intensity @xmath22 becomes @xmath60 \tilde{i}(\vec{u ' } ) .\end{aligned}\ ] ] the math leading to a solution for @xmath22 may be found in the appendix , section [ math1 ] . very briefly , the integral is solved by conveniently redefining the ionospheric modes @xmath61 ( as given below ) , performing a multinomial expansion on the term raised to the power @xmath62 , recognizing that the final product of this expansion leaves the @xmath8 integral in the form a delta function , and then using that delta function to evaluate the integral over @xmath38 . the final solution is @xmath63 where @xmath64 @xmath65 and the summation over all @xmath66 ( with @xmath67 ) is a restricted sum such that @xmath68 and @xmath69 . this equation might look a little daunting , but it may be thought of simply as the addition of many delta functions of varying amplitudes , with those further from the point @xmath1 in question tending to contribute less to the sum . ( this is , in fact , similar to what one sees with intermodulation distortion ) . notice that although there are @xmath70 modes distorting the sky , the sums and products above involve @xmath67 . thus , there appear to be @xmath71 _ effective modes _ distorting the sky . this factor of 2 comes from the constraint that @xmath48 be real , as may be more easily seen by following the math provided in the appendix , section [ math1 ] . another important feature of this solution is that it is inherently non - local , with the corrected intensity @xmath22 at a given @xmath1 depending on the values of the perturbed intensity @xmath27 at the appropriate neighboring points @xmath72 . this non - locality , which is also evident in the most general form for @xmath22 ( equation [ iupure ] ) , will create edge effects because the @xmath0-plane is finite in all practical applications , as will be more easily seen and understood later in section [ trunc10 ] . it should also be pointed out that this equation for @xmath22 is the result of a double expansion : the taylor - series expansion of the exponential containing @xmath47 ( see equation [ n1 ] ) , which is now evident in the summation over @xmath62 , and the expansion of the ionospheric perturbation @xmath47 itself into sinusoidal modes ( see equation [ dtsine ] ) , which is now evident in the restricted sum over @xmath66 . throughout this paper we will assume that this second expansion is perfect " ; that is , we will assume that we are able to perfectly model the ionosphere with the @xmath70 modes that we assume are provided for us . we will instead study the errors created by truncating the first expansion . as previously stated , the main goal of the @xmath0-plane correction is to correct for the ionosphere in a less computationally intensive manner than that required for the real space correction . our final expression for @xmath22 , however , contains an infinite sum over @xmath62 . clearly , making the @xmath0-plane correction computationally feasible will require truncating this sum after a finite number of terms . the next section will explore the effect of such a truncation . but even truncating this sum over @xmath62 can not guarantee the computational feasibility of the @xmath0-plane correction because of the second expansion over sinusodial modes and its resulting restricted summation . more specifically , the number of terms in the restricted sum over all possible combinations of @xmath66 such that @xmath68 may be calculated through the following trick : if @xmath71 represents the total number of effective modes , then consider the problem of arranging @xmath62 balls and @xmath73 partitions in a straight line . here the @xmath74-th partition marks the stopping point where @xmath75 ends and @xmath76 begins . for example , if for a particular arrangement 7 balls lie between the 4th and 5th partitions , then @xmath77 = 7 for that arrangement . the total number of ways to arrange these @xmath78 objects is @xmath79 . of course , exchanging the position of any two of the same object ( ball or partition ) does not lead to a different arrangement , so the total number of terms in the restricted sum such that @xmath80 is given by @xmath81 as an example , suppose that we wish to calculate this sum for 10 modes ( @xmath82 effective modes ) to the 40th order in @xmath62 . using the above formula , we calculate that such a sum has approximately 70 trillion terms . from this we see that the @xmath0-plane correction is only computationally feasible if the number of modes necessary to model the sky @xmath70 _ and _ the number of orders necessary in the expansion of the exponential @xmath62 are relatively small . as stated in the previous section , the @xmath0-plane correction is only computationally feasible if we truncate the infinite sum over @xmath62 . at this point we pause to study the results of truncating this sum after a finite order of correction , @xmath83 . in order to study the qualitative effects of truncating the sum over @xmath62 , we created a simple sky and perturbed it with a simple mode , and then used numerical code to correct for this ( known ) perturbation in the @xmath0-plane by using the mathematical formula found above in equation [ finaliu ] . the initial , unperturbed sky is shown in the top panel of figure [ pureandbad ] . we will refer to this sky throughout this paper as the _ simple sky_. it contains a @xmath84 pixel array with a spacing of 4 arcmin between pixels , which is the approximate resolution we expect for the final mwa array ( the axes in this and all the following real space figures are labeled in radians ) . this pure sky is a single source sky : the value of the intensity at all pixels is set to 0.0 except at one pixel where the value 1.0 . ( the important qualitative results found below would not be altered by including side lobes , so we will leave them out to keep things simpler . ) figure [ uv ] shows the real part of the intensity in the @xmath0-plane for this pure sky plotted in the third direction ( which is determined by the color scale ) and demonstrates that the uncorrupted , @xmath0-sky is a simple sinusoid ( as one would expect for the fourier - transform of a delta function ) . in contrast to this plot , the color scale for all the following @xmath0-space plots is representative of the magnitude of the intensity @xmath22 in the @xmath0-plane ( although @xmath22 is complex , the important @xmath0-plane results found below do not require phase information to understand graphically ) . the important feature to take away from this plot is that the absolute magnitude of the intensity is constant and of the order @xmath85 at all points in the @xmath0-plane ( although we ve only plotted the real part here ) . we then perturb this simple sky with a rather strong mode , as shown in the bottom panel of figure [ pureandbad ] . this distorting mode has @xmath86 radians squared and @xmath52 with a magnitude of 378.0 inverse radians and oriented in the @xmath87 direction as measured from the @xmath88 axis . ( these values were chosen for the sole reason that they produce a strong shift of a few pixels ( tens of arcminutes ) , and thus accentuate the qualitative features of the ionospheric correction as seen below . a more realistic distortion will be discussed later in section [ realistic ] . ) notice that the intensity is still 1.0 at exactly one pixel and 0.0 at every other pixel , but now the location of this pixel has slightly shifted in the @xmath87 direction . ( it should be noted that this ionospheric shift was applied to the continuous sky with a delta function at one point , and not to its pixelized represention shown in the top panel of figure [ pureandbad ] ) . although we have not included the plot , the magnitude of the intensity in the perturbed @xmath0-plane remains of the order of @xmath85 , its value in the pure visibility sky . we now begin correcting for this simple one - mode distortion using various values of @xmath89 . we ( quite naturally ) begin with the first order correction , @xmath90 ( @xmath91 leads to no correction , see equation [ finaliu ] ) . after correcting to first order in the @xmath0-plane , we inverse fourier transform back to the real space sky shown in the top panel of figure [ order1 ] in order to determine the effect that this first order correction has had on the real sky ( in particular , we would like to know whether it has successfully shifted the single source back to its unperturbed location ) . as it turns out , the correction to one order has not shifted the star from its perturbed location . the cross - like pattern of the star is somewhat interesting , but what is most important about this figure is that the maximum in the intensity has now doubled from 1.0 to 2.0 . a potential clue to this behavior is found by studying the visibility space sky corrected to first order , as shown in the bottom panel of figure [ order1 ] . from this figure we see that the first order correction in the @xmath0-plane has created an increase of an order of magnitude in the absolute value of the intensity at those points in the @xmath0-plane furthest from the origin . another important feature of this figure ( although , as it turns out , it is not the cause of the increase in the real space intensity ) is the ring around the outside edge of the figure . this ring is caused by the previously mentioned fact that the correction in the @xmath0-plane is non - local ( see equation [ finaliu ] ) , combined with the finite nature of our numerical @xmath0-plane . more specifically , points near the edge of our @xmath0-plane may not obtain the full correction in visibility space , because doing so requires pulling values of the intensity that are off the edge of the grid . therefore , values near the edge are never fully corrected . we will later see that the result of this is a small spreading of the initial source ( ie , a loss of precision ) , in real space . we now continue on to the second order correction ( @xmath92 ) . after inverse fourier transforming , we obtain the real space sky shown in figure [ order2 ] ( top panel ) . from this figure we see that the maximum in the intensity has increased even more , and is now 12 times its unperturbed and uncorrected value . the visibility sky after two orders of correction , as shown in the bottom panel of figure [ order2 ] , has a maximum in the intensity that is now 100 times the value of the pure @xmath0-sky . correcting to higher orders , we see that the problem with intensities that are too high not only persists , but continues to get worse . after 5 orders , the intensity in the real space sky ( figure [ order5 ] , top panel ) is 120 times too large , and that in the visibility space sky ( figure [ order5 ] , bottom panel ) is @xmath93 times too large ; after 10 orders , the intensity in the real space sky ( figure [ order10 ] , top panel ) is 600 times too large , and that in the visibility space sky ( figure [ order10 ] , bottom panel ) is @xmath94 times too large . in addition to the increase in the maximum in real space intensity , the source is also beginning to spread out and look less like a single point source . before continuing , let s pause to develop an intuition of what is happening here . consider a simple exponential , @xmath95 suppose that we want to approximate this exponential using a taylor expansion , @xmath96 it makes sense that a decent approximation to the original exponential should be possible by truncating this sum after a finite number of terms . but how many terms are necessary ? let s first consider the zeroth order approximation , in which only the @xmath97 term is kept , @xmath98 notice that the zeroth order approximation gives us the right magnitude of 1 , but all information about the phase has been lost . if we instead correct to first order , we obtain @xmath99 this is not even close to the correct answer not only does this not contain the correct phase , but the magnitude is now not even close to being correct . correcting to second order gets us even further from the correct answer , @xmath100 this pattern continues for higher orders as well . in fact , the approximation wo nt begin to look decent until @xmath101 . even more relevant to our observations in the previous section , notice that adding subsequent terms to the approximation does not necessarily make the approximation better until @xmath101 . before then , adding subsequent terms actually makes the approximation worse . drawing from these observations , we expect that the trend we ve seen so far is the result of under correcting in the @xmath0-plane , and that by going to more and more orders we will eventually obtain a decent correction . we now verify that this intuition is correct by studying higher order corrections . if correct , we expect to see the results gradually get better . we now consider the fifteenth order correction , @xmath89 = 15 , as shown in figure [ order15 ] . the real space sky after 15 orders ( top panel ) is now only 500 times too intense ( versus 600 for 10 orders ) , while the visibility space sky ( bottom panel ) is still approximately @xmath94 times too intense . from this it is unclear that things are getting better , but in the very least the intensities are not getting worse . the shape of the source , however , continues to grow further from a point source . moving on to @xmath102 ( figure [ order20 ] ) is a bit more reassuring . the maximum in the real space sky intensity is now approximately only 50 times its actual value ( top panel ) , although the maximum in the visibility space intensity is still four orders too high ( bottom panel ) . the gradual improvement continues when we skip ahead to 25 orders ( figure [ order25 ] ) . the real space intensity is now only 4 or 5 times too large ( top panel ) , while the visibility space intensity has now dropped to 1000 times too large ( bottom panel ) . the shape of the source , however , continues to grow worse . skipping ahead next to 30 orders of correction shows a dramatic improvement . the top panel of figure [ order30 ] shows that the single source now appears to be a single source of the right order of magnitude in intensity . and in addition , the most intense pixel is now located exactly where it was for the pure sky , so the @xmath0-plane correction has ( at least in terms of location of the max ) successfully corrected for the shift by the ionosphere . the bottom panel of figure [ order30 ] shows that the maximum in the visibility space sky intensity ( which , as always , occurs near the edge of the grid ) is now only an order of magnitude too big . it appears as if we ve gone over the hump , and are now on our way to decent results . the correction to 35 orders shows minor improvement in real space ( figure [ order35 ] , top panel ) . in the visibility space sky ( figure [ order35 ] , bottom panel ) , however , the entire grid now has the correct order of magnitude of @xmath85 , including the most extreme pixels . we therefore now see some of the finer patterns caused by the non - locality of the correction and finite nature of the @xmath0 grid ( as mentioned previously in section [ trunc10 ] ) , which had previously been hidden by the extreme intensities at the corners . it should be noted that the most intense pixel in this fully corrected real space sky in the top panel of figure [ order35 ] still lies at the location of the single source in the original , pure sky . in other words , the @xmath0-plane correction has successfully shifted the reflected source back to its initial position , at the cost of minor spreading over a few neighboring pixels . this spreading , which can not be eliminated by correcting to still higher orders , is caused by the finite nature of the @xmath0-plane and thus can not be avoided . it turns out that corrections to higher orders show negligible improvement over the correction to 35 orders , so the resulting skies , identical to those of figure [ order35 ] , are not shown . so far we have only used a particularly simple sky with one star . figure [ 10stars ] shows ( the absolute value of ) the residual between a more complicated pure sky with 10 stars , and the real space sky obtained after perturbing this pure sky and then correcting in the @xmath0-plane to 40 orders . for comparison s sake the perturbation used here was the same ionospheric mode used above to perturb the simple sky of figure [ pureandbad ] ( ie , that used throughout this section ) . this figure shows the kinds of errors we may expect from the @xmath0-plane correction . the residual from the star on the top right shows a light cross pattern , indicative of a spreading of the source caused by the process of perturbing the star and then applying the @xmath0-plane correction . places in the plot with two consecutive pixels of high intensity represent stars which were not shifted back to exactly the same pixel that they started at , but rather to a neighboring pixel . recall that the stars are typically initially shifted 3 or 4 pixels by the ionosphere , so the @xmath0-plane correction is still providing some improvement with these stars . there are two main points to be taken from our analysis so far . first , from section [ numterms ] we learned that either needing too many modes to model the ionospheric correction or too many orders of correction leads to an unreasonable number of numerical calculations . second , from section [ nmax ] we learned that under correcting in the @xmath0-plane is a huge mistake and a lot worse than not correcting at all . hence the dilemma : choosing @xmath89 too small leads to the destruction of the data , while choosing @xmath89 too large leads to a computationally infeasible problem . it is therefore advantageous to develop a theoretical prediction of how many orders of correction are necessary . as it turns out , the result will lead to a few tricks which make the problem more reasonable . if we correct to only @xmath89 orders , then the magnitude error @xmath104 in our result must be the absolute value of the sum of all the terms we left out ; more specifically , @xmath105 the steps leading to an upper bound @xmath103 on this error may be found in the appendix , section [ math2 ] . the result is @xmath106 where @xmath107 @xmath108 and @xmath70 is the number of modes , as always . [ see equations [ c_qdef ] and [ d_qdef ] for reminders on how effective modes @xmath109 are related to actual modes @xmath110 . this is our final result for a strict upper bound on the error . unfortunately , this formula is not too enlightening . in order to obtain a theoretical estimate for @xmath89 , we must make a few further approximations . as one would expect , the optimal value of @xmath89 , which represents the number of orders necessary to obtain some level of accuracy in the @xmath0-plane , is dependent upon the level of accuracy desired . to quantify this , define @xmath111 to be the value of the intensity in the @xmath0-plane after correcting up through @xmath112 , and @xmath113 to be the value of the intensity in the @xmath0-plane that one would obtain by employing the full correction and not truncating the sum ( ie , @xmath114 ) . ( it should be noted that @xmath113 here also assumes a @xmath0-plane infinite in extent . this will have effects seen later . ) the fractional error @xmath115 in the @xmath0-plane correction caused by truncating the sum is then @xmath116 in the appendix ( section [ math3 ] ) you will find the steps leading up to a theoretical prediction of the value of @xmath89 at a given @xmath1 in the @xmath0-plane necessary to obtain a fractional error less than or equal to @xmath117 if given the ionospheric effective modes distorting the sky , @xmath109 . the result is that the optimal value of @xmath89 is estimated by @xmath118 where @xmath70 is the number of modes and @xmath119 is defined as it was above in equation [ vu ] . this formula is a little hard to digest , so some values for @xmath89 given @xmath120 and @xmath117 are provided in table [ nmaxtab ] . for the example sky and perturbation used in section [ nmax ] , the theoretical predictions for the number of orders necessary is shown graphically in the top panel of figure [ nmaxtheory ] . this figure has @xmath121 , although @xmath89 does not change too significantly when varying @xmath117 , as is seen in table [ nmaxtab ] . recall that the single distorting mode @xmath122 is in the @xmath87 direction , which defines the favored direction seen in this figure . as a check of these theoretical predictions , we used matlab to numerically compute the number of orders necessary to obtain the desired fractional error of @xmath123 . the bottom panel of figure [ nmaxtheory ] shows the difference between these computational results and the theoretical predictions shown in the top panel of figure [ nmaxtheory ] . more specifically , it represents the number of orders of correction theoretically predicted minus the number found computationally . this figure suggests that for a bulk of the @xmath0-plane , the theoretical prediction is quite accurate , predicting the number of orders to within 5 . near the extremes , however , the finite nature of the @xmath0-plane causes problems ( remember that the theoretical estimate assumed an infinite @xmath0-plane ) . in fact , the cool - colored pixels near the corners are pixels which never obtained a fractional error of @xmath124 ( the numerical code cutoff after 50 orders ; all points with fractional errors too high at that point were assigned a value of 50 orders ) . the theory predicts that about 35 orders are required to correct at the most extreme points in the @xmath0-plane , which is what our previous numerical computations found . an important feature of the theoretical predictions shown in figure [ nmaxtheory ] that is characteristic of all skies is that the necessary number of orders of correction varies with @xmath1 , and in particular it increases as @xmath1 increases along the direction of the mode . therefore , points closer to the origin are corrected in less orders than those further away . [ nmaxtab ] the accuracy of the theoretical prediction here is in no small part due to the existence of only one ionospheric mode in our simple sky model . this reason for this is that the above theoretical estimate ( equation [ finalnmax ] ) was derived from an expression for the upperbound on the error ( equation [ upper ] ) which assumes that all the ionospheric modes in the sky are as strong as the strongest mode at @xmath1 and add constructively ( which may be seen in the appendix , section [ math2 ] , near equation [ uu ] ) . as such , the result is not a bad prediction for only one distorting mode , but tends to ( perhaps significantly ) overestimate the necessary number of orders for multiple distorting modes . in short , the above mentioned theoretical estimate may perhaps be more accurately called a theoretical _ _ over__estimate . given the results seen in section [ nmax ] ( more specifically , the terrible consequences of undercorrecting in the @xmath0-plane ) , this was done intentionally to ensure that our @xmath0-plane was adequately corrected . still , it may be useful to obtain a more accurate estimate of the number of orders necessary . one such estimate would be a _ strongest mode approximation _ , in which we assume that at any given @xmath1 , the only significant contribution comes from the strongest mode at that point . the contributions from the other modes are assumed to be weak and negligible . this approximation ultimately boils down to setting @xmath125 in the final equation determining @xmath89 from the previous section ( equation [ finalnmax ] ) . this approximation may provide a more accurate estimate of @xmath89 , but it also runs a high risk of underestimating the correct number of orders , which should be avoided if possible . an alternate approximation would be a _ significant modes approximation _ , in which only modes at a given @xmath1 with strengths within a certain critical fraction of the strongest mode s are included in the value for @xmath70 used in equation [ finalnmax ] . with a closer study of perturbations from more realistic ionospheric modes , it may be possible to set this critical fraction in such a way as to fairly accurately predict the number of orders necessary . the above analysis suggests two methods for making the @xmath0-plane correction less time intensive : 1 ) the points furthest out in the @xmath0-plane take the most time to correct . eliminating them decreases computation time , but at the cost of resolution in the real space sky . 2 ) different points in the @xmath0-plane require different numbers of orders of correction , so write a code that corrects to different numbers of orders depending on the point in the @xmath0-plane . ( in other words , do nt waste time correcting to 35 orders near the origin when 2 is enough . ) for this method , we eliminate the problems caused by under correcting at the extremes in the @xmath0-plane by setting the values at those extremes to 0 . take , as a visual example , figure [ uvshave ] , in which we have set the values of the pixels in the 25 diagonal rows from the corner to zero ( we _ shaved _ 25 pixels from the corner ) . as a reminder of the real space sky after 25 orders without edge shaving , consider the top panel of figure [ corner25 ] ( a reproduction of figure [ order25 ] , top panel ) . notice that the intensity is approximately 4 or 5 times too high at the brightest points and , even worse , our single point source has turned into some sort of supernova explosion . compare this to the edge - shaved version of the real space sky ( figure [ corner25 ] , bottom panel ) , in which we see a sky that looks almost identical to our fully corrected sky after 35 orders ( top panel , figure [ order35 ] ) . to see this more clearly , consider figure [ residshave ] , which represents ( the absolute value of ) the residual between the real space sky corrected to 40 orders with no edge shaving and the real space sky corrected to 25 orders with edge shaving . from this figure we see that the result of the edge shaving was to create a small spread around the star , but of an intensity about an order of magnitude lower than the maximum intensity of the star . the relative success of this scheme leads to the question of how low we may push the number of correction orders when edge shaving is introduced . figure [ realshave15 ] shows the result of only correcting to 15 orders , but shaving 75 rows of pixels from the corners in the @xmath0-plane . without edge shaving , the real space sky corrected to 15 orders had a maximum intensity of about 500 ( figure [ order15 ] , top panel ) . now , the total intensity is approximately 1 as it should be , but it is spread over a number of pixels . so while the results are a dramatic improvement over what they had been , for the sake of precision it might be a good idea to correct to higher orders and shave less . the moral : the process works , but be careful about trying to shave too much . for this method , we correct to different numbers of orders at different @xmath1 . to test this method , we re - wrote our matlab code so that the number of orders of correction at a given @xmath1 was determined by the theoretical estimate from section [ theorynmax ] ( more specifically , equation [ finalnmax ] ) . we then corrected the distortion for the same simple sky used in section [ nmax ] . figure [ residmeth2 ] shows the residual between the real space sky corrected to 40 orders at all points in the @xmath0-plane and the real space sky corrected to different orders in the @xmath0-plane . not surprisingly , the residual is incredibly small 5 or 6 orders of magnitude less than the maximum in the intensity of the source . ( this is , of course , another sign that the theoretical prediction of the number of orders of correction is pretty good ) . however , this new matlab code presented a small problem : matlab is so much better at manipulating matrices than running for - loops that this second code , which theoretically requires less computation , takes approximately 15 times as long to run . of course , if the @xmath0-plane correction is eventually used in mwa , a programming language more adept at loops will undoubtfully be used , and this method will potentially save time . previously , we have assumed that @xmath126 took the form @xmath127 given the relative complexity of the results above , it pays to investigate an alternate choice of @xmath126 . the main reason that this was chosen was because it allowed for an analytic solution for @xmath22 , where @xmath128 as shown previously in equation [ iupure ] . without an analytic solution to these integrals for @xmath22 , the @xmath0-plane correction becomes more computationally intensive than inverse fourier transforming to real space , correcting for the ionosphere there , and then fourier transforming back to visibility space . therefore , an analytic solution is required for any choice suitable choice of @xmath126 . unfortunately , choices for @xmath126 which allow such analytic solutions are hard to find . there is , however , at least one other such distortion : a polynomial expansion , given by @xmath129 where @xmath130 and @xmath70 are the number of terms in the @xmath88 and @xmath131 directions , respectively , necessary to accurately model the distortion by the ionosphere . the @xmath132 and @xmath133 shown here are real . the analytic derivation of @xmath22 is given in the appendix , section [ math4 ] . the result is @xmath134 where @xmath135 @xmath136 and the sums over @xmath137 and @xmath138 are restricted so that @xmath139 and @xmath140 . while we ve confined the analytic solution of this to the appendix , it should be mentioned that en route to this solution the exponential in @xmath126 from equation [ second ] was taylor - expanded ( which , you may recall , was also the case for the other choice of @xmath126 as a sum over sinusoidal modes , and is explicitly shown in equations [ n1 ] and [ n2 ] ) . in other words , this solution is likewise characterized by the double expansion mentioned previously at the end of section [ sinmodes ] : one expansion over @xmath141 from expanding the exponential in @xmath126 , and one expansion in @xmath142 resulting from our model for the ionosphere . as such , this solution shows many of the unfortunate characteristics of the sinusoidal choice . in particular , there are still restricted sums which contain a number of terms comparable to that calculated in section [ numterms ] , so this choice has the same problem of making the @xmath0-plane correction unreasonable if too many orders of correction or ionospheric modes are needed . in addition , the correction is still not strictly local : numerical computation of derivatives requires neighboring pixels , with higher orders requiring more neighbors . moreover , numerical computation of derivatives for a finite data set introduces its own set of additional errors , and thus makes this choice much less appealing than the previous one with sinusoidal modes . the ugliness of both of these choices ultimately stems from the inability to solve for @xmath22 analytically without expanding the exponential containing @xmath126 . unless a choice is found which may be solved analytically without this first expansion , it is doubtful that a better choice than the sinusoidal modes will be found . the ionospheric distortion presented throughout this paper was chosen because its particularly strong nature accentuated the subtleties of the @xmath0-plane correction . the strength of this mode made the @xmath0-plane correction appear computationally infeasible : any ionosphere which required even 10 of these modes to accurately model would require too much computation ( see section [ numterms ] ) . however , an ionospheric mode which shifts sources on the sky by tens of arcmin is somewhat unrealistic . we conclude by considering a more realistic distortion . the computational feasibility of the @xmath0-plane correction is determined by the largest value of @xmath89 ( given by equation [ finalnmax ] ) required for any @xmath1 . to calculate this for a realistic sky , we need to know the largest possible value of @xmath119 . it is possible to cast the largest value of @xmath119 , which we label @xmath143 , in a form which better elucidates its physical significance . more specifically , notice that @xmath144 where @xmath145 is the angle between @xmath1 and @xmath122 . for an arbitrary choice of @xmath1 , this cosine term may be significant . however , if we wish to calculate @xmath143 , we may set @xmath146 and @xmath147 = @xmath148 , where @xmath148 is the greatest distance from the origin in the @xmath0-plane that our antenna s @xmath0-coverage allows . this is valid for mwa because the @xmath0-coverage is approximately circular , so that the strongest ionospheric distorting mode is guaranteed to lie along a direction which possesses this maximum displacement in the @xmath0-plane . with these changes , and based upon our previous definitions of the effective modes @xmath149 in terms of the actual modes @xmath150 ( see equations [ c_qdef ] and [ d_qdef ] ) , we may write @xmath143 as @xmath151 but , @xmath152 let s define @xmath153 in words , @xmath154 is the maximum deflection caused by a single mode that we might observe . in addition , let @xmath155 denote the length of our antenna array s longest baseline . the maximum @xmath0-plane displacement @xmath148 is then this length measured in units of the wavelength @xmath156 that our antenna is detecting , @xmath157 [ the extra factor of @xmath158 is the result of our convention for fourier transforms ( see equation [ fourierdef ] ) , which differs from that conventionally used in radio astronomy ] . with these substitutions , @xmath159 therefore , in terms of these parameters , the number of orders of correction necessary is ( adapted from equation [ finalnmax ] ) @xmath160 where @xmath117 is ( as before ) the fractional error desired for the correction . it should be noted that this form is only valid for determining the largest @xmath89 among all @xmath1 . for calculating @xmath89 for a particular @xmath1 , equation [ finalnmax ] must be used . for mwa , a typical frequency detected will be about 140 mhz , corresponding to @xmath161 meters ( this represents the 21 cm emission for a red shift of @xmath162 . ) we expect the ionosphere to deflect such a wave approximately 0.6 arcmin = @xmath163 radians ( @xcite , value is for the night ) . if we consider baselines of approximately 400 meters , then @xmath164 if given the number of modes necessary to accuately model the ionosphere @xmath70 ( which is as of yet undetermined ) , then @xmath89 for the full array may be determined from table [ nmaxtab ] by substituting @xmath165 for @xmath166 . as an example , if a fractional error of @xmath121 is desired and @xmath167 , then @xmath168 and table [ nmaxtab ] shows that 12 orders of correction are necessary . whether such a result is computationally feasible is dependent upon how much time is alloted for the ionospheric correction and the quality of the computers used . as such , it may not be determined here . what is clear , however , is that such a correction is not obviously ruled out on computational grounds ( especially if a technique such as edge shaving is used to reduce @xmath89 ) . [ quick aside : edge shaving alters the above results by substituting the largest @xmath147 left unshaved in place of @xmath148 in the above calculations . ] it should be noted that the values that went into calculating @xmath165 above were estimates , and certainly not set in stone . in particular , we once again emphasise that throughout this paper we have remained ignorant of the details involved in the expansion of the ionosphere , and have no knowledge of how many modes @xmath70 are necessary to sufficiently model the effect of the ionosphere . in addition , the ionospheric deflection @xmath169 is proportional to @xmath170 , with longer wavelengths experiencing greater shifts @xcite . therefore , we expect longer wavelengths than the above to require more orders of correction and shorter wavelengths , fewer . if @xmath165 is in fact lower by a factor of 5 , for example , then @xmath171 and the @xmath0-plane correction is certainly a viable candidate for correcting the ionosphere . in particular , if the strongest mode approximation discussed in section [ strongest ] turns out to be a good approximation , then even with baselines of 1.5 km we may expect a good correction after only 4 orders for the wavelength given above . on the other hand , if @xmath165 is raised by a factor of 5 , then @xmath172 and the @xmath0-plane correction is clearly computationally infeasible for any reasonable value of @xmath70 . as stated previously , correcting for the ionosphere in the @xmath0-plane entails multiplying the perturbed data @xmath173 by the ionospheric correction operator a@xmath40@xmath174 , @xmath175 where we have now made the time dependence of these quantities explicit . the above may be thought of as a matrix equation , where @xmath176 and @xmath177 represent our uncorrected and corrected ( respectively ) data arrays , and @xmath178 represents a correction matrix . the entries of this correction matrix are calculated by the appropriate binning of the coefficients in our correction equation ( reproduced from [ finaliu ] ) , @xmath179 where the coefficients " are the quantities preceeding @xmath180 on the right hand side of the equation . if the timescale within which one wishes to recalculate the effect of the ionosphere is small compared to the timescale within which the ionosphere significantly changes , then it is possible that the correction matrix a@xmath40*@xmath181 has changed very little from that previously calculated . more specifically , if one wishes to calculate the correction matrix at a time @xmath182 shortly after having calculated it at time @xmath183 ( ie , if @xmath184 where @xmath185 is the time scale of significant change in the ionosphere ) , then @xmath186 where @xmath187 represents a small correction matrix . in this regime , it is computationally much easier to calculate the small correction @xmath187 and add it to the previously calculated @xmath188 then to calculate @xmath189 from scratch . therefore , in such a scenario the process of updating the correction matrix is computationally favorable . the first order correction @xmath187 may be analytically calculated as follows : let s assume that we model the ionosphere using the effective modes @xmath190 that we have been throughout this paper ( see equations [ c_qdef ] and [ d_qdef ] ) , and that the @xmath122 represent part of a fixed fourier basis while the @xmath191 are our fitting parameters . ( in other words , the @xmath122 are fixed and time independent , while the @xmath191 fluctuate with time ) . let s assume that we ve calculated the correction at a time @xmath183 . more specifically , assume that for all @xmath1 we ve calculated and stored all the relevent terms in the correction equation for @xmath192 , @xmath193 where the time dependence of @xmath194 , @xmath195 , and @xmath191 is now explicit . now we want to correct for the ionosphere at a later time @xmath182 . to compose the correction matrix at time @xmath182 , we could start with the full correction formula , @xmath196 and then construct the new matrix @xmath189 . instead , however , let s assume that we re in the regime of small ionospheric changes , so that to first order @xmath197 where @xmath198 is small . substituting this into the full correction and only keeping terms to first order we obtain @xmath199 in this final equation , the coefficients in this first group of terms exactly replicate those coefficients from time @xmath183 ( see equation [ t_0 ] ) . in other words , these terms represent the previously determined correction matrix @xmath188 . the second group represents the small adjustment @xmath187 to the correction matrix at @xmath183 . notice that these terms have been there own maximum cutoff for @xmath62 , labeled as @xmath200 in the above equation . if the quantity @xmath201 is small , as assumed , then the individual terms in this second sum are also small , and thus a smaller value of @xmath200 is necessary to obtain a desired fractional error for the intensity . in this case , it is computationally favorable to update the correction matrix rather than derive it from scratch . it should also be noted that if @xmath202 , then the nonzero entries of the matrix @xmath187 form a subset of the nonzero entries of the matrix @xmath188 , and thus their sum ( which represents @xmath189 ) is equally as sparse as @xmath188 . in other words , the matrix @xmath203 does not become less sparse through this process of correction ( a fact which is important for large numerical matrix manipulations ) . it is worth mentioning that whether updating the correction matrix is a viable method depends on the time scales of the changing ionosphere . more specifically , the above was calculated keeping only terms to first order in @xmath201 . it is possible that for the time scales considered higher terms are also necessary , or that @xmath198 is not small compared to @xmath204 ; the former situation complicates the math but does not necessarily outrule this method , while the latter pretty much requires that the correction matrix be built from scratch every time . the @xmath0-plane correction only makes computational sense if the model for the ionospheric perturbation allows for an analytic solution to @xmath22 ( equation [ iupure ] ) . one such model is a sum over sinusoidal modes ( equation [ dtsine ] ) . by running numerical codes with this choice , the most important result discovered was that under correcting in the @xmath0-plane is worse than not correcting at all ( section [ nmax ] ) . but in addition to this , correcting to too many orders or requiring too many modes to model the effect of the ionosphere may lead to a computationally unreasonable problem ( section [ numterms ] ) . to help avoid this issue , a theoretical estimate of the number of orders of correction necessary ( which agrees well with the sample sky provided in this paper ) may be used ( section [ theorynmax ] ) . this estimate reveals that the number of orders of correction necessary varies in the @xmath0-plane . this , however , suggests two methods for alleviating the problem : eliminating those points in the @xmath0-plane which are particularly troublesome at the cost of precision for the real space sky ( section [ meth1 ] ) and correcting to different orders at different points in the @xmath0-plane ( section [ meth2 ] ) . both techniques prove successful and make the problem of correcting in the @xmath0-plane more feasible . in addition , depending on how often the ionosphere s effect is updated compared to the timescales of change in the ionosphere , it may be compuationally favorable to update the previously determined effect of the ionosphere rather than rederive its full effect from scratch each time . the authors would like to thank matias zaldarriaga for helpful conversations . the purpose of this appendix is to rigorously derive some of the mathematical formulas merely stated within the main text . it is included for completeness and for the curious reader ; no new results are derived . in this section , we solve for the unperturbed intensity @xmath22 , @xmath205 using a sum over sinusoidal modes for our ionospheric deflection @xmath48 , @xmath206 with a bit of algebra ( and keeping in mind that @xmath52 and @xmath8 are real but @xmath53 is complex ) , @xmath48 may be written as @xmath207 where @xmath208 is the complex conjugate of @xmath53 . with this choice of @xmath48 , the expression for @xmath22 becomes @xmath209 before proceeding , it is convenient to convert the summation over @xmath210 as follows : @xmath211 where @xmath212 and @xmath213 with this form , we see that although there are @xmath70 modes distorting the sky , there are @xmath71 terms in the sum . this extra factor of 2 comes from the above constraint that @xmath48 be real . we will refer to these modes labeled by ( @xmath214 ) as _ effective modes_. writing the intensity @xmath22 in terms of effective modes gives @xmath215 the individual terms inside the summation over @xmath62 may be manipulated using the multinomial expansion to give @xmath216 where @xmath217 denotes a restricted sum such that @xmath68 and @xmath69 . the expression for @xmath22 now becomes @xmath218 because of the above performed multinomial expansion , the integral over @xmath8 ( once brought inside the summation ) now takes on the familiar form of a delta - function , and is easily performed to yield @xmath219 the integral over @xmath38 is now a simple delta - function integral , and its integration gives @xmath220 in this section , we estimate the error in the @xmath0-plane accumulated by truncating the infinite sum over @xmath62 after the term @xmath221 . recall that the total correction term is @xmath222 when correcting through @xmath83 , the magnitude error @xmath104 in the correction is equal to the absolute value of the sum over all terms left out . more specifically , @xmath223 we now attempt to determine an upper bound @xmath103 on this error . to begin , we bring the absolute value inside the sum , so that all terms now add constructively , @xmath224 we expect the magnitude of @xmath22 will be approximately the same at all points in the @xmath0-plane . denoting the maximum value of @xmath225 for the uncorrected sky as @xmath226 , we find @xmath227 next , consider the terms in the sum of the form @xmath228 . define @xmath119 to be the maximum value of @xmath229 for a given value of @xmath1 and the effective modes in question . then , @xmath230 where the equality in the equation above occurs because the restricted sum over @xmath66 requires that @xmath231 . with this substitution , the upper bound on our error function becomes @xmath232 now concentrate on the inner summation . define @xmath233 according to @xmath234 i assert that @xmath235 [ aside : before providing the proof of this , we should point out that the fraction @xmath236 is not guaranteed to be an integer , and therefore this factorial and the ones given hereafter should be taken to be given by the gamma - function @xmath237 , @xmath238\ ] ] the proof is quite short : \i ) start with the given form of @xmath233 , corresponding to @xmath239 , for any @xmath240 . \ii ) any value of @xmath241 which corresponds to an alternative choice for the @xmath66 under the constraint that @xmath231 may be obtained by multiplying this value for @xmath233 by a finite number of factors whose magnitudes are all less than 1 . therefore , @xmath233 is the maximum . an example may be quite useful here : consider the scenario with @xmath242 and @xmath243 . in this case , we assert that @xmath244 which corresponds to the choice @xmath245 for all four @xmath66 . now , let s pick an alternative choice for the @xmath66 ; let s say @xmath246 . for this choice we obtain , @xmath247 but this may be re - written as @xmath248 where the first factor transforms @xmath249 from ( 3,3 ) to ( 2,4 ) and the second transforms @xmath250 from ( 3,3 ) to ( 5,1 ) . therefore @xmath251 . our new knowledge of @xmath233 , when combined with our previous determination of the number of terms in the restricted sum over @xmath66 ( see section [ numterms ] ) , leads us to conclude that @xmath252 therefore , our new upper bound @xmath103 on the error @xmath104 becomes @xmath253 this is our final result for a strict upper bound on the total error . notice that this final step is equivalent to assuming that the contributions from all modes are as strong as the strongest , and add constructively . as such , @xmath103 is clearly an upperbound on the error . for one mode , this last step does not lead to that great of an overestimate . with the addition of more modes , however , this step overemphasises the contribution from weaker modes , and leads to a ( potentially much ) larger overestimate of the error . the goal of this section is to determine the value of @xmath89 necessary to obtain a precision in the @xmath0-plane equal to @xmath117 if given @xmath1 and the ionospheric modes distorting the sky . based upon the terrible consequences which result from undercorrecting in the @xmath0-plane ( see section [ nmax ] ) , we begin with the expression just derived for the upperbound on the error in hopes of avoiding this pitfall . in order to make sense of our expression for an upper bound @xmath103 ( equation [ uu ] ) and derive from it the optimal choice of @xmath89 , we must make a few further approximations . some of these approximations will actually slightly decrease the expression for the error relative to @xmath103 , but are necessary in order to make sense of this ugly expression . to begin , we define @xmath254 so that @xmath255 next , we calculate the ratio @xmath256 and find that @xmath257 where we have employed stirling s approximation , @xmath258 stirling s approximation is best suited for large @xmath62 , but is actually quite accurate for small @xmath62 as well . it gives an answer within 8% of the actual value for @xmath259 , and within 1% for @xmath260 . in other words , by using this approximation we greatly simplify our expression and sacrifice only a little in terms of accuracy . using the fact that @xmath261 we see that for large @xmath62 this ratio reduces to @xmath262 this expression shows that at a critical value of @xmath62 , namely , @xmath263 this ratio is approximately equal to 1 , and @xmath264 . for @xmath265 , @xmath266 ; and for @xmath267 , @xmath268 . in other words , @xmath269 is an increasing function of @xmath62 until @xmath270 , and then decreases from then on . strictly speaking , these results are only valid for large @xmath271 . however , in order to get an approximate expression for @xmath89 , we now extend these results to all @xmath62 . the hope is that the approximate value of @xmath89 is varied only slightly by this extension to small @xmath62 . but even if this approximation causes large enough error to raise doubts about our quantitative results for @xmath89 , it should still be good enough to learn something about the qualitative behavior of @xmath89 . recall that setting @xmath91 is the same thing as not correcting in the @xmath0-plane ( see equation [ finaliu ] ) ; @xmath272 put differently , the @xmath273 term in the sum is of the order of the uncorrected @xmath0-plane , @xmath274 . according to the above , successive corrections @xmath269 differ in magnitude from the previous term by a factor of @xmath275 . therefore , the approximate magnitude of the term @xmath269 is @xmath276 we expect the distortions created by the sky to alter the magnitude of the intensity only very slightly , so that @xmath277 . furthermore , for @xmath278 ( which is the case for @xmath89 ) we approximate that the ratio @xmath256 falls quick enough that we may approximate the total remaining error as being enirely due to @xmath269 , @xmath279 , where @xmath280 is the intensity in the @xmath0-plane after being corrected to @xmath62 orders . therefore , in order to obtain an fractional error @xmath115 less than @xmath117 for our @xmath0-plane correction , we must correct to enough orders @xmath62 so that @xmath281 therefore the optimal value of @xmath89 is given by @xmath282 some values for @xmath89 given @xmath283 and @xmath117 are given in the table embedded within the main text , table [ nmaxtab ] . a bulk of this paper has assumed that @xmath126 takes the form where @xmath130 and @xmath70 are the number of terms in the @xmath88 and @xmath131 directions , respectively , necessary to accurately model the distortion by the ionosphere . the @xmath132 and @xmath133 here are real . similar to last time , this choice is chosen because it allows an analytic solution to @xmath22 . with this choice , our earlier equation for @xmath22 , @xmath284 becomes @xmath285 where @xmath286 first , we focus on evaluating @xmath287 . to do this , we first taylor - expand the second exponential , @xmath288 where the sum over @xmath137 is a restricted sum such that @xmath289 . plugging this into our expression for @xmath287 gives @xmath290 where we have assumed that the summations and the integrals may be freely interchanged . to solve this integral , we first introduce an additional parameter @xmath156 ( which we will eventually set to 1 ) and notice that @xmath291 therefore , if we define @xmath292 , then @xmath293 but , @xmath294 thus , @xmath295 plugging this into our earlier expression for @xmath22 , we obtain @xmath296 where @xmath297 but @xmath298 here is of the same form as @xmath287 earlier , and therefore @xmath299 where @xmath135 @xmath136 and the sums over @xmath137 and @xmath138 are restricted so that @xmath139 and @xmath140 . mitchell , d.a . ; greenhill , l.j . ; wayth , r.b . ; sault , r.j . ; lonsdale , c.j . ; cappallo , r.j . ; morales , m.f . ; ord , s.m . , `` real - time calibration of the murchison widefield array , '' selected topics in signal processing , ieee journal of , vol.2 , no.5 , pp.707 - 717 , oct . 2008 url : http://ieeexplore.ieee.org / stamp / stamp.jsp?arnumber= \linebreak 4703504&isnumber=4703300[http://ieeexplore.ieee.org / stamp / stamp.jsp?arnumber= \linebreak 4703504&isnumber=4703300 ]	as is common for antenna arrays in radio astronomy , the output of the mwa s correlator is the intensity measured in visibility space . in addition , the final power spectrum will be created in visibility space . as such , correcting for the ionosphere in visibility space instead of real space saves the computation required to inverse fourier transform to real space and then fourier transform back ( a significant decrease in computation for systems operating in real time such as the mwa . ) in this paper , we explore this problem of correcting for ionospheric distortions in the @xmath0-plane . the mathematical formula for obtaining the unperturbed data from that reflected by the ionosphere is non - local , which in any practical application creates edge effects because of the finite nature of the @xmath0-plane ( section [ ic ] ) . in addition , obtaining an analytic solution for the unperturbed intensity is quite difficult , and can only be done using very specific expansions of ionospheric perturbations . we choose one of these models ( with perturbations as sinusoidal modes , section [ sinmodes ] ) and run numerical codes to further study the correction . numerically implementing this correction to too few orders distorts the data in such a way as to be worse than not correcting at all ( section [ nmax ] ) . it is therefore critical to correct to the correct number of orders , and we present an analytic estimate for the optimal order ( section [ theorynmax ] ) . this analytic estimate shows that the optimal number of orders varies with @xmath1 , and in particular increases as @xmath1 increases along the direction of an ionospheric distorting mode . based on this observation , we then investigate a couple of methods which save computation ( section [ feasible ] ) . these methods are ( a ) eliminating the intensity at values of @xmath1 which require too many orders , and ( b ) correcting to different orders at different @xmath1 . both methods prove successful , although the first creates a loss of some precision in the real space sky . we conclude by considering an alternate form with which to model ionospheric perturbations ( section [ choice2 ] ) . this alternate form was once again chosen because it lends itself to an analytic solution , but contains as many ( if not more ) downfalls than the original choice .
the electronic fpi , discussed in some detail first by chamon et al . @xcite , was realized in a form of a hall bar perturbed along the current flow by two constrictions ; each formed by a biased split - gate quantum point contact ( qpc ) , with a plunger gate that controlled the area of the fpi ( shown in fig . current flows chirally in the form of distinct edge channels , with the lower right - moving channel biased ( at voltage ) relative to the upper left - moving one ( kept at [ fig : setup ] ground potential ) . when the bulk is in a qhe state , the two - terminal conductance , in the absence of the constrictions , is quantized at @xmath0 , with @xmath1 being the filling factor , where @xmath2 is the areal electron density , @xmath3 the flux quantum , @xmath4 the electron charge , @xmath5 the planck constant , and @xmath6 the magnetic field . each constriction induces backscattering from a forward propagating edge channel to a backward moving one . circulating edge currents in the fpi acquire a phase determined by the ab effect ( proportional to the enclosed magnetic flux ) and the number of enclosed quasiparticles . the ab phase can be controlled by either changing the magnetic field or the area of the interferometer ( via biasing the plunger gate ) . our interferometers were fabricated on heterostructures embedding 2deg with a few different areal electron densities , @xmath75.6@xmath810@xmath9@xmath10 , 8.6@xmath810@xmath9@xmath10 and 2.0@xmath810@xmath11@xmath10 , all having low temperature mobility in excess of 10@xmath810@xmath12@xmath13v - s . measurements were done on two different sizes fpis , one with an area @xmath144.4@xmath15m@xmath16 and one with an area @xmath1417.6@xmath17m@xmath16 , with plunger gate width @xmath18 . here we present data mainly taken with the smaller fpi ( unless otherwise specified ) , fabricated on the wafer with electron density 8.6@xmath810@xmath9@xmath10 ( see fig . samples were cooled in a dilution refrigerator to an electron temperature @xmath1410mk ( verified by temperature sensitive shot noise measurements ) with conductance measurements taken at a typical excitation voltage of about 1@xmath15vp - p at 800khz . the parameters of the system are as follows : the area of the interferometer , @xmath19 ; the magnetic field , @xmath6 ; the qpcs split - gate voltages , and ; the plunger gate voltage , ; the bias on the source , ; the electron temperature , @xmath20 ; and in some devices the voltage applied to a top - gate , . the cell of the fpi , is defined by depleting the electrons underneath the metallic gates that surrounds it while assuring that the two constrictions stay open . then , the magnetic field is tuned to the approximate center of a conductance plateau ; with the filling factor in the fpi cell similar to that in the bulk ( verified by multi - terminal hall measurements across the interferometer ) . the constrictions are then being carefully pinched ( via the split - gate voltages , & ) , resulting in transmissions of the interfering edge channel @xmath21 and @xmath22 , respectively . we denote the number of fully reflected edge channels , belonging to the highest landau levels , being also the number of circulating edge channels within the fpi cell , by , and the number of fully transmitted edge channels , belonging to the lowest landau levels , by . together with the interfering channel the total number of channels is = + 1 + ; being determined by the bulk filling factor @xmath23 ( for iqhe , = @xmath23 ) . we notate each configuration as ( @xmath23 , ) @xmath24(bulk filling factor , number of fully transmitted edge channels ) . the two terminal differential conductance @xmath25 was measured as function of various parameters , @xmath26 , from which the periodicities of the ab oscillations were extracted , @xmath27 ( @xmath28 ) and @xmath29 . we selected to present here only a subset of our results , which , however , portray a rather complete picture of the behavior of the fpi . in the iqhe regime , interference was measured for all possible configurations ( @xmath23 , ) , with = @xmath30 and @xmath31 through 6 and at different settings of transmissions @xmath21 and @xmath22 . as expected , for @xmath21 , @xmath32 ( weak coupling to the leads ) multiple reflected trajectories within the fpi interfered , and the two terminal conductance peaked sharply as function of flux ; while at high transmissions @xmath21 and @xmath22 , the conductance was almost sinusoidal as function of flux . however , and this is important to note , the periodicities of the interference were independent of the transmissions . we start with @xmath33 , namely , a single propagating edge channel ( the spin resolved first landau level ) , with an oscillating conductance as function of , with periodicity @xmath342.35mv ( inset in fig . [ fig : b_indep ] remarkably , the conductance in this case ( and universally for all cases of = 0 ) was found to be independent of magnetic field for many added flux quanta ( the magnetic field corresponding to one flux quantum threading 4.4@xmath15m@xmath16 is @xmath149.8gauss ) . so here we had an extraordinary situation where ab oscillations were responsive to a change in area but not to a change in the threading magnetic field . in contrast , profound oscillations as function of both @xmath6 and were always observed for @xmath350 , with a typical behavior for case @xmath36 shown in fig . 3 . the periodicity with was found to depend only on ( in both , the iqhe and fqhe regimes ) , increasing when increased . unexpectedly , the slope of the constant phase lines in the @xmath6- plane is counter intuitive : _ an increase _ in @xmath6 _ must be accompanied by an increase _ in in order to keep the phase constant ( as was also observed in ref . this suggests that an increase in @xmath6 leads to a decrease of the enclosed flux - an opposite behavior to that observed before in the electronic mzi @xcite . an exception to this behavior , observed in our large fpi , will be discussed separately . in the example in fig . 3a [ case @xmath36 ] , the dependence of the transmission on @xmath6 and on was sinusoidal for large @xmath21 and @xmath22 ; becoming an array of periodic sharp peaks when the constrictions were pinched ( fig . 3b ) , however , the periodicities did not change . [ fig : t_indep ] similar conductance oscillations were observed as function of a voltage applied to an added small top - gate ( @xmath38 ) in the center of the fpi cell , with area @xmath140.75@xmath15m@xmath16 and circumference @xmath143@xmath15 m ( see fig . again , the periodicities in for @xmath33 and @xmath39 , are identical - as summarized in fig . 4b . for @xmath35 - 135mv , the area under the top - gate is being only partly depleted , and the periodicity observed is @[email protected] . a trivial two - plate capacitor model , with @xmath410.66ff calculated for our specific device , suggests an expulsion of one electron in a single period . for @xmath42 - 135mv , the 2deg under the top - gate is fully depleted and the depletion region extends away from the top - gate . the capacitance of the top - gate to the 2deg slowly increased with gate voltage , starting at a periodicity of 0.23mv and increasing to @xmath140.6mv . the periodicity measured with a plunger gate for the = 0 cases if about 2.3mv . dividing it by the ratio between the top - gate length and its own ( @xmath140.4@xmath15m@xmath16 ) gives about 0.30mv . this is consistent with @xmath40 measured at beyond depletion voltages . a rather interesting behavior is observed in the - plane when looking at the slope of the constant phase lines in fig . 4c ; both gates , plunger and top , lowered the area when biased negatively . we address this issue later . [ fig : antidot ] we now discuss the fqhe regime , where interference was measured for @xmath23=4/3 , @xmath23=2/3 , @xmath23=2/5 and @xmath23=1/3 . due to space limitation we discuss here only the latter two cases . like in the iqhe , no interference oscillations were observed as function of @xmath6 for cases with = 0 , namely , at @xmath43 and @xmath39 . however , periodic oscillations of the conductance , as function of both @xmath6 and , were observed at @xmath44 as shown in fig . [ fig : one_third]a . while the quality of the data deteriorated somewhat , a fast fourier transform of the bare data shows a single , reproducible , distinct peak , corresponding to @xmath45=9.2gauss and @xmath40=1.13mv ( see fig . [ fig : one_third]b ) . we return to the analysis of this periodicity in the discussion section . [ fig : one_third ] the results discussed so far can be encapsulated as follows : 1 . in each filling factor , any of the edge channels can be made to interfere independently of the others . the periodicities @xmath45 and @xmath40 do not directly depend on the magnetic field ( or the filling factor ) nor the transmission coefficients @xmath21 and @xmath22 , but only on . the periodicity in magnetic field is solely determined by a and : @xmath46 as predicted in ref . @xcite for the case @xmath21 , @xmath32 ; here , however , we found it to hold for all values of the transmissions . similarly , the periodicity in plunger gate voltage depends only on : @xmath47 with @xmath40 monotonically increasing with increasing ( aside from two exceptions to be discussed below ) . 6 summarizes the measured periodicities @xmath45 and @xmath40 in cases of iqhe and a single case of fqhe ; 3 . for = 0 , the enclosed flux is independent of the magnetic field , indicating that the relevant area of the interferometer is proportional to @xmath48 ; 4 . for @xmath350 , the enclosed flux decreases with increased magnetic field , indicating that the interferometer area shrinks faster than @xmath48 ( hence , the minus sign in eq . [ eq : b_periodicity ] ) . [ fig : summary ] the first , and most important , exception to these general statements is the periodicity @xmath40 in the case @xmath44 : it is @xmath49 of the measured value for all other configurations with = 1 . the second exception takes place for high filling factors , when a large number of channels are being trapped ; such as @xmath50 . as seen in the example @xmath51 in fig . 7 , a ` lattice - like ' structure in the @xmath6- plane appears instead of the more ubiquitous ` pyjama - like ' stripes . the dependence on @xmath6 and is still periodic , but it can not be described by a periodic function of a linear combination of @xmath6 and . this behavior is not yet understood . [ fig : exotic ] the empirical eqs . [ eq : b_periodicity ] and [ eq : vg_periodicity ] may be understood by means of a single physical assumption : the relevant area of the interferometer varies as the magnetic field is varied , namely , the magnetic flux through fpi area is @xmath52 , with a depending on and @xmath6 . then , the phase evolves as function of @xmath6 and : @xmath53 with the lines of constant phase ( constant flux ) in the @xmath6- plane given by : @xmath54 in earlier experiments with an electronic two - path mzi , we found that @xmath55 , with the lines of constant phase having an opposite slope from the one measured with the fpi @xcite . assume that all fully transmitted channels enclose a similar area @xmath56 . this area is expected to be field independent since the channels are in equilibrium with the leads . the inner channels are trapped and thus maintain a discrete and constant number of electrons @xmath57 for small changes in @xmath6 , while a smooth electron exchange can take place between the interfering channel and the fully transmitted ones . the total amount of charge in the interferometer cell is @xmath58 $ ] . if charge neutrality is a major consideration then @xmath59 should be independent of @xmath6 , namely , @xmath60 . assuming @xmath61 , @xmath62 implying that the flux , and hence the ab phase , indeed _ decreased _ with increasing magnetic field at a rate proportional to , and were independent of @xmath6 for = 0 ; as observed in both the integer and fractional cases . for = 0 , increasing @xmath6 is followed by a decrease in the area @xmath19 in such a way that the flux , and hence the number of occupied states , are kept constant , but the charge density at the center of the fpi cell grows . a spatial imbalance between electrons and ionized donors takes place . increasing the field further must lead eventually to relaxation of that imbalance by creation of a hole within the interfering landau level if this is also the only edge channel , or electron tunneling from the trapped channels otherwise . in any case , this leads to abrupt increase of one quantum flux to the interfering loop ( hence , invisible in the interference pattern ) . the area ` breathes ' with magnetic field . it decreases monotonically and increases abruptly while keeping the average constant . the same holds for @xmath350 . the only difference is that the area decreases with increasing @xmath6 even faster , since the interfering channel loses charge monotonically to the lower filled landau levels ( whose density increases with field ) leading to the positive slope of the constant phase lines in the @xmath6- plane . the periodicity in gate voltage @xmath40 , and in particular its relation to the fractional filling factor @xmath23 , provide an insight to the charge of the interfering quasiparticle . the dependence on can be understood if we assume that the capacitance @xmath63 between the plunger gate and the interfering channel depends only on the number of fully transmitted channels ( that screen the plunger gate voltage ) , namely , @xmath64 . the interfering channel flows at the interface between two areas with different filling factors , @xmath65 and @xmath66 , each with charge @xmath67 per flux quantum @xmath68 . as long as the sole function of biasing the plunger gate by @xmath69 is to move the interface between the two filling factors by area @xmath70 , and thus expel charge @xmath71 from within the interferometer , we may write : @xmath72 assuming that the change in area does not change the number of quasiparticles enclosed by the interfering loop ( as their number is very small near the center of the conductance plateau ) , the change of area leads to a change in the ab phase , @xmath73 using eq . [ eq : capacitance ] we get a relation between the interfering quasiparticle charge @xmath74 and the periodicity @xmath40 : @xmath75 with @xmath76 the gate voltage needed to expel one electron . look first at cases @xmath39 and @xmath43 ; both with a gate voltage periodicity @[email protected] ( which is the similar value as for = 0 in the integer cases ) . evidently , with the assumption @xmath64 , with @xmath63 being independent of the filling factor , the measured expelled charge per period must be in both cases @xmath4 . since the interfering edge channel in both cases belongs to the @xmath49 fractional state , @xmath77 , and the interfering quasiparticle charge , deduced directly from eq . [ eq : charge ] , must be @xmath78 ; agreeing with our expectations . what happens in the case @xmath44 , where the interfering channel belongs to the @xmath79 fractional state ? the observed periodicity is @[email protected] ( see fig . 5 ) - nearly @xmath49 of the period observed in the integer cases @xmath80 . since the capacitance @xmath81 was found to be independent of filling factor , and the measured periodicity is a third of that observed in the integer regime , the expelled charge per period of the oscillation must be @xmath82 ( directly proportional to @xmath40 ) . again , for @xmath83 in this case , and the above periodicity , the interfering charge is @xmath84 . this is a striking example of an expelled charge @xmath82 per period of gate voltage , while the interfering quasiparticles carried charge @xmath85 . since the periodicity of the oscillations is independent of @xmath21 and @xmath22 , one can invoke a simpler explanation @xcite for the expelled charge in each period , which is valid only in the coulomb blockaded limit , namely , when @xmath21 & @xmath32 . in that limit , the charge enclosed by the constrictions must be quantized in units of the elementary charge in the incompressible regions within the two constrictions . for cases @xmath39 and @xmath43 , the filling factor within the constrictions is zero and the expelled charge is that of the electron ; as indeed was observed . alternatively , for @xmath44 , the filling factor in the constrictions is @xmath49 and the expelled charge must be @xmath82 ; again , agreeing with the observation . [ fig : top_gate ] the significance of coulomb interactions in the fpi was tested by adding a grounded metal gate that fully covered the fpi s area ( ` large top - gate ' , or ltg ) ; as was also done in ref . this gate effectively increased the screening of the uncompensated charges . while the addition of the ltg to the smaller fpi ( area 4.4@xmath15m@xmath16 ) did not affect the orientation of the constant phase lines ( it did increase the periodicity in plunger gate voltage , which is expected as the plunger gate is less effective due to screening provided by the ltg biasing ) , adding a ltg to the larger fpi ( area 17.6@xmath15m@xmath16 , as seen in fig . 8a ) led to markedly different results . first , unlike all our previous results , ab oscillations were measured as function of magnetic field also for cases with = 0 , such as @xmath86 & @xmath33 . second , as seen in fig . 8b for the case @xmath87 , the orientation of the constant phase lines in the @xmath6- plane was reversed , indicating an increased flux with increasing magnetic field ( see also @xcite ) . however , the area determined from the ab periodicities in @xmath6 still did not match the expected area . our experiments provide a detailed first step study towards the realization of interferometers that might give direct evidence of the statistics of fractionally charged quasiparticles . our studies covered interference of the different edge channels of complex edge structures in integer and a few fractional states in the qhe regime . we found that the interferometers are very strongly affected by coulomb interactions , which radically modify the interferometer area with magnetic field - independent of the coupling strength to the leads ( in or away from the coulomb blockade regime ) . interfering charges of @xmath88 , @xmath82 and @xmath85 where deduced from the interference patterns . we are grateful to b. i. halperin , c. marcus , i. neder , b. rosenow , a. yacoby and y. m. zhang for helpful discussions . we acknowledge the partial support of the israeli science foundation ( isf ) , the minerva foundation , the german israeli foundation ( gif ) , the german israeli project cooperation ( dip ) , the european research council under the european community s seventh framework program ( fp7/2007 - 2013)/erc grant agreement # 227716 , the us - israel bi - national science foundation , and microsoft s station q. n. ofek acknowledges support from the israeli ministry of science and technology . f. e. camino , w. zhou , v. j. goldman ( 2007 ) , _ quantum transport in electron fabry - perot interferometer _ , b 76:155305 ; d. m. godfrey et al . ( 2007 ) _ aharonov - bohm - like oscillations in quantum hall corrals _ , arxiv:0708.2448 . f. e. camino , w. zhou , v. j. goldman ( 2005 ) , _ aharonov - bohm superperiod in a laughlin quasiparticle interferometer _ rev . letts . 95 , 246802 ( 2005 ) ; f. e. camino , w. zhou , v. j. goldman ( 2005 ) , _ realization of a laughlin quasiparticle interferometer : observation of fractional statistics _ , phys . b 72:075342 .	interference of edge channels is expected to be a prominent tool for studying statistics of charged quasiparticles in the quantum hall effect ( qhe ) @xcite . we present here a detailed study of an electronic fabry - perot interferometer ( fpi ) operating in the qhe regime @xcite , with the phase of the interfering quasiparticles controlled by the aharonov - bohm ( ab ) effect . our main finding is that coulomb interactions among the electrons dominate the interference , even in a relatively large area fpi , leading to a strong dependence of the area enclosed by the interference loop on the magnetic field . in particular , for a composite edge structure , with a few independent edge channels propagating along the edge , interference of the outmost edge channel ( belonging to the lowest landau level ) was _ insensitive _ to magnetic field ; suggesting a constant enclosed flux . however , when any of the inner edge channels interfered , the enclosed flux _ decreased _ when the magnetic field _ increased_. by intentionally varying the enclosed area with a biased metallic gate and observing the periodicity of the interference pattern , charges e ( for integer filling factors ) and e/3 ( for a fractional filling factor ) were found to be expelled from the fpi . moreover , these observations provided also a novel way of detecting the charge of the interfering quasiparticles . = 1 a considerable amount of work has been focused in recent years on interference of quantum particles , aiming at understanding ` single - particle ' as well as ` correlated - particles ' physics . experiments were designed to measure the quantum coherence time @xcite ; to determine phase of the scattering amplitudes @xcite ; to test entanglement between a pair of quantum particles @xcite ; and , more recently , to probe their charge and coherence properties integer and fractional qhe regimes ( iqhe & fqhe , respectively ) @xcite . however , even though the statistics of identical quantum particles upon exchange is at the core of many physical phenomena , experiments to directly observe this intrinsic property are scarce @xcite . recent interest in topological quantum computation @xcite led to a series of theoretical proposals aimed at demonstrating the statistics of quasiparticles in the fqhe @xcite . the ideas are primarily based on performing interference in an electronic version of a fabry - perot interferometer ( fpi ) @xcite or in a similar mach - zehnder interferometer ( mzi ) @xcite ; with distinct ` fingerprints ' of the interfering quasiparticles . the mzi , while being more difficult to fabricate , is a ` true two - path ' interferometer . on the other hand , in the fpi , which is a simpler device ; many trajectories contribute to the interference . understanding the physics governing these interferometers is crucial if they are to be used to probe the statistics of quasiparticles . we performed a detailed study of an electronic fpi operating in the iqhe and the fqhe regimes . different from previous works , our devices allowed the study of interference of any of the propagating edge channels ; at integer and fractional filling factors ; for a few areal carrier densities ; the dependence on magnetic field ; as function of the fpi area ; and at different coupling strengths to the reservoir leads . previous works @xcite reported on more restricted phenomena ( mostly on interference of the innermost channel ) , hence , missing some of the more striking effects we report here . moreover , our measurements in the fqhe regime are strikingly different from those of goldman s group @xcite , which have all along been difficult to interpret @xcite . our data aims at attaining a rather complete picture of the behavior the fpi in the qhe regime , with a particular focus on the role of electron - electron interaction that seem to dominate the interference .
in order to understand the superconductivity mechanism for high-@xmath0 superconductors ( htsc ) , many experiments have attempted to find out possibles correlations between the superconductivity temperatures @xmath0 and some physical parameter . some of this experiments deal with the interplay among @xmath0 , hole contents and pressure effects which have revealed interesting features like the maximum attained @xmath0@xcite and have attracted a lot of attention as it is summarized by some review articles@xcite . on the other hand , neither one of the several microscopic theoretical models have been able to deal successfully with all these observed basic properties and therefore the nature of the condensate pair has not yet been established . as a consequence , some phenomenological approaches have been advanced in an effort to account for the large experimental data on htsc . for instance , the @xmath0 dependence on the number of carriers per @xmath2 follows the so - called universal inverse parabolic curve , i.e. , @xmath3 $ ] where @xmath4 is the optimum carrier density @xmath5 and @xmath6 is an adjustable parameter which depends on the type of compound . recently we have introduced an intermediate approach@xcite which is neither pure microscopic nor pure phenomenological : it starts with the microscopic extended hubbard hamiltonian ( t - u - v ) on a square lattice of side @xmath7 @xmath8 where @xmath9 is the nearest neighbor hopping integral , u is the on - site repulsion and v a phenomenological attractive interaction . several authors have argued that this hamiltonian accounts for the physics of the charge carriers@xcite in htsc . the strength of the phenomenological potential coupling constants are determined _ a posteriori _ in comparison with the experimental results for @xmath1 . the hopping integral @xmath9 is obtained by comparing with band structure calculations@xcite . using the bcs method and the fourier transform of the potential we have derived zero and finite temperature gap equations which are integrated up to region where pair formation has a non - negligible probability . in this way we have been able to relate @xmath0 , hole content @xmath5 and the strength of the zero temperature gap @xmath10 as described in detail in ref.@xcite . as concerns the phase diagram , its optimal doping is very sensitive if we move v from its original nearest neighbor position for the @xmath11-channel but does not change its optimal position for @xmath12-wave . a study of these and others channels as well as their complex mixtures and interplay as function of the position of v will be published elsewhere@xcite . this method was successfully applied to the experimental measurements of the curves @xmath1 for the @xmath13 and @xmath14 compounds@xcite . on the other hand if we study the typical parameter which characterizes the strength of the attractive interaction for a square well in 2d with range @xmath15 and strength@xcite v , @xmath16 , we see that for the above two compounds and hg1201 and , it varies between 10 - 3000 and such high values indicate that the size of the bound states ( or cooper pairs ) is related to the position of the minimum and the range of the attractive potential . this is the main motivation to change the position of the attractive potential v from the usual nearest neighbor and it is very interesting to find out that the curves for @xmath1 yield a different @xmath11-wave optimal doping value depending on the size of v while it remains fixed for @xmath12-wave . another interesting point from the theoretical point of view is that it provides a natural interpretation for the order parameter expansion introduced by the spin - fluctuation theory@xcite . in this approach it has been proposed a @xmath12-wave gap of the form @xmath17 where @xmath18 is expanded in powers of @xmath19 . analising this expansion , one can easily verify that it contains terms like @xmath20 which are proportional to @xmath21 and what can be seen as a type of d - wave gap symmetry that arises from a potential like @xmath22 . by the same token , we can find terms proportional to @xmath23 which can be originated by a potential @xmath24 and so on . based on this result , we adopt below a notation @xmath25 , with @xmath26 . as one expects , the more distant is the position of v , that is the larger is @xmath27 , smaller is the value of the density necessary to start the superconductivity process of pair formation . in this way we have derived @xmath1 curves@xcite that agree well with the experimental data when v was placed at @xmath28 ( s - wave ) and @xmath29 ( d - wave ) neighbors for @xmath13 and @xmath14 compounds respectively which by the way is very close the ratios of their measured coherence length @xmath30@xcite . thus the method provides a very interesting and direct interpretation for the measured values of the coherence length for the htsc compounds besides the fact that it gives a hint for the superconducting fundamental interaction in htsc . we have also extended this approach to study the pressure effects@xcite on htsc . we have noticed that @xmath31 is very sensitive to changes in the zero temperature gap amplitude @xmath10 which is used as an adjustable parameter . consequently we have introduced the assumption that the pressure induces changes on the attractive potential end ultimately on @xmath10@xcite . this assumption together with the well documented fact that the applied pressure induces a charge transfer from the reservoir layers to the @xmath2 planes , has enabled us to derive an expansion for @xmath32 in power of the pressure . this expansion was applied to the hg1201 , 1212 and 1232 compounds and has reproduced the measured values of @xmath0 for a large set of hole content and applied pressures up to 45 gpa@xcite . we should stress that since the principal interaction mechanism which forms cooper pairs has yet to be identified , the connection between @xmath0 , hole content , pressure effects and a microscopic hamiltonian should provide some clues for this interaction and it is a efficient method to obtain information on the microscopic mechanism for htsc . thus our study provides a guide to the strength and range of this fundamental mechanism . in this paper we shall clarify and illustrate the main physical ideas of the method mentioned above and to show that it is general and can be applied to other htsc than the hg - based compounds , as we apply it to the experimental phase diagram and pressure data on bi2212 . the @xmath33 family has been studied by several groups and in particular , we shall discuss firstly the measurements of the dependence of the critical temperature @xmath0 on the hole content @xmath5@xcite to set the pertinent parameters to be used afterwards on the study of the variation of @xmath0 under applied external pressures@xcite . the is strongly anisotropic , its resistivity on the ab plane is @xmath34 larger than its similar along the c - axis@xcite which justify the use of a 2d model and has one of the shortest coherence length@xcite among all cuprates superconductors and this property will be taken into account to set the parameter @xmath27 in our calculations . following the bcs method@xcite we use a many - body wave function which describes the formation of pairs of charge carriers @xmath35 where @xmath36 is the empty band state and @xmath37 . in connection with the hamiltonian of eq . ( [ hamiltonian ] ) , and the variational approach@xcite , we obtain @xmath38 where @xmath39 is the fourier transform of the potential of eq . ( [ hamiltonian ] ) , and following along the lines of ref.@xcite may be written in a `` separable '' form , @xmath40 . a similar approach which takes next - nearest neighbor hopping into account was developed by angilella et al@xcite with the same purpose to study pressure effects on compounds but they only deal with nearest neighbor attractions . with the use of the expressions for @xmath41 and @xmath42 , we get an equation for the zero temperature gap , @xmath43 where @xmath44 , thus since the gap has the same functional form of the potential , namely , @xmath45 , where the plus sign is for the s - wave and the minus sign is for d - wave channel . the chemical potential @xmath46 yields the density @xmath5 that must be calculated self - consistently@xcite but as it concerns the superconducting phase boundary it suffices to approximate it by the @xmath47k value of the maximum energy ( concentration dependent ) in the tight - binding band . in order to study pair formation , we calculate the probability of finding a hole pair , that is , the condensation amplitude @xmath48 . it is a general result@xcite that it has a maximum at @xmath49 and drops very rapidly for @xmath50 . according to these considerations and taking @xmath51 in eq.([gap ] ) , we obtain for s - wave @xmath52 where @xmath53 and @xmath54 and @xmath55 . @xmath56 and @xmath57 and @xmath58 are the maximum @xmath47 occupied values ( like a fermi momentum ) that depend on the density of holes @xmath5 . the parameter @xmath27 is used to set the position of the attractive potential v , thus * @xmath59 for third nearest neighbor and so on. we use @xmath60 and @xmath61 . the integrations are performed up to @xmath62 and @xmath63 , which are determined at values where the condensation amplitude becomes very small , namely , @xmath64 . this is usually attained for @xmath65 . for @xmath66 , the excitations with their respective probability must be taken into account . the derivation of a self - consistently temperature - dependent gap equation is analogous to that which leads to eq . ( [ gap0 ] ) . at this point , we again follow the bcs approach@xcite and assume that @xmath67 vanishes at the critical temperature @xmath0 , which yields the following equation for s - wave symmetry , @xmath68 where we again integrate up to @xmath62 and @xmath63 now , if we had integrated eqs . ( [ gap0 ] and [ gapt ] ) over the whole brillouin zone and solve them self - consistently , we would get @xmath69 . instead of this procedure , we find , for a given density @xmath5 , the value of @xmath0 which is compatible , i.e. , make the equality @xmath70 for a given @xmath71 and @xmath27 . this is the basic procedure of our method . the same approach can be carried out for d - wave symmetry but in this case we take @xmath72 in eqs.([gap0 ] and [ gapt ] ) . thus , with only one given value for the parameter @xmath27 and another one for @xmath71 , we can find the value of @xmath0 for a given density of carriers @xmath5 and therefore obtain the curve @xmath1 . now we turn our attention to the study of the htsc under pressure . one of the effects of the pressure which is generally accepted and well documented in certain materials is an increase of the carrier concentration on the @xmath2 planes transferred from the reservoir layers@xcite . such pressure induced charge transfer ( pict ) has been confirmed by hall effect and thermoelectric power measurements on several compounds@xcite . therefore this effect combined with an assumption of an intrinsic variation of @xmath0 ( linear on the pressure ) independently of the charge transfer was largely explored to account for the quantitative relation between @xmath0 and the pressure p and it gave origin to many models@xcite . however , the various pressure data taken with the mercury family of compounds@xcite for underdoped and overdoped compounds of @xmath73 ( hg1201 ) has some features that could not be interpreted@xcite by these pure charge transfer models by a single set of parameters . on the other hand we could explain the hg - family data using a single set of parameter and this is an achievement of our method described below . the study of @xmath74 which is necessary to find the appropriated values of @xmath27 and @xmath71 ( at @xmath75 ) demonstrated the direct proportionality between @xmath71 and @xmath0 which varies with the hole content @xmath5 and it is maximum at the optimal doping . taking this result into account and the discussion of the above paragraph , we are led to propose that the effects of pressure are two - fold : ( i)- the well accepted pict and an additional assumption ; ( ii)- a change in the attractive potential v as proposed previously by angilella et al@xcite which also implies on a change in the zero temperature gap @xmath71 . the assumption `` ii '' came from the analysis of the curves @xmath1 which have revealed that @xmath0 is proportional to @xmath71 but with a large constant of proportionality around the optimal doping@xcite @xmath4 and smaller near the both underdoped and overdoped extremes of the @xmath1 curve . the quantitative changes on v due to the applied pressure were estimated in ref.@xcite for the using the measured values@xcite of the compressibility tensor under the linear approximation and the self - consistently equations . thus the pict ( i ) implies that @xmath76 and the assumption of a pressure dependent gap ( ii ) implies that @xmath77 and both equations lead to @xmath78 . therefore to estimate @xmath0 for a given compound with a nominal value of @xmath5 and under a given pressure p , we perform an expansion of @xmath32 in terms of p. with the assumption of the linear variation of @xmath5 and @xmath79 ( or @xmath71 ) on the pressure , we obtain only terms up to third order , that is , [ allequations ] @xmath80 with @xmath81 and @xmath82 is just the @xmath75 term and where the derivatives @xmath83 and @xmath84 are determined by comparing with the experimental data ; @xmath83 can be estimated by a set of values @xmath85 for a given density @xmath5 and @xmath84 can be determined by hall coefficients or thermoelectric power measurements . after determined these parameters , one can derive analytical expressions for each coefficient as function of @xmath5 ; using the universal parabolic fitting we can obtain a analytical expression for @xmath86 and studying the phase diagram for different values of @xmath10 , we can calculate @xmath87 which calculates the @xmath0 dependency on p ( assumption ii ) . this procedure gives an intrinsic term which has a clear interpretation since it comes from the changes in @xmath71 as well as a new third order term . it is remarkable that with a single set of parameters calculated for hg1201 , with just the change in @xmath79 , we were also able to apply the same expansion ( with same coefficients ) to hg1212 and hg1223 . in order to apply the entire approach described above , we start with the @xmath75 curve @xmath1 for which allows us to determine the initial parameters @xmath27 and @xmath71 . we have used the values @xmath88 and v@xmath89 given in ref.@xcite . the coherence length estimate for is @xmath90 which is about 10% less than the @xmath30 for ybacuo@xcite . since we have used @xmath91 for fitting the phase diagram for these compounds , we use here @xmath92 in order to obtain @xmath1 for . the results are displayed on fig.1 and one can see that it provides a very reasonable fitting for the experimental points of allgeier et al@xcite . in order to match the @xmath93 with v@xmath89 , we used @xmath94k which is the same for all values of @xmath5 . this value is very close @xmath95k used to the hg1210 compounds because both compounds have very closed values for @xmath96k and we used for both a @xmath11 gap function . we have also plotted a parametric curve which has been applied to fit the experimental points , @xmath3 $ ] with @xmath93k and @xmath97 . however the data of huang et al@xcite appears to have smaller @xmath79 , displaced toward higher hole densities and they are fitted with @xmath98 . the parabolic is a pure phenomenological fitting which is used here only because it provides a analytical expression that can be used in the pressure expansion to estimate @xmath99 . now that _ @xmath27 and @xmath71 has been determined _ let us study the compound under applied pressure . from the analysis of the phase diagram@xcite shown in fig.1 and by varying the value of @xmath71 around the @xmath75 value of @xmath94k , we can infer that @xmath100 at @xmath101 which is near the optimal doping value of hole content used by angilella et al@xcite . from the hall coefficient measurements@xcite , we get @xmath102 and since their results seens to be shifted to larger values of hole contents , we estimate @xmath103 _ which is two order of magnitude higher than previously derived @xmath104 for the pressure effects on hg compounds@xcite . _ from the low pressure data we can get @xmath105 which allows us to derive the value @xmath106 which is necessary to determine all the high order coefficients . we noticed that this value is very close to @xmath107 derived for the hg - based htsc . now , with the values of @xmath108 and @xmath109 determined , we can calculate the high order terms of eq.[coeff ] and in particular , we obtain @xmath110 . the third order coefficient was also calculated but it does not give any appreciable contribution in the range of pressure below @xmath111 . thus with all the coefficients of eq.[coeff ] determined , we are all set to calculate the values of @xmath32 . the calculated results are in excellent agreement with the experimental data@xcite as it is shown in fig.2 . the large value of @xmath109 yields a very large negative second order coefficient @xmath112 what makes the curve change its derivative at @xmath113 gpa . for the hg compounds the change in sign of @xmath114 occurs at much larger @xmath115 gpa and from the above analysis we can see that this difference in behavior is due to the large difference on the value of the charge transfer @xmath109 for this two compounds . thus , we conclude this work pointing out that our novel calculations based on a bcs type mean field and on the extended hubbard hamiltonian with the position of v as a variable parameter , is appropriate to describe the interplay between the @xmath0 and hole content for htsc . the change in the position of the attractive interaction provides informations on the range of the cooper pair attractive mechanism . our method yields also a novel and systematic way to study the effects of the pressure and gives indication that the pressure induces a variation on the attractive potential which in turn , gives an intuitive interpretation on the origin of the intrinsic term . we should also emphasize that our procedure is _ based and has its starting point on a microscopic hamiltonian _ and it differs from some pure phenomenological ways to fit the data@xcite that uses a parabolic fitting to the experimental pressure data and are in the same category of the inverse parabolic fitting for the @xmath1 curves mentioned above that are useful to obtain information on the different sets of data but has not any microscopic implications which is our main concern here . by applying it to the superconductors and comparing the results with our previous calculations on hg - based htsc we have seen that the method not only deals successfully with the the experimental results but also displays clearly the physical properties that accounts for their different experimental behavior under pressure . we acknowledge partial financial support from the brazilian agencies capes and cnpq . l.gao et al , phys.rev . b50 * , 4260(1994 ) . r.j.wijngaarden and r. griessen , `` high pressures studies '' in `` studies of high temperature superconductors '' , vol.2 , p.29 edited by a.v . narlikar ( nova science , n.y.,1989 ) . h.takahashi and n.mori , in `` studies of high temperature superconductors '' , vol.16 , edited by a.v .narlikar,(nova science , n.y.,1996 ) . j.s.schilling and s.klotz , in `` physical properties of high temperature superconductors'',edited by d.m.ginsberg , vol.ii , ( world scientific , singapore 1992 ) . e.v.l.de mello , physica * c259 * , 109(1996 ) and check . j. phys.46 , 945 ( 1996 ) . r.micnas , j.ranninger and s.robaszkiewicz , rev.mod.phys . * 62 * , 113 ( 1990 ) . angilella , r.pucci and f.siring , phys . rev . * b54 * , 15471 ( 1996 ) . mello submitted to the physical review letters . `` mcanique quantique '' , l. landau and i. lifshitz , ( ed . mir , moscow 1966 ) , page 190 . p. monthoux , a. v. balatsky and d. pines , phys . lett . * 67 * , 3448 ( 1991 ) . m.cyrot and d.pavuna , `` introduction to superconductivity and high - tc materials '' , world scientific , singapore(1992 ) . de mello and c.acha , phys . rev * b56 * , 466 ( 1997 ) and physica * c 282 - 287 * , 1819 , ( 1997 ) . c. allgeier and j. s. schilling , physica * c168 * , 499 ( 1990 ) . t. huang et al , phys * b48 * , 7712 ( 1993 ) . gennes , `` superconductivity of metals and alloys '' , w.a.benjamin , new york , 1966 . r.j.wijngaarden , e.n.van eenige , j.j.scholtz and r. griessen , physica * c185 - 189 * , 787 ( 1991 ) . almasan , phys . * 69,680 ( 1992 ) . neumeier and h.a . zimmermann , phys . b47 * , 8385 ( 1993 ) . r.p . gupta and m. gupta , phys . rev.b51 11760 ( 1995 ) . m. kosuge et al , phys . rev . * b45 * , 10713 ( 1992 ) . y.cao , q.xiong , y.y.xue and w.chu , phys . rev * b52 * , 6854 ( 1995 ) . d.tristan jover , h.wilhelm , r.j.wijngaarden and r.s.liu , phys . rev.b55 , 11832 ( 1997 ) . * fig.1 * calculations for the bi phase diagram . the continuous line is the phenomenological fitting and the dashed line are the calculations described in the article . the squares are experimental points taken from ref.13 . + * fig.2 * calculations for the variations of @xmath0 for the bi compound under pressure . the circles are experimental points taken from ref.14 .	we use the two dimensional extended hubbard hamiltonian with the position of the attractive potential as a variable parameter with a bcs type approach to study the interplay between the superconductor transition temperature @xmath0 and hole content for high temperature superconductors . this novel method gives some insight on the range and intensity of the cooper pair interaction and why different compounds have different values for their measured coherence lengths and it describes well the experimental results of the superconducting phase diagram @xmath1 . the calculations may also be used to study the effect of the applied pressure with the assumption that it increases the attractive potential which is accompanied by an increase in the superconductor gap . in this way we obtain a microscopic interpretation for the intrinsic term and a general expansion for @xmath0 in terms of the pressure which reproduces well the experimental measurements on the superconductors . : bcs model , hubbard hamiltonian , phase diagram , pressure effects +
the rotation of a quantum fluid leads to many interesting problems . in particular , the rotation of ultracold bose gases has motivated a lot of scientific work within the last few years . + the rotation has the effect of adding an angular momentum to the system so that it can be compared with the motion of a charged particle in a magnetic field . initial experiments @xcite have shown that , analogously to the superfluid helium ii @xcite , vortices nucleate at some critical rotation frequency . the main difference between earlier studies about quantum fluids and bec s is the trap in which the condensate is confined . this inspired a.l . fetter @xcite to suggest adding a quartic term to the harmonic trap potential . his idea was to rotate the condensate so fast that the centrifugal force may overcompensate the harmonic trapping potential . in a harmonic trap , the condensate gets lost when the rotation frequency comes close to the harmonic trap frequency @xcite , but the additional anharmonicity ensures the confining of the condensate . such a trap was realized in an experiment in paris at the cole normale suprieure ( ens ) in the group of j. dalibard in 2004 @xcite . + the experimental setup was a usual magneto - optical trap with the frequencies @xmath0 hz and @xmath1 hz . the additional quartic anharmonicity was generated with a gaussian laser beam propagating in @xmath2-direction which created a potential @xmath3 with the perpendicular radius @xmath4 , the laser s waste @xmath5 @xmath6 m , and the intensity @xmath7 nk . due to the experimentally realized condition @xmath8 , this potential is well approximated by @xmath9 . in ref . @xcite an amount of @xmath10 atoms of @xmath11rb is set into rotation with another laser beam acting as a stirrer . the laser creates an anisotropic potential in the @xmath12-plane which rotates with frequency @xmath13 . in the corotating frame , the resulting trapping potential can be written as @xmath14 where @xmath15 hz and @xmath16 is the atomic mass . the last term in ( [ vrot1 ] ) corresponds to the quartic anharmonicity with @xmath17 jm@xmath18 . + in the following we treat the bose gas in the anharmonic trap ( [ vrot1 ] ) within the grand - canonical ensemble and determine the critical temperature , the condensate fraction , and the heat capacity of the bose gas within a semiclassical approximation . in our discussion the rotation frequency @xmath13 appears as a control parameter . the paris experiment @xcite allows rotation frequencies @xmath13 up to @xmath19 . however , in the present theoretical discussion , we consider arbitrarily large rotation frequencies . in fig . [ pot ] we depict how the trapping potential ( [ vrot1 ] ) varies with increasing rotation frequency @xmath13 . for small rotation frequencies @xmath20 , the potential ( [ vrot1 ] ) is convex , for the critical rotation frequency @xmath21 it is purely quartic in the perpendicular plane , and for a fast rotation frequency @xmath22 the trap has the shape of a mexican hat . in our discussion , we focus on two particular rotation frequencies , namely the critical rotation frequency @xmath21 and the limit of an infinite fast rotation frequency @xmath23 as the potential reduces to power laws in both cases . such potentials were investigated some time ago as they lead to analytic formulas for the respective thermodynamical properties @xcite . note that the case @xmath23 corresponds to a trap where the bosons are confined to a cylinder of radius @xmath24 . thus , the above mentioned experimental restriction @xmath25 allows to determine a maximum rotation frequency @xmath26 for which the anharmonic potential ( [ vrot1 ] ) of the paris experiment is valid . the resulting value @xmath27 shows that considering an infinite rotation frequency @xmath23 is not suitable for this experiment . ) in @xmath28-direction for the values of the paris experiment @xcite and varying rotation frequencies @xmath29 , @xmath30 , @xmath31 . the latter is the largest experimentally realized rotation frequency.,title="fig : " ] ( -3.5,-.3)@xmath32 $ ] ( -7,1 ) ( -3.3,2.3)@xmath33 ( -2,2)@xmath34 ( -1.5,1.1)@xmath35 we consider @xmath36 particles of an ideal bose gas which are distributed over various quantum states @xmath37 of the system . these states are characterized through the population @xmath38 of the one - particle state @xmath39 of the trap ( [ vrot1 ] ) , such that the energy levels are @xmath40 , where @xmath41 denotes the one - particle energy . correspondingly , the number of particles in state @xmath37 are given by @xmath42 . the resulting grand - canonical ensemble is determined by its partition function @xmath43}\,,\ ] ] where @xmath44 denotes an inverse temperature , @xmath45 is boltzmann s constant , and @xmath6 is the chemical potential . the corresponding grand - canonical free energy @xmath46 allows to calculate all relevant thermodynamical quantities @xcite . around the minimum , the trap ( [ vrot1 ] ) has a small curvature so that the energy levels are close to each other . with increasing rotation frequency , the curvature decreases until the critical rotation frequency @xmath21 is reached . not until the rotation frequency overcompensates the harmonic part of the trap , the curvature increases again . thus , for all experimentally realized rotation frequencies @xmath47 , our system can be described by the discrete ground state @xmath48 , which must be retained quantum - mechanically , plus a continuum of states above @xmath48 . within this semiclassical approximation , we can set the ground - state energy @xmath48 to zero so that the grand - canonical free energy of the ideal bose gas reads @xmath49\right\}}\,,\end{aligned}\ ] ] where the energy levels are replaced by the classical hamiltonian @xmath50 the critical chemical potential @xmath51 , where the condensation emerges , is determined by the condition @xmath52 and is therefore given by @xmath53 . due to ( [ vrot1 ] ) it reads explicitly . the highest lying line ( solid ) corresponds to the data of the paris experiment @xcite , see data below eq . ( [ vrot1 ] ) . for the deeper lying lines we varied the anharmonicity parameter @xmath54 : @xmath55 ( long dashes ) , @xmath56 ( short dashes ) , and the harmonic limit @xmath57 ( gray solid).,title="fig : " ] ( -4,-.3)@xmath58 ( -7,1.6 ) @xmath59 -{\displaystyle \frac{m^2}{4k}\left(\omega_\bot^2-\omega^2\right)^2}&;\,\omega>\omega_\bot\ , . \end{array}\right.\ ] ] performing the phase - space integral in ( [ fg1 ] ) , we obtain @xmath60 where we have introduced the generalized @xmath61-function @xmath62 } \mathrm{erfc}\left[\sqrt{j \gamma_t } \left(\omega_\bot^2-\omega^2\right)\right]\end{aligned}\ ] ] with the complementary error function @xmath63 we have also shortened the notation with using @xmath64 as another inverse temperature . we remark that in the limit of a vanishing anharmonicity @xmath57 with an undercritical rotation frequency @xmath20 the generalized @xmath61-function ( [ zeta1 ] ) reduces to the polylogarithmic function @xmath65 which is related to the riemann @xmath61-function via @xmath66 furthermore , we note that in the limit of the critical rotation frequency @xmath67 , the generalized @xmath61-function ( [ zeta1 ] ) reads @xmath68 and in the limit of an infinite fast rotation it is approximated by @xmath69 ) of a bose gas in the trap ( [ vrot1 ] ) for the rotation frequency @xmath70 and the parameters of the paris experiment @xcite . the dashed line corresponds to the condensate fraction at the critical @xmath21 and at the infinite fast @xmath23 rotation frequency given by eq . ( [ cf1]).,title="fig : " ] ( -7,1.7 ) ( -4,-0.3)@xmath71 from the grand - canonical free energy ( [ fg2 ] ) we read off that the number of particles @xmath72 of an ideal bose gas is given by a sum of @xmath73 of particles in the ground state @xmath74 and particles in excited states @xmath75 : @xmath76 the critical temperature @xmath77 at which the condensation emerges can be found from eq . ( [ n1 ] ) by setting @xmath78 and @xmath79 . for undercritical rotation frequencies @xmath20 and vanishing anharmonicity @xmath57 , we apply ( [ zeta2 ] ) so that the critical temperature reads @xmath80^{1/3}\,;\,\,k=0\,.\ ] ] for a non - vanishing anharmonicity @xmath54 , the critical temperature can not be determined explicitly because it appears in eq . ( [ n1 ] ) transcendentally . however , there are two special cases in which we obtain an analytical expression for the critical temperature . at first , for the critical rotation @xmath21 , we find with ( [ zeta5 ] ) @xcite : ) is evaluated for various rotation frequencies @xmath13 , the solid line corresponds to @xmath70 to which the temperature is normalized . the other lines are for @xmath81 ( short dashes ) , @xmath21 ( longer dashes ) , and @xmath82 ( long dashes).,title="fig : " ] ( -7,1.7 ) ( -4,-0.3)@xmath83 @xmath84 secondly , the limit of an infinite fast rotation frequency @xmath23 leads with ( [ zeta6 ] ) to the critical temperature @xmath85 which is by a factor @xmath86 smaller than the previous one . a numerical evaluation of the critical temperature obtained from ( [ n1 ] ) is shown in fig . [ ptc1 ] for the values of the paris experiment @xcite . for the non - rotating trap , we see that the anharmonicity only slightly affects the critical temperature . with increasing rotation frequency @xmath13 , the critical temperature decreases and the difference between the harmonic and the anharmonic trap is clearly seen . at the critical rotation frequency @xmath21 , the critical temperature is @xmath87 nk which is about three times smaller than the one estimated for the paris experiment @xcite . + from the number of particles ( [ n1 ] ) , we also obtain the condensate fraction in the temperature regime @xmath88 . here the chemical potential coincides with the critical one given by eq . ( [ mu1 ] ) . for undercritical rotation frequencies @xmath20 and vanishing anharmonicity @xmath57 , we use ( [ zeta2 ] ) so that the condensate fraction is given by @xmath89 furthermore , applying ( [ zeta5 ] ) and ( [ zeta6 ] ) in the cases @xmath21 and @xmath23 , respectively , yields with the critical chemical potential ( [ mu1 ] ) the following condensate fraction : @xmath90 here , @xmath77 is given by ( [ tc0 ] ) and ( [ tc1 ] ) , ( [ tc2 ] ) , respectively . in general , the condensate fraction is given by @xmath91 where @xmath92 . in the low - temperature limit @xmath93 , the condensate fraction shows a power - law behavior which is different in the two regimes of undercritical rotation @xmath20 and overcritical rotation @xmath22 . due to ( [ zeta2 ] ) and ( [ zeta6 ] ) , we obtain @xmath94 { \displaystyle 1-\frac{m\sqrt{\pi}k_b^{5/2}\zeta(5/2)}{n\sqrt{k}\hbar^3\omega_z}\,t^{5/2 } } & \!\!;\ , \omega > \omega_\bot\ , . \end{array}\right.\ ] ] the temperature dependence of the condensate fraction @xmath95 following from ( [ cf2 ] ) is shown in fig . [ pcf1 ] and fig . [ pcf2 ] . from this we read off that the temperature dependence of the condensate fraction depends crucially on the rotation frequency and is thus not universal . the heat capacity follows from the grand - canonical free energy @xmath96 , where @xmath97 is the internal energy and @xmath98 is the entropy , according to @xmath99 within the grand - canonical ensemble , the heat capacity has to be determined separately in the two regimes @xmath100 and @xmath88 . at first , we treat the gas phase where @xmath78 and determine the entropy via the thermodynamical relation @xmath101 : @xmath102 & & -\left[\beta\mu+\gamma_t(\omega_\bot^2-\omega^2)^2\right]\end{aligned}\ ] ] the internal energy @xmath103 then follows from ( [ fg2 ] ) , ( [ n1 ] ) , ( [ s1 ] ) and reads @xmath104\,.\end{aligned}\ ] ] finally , the heat capacity ( [ hc1 ] ) for temperatures above @xmath77 is given by ) , solid lines , and its approximation ( [ hc2dp ] ) , dashed lines , to the dulong - petit law , horizontal line.,title="fig : " ] ( -7,1.5 ) ( -4,-0.3)@xmath83 ( -4.5,.8)@xmath105 ( -3.5,2.8)@xmath70 @xmath106 & & \hspace{-15mm}+\gamma_t(\omega_\bot^2-\omega^2)^2\left[\frac{11\zeta_3(e^{\beta\mu})}{2\zeta_3(e^{\beta\mu},\omega)}-5\frac{\zeta_2(e^{\beta\mu})}{\zeta_2(e^{\beta\mu},\omega)}\right ] \nonumber\\[2 mm ] & & \hspace{-15mm}+\gamma_t^2(\omega_\bot^2-\omega^2)^4\frac{\zeta_2(e^{\beta\mu})\zeta_2(e^{\beta\mu},\omega)-\zeta_2 ^ 2(e^{\beta\mu})}{\zeta_2(e^{\beta\mu},\omega)\zeta_3(e^{\beta\mu},\omega)}\,.\end{aligned}\ ] ] to obtain this result , we have determined the derivative @xmath107 from ( [ n1 ] ) . in the limit @xmath108 of a harmonic trap , the heat capacity ( [ hc2 ] ) reduces to the well - known result @xmath109 again , both cases @xmath21 and @xmath23 yield the same analytic expression : @xmath110 & & -\frac{25}{4}\frac { \zeta_{5/2}\left ( e^{\beta\mu+\gamma_t(\omega_\bot^2-\omega^2)^2 } \right ) } { \zeta_{3/2}\left ( e^{\beta\mu+\gamma_t(\omega_\bot^2-\omega^2)^2 } \right ) } \ , .\end{aligned}\ ] ] now , we investigate the heat capacity ( [ hc2 ] ) in the high temperature limit @xmath111 . to this end we use a large @xmath112-expansion of the generalized @xmath61-function ( [ zeta1 ] ) : ) of the trap ( [ vrot1 ] ) with @xmath113 . the dashed line corresponds to the harmonic trap heat capacity ( [ hc2l ] ) , ( [ hc3l ] ) for the paris trap ( [ vrot1 ] ) with @xmath113 . the black solid line is the heat capacity ( [ hc2 ] ) , ( [ hc3 ] ) for the anharmonic paris trap @xcite . the horizontal lines correspond to the dulong - petit law : harmonic trap ( dashed ) and anharmonic trap ( gray).,title="fig : " ] ( -4,-.3)@xmath114 ( -7,1.5 ) ( -4.4,2.8)@xmath113 ( -2.9,2.3)@xmath115 @xmath116\nonumber\\ & & \hspace{-2 cm } + \frac{e^{2\beta\mu}}{2^\nu}\!\left[\ ! \sqrt{2\gamma_t}(\omega_\bot^2-\omega^2 ) - 4\gamma_t(\omega_\bot^2-\omega^2)^2 \!+ ... \right]+ ... \,.\end{aligned}\ ] ] inserting the expansion ( [ zeta4 ] ) into the number of particles ( [ n1 ] ) , we find for the first order of the fugacity @xmath117 , so that the heat capacity ( [ hc2 ] ) behaves like @xmath118 thus , the heat capacity approaches the dulong - petit law in an anharmonic trap @xmath119 which is @xmath120 smaller than the corresponding one in the harmonic trap . furthermore , the first @xmath13 dependent term in ( [ hc2dp ] ) changes its behavior , from being larger ( @xmath20 ) than the limit to being smaller ( @xmath22 ) , see fig . + at the critical point , the harmonic heat capacity ( [ hc2l ] ) reduces for small rotation frequencies @xmath20 to @xmath121 in both limits @xmath21 and @xmath23 , the heat capacity ( [ hc2rl ] ) at the critical point is given by @xmath122 now we turn to the condensate phase @xmath88 where the chemical potential is given by ( [ mu1 ] ) . for the entropy , we obtain @xmath123 & & \hspace{-10mm}\left.+\gamma_t(\omega_\bot^2-\omega^2)^2 \left[\frac{\zeta_3(e^{\beta\mu_c})-\zeta_3(e^{\beta\mu_c},\omega)}{\zeta_3(e^{\beta_c\mu_c},\omega)}\right ] \right\}.\end{aligned}\ ] ] the internal energy below the critical temperature reads @xmath124 & & \hspace{-2cm}\left.\times\left[\frac{\zeta_3(e^{\beta\mu_c})-\theta(\omega_\bot-\omega)\zeta_3(e^{\beta\mu_c},\omega)}{\zeta_3(e^{\beta_c\mu_c},\omega ) } \right]\right\}+\mu_c n\,,\end{aligned}\ ] ] where @xmath125 denotes the heaviside function . thus , we find for the heat capacity below the critical temperature with help of eq . ( [ mu1 ] ) @xmath126\nonumber\\[2 mm ] & & \hspace{-7mm}+\,\gamma_t^2(\omega_\bot^2-\omega^2)^4\left[\theta(\omega-\omega_\bot)\frac{\zeta_2(e^{\beta\mu_c})}{\zeta_3(e^{\beta_c\mu_c},\omega)}\right.\nonumber\\[2 mm ] & & \hspace{-7mm}-\left.\left.\theta(\omega_\bot-\omega)\frac{\zeta_2(e^{\beta\mu_c})-\zeta_2(e^{\beta\mu_c},\omega)}{\zeta_3(e^{\beta_c\mu_c},\omega ) } \right]\right\}\,.\end{aligned}\ ] ] in the limit @xmath108 , eq . ( [ hc3 ] ) simplifies to @xmath127 thus , at the critical point it has the value @xmath128 in both limits @xmath21 and @xmath23 , the heat capacity is given by of the anharmonic trap ( [ vrot1 ] ) for varying rotation frequencies . the lines correspond to @xmath70 ( solid ) , @xmath21 ( long dashes ) , and @xmath129 ( short dashes ) . the gray solid line corresponds to the dulong - petit law , the first term of ( [ hc2dp ] ) . we note that for @xmath129 and @xmath100 the heat capacity approaches the dulong - petit limit from below ( see fig . [ phc3]).,title="fig : " ] ( -7,2 ) ( -4,-0.3)@xmath83 @xmath130 so that the heat capacity ( [ hc3 ] ) at the critical point reduces to @xmath131 in the low temperature limit @xmath132 , the heat capacity ( [ hc3 ] ) tends to zero in accordance with the third law of thermodynamics . we note that the low - temperature limit of the heat capacity ( [ hclt ] ) has the same power - law behavior as the corresponding one of the condensate fraction ( [ cf3 ] ) : @xmath133 { \displaystyle \frac{35m\sqrt{\pi}k_b^{5/2}\zeta(7/2)}{4n\sqrt{k}\hbar^3\omega_z}\,t^{5/2 } } & \!\!;\,\omega>\omega_\bot\ , . \end{array}\right.\ ] ] fig . [ phc1 ] shows the temperature dependence of the heat capacity for the values of the paris experiment without rotation @xcite . we see that the effect of the anharmonicity is rather small . according to ehrenfest s classification , the discontinuity at the critical temperature characterizes the phase transition to be of second order . in fig . [ phc2 ] we show how the heat capacity depends on the rotation frequency @xmath13 . here , the rotation has a huge influence on the temperature dependence of the heat capacity . we have determined a semiclassical approximation for the critical temperature @xmath77 at which the condensation of a rotating ideal bose gas occurs in the anharmonic trap ( [ vrot1 ] ) . we have found that condensation is possible even in the overcritical rotation regime @xmath22 which is in contrast with the harmonic trap where the condensate gets lost , when the rotation frequency @xmath13 gets close to the trap frequency @xmath134 . our value for the critical temperature @xmath135 nk at the critical rotation frequency @xmath21 yields @xmath136 which justifies a posteriori our semiclassical treatment . however , the value @xmath135 nk is about three times smaller than the one estimated in the paris experiment @xcite . this huge discrepancy could not be explained with the circumstance that our semiclassical analysis of the rotating ideal bose gas neglects three important aspects , namely finite - size corrections , interactions between the particles , and the effect of vortices in the condensate . all three should have the effect of lowering the critical temperature . in a harmonic trap , it was found numerically that the finiteness of the system slightly lowers the critical temperature @xcite which was also shown analytically @xcite . furthermore , an additional weak repulsive two - particle contact interaction leads to a negative shift of just a few percent @xcite . therefore , we conclude that our semiclassical findings ask for a revised experimental measurement of the critical temperature for a rotating bose gas . + for the condensate fraction , we can state that the low - temperature behavior crucially depends on the rotation frequency @xmath13 . it shows a non - uniform temperature dependence which is in between the two power - laws @xmath137 and @xmath138 . + the heat capacity of the rotating ideal bose gas is discontinuous at the critical temperature . it tends in agreement with the third law of thermodynamics to zero in the low temperature limit @xmath132 , and approaches the dulong - petit law ( [ hc2dp ] ) in the high temperature limit @xmath111 from above ( @xmath139 ) or from below ( @xmath22 ) . we thank jean dalibard , hagen kleinert , and sabine stock for critical reading of the manuscript as well as robert graham for the hospitality at the university duisburg - essen . furthermore , support from the dfg priority program spp 1116 _ interaction in ultracold gases of atoms and molecules _ is acknowledged . 199 f. chevy , k.w . madison , and j. dalibard , phys . lett . * 85 * , 2223 ( 2000 ) c. raman , j.r . abo - shaeer , j.m . vogels , k. xu , and w. ketterle , phys . * 87 * , 210402 ( 2001 ) r.j . donelly , _ quantized vortices in helium ii _ ( cambridge univ . press , cambridge , 1991 ) a.l . fetter , phys . a * 64 * , 063608 ( 2001 ) p. rosenbusch , d.s . petrov , s. sinha , f. chevy , v. bretin , y. castin , g. shlyapnikov , and j. dalibard , phys . lett . * 88 * , 250403 ( 2002 ) v. bretin , s. stock , y. seurin , and j. dalibard , phys . rev . lett . * 92 * , 050403 ( 2004 ) v. bagnato , d.e . pritchard , and d. kleppner , phys . a * 35 * , 4354 ( 1987 ) ph . w. courteille , v.s . bagnato , and v.i . yukalov , las . phys . * 11 * , 659 ( 2001 ) l.p . pitaevskii and s. stringari , _ bose - einstein condensation _ ( science pub . , oxford , 2003 ) c.j . pethick and h. smith , _ bose - einstein condensation in dilute gases _ ( cambridge univ . press , cambridge , 2002 ) s. stock , b. battelier , v. bretin , z. hadzibabic , and j. dalibard , las . * 2 * , 275 ( 2005 ) s. stock , _ quantized vortices in a bose - einstein condensate : thermal activation and dynamic nucleation _ , phd thesis , universit paris 6 , france , january 2006 w. ketterle and n.j . van druten , phys . rev . a * 54 * , 656 ( 1996 ) s. grossmann and m. holthaus , phys . a * 208 * , 188 ( 1995 ) t. haugset , h. haugerud , and j.o . andersen , phys . a * 55 * , 2922 ( 1997 ) s. giorgini , l.p . pitaevskii , and s. stringari , phys . a * 54 * , r4633 ( 1996 ) f. gerbier , j.h . thywissen , s. richard , m. hugbart , p. bouyer , and a. aspect , phys . * 92 * , 030405 ( 2004 )	in a recent experiment , a bose - einstein condensate was trapped in an anharmonic potential which is well approximated by a harmonic and a quartic part . the condensate was set into such a fast rotation that the centrifugal force in the corotating frame overcompensates the harmonic part in the plane perpendicular to the rotation axis . thus , the resulting trap potential became mexican - hat shaped . we present an analysis for an ideal bose gas which is confined in such an anharmonic rotating trap within a semiclassical approximation where we calculate the critical temperature , the condensate fraction , and the heat capacity . in particular , we examine in detail how these thermodynamical quantities depend on the rotation frequency .
in nonleptonic hyperon decays such as @xmath4 it is possible to search for @xmath0 violation by comparing the angular distribution with that of the corresponding anti - hyperon decay @xcite . the fermilab experiment hypercp is currently analyzing data searching for @xmath0 violation in such a decay . the reaction of interest for hypercp is the decay of a polarized @xmath5 , with known polarization @xmath6 , into a proton ( whose polarization is not measured ) and a @xmath7 with momentum @xmath8 . the interesting observable is a correlation in the decay distribution of the form @xmath9 the branching ratio for this mode is @xmath10 , and the parameter @xmath11 has been measured to be @xmath12 @xcite . the @xmath0 violation in question involves a comparison of the parameter @xmath11 with the corresponding parameter @xmath13 from the reaction @xmath14 to obtain polarized @xmath5 s with known polarization , it is necessary to study the double decay chain @xmath15 @xcite . this eventually leads to the experimental observable being sensitive to the _ sum _ of @xmath0 violation in the @xmath16 decay and @xmath0 violation in the @xmath5 decay . in both reactions , @xmath17 and @xmath18 the final state can be reached from the initial state via @xmath19 or @xmath20 transitions . it is known that due to the existence of a strong @xmath19 rule for nonleptonic hyperon decay , the dominant contribution to the @xmath0-violating asymmetries arises from interference between an @xmath21-wave and a @xmath22-wave within the @xmath19 transition @xcite . one can define the @xmath0-violating asymmetries @xmath23 for the @xmath5 and @xmath24 decays , respectively . the experimental observable is then @xcite @xmath25 approximate expressions have been obtained for @xmath26 in the case of @xmath19 dominance @xcite , namely @xmath27 here , @xmath28 @xmath29 is the strong @xmath21-wave ( @xmath22-wave ) @xmath30 scattering phase - shift at @xmath31 and @xmath32 @xmath33 is the strong @xmath21-wave ( @xmath22-wave ) @xmath34 scattering phase - shift at @xmath35 moreover , @xmath36 @xmath37 are the @xmath0-violating weak phases induced by the @xmath38 @xmath39 interaction in the @xmath21-wave ( @xmath22-wave ) of the @xmath40 and @xmath41 decays , respectively . experimentally , the current published limit is @xmath42 from e756 @xcite , and the expected sensitivity of hypercp is @xmath43 @xcite . in addition , hypercp has recently obtained a preliminary measurement of @xmath44 @xcite . previous estimates for @xmath45 indicated that it occurs at the few times @xmath46 level within the standard model @xcite and that it can be as large as @xmath47 beyond the standard model @xcite . the larger asymmetries occur in models with an enhanced gluon - dipole operator that is parity - even and thus does not contribute to the @xmath48 parameter in kaon decay . the @xmath47 upper bound corresponds to the phenomenological constraint from new contributions to the @xmath49 parameter in kaon mixing . this illustrates the relevance of the hypercp measurement which complements the @xmath48 experiments in the study of @xmath0 violation in @xmath50 transitions . the strong @xmath51 scattering phases needed in eq . ( [ approxa ] ) have been measured to be @xmath52 and @xmath53 with errors of about @xmath54 @xcite . in contrast , the strong @xmath34 scattering phases have not been measured . modern calculations based on chiral perturbation theory indicate that these phases are small , with @xmath55 being at most @xmath56 @xcite . for our numerical estimates , we will allow the @xmath34 phases to vary within the range obtained at next - to - leading order in heavy - baryon chiral perturbation theory @xcite , @xmath57 one could choose to be less constrained and include the larger @xmath58 found in ref . @xcite , but this would only enlarge the @xmath59 range and hence the uncertainty of the predicted asymmetry . in any case , eventually these phases can be extracted directly from the measurement of the decay distribution in @xmath60 @xcite . recently e756 has reported a preliminary result of @xmath61 @xcite . in this paper , we estimate the weak phases that appear in @xmath62 and @xmath63 within the standard model . in sec . [ cpt ] , we present a calculation of the weak phases guided by heavy - baryon chiral perturbation theory in terms of three unknown weak counterterms . in sec . [ ect ] , we estimate the value of these counterterms by considering contributions that arise from the factorization of the penguin operator and also nonfactorizable contributions estimated in the mit bag model . [ results ] contains the resulting weak phases and @xmath0-violating asymmetries . finally , in sec . [ disc ] , we compare our results to those of previous work and present our conclusions . for completeness , we also provide in an appendix the results for the corresponding asymmetries in @xmath64 decays . the chiral lagrangian that describes the interactions of the lowest - lying mesons and baryons is written down in terms of the lightest meson - octet , baryon - octet , and baryon - decuplet fields @xcite . the meson and baryon octets are collected into @xmath65 matrices @xmath66 and @xmath67 , respectively , and the decuplet fields are represented by the rarita - schwinger tensor @xmath68 , which is completely symmetric in its su(3 ) indices ( @xmath69 ) . the octet mesons enter through the exponential @xmath70 where @xmath71 is the pion - decay constant . in the heavy - baryon formalism @xcite , the baryons in the chiral lagrangian are described by velocity - dependent fields , @xmath72 and @xmath73 . for the strong interactions , the leading - order lagrangian is given by @xcite @xmath74 \right\rangle \nonumber \\ & & -\,\ , \bar t_v^\mu { \rm i } v\cdot { \cal d } t_{v\mu}^ { } + \delta m\ , \bar t_v^\mu t_{v\mu}^ { } + { \cal c } \left ( \bar t_v^\mu { \cal a}_\mu^ { } b_v^ { } + \bar b_v^ { } { \cal a}_\mu^ { } t_v^\mu \right ) + 2{\cal h}\ , \bar t_v^\mu s_v^{}\cdot{\cal a } t_{v\mu}^ { } \,\ , , \end{aligned}\ ] ] where @xmath75 denotes @xmath76 in flavor - su(3 ) space , @xmath77 is the spin operator , and @xmath78 with further details given in ref . @xcite . in this lagrangian , @xmath79 , @xmath80 , @xmath81 , and @xmath82 are free parameters , which can be determined from hyperon semileptonic decays and from strong decays of the form @xmath83 fitting tree - level formulas , one extracts @xcite @xmath84 whereas @xmath82 is undetermined from this fit . from the nonrelativistic quark models , one finds the relations @xcite @xmath85 which are well satisfied by @xmath79 , @xmath80 , and @xmath81 , suggesting the tree - level value @xmath86 in our numerical estimates , we use eqs . ( [ sptree ] ) and ( [ htree ] ) for the leading - order results and the estimate of their uncertainty from one - loop contributions , with @xmath81 and @xmath82 only appearing in loop diagrams involving decuplet baryons . as another estimate of the uncertainty in these results , we will evaluate the effect of varying @xmath79 and @xmath80 between their tree - level values above and their one - loop values to be given later . at next - to - leading order , the strong lagrangian contains a greater number of terms @xcite . the ones of interest here are those that explicitly break chiral symmetry , containing one power of the quark - mass matrix @xmath87 for our calculation of the factorization of the penguin operator , we will need these terms in the form @xmath88 \right\rangle + \frac{b_0^{}}{2 b_0^ { } } \left\langle \chi_+^ { } \right\rangle \left\langle \bar b_v^ { } b_v^ { } \right\rangle \nonumber \\ & & + \,\ , \frac{c}{2 b_0^{}}\ , \bar t_v^\mu \chi_+^ { } t_{v\mu}^ { } - \frac{c_0^{}}{2 b_0^ { } } \left\langle \chi_+^ { } \right\rangle \bar t_v^\mu t_{v\mu}^ { } \,\ , , \end{aligned}\ ] ] where we have used the notation @xmath89 to introduce coupling to external ( pseudo)scalar sources , @xmath90 such that in the absence of the external sources @xmath3 reduces to the mass matrix , @xmath91 as will be discussed in the next section , we also need from the meson sector the next - to - leading - order lagrangian @xmath92 where only the relevant term is explicitly shown . in eqs . ( [ ls2 ] ) and ( [ ls4 ] ) , the constants @xmath93 , @xmath94 , @xmath95 , @xmath96 , and @xmath97 are free parameters to be fixed from data . as is well known , the weak interactions responsible for hyperon nonleptonic decays are described by a @xmath50 hamiltonian that transforms as @xmath98 under su(3@xmath99@xmath100su(3@xmath101 rotations . it is also known from experiment that the octet term dominates the 27-plet term , as indicated by the fact that the @xmath19 components of the decay amplitudes are larger than the @xmath20 components by about twenty times @xcite . we shall , therefore , assume in what follows that the decays are completely characterized by the @xmath102 , @xmath19 interactions . the leading - order chiral lagrangian for such interactions is @xcite @xmath103 \right\rangle + h_c^{}\ , \bar t_v^\mu\ , \xi^\dagger h \xi\ , t_{v\mu}^ { } \nonumber \\ & & + \,\ , \gamma_8^ { } f^2 \left\langle h\ , \partial_\mu^ { } \sigma\ , \partial^\mu \sigma^\dagger \right\rangle \,\,+\,\ , { \rm h.c . } \,\ , , \label{weakcl } \end{aligned}\ ] ] where @xmath104 is a 3@xmath1003 matrix with elements @xmath105 and the parameters @xmath106 and @xmath107 contain the weak phases to be discussed below . the weak lagrangian in eq . ( [ weakcl ] ) is thus the leading - order ( in @xmath3pt ) realization of the effective @xmath50 hamiltonian in the standard model @xcite , @xmath108 where @xmath109 is the fermi coupling constant , @xmath110 are the elements of the cabibbo - kobayashi - maskawa ( ckm ) matrix @xcite , @xmath111 are the wilson coefficients , and @xmath112 are four - quark operators whose expressions can be found in ref . later on , we will express @xmath110 in the wolfenstein parametrization @xcite . it follows that @xmath113 at lowest order in @xmath114 . for our numerical estimates , the relevant parameters that we will employ are @xcite @xmath115 in the next section , we match the penguin operator @xmath116 in the short - distance hamiltonian of eq . ( [ weaksd ] ) with the corresponding lagrangian parameters in eq . ( [ weakcl ] ) . we now have all the ingredients necessary to calculate the weak decay amplitudes in terms of the four parameters @xmath106 and @xmath107 ( only the first two are needed at leading order ) . in the heavy - baryon formalism , the amplitude for the weak decay of a spin-@xmath117 baryon @xmath67 into another spin-@xmath117 baryon @xmath118 and a pseudoscalar meson @xmath119 has the general form @xcite @xmath120 where the superscripts refer to the @xmath21- and @xmath22-wave components of the amplitude . to express our results , we also adopt the notation @xcite @xmath121 with the lagrangians given above , one can derive the amplitudes at leading order , represented by the diagrams in fig . [ tree0 ] . [ tree0](a ) indicates that the @xmath21-wave is directly obtained from a weak vertex provided by eq . ( [ weakcl ] ) . the leading contribution to the @xmath22-wave arises from baryon - pole diagrams , as in fig . [ tree0](b ) , which each involve a weak vertex from eq . ( [ weakcl ] ) and a strong vertex from eq . ( [ ls1 ] ) . thus the leading - order results for amplitudes not related by isospin are @xcite + the leading nonanalytic contributions to the amplitudes arise from one - loop diagrams , with @xmath124 only appearing in those involving decuplet baryons . these contributions have been calculated by various authors @xcite , and we will adopt the results of ref . @xcite for the numerical estimate of our uncertainty . leading - order diagrams for ( a ) @xmath21-wave and ( b ) @xmath22-wave hyperon nonleptonic decays . in all figures , a solid ( dashed ) line denotes a baryon - octet ( meson - octet ) field , and a solid dot ( hollow square ) represents a strong ( weak ) vertex , with the strong vertices being generated by @xmath125 in eq . ( [ ls1 ] ) . here the weak vertices come from the @xmath126 terms in eq . ( [ weakcl ] ) . ] in fig . [ tree1 ] , we show the kaon - pole diagram to be discussed later on . in this diagram , there is a strong vertex from eq . ( [ ls1 ] ) followed by a kaon pole and a weak vertex from the @xmath107 term in eq . ( [ weakcl ] ) . notice that this term is not only subleading in the chiral expansion , but also suppressed by an @xmath127 factor ( and hence vanishing in the @xmath128 limit ) . once the values of the weak couplings @xmath126 are specified , the formulas in eq . ( [ cptamp ] ) determine the leading - order amplitudes . it is well known that this representation does not provide a good fit to the measured @xmath22-wave amplitudes , and that higher - order terms are important @xcite . the procedure that we adopt for estimating the weak phases is to obtain the real part of the amplitudes from experiment ( assuming no @xmath0 violation ) and to use eq . ( [ cptamp ] ) to estimate the imaginary parts . the dominant @xmath0-violating phases in the @xmath19 sector of the @xmath50 weak interaction occur in the wilson coefficient @xmath129 associated with the penguin operator @xmath116 . our strategy will be to calculate within a model the imaginary part of the couplings @xmath106 and @xmath107 induced by @xmath116 . as a numerical result , we propose a central value from leading - order @xmath3pt [ eq . ( [ cptamp ] ) ] and an estimate of the error from the nonanalytic corrections obtained with the expressions given in ref . @xcite . to end this section , for later use we collect in table [ spx ] the experimental values of the @xmath21- and @xmath22-wave amplitudes of interest , reproduced from ref . the numbers are extracted ( neglecting strong and weak phases ) from the measured decay width @xmath130 and decay parameter @xmath11 by means of the relations @xmath131 the @xmath132 and @xmath133 amplitudes are related to those in eq . ( [ m , hb ] ) above by expression has the opposite sign , @xmath134 but this turns out to be inconsistent with the amplitude formula from which both @xmath130 and @xmath11 are derived . nevertheless , the sign flip does not affect the conclusions of refs . @xcite , as the fits therein were performed to the @xmath21-waves and the @xmath22-waves were poorly reproduced regardless of the sign of @xmath133 . ] @xmath135 we may , therefore , conclude that the uncertainties of the weak phases are all of order 200@xmath137 . this is reflected in our prediction of the phases , which are collected in table [ phases_sigma ] . the corresponding strong phases have been measured @xcite and their values have also been included in this table . @c@c@c@ c@\|\|@c@c@ c@c@ @xmath138@xmath139 & @xmath140 & @xmath141 & @xmath142 & @xmath143 & @xmath144 & @xmath145 & @xmath146 + & & & + @xmath147 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 & @xmath153 + from the central values of the isospin amplitudes and the phases in eq . ( [ spsigma ] ) and table [ phases_sigma ] , respectively , we obtain @xmath154 where we have used @xmath155 as before . in this case our estimate is a very rough one , as its uncertainty is larger than those for the other hyperons . this is due to the ( apparently accidental ) smallness of @xmath156 and its large experimental error , indicated in eq . ( [ spsigma ] ) , as well as to the already sizable uncertainties quoted in table [ phases_sigma ] . in order to have a more quantitative estimate of the uncertainties , these modes will have to be revisited when better measurements of the amplitudes become available . s. okubo , phys . rev . * 109 * , 984 ( 1958 ) ; a. pais , phys . lett . * 3 * , 242 ( 1959 ) . groom _ et al . _ [ particle data group collaboration ] , eur . j. c * 15 * , 1 ( 2000 ) . luk _ et al . _ [ e756 collaboration ] , phys . lett . * 85 * , 4860 ( 2000 ) . luk _ et al . _ [ e756 and hypercp collaborations ] , arxiv : hep - ex/0005004 . donoghue and s. pakvasa , phys . * 55 * , 162 ( 1985 ) . donoghue , x .- he , and s. pakvasa , phys . d * 34 * , 833 ( 1986 ) . x .- g . he , h. steger , and g. valencia , phys . b * 272 * , 411 ( 1991 ) . p. zyla [ hypercp collaboration ] , talk given at the 5th international conference on hyperons , charm and beauty hadrons , vancouver , canada , 25 - 29 june 2002 . iqbal and g.a . miller , phys . d * 41 * , 2817 ( 1990 ) . d. chang , x .- he , and s. pakvasa , phys . lett . * 74 * , 3927 ( 1995 ) ; x .- g . he and g. valencia , phys . rev . d * 52 * , 5257 ( 1995 ) ; x .- g . he , h. murayama , s. pakvasa , and g. valencia , _ ibid . _ * 61 * , 071701 ( 2000 ) . roper , r.m . wright , and b. feld , phys . rev . * 138 * , 190 ( 1965 ) ; a. datta and s. pakvasa , phys . d * 56 * , 4322 ( 1997 ) . m. lu , m.b . wise , and m.j . savage , phys . b * 337 * , 133 ( 1994 ) . a. datta and s. pakvasa , phys . b * 344 * , 430 ( 1995 ) . kamal , phys . d * 58 * , 077501 ( 1998 ) . j. tandean , a.w . thomas , and g. valencia , phys . d * 64 * , 014005 ( 2001 ) . meissner and j.a . oller , phys . d * 64 * , 014006 ( 2001 ) . n. kaiser , phys . c * 64 * , 045204 ( 2001 ) . a. chakravorty [ e756 collaboration ] , talk given at the meeting of the division of particles and fields of the american physical society , williamsburg , virginia , 24 - 28 may 2002 . e. jenkins and a.v . manohar , phys . b * 255 * , 558 ( 1991 ) . e. jenkins and a. manohar , phys . b * 259 * , 353 ( 1991 ) . a. abd el - hady , j. tandean , and g. valencia , nucl . * a651 * , 71 ( 1999 ) . n. kaiser , p.b . siegel , and w. weise , nucl . * a594 * , 325 ( 1995 ) ; n. kaiser , t. waas , and w. weise , _ ibid . _ * a612 * , 297 ( 1997 ) ; j. caro ramon , n. kaiser , s. wetzel , and w. weise , _ ibid . _ * a672 * , 249 ( 2000 ) ; c.h . lee , g.e . brown , d .- min , and m. rho , _ ibid . _ * a585 * , 401 ( 1995 ) ; j.w . bos _ et al . d * 51 * , 6308 ( 1995 ) ; _ ibid . _ * 57 * , 4101 ( 1998 ) ; g. mller and u .- g . meissner , nucl . * b492 * , 379 ( 1997 ) . overseth , in review of particle properties , _ phys . lett . _ * 111b * , 286 ( 1982 ) . e. jenkins , nucl . phys . * b375 * , 561 ( 1992 ) . g. buchalla , a.j . buras , and m.k . harlander , nucl . phys . * b337 * , 313 ( 1990 ) . g. buchalla , a.j . buras , and m.e . lautenbacher , rev . * 68 * , 1125 ( 1996 ) . n. cabibbo , phys . * 10 * , 531 ( 1963 ) ; m. kobayashi and t. maskawa , prog . phys . * 49 * , 652 ( 1973 ) . l. wolfenstein , phys . * 51 * , 1945 ( 1983 ) . a. hocker , h. lacker , s. laplace , and f. le diberder , eur . j. c * 21 * , 225 ( 2001 ) . springer , arxiv : hep - ph/9508324 ; phys . b * 461 * , 167 ( 1999 ) ; b. borasoy and b.r . holstein , eur . j. c * 6 * , 85 ( 1999 ) . a. abd el - hady and j. tandean , phys . d * 61 * , 114014 ( 2000 ) . b. borasoy and b.r . holstein , phys . d * 59 * , 094025 ( 1999 ) ; e.m . henley , w.y . hwang , and l.s . kisslinger , nucl . * a706 * , 163 ( 2002 ) . chivukula , j.m . flynn , and h. georgi , phys . b * 171 * , 453 ( 1986 ) . a recent review can be found in e. de rafael , tasi 94 lectures , hep - ph/9502254 . j. gasser and h. leutwyler , annals phys . * 158 * , 142 ( 1984 ) . e. jenkins , nucl . * b368 * , 190 ( 1992 ) . donoghue , e. golowich , w.a . ponce , and b.r . holstein , phys . d * 21 * , 186 ( 1980 ) . m. ciuchini and g. martinelli , nucl . suppl . * 99b * , 27 ( 2001 ) . r. nath and a. kumar , nuovo cimento * xxxvi * , 1949 ( 1965 ) .	we calculate the @xmath0-violating asymmetries @xmath1 and @xmath2 in nonleptonic hyperon decay within the standard model using the framework of heavy - baryon chiral perturbation theory ( @xmath3pt ) . we identify those terms that correspond to previous calculations and discover several errors in the existing literature . we present a new result for the lowest - order ( in @xmath3pt ) contribution of the penguin operator to these asymmetries , as well as an estimate for the uncertainty of our result that is based on the calculation of the leading nonanalytic corrections . november 2002
measurements of the cosmic infrared background ( cirb ) probe the integrated emission from distant galaxies @xcite . the brightness and spectrum of the cirb is sensitive to galaxy number density , luminosity and evolution , as well as cosmology @xcite . however , a search for the cirb must account for the local zodiacal and stellar foregrounds . these have proven difficult to eliminate , even at @xmath14 @xmath0 m where the local foreground from thermal emission and scattering by zodiacal dust is minimal . dirbe reported upper limits on the cirb at near - infrared wavelengths @xcite , limited by foreground subtraction . the total astrophysical emission observed by dirbe in a @xmath15 beam contains a significant emission component from stars . dwek & arendt ( 1998 ) used ground - based @xmath11-band galaxy counts along with the dirbe data to obtain a tentative detection of the cirb . gorjian , wright & chary ( 2000 ) independently claimed a tentative detection of the cirb using ground - based @xmath16-band counts to account for stellar emission in the dirbe beams . wright & reese ( 2000 ) reported a detection using a different zodiacal model than used by hauser et al . matsumoto ( 2000 ) analyzed observations with the near infrared spectrometer ( nirs ) on irts , with an @xmath17 field of view that is significantly smaller than that of dirbe , and reported an isotropic near - infrared component peaking at @xmath18 1.5 @xmath0 m . brightness fluctuations also probe the cirb . the fluctuations contain information on the number counts and clustering properties of sources below the detection limit @xcite . poissonian fluctuations ( `` confusion noise '' ) are predicted based on models of galaxy evolution , including stellar and cirrus components from our galaxy @xcite . the brightness distribution depends on the index , @xmath19 , of integrated source counts @xmath20 , and can thus be used to probe the density of sources below the detection limit @xcite . the spatial distribution of fluctuations measures the clustering of sources @xcite on size scales of galaxy clustering . for example , nirs / irts observations show brightness fluctuations prominent at @xmath21 , appearing at the level of @xmath18 5% of the total diffuse sky brightness @xcite . confusion noise sets the limiting sensitivity for future space - borne infrared observations , such as for sirtf @xcite and astro - f @xcite . we present integrated source counts and a statistical analysis of residual brightness fluctuations in wide - field images obtained with a rocket - borne near - infrared camera at @xmath22 @xmath0 m . these images provide a direct account of the point source contribution to the near - infrared sky brightness , and an indirect measure of the brightness contribution from sources below this detection limit . the near - infrared telescope experiment ( nite ) consists of a 16 @xmath23 rocket - borne telescope cooled by supercritical helium and a 256 @xmath24 256 low - background insb focal plane array . the array is located at the focus of an @xmath25 gregorian telescope providing a @xmath26 field of view with a plate scale of @xmath2 per pixel . a 3.5 @xmath0 m long - wavelength pass infrared filter and the intrinsic long wavelength cutoff of the insb focal plane array defines the 3.5 to 5 @xmath0 m spectral band of the instrument . the array , sampled continuously and non - destructively `` up the slope '' at 3.8 @xmath27 over a 10 @xmath28 reset interval , gives near - background - limited noise performance and maximum flexibility in data reduction . the telescope and cryogenic baffle tube were cooled by supercritical liquid helium to eliminate thermal emission from the instrument . the focal plane array was operated at @xmath18 30 @xmath11 for optimal noise performance . further details on the instrument may be found in bock et al . ( 1998 ) . nite was designed to probe diffuse near - infrared emissions about nearby edge - on spiral galaxies , arising from possible low - mass stars in the massive halo . analysis of data from two flights , on 1997 may 28 centered about ngc 4565 , and on 1998 may 22 centered about ngc 5907 , produced strict upper limits on the near - infrared halo brightness @xcite . for purposes of assessing point source contributions to the near - infrared background , the two fields allow us to evaluate the relative contributions of point sources in the ngc 4565 ( @xmath29 , @xmath30 ; @xmath31 , @xmath32 ) and the ngc 5907 ( @xmath33 , @xmath34 ; @xmath35 , @xmath36 ) fields . we observed each region by stepping the field of view @xmath37 in a square pattern , and integrating for 20 @xmath28 in two 10 @xmath28 exposures at each position . this sequence produces a complete map consisting of 8 individual exposures . the observation sequence repeats with a slight shift of @xmath38 to sample the sky with different pixels . the central @xmath39 region around the target edge - on galaxy is thus observed in each of the 16 exposures . bright calibration stars , observed at the beginning and end of the sequence , and bright stars in the field allow us to calibrate the responsivity and point spread function ( psf ) of the instrument . further details on the observations for each flight may be found in uemizu et al . ( 1998 ) and yost et al . we present a statistical analysis from the 1997 observation about ngc 4565 . the 1998 observation of ngc 5907 , at lower galactic latitude ( @xmath33 ) , suffers from a relatively larger contribution from stars . furthermore , the 1998 observation has only 8 useful exposures due to a large terrestrial signal that appeared in the latter portion of the flight . we use the 1998 observations to assess the relative abundance of detected point sources at lower galactic latitude . the 1997 data has a noisy detector readout that samples every fourth column on the array . the cause of the excess noise , a loose wire - bond to the focal plane array , was subsequently repaired after the flight . one edge of the array was vignetted by a misalignment with the field stop . we carefully omit data from the noisy column and the vignetted region throughout the fluctuation analysis ( sections 3.2 , 3.3 , 3.4 , and 4.2 ) . we adopt the same reduction procedure as in uemizu et al . ( 1998 ) , briefly summarized as follows . we correct each of the 16 exposures for dark current by subtracting a master dark frame with a median dark current of 1.55 @xmath40 , obtained by closing an internal cryogenic ( 5 @xmath11 ) shutter just before launch . the flat field response is derived by first normalizing each individual exposure on the sky to its mean value . the mean is determined by applying an iterative 3@xmath41 clip to remove point sources . a flat field frame for each quadrant is then obtained by evaluating the median value in each pixel over 12 exposures . thus 4 flat field frames are generated for each of the 4 pointing positions used to observe the central galaxy . the flat field for each pixel in this reduction procedure is thus derived from observations of numerous patches of sky separated by @xmath18 @xmath42 from the observation position . we divide all of the exposures by their corresponding flat field frames , and subtract a small constant value from each exposure to account for a slowly varying terrestrial signal . the time - varying signal , @xmath18 17 @xmath40 , is small compared with the total photocurrent of 160 @xmath40 . the exposures are aligned to bright stars in the field , and combined to form a @xmath1 mosaic . we constructed a star mask from the mosaic , masking all pixels 3@xmath41 above the average over the image . bright point sources are each masked out to regions such that their extended emission is no more than 2% above the total sky background based on the psf . ngc 4565 is similarly masked based on the results of uemizu et al . ( 1998 ) . as the combined mosaic has a higher sensitivity to point sources than the individual exposures , we re - evaluate the flat field frames using the same procedure as with the mask obtained from the mosaic . the new flat field frames are then combined to produce a final mosaic . the resulting gray - scale image is shown in fig . [ fig : fov97 ] , including data from all four array readouts . the regions on the corners of the image are obtained with only 2 pointings ( 4 exposures ) , while the central region of the image is well sampled with 8 pointings ( 16 exposures ) . the image in fig . [ fig : fov97 ] has prominent stripes due to data from the noisy column ( for further evaluation of the statistical impact of the noisy columns see fig 4 ) . the remaining 3 array readouts are separately binned and found to have very similar noise properties , with read noises of 52 @xmath13 0.7 @xmath43 , 53 @xmath13 1 @xmath43 , and 53 @xmath13 0.7 @xmath43 for 10 @xmath28 integration time . we carefully omit data from the noisy column and the vignetted regions throughout the fluctuation analysis ( sections 3.2 , 3.3 , 3.4 , and 4.2 ) . for the sake of simplicity , we confine the variation analysis in sections 3.2 and 3.3 to pixels with 100% observing time . to improve our statistical power , we include pixels with 75% or more observing time for the analysis in sections 3.4 and 4.2 . identification of point sources in our field of view ( fig . [ fig : fov97 ] ) is conducted with a psf - fitting using daophot . bright calibration stars , observed at the beginning and end of the flight sequence , and bright stars in the field allow us to calibrate the responsivity and psf of the instrument . we iteratively extract sources by locating and masking the brightest sources in our field of view . the identified sources are individually masked out to regions where the extended psf brightness is @xmath44 2% of the mean sky brightness , @xmath45 @xmath4 @xmath5 @xmath6 . we use daophot to extract fluxes from identified sources . for consistency check , we compare the daophot fluxes with aperture photometry . in the case of very bright sources , we find some discrepencies , since a few are nearby extended galaxies . we thus use aperture photometry for the bright and extended sources . the resulting integrated source counts , detected with @xmath46 4@xmath41 significance , are displayed in fig . [ fig : num_count ] for the 1997 and 1998 fields . the number counts are not corrected for incompleteness at the faint end . the large error bars at the bright end are due to the small number of sources . we produce two independent sky images @xmath47 and @xmath48 from the first 8 and second 8 exposures respectively , to discriminate sources of noise from the instrument , sky , and variations in the gain matrix ( `` flat - field '' errors ) . we estimate the displacement between @xmath47 and @xmath48 accurately by comparing the centroids of bright stars . the resulting longitudinal and transverse displacements are @xmath49 0.3 and 0.2 @xmath50 respectively . as we shall see , subtracting the two images allows us to distinguish noise associated with the array , compared to fluctuations associated with a fixed pattern on the sky . the measured noise in a single 10 @xmath28 integration is 53 @xmath43 , in good agreement with photon noise , @xmath51 , and readout electronics noise , @xmath18 30 @xmath43 as determined form dark frames before launch , combined in quadrature , @xmath52 . instrument noise may also include various non - ideal noise sources at a low level , such as a fixed pattern noise on the array , low - level cosmic rays , incomplete subtraction of dark current , and time - varying signals . subtracting the @xmath47 and @xmath48 images , which are shifted in position relative to one another on the sky , therefore provides an accurate assessment of all sources of instrument noise which are necessarily correlated with array coordinates . to evaluate sky fluctuations statistically , we confine our analysis to the 1997 field about ngc 4565 . @xmath47 consists of a mosaiced image from the first eight exposures , and @xmath48 consists of a mosaiced image from the second eight exposures . we denote @xmath53 , the average of @xmath47 and @xmath48 with respect to sky coordinates , as the combined sky image . we mask @xmath47 and @xmath48 for 4@xmath41 sources with a common mask derived from @xmath54 , taking advantage of the higher sensitivity to point sources in the combined image @xmath54 . both @xmath47 and @xmath48 are corrected for flat - field variations as described above with a common gain matrix . we consider three sources of noise , instrument noise ( defined to include all un - correlated noise between exposures ) , errors in the gain matrix , and fluctuations on the sky . because the gain matrix is determined from observations of the sky , it can introduce statistical correlations in the data . we confine our analysis to the central @xmath55 region to eliminate statistical correlations and to minimize gain errors , `` flat - field noise '' , as follows . the majority of the pixels in the inner region of the combined image @xmath54 are obtained with 8 independent pixels on the array . the flat field for each of these pixels is obtained from 6 different patches of sky . in constructing @xmath54 , each point in the inner region is compared with 48 patches of sky located @xmath56 outside the inner region , 24 of which are independent . in the limit that the combined image @xmath54 is dominated by instrument noise , the variance of each pixel in the central region is increased by a factor of @xmath57 due to flat - field variations . in the limit that each exposure is dominated by sky fluctuations , the variance is increased by a factor of @xmath58 . as sky fluctuations and instrument noise are actually of similar amplitude in @xmath54 , the systematic variance of each pixel in the central region is increased by a factor of @xmath59 . the factor of @xmath59 is derived by noting that flat - field errors arise from the errors in the off beams , with instrument variance dominating over sky variance in the off beams by a factor of 4 . however , the variance in @xmath54 is @xmath18 2 times larger than instrument noise , so the relative flat - field variance is @xmath60 , or approximately @xmath59 . thus variations arising from flat field errors should generally be small in comparison to sky fluctuations in the inner region , and do not contribute to statistical correlations for @xmath61 . we assume the variance in @xmath62 arises from a combination of instrument noise ( @xmath63 ) , flat - field variations ( @xmath64 ) , and sky fluctuations ( @xmath65 ) . for the sake of simplicity , we further confine the following analysis to pixels with full observing time in the central @xmath55 region . the measured variance in @xmath54 is given by @xmath66 we note that sky fluctuations will have the same amplitude in @xmath47 and @xmath48 because @xmath47 and @xmath48 share a common flat field matrix . however , the gain for each pixel is derived from an independent arrangement of `` off '' positions , so flat field variations are independent from pixel to pixel . we assume that instrument noise is uncorrelated in the two images . the variance in @xmath47 and @xmath48 is thus given as @xmath67 we can separate these contributions by using the fact that @xmath47 and @xmath48 are displaced from one another by @xmath38 on the sky . by subtracting @xmath47 and @xmath48 with respect to sky coordinates , the variance in the differenced image @xmath68 becomes @xmath69 comparing the variance of @xmath54 and @xmath70 thus gives a direct measure of the sky fluctuations , @xmath71 @xmath72 @xmath73 @xmath74 . we evaluate the instrument noise contribution by subtracting image pairs viewing the sky 10 @xmath28 apart , and determine the variance to be @xmath75 @xmath72 @xmath73 @xmath74 . as 16 exposures comprise @xmath70 , the estimated contribution from instrument noise is @xmath76 we then estimate the flat - field contribution to be @xmath77 @xmath72 @xmath73 @xmath74 . this result agrees within experimental error with our expectation that the variance from flat - field errors is @xmath78 , as described above . furthermore , we can see that the instrument noise averages down with the number of exposures as expected . if the array had a fixed pattern noise that did not average down , it would be accounted as flat - field noise . however since the flat - field noise is in agreement with our expectation , such sources of excess noise must be small . the variance in each pixel of the combined image @xmath54 is thus marginally dominated by sky fluctuations , as shown in fig . [ fig : histogram ] . we may test the reliability of our estimation by differencing the same two masked sky images with respect to array coordinates . the resulting difference image @xmath79 is free of flat - field variation , since @xmath47 and @xmath48 share a common gain matrix . its variance becomes @xmath80 in agreement with our expected variance of 9.55 @xmath13 0.60 @xmath72 @xmath73 @xmath74 , obtained from eq . ( [ eqn : vara ] ) , ( [ eqn : varb ] ) , and ( [ eqn : varc ] ) above . this consistency check shows that flat - field errors , or a non - ideal fixed pattern noise associated with the array , are both small and accurately estimated . it also shows that any effect from possible sub - pixel displacement between @xmath47 and @xmath48 is small comparing to the statistical uncertainty in our variance estimation , since eq . ( [ eqn : varcheck ] ) does not require alignment on the sky between @xmath47 and @xmath48 . the detection of sky fluctuations can be shown graphically by evaluating the variance in @xmath81 as a function of the positional shift between @xmath47 and @xmath48 , as shown in fig . [ fig : difplot ] . pixels with 75% or more observing time were used in fig . [ fig : difplot ] to get more complete coverage , although this slightly increases @xmath63 above the value obtained in eq . ( [ eqn : varc ] ) . the variance is minimized when the two images are differenced according to sky coordinates , and does not appreciably decrease when differenced according to array coordinates ( correspondingly marked by the `` + '' in the @xmath82x -@xmath82y plane , fig . [ fig : difplot ] ) . the width of the minimum in the variance is consistent with the psf of the instrument . therefore the observed sky fluctuations are unlikely to arise from improperly masking the extended emission of ngc 4565 or scattered light from bright stars . as a check , we changed the masking level and found no significant change in the above results . also note that the striped pattern in the variance evaluation shown in fig . [ fig : difplot ] , an artifact arising from omitting data obtained with noisy readout columns , is small ( @xmath18 5% in amplitude ) . we estimate the spatial structure of the sky fluctuations using an auto - correlation analysis using the central pixels with full observing time . the two - point auto - correlation function @xcite , @xmath83 where @xmath84 , can be used to probe fluctuations on angular scales from @xmath2 to @xmath85 . we evaluate the correlation function on the summed ( @xmath54 ) and differenced ( @xmath70 ) images , described in section 3.1 . a constant offset is removed from the summed image before auto - correlation analysis , obtaining an ensemble average of zero in @xmath54 . sky , instrument , and flat - field variance contribute to the variance of @xmath54 , whereas @xmath70 contains no sky variance , only instrument and flat - field variance . since we only analyze the central @xmath86 region , we can remove any correlated structure as a result of flat - fielding . the resulting correlation functions are shown in fig . [ fig : autocorrelation ] . the correlation function at zero - lag , c(0 ) , gives the variance resulting from instrument , sky fluctuation , and flat - field contributions as described in section 3.2 . the gradual rise in @xmath87 near @xmath88 is consistent with a combination of instrument noise and sky fluctuations arising from unresolved sources , spatially varying as the psf of the instrument . @xmath89 shows no gradual rise , as expected for variance arising from instrument noise . we see no evidence for sky fluctuations at larger angular scales , @xmath90 . we constrain @xmath91 @xmath72 @xmath73 @xmath74 ( 2@xmath41 ) , averaged from @xmath92 . our non - detection of correlated structure can not exclude the detection by matsumoto ( 2000 ) on larger angular scales ( @xmath93 ) and shorter wavelengths ( 1.4 @xmath94 2.1 @xmath0 m ) , where rms fluctuations are significantly brighter . we have shown the presence of statistical fluctuations that are in excess of the instrument and flat - field noise sources . these fluctuations correlate with sky coordinates , not array coordinates . furthermore the correlation analysis shows these fluctuations are spatially described by point sources . no known non - ideal noise source on the detector array can mimic these properties . the correlation analysis also shows that sky fluctuations can not arise from structure in the zodiacal foreground . further information on the nature of the sky fluctuations may be evaluated from the statistical ensemble of pixels in the central region of the image . the brightness distributions of the 5000 central region pixels ( with 75% and more observing time ) are shown in fig . [ fig : histogram ] , after masking the galaxy , and detected point sources . the brightness distribution of summed image @xmath95 , @xmath96 , is noticeably skewed towards positive sources , and significantly wider than that of the differenced image @xmath97 , @xmath98 . the differenced histogram @xmath98 is well - described by a gaussian , which eliminates noise sources on the array that could produce a non - gaussian `` tail '' such as cosmic rays . the source counts presented in fig . [ fig : num_count ] appear to be dominated by stars . the counts may be parameterized as @xmath99 where @xmath100 @xmath6 , @xmath101 @xmath4 @xmath5 , and @xmath102 for the ngc 4565 field . the index of the counts , @xmath102 , is much shallower than that of deeper @xmath11-band galaxy counts , @xmath103 @xcite . the extended surface brightness due to detected point sources amounts to @xmath104 @xmath4 @xmath5 @xmath6 in the ngc 4565 field , and 10.0 @xmath4 @xmath5 @xmath6 in the ngc 5907 field . the ratio of the surface brightness of the two fields from detected sources , @xmath105 , is 1.50 , somewhat larger than that predicted by the stellar model , 1.38 , but consistent with the counts being largely stars and not galaxies . we are able to cross - identify 548 sources in the ngc 4565 field with digitized sky survey images . of the 548 cross - identified sources , 27 appear to be galaxies based on their measured fwhm . we can compare the source counts in the ngc 4565 and ngc 5907 fields to the counts from the model of wright & reese ( 2000 ) based on the wainscoat et al . ( 1992 ) galactic model , hereafter referred to as the w&r stellar model . nite represents a significant check on the model , probing the counts down to @xmath12 @xmath4 @xmath5 ( 0.8 @xmath106 ) in a wide field . as the w&r stellar model computes differential counts in standard @xmath16 and @xmath107 bands , we note that @xmath108 over the range @xmath16 = 5 to 15 . thus in order to estimate number counts in the nite band , we compute @xmath109 , where the flux @xmath110 is converted to @xmath4 @xmath5 in the nite band in the same manner as the calibration stars . the integrated source counts agree reasonably well with the model , and are shown in fig . [ fig : num_count]a . the integrated surface brightness , @xmath111 , where the flux @xmath110 is in @xmath112 units , is compared with the measured @xmath113 in fig . [ fig : num_count]b . an offset has been subtracted from the model by setting @xmath113 to the measured value at @xmath114 @xmath4 @xmath5 , the source detection limit . the offset must be subtracted to account for the surface brightness of bright , but rare , stars that increase the surface brightness of the model but were not present in the nite field . the w&r stellar model predicts that the surface brightness contribution for all sources below the detection threshold amounts to @xmath115 @xmath4 @xmath5 @xmath6 for the ngc 4565 field . thus we estimate that our infrared images detect 90% of the stellar surface brightness contribution . the combination of a good statistical understanding of sky fluctuations and steep number counts allows us to deeply probe the number counts of undetected sources . following condon ( 1974 ) , we may recover information on the density of sources below the detection limit , given our large statistical ensemble of 5000 pixels ( section 3.4 ) . we model the distribution as arising from gaussian instrument and flat - field noise , estimated in section 3.2 , and unresolved sources . the surface brightness fluctuations due to poissonian variations in a beam @xmath116 , where @xmath117 @xmath118 , @xmath119 may be used to constrain the surface brightness , @xmath120 , due to sources below the detection limit , @xmath121 thus for a source distribution described as a power law ( eq . [ [ eqn : power ] ] ) , fluctuations probe deeply into the background if the source counts are steep . sources brighter than the cutoff limit , @xmath122 , produce a measurable change in the brightness histogram according to @xmath123 where @xmath124 is the point source detection limit and @xmath125 is the change in the variance produced by all sources fainter than @xmath122 . since our measurement of the fractional uncertainty in the sky variance is @xmath126 ( @xmath127 @xmath72 @xmath73 @xmath74 , section 3.2 ) , the histogram is sensitive to sources much fainter than the detection limit @xmath124 . for example @xmath128 in the case of euclidean counts ( @xmath129 ) . if @xmath130 , fluctuations will be dominated by sources where the power law relationship eventually rolls over , in principle well below @xmath124 . the width and near - gaussian shape of the @xmath96 in fig . [ fig : histogram ] indicate a rapid increase in the source counts below the detection threshold . comparison with @xmath11-band galaxy counts ( @xmath131 ) also suggests the need for higher @xmath19 . therefore , we fit the data to two components , stellar counts taken from the w&r stellar model , and a power law population ( eq . [ [ eqn : power ] ] ) with undetermined @xmath132 and @xmath19 . we assume that the latter component represents galaxies . assuming the source counts are dominated by stars taken from the w&r stellar model , our simulation results in a poor approximation to the observed fluctuations , as shown by the solid line in fig . [ fig : bestfit ] . we carry out a p(d ) analysis to estimate the density of sources below the source detection limit , and constrain the associated background produced by these sources . we created a monte carlo sky to best estimate faint galaxy source count parameters @xmath132 and @xmath19 ( eq . [ [ eqn : power ] ] ) . the simulated sky contains point sources with flux @xmath133 . according to eq . ( [ eqn : cutoff ] ) , we must include all sources brighter than @xmath122 . including sources fainter than @xmath122 has no appreciable affect on the results . we begin our simulation with a @xmath134 field , gridded with @xmath135 pixels . we populate the grid with randomly placed delta functions , which we generate assuming a galaxy power law source counts ( eq . [ [ eqn : power ] ] ) , and the w&r stellar model , selected in steps of 0.1 magnitude ( corresponding to a multiplicative factor of @xmath136 @xmath4 @xmath5 ) . we then convolve the simulated image with the psf , and re - bin to @xmath2 pixels to match that of the detector array . finally we add gaussian noise with dispersion @xmath137 to each pixel to simulate instrument and flat - field noise . the simulated sky is then compared with the observed histograms @xmath96 and @xmath98 ( see fig . [ fig : histogram ] ) . we fit the histograms jointly , allowing @xmath132 , @xmath19 , @xmath138 , and @xmath137 to be free parameters . @xmath138 accounts for the difference in mean between the simulated and the observed skies . we must let @xmath138 be undetermined since our data contains a significant undetermined contribution from zodiacal emission . letting @xmath138 vary also means that our simulation is insensitive to the faintest sources ( @xmath139 ) applied in the monte - carlo simulation , since these sources add surface brightness but give no appreciable change in the histogram . we let @xmath137 vary to allow for uncertainty in the instrument and flat - field noise . the quantity @xmath137 has an experimental error ( see eq . [ [ eqn : varb ] ] ) that affects both @xmath96 and @xmath98 . fitting @xmath96 and @xmath98 jointly properly accounts for the uncertainty in @xmath137 , and essentially constrains @xmath137 to values within experimental errors . we determine the best fit ( reduced @xmath140 ) to the histograms for a given set of @xmath132 and @xmath19 values , marginalizing over @xmath137 and @xmath138 . we semi - analytically truncate the analysis to @xmath141 , based on eq . ( [ eqn : cutoff ] ) , to handle the case where the histogram is sensitive to sources fainter than our minimum monte - carlo source flux . we repeat simulations multiple times to accumulate statistical power and to reduce the numerical error in the simulation . we examine the goodness - of - fit over a wide range of @xmath132 and @xmath19 , as shown in fig . [ fig : error_ellipse ] , with 50 simulations performed for each point in the @xmath142 parameter space . our best - fit result , as shown by the dashed line in fig . [ fig : bestfit ] , and marked by the `` + '' in fig . [ fig : error_ellipse ] , gives @xmath143 @xmath6 , and @xmath144 , with @xmath145 fixed to the numerically convenient value of @xmath146 @xmath4 @xmath5 . the error ellipse shown in fig . [ fig : error_ellipse ] may be translated into @xmath147 and @xmath148 vs. @xmath110 plots by applying eq . ( [ eqn : power ] ) and ( [ eqn : cutoff ] ) , and propagating the errors appropriately . we evaluate the `` true '' cutoff fluxes using a monte - carlo simulation , and find them to be brighter than expected from eq . ( [ eqn : cutoff ] ) , presumably due to the addition of the fitted parameters @xmath138 and @xmath137 . we adopt the simulated cutoff fluxes for the 1@xmath41 and 2@xmath41 error contours , rather than those from eq . ( [ eqn : cutoff ] ) . we truncate the error bars when any portion of the error ellipse falls below the corresponding cutoff flux . the resulting constraints are shown in fig . [ fig : error_bands ] , along with the source count data from the ngc 4565 field . the associated errors on @xmath147 and @xmath148 , also shown in fig . [ fig : error_bands ] , improve at fainter fluxes because the error ellipse rotates to make the projected error in @xmath147 and @xmath148 smaller . we thus determine the integrated source counts @xmath10 @xmath6 , contributing @xmath8 @xmath4 @xmath5 @xmath6 to the cibr , evaluated at @xmath9 @xmath4 @xmath5 . we note that this monte - carlo technique is limited by the statistical power of the data . a more precise statistical determination of the histograms @xmath96 and @xmath98 would produce a smaller error ellipse in fig . [ fig : error_ellipse ] . this would allow us to exclude the smallest values of @xmath19 currently in the error ellipse , and consequently carry the curves in fig . [ fig : error_bands ] deeper since the maximum allowed cutoff flux would be lower . the values of @xmath132 and @xmath19 agree with the observed galaxy counts at @xmath11-band ( @xmath149 @xmath6 at @xmath150 @xmath4 @xmath5 , and @xmath131 ; saracco et al . 1997 ; bershady et al . 1998 ; saracco et al . 1999 ; jarrett et al . 2001 ) , giving an average color @xmath151 . the @xmath11-band counts show a roll - off at @xmath152 , which is not included in our source count model . alternately , we can assume reddened @xmath11-band galaxy counts ( @xmath151 ) in our simulation . the result is consistent with the histogram data and produces a background of 2 @xmath4 @xmath5 @xmath6 . as a check , we fit to the histogram @xmath96 and employ three point gauss - hermite quadrature summation @xcite to estimate the errors arising from variation in @xmath137 . the best - fit result is consistent with the joint analysis . dirbe produced an absolutely calibrated all sky map at 10 different ir wavelength from 1.25 @xmath0 m to 240 @xmath0 m . unfortunately , dirbe s @xmath153 beam is contaminated by galactic foreground stars . since it is difficult to observe large regions of sky to sufficient depth in the thermal near - infrared , models and fits of the galactic stellar flux have been the primary method used to remove the contamination from the dirbe maps @xcite . recently the two micron all sky survey ( 2mass ) provided the needed large - scale , high - resolution data to help determine the background at 1.25 @xmath0 m and 2.2 @xmath0 m @xcite . gorjian , wright , & chary ( 2000 ) also determined the brightness contribution from foreground stars at @xmath154 in selected dirbe fields , down to a limiting flux of @xmath155 @xmath4 @xmath5 . in comparison , these observations are 50 times deeper in @xmath112 than gorjian , wright , & chary ( 2000 ) and cover a wider field . galactic dust emission appears to be negligible in our field . arendt et al . ( 1998 ) established an empirical relation between residual galactic dust 100 @xmath0 m emission not associated with neutral hydrogen , and emission at other wavelengths . we use 16 dirbe pixels on the perimeter of our field to derive a 100 @xmath0 m intensity of 52.12 @xmath4 @xmath5 @xmath6 , of which 32.32 @xmath4 @xmath5 @xmath6 remains after extrapolating to @xmath156 . the inferred contribution from dust emission is completely negligible , 0.06 @xmath4 @xmath5 @xmath6 at 3.5 @xmath0 m and 0.09 @xmath4 @xmath5 @xmath6 at 4.9 @xmath0 m . likewise the emission from the edge - on spiral galaxy in the center of our field makes a small contribution to the surface brightness . although technically part of the extragalactic background , we exclude the flux , @xmath157 @xmath4 @xmath5 for ngc 4565 and @xmath158 @xmath4 @xmath5 for ngc 5907 , amounting to 0.75 and 0.43 @xmath4 @xmath5 @xmath6 averaged over the @xmath1 field , respectively , from the cirb . the image obtained by nite provides us with a large scale , high - resolution image , and so we can use it to subtract stellar flux from a dirbe map with zodiacal subtraction @xcite . we align the nite and dirbe fields , and use the nine dirbe pixels completely contained within the nite image to calculate the cirb . in table 1 we account for the stellar foreground in our two fields . we remove the small flux from galaxies resolved in poss plates from the stellar foreground . the color of the stellar foreground produced by @xmath159 stars is @xmath160 in the wright & reese ( 2000 ) model ( note the value for @xmath113 is slightly different than for @xmath161 ) . undetected stars , as noted in sections 4.1 and 4.2 , increase the stellar surface brightness by only 10% based on the stellar model . assuming this color , the total estimated stellar foreground is 11.6 @xmath13 0.5 ( 4.1 @xmath13 0.2 ) @xmath4 @xmath5 @xmath6 at 3.5 ( 4.9 ) @xmath0 m for the ngc 4565 field , and 17.4 @xmath13 0.8 ( 6.2 @xmath13 0.3 ) @xmath4 @xmath5 @xmath6 at 3.5 ( 4.9 ) @xmath0 m for the ngc 5907 field . in table 1 we estimate the residual emission , subtracting the stellar foreground from the zodi - subtracted dirbe sky brightness in our fields . we also account for the small contributions from detected galaxies , undetected stars , and the central edge - on galaxy . the resulting residuals are 24.0 @xmath13 2.7 ( 36.8 @xmath13 6.0 ) @xmath4 @xmath5 @xmath6 at 3.5 ( 4.9 ) @xmath0 m for the ngc 4565 field , and 14.0 @xmath13 2.6 ( 22.2 @xmath13 5.9 ) @xmath4 @xmath5 @xmath6 at 3.5 ( 4.9 ) @xmath0 m for the ngc 5907 field . the errors are dominated by the quoted zero - point errors from the zodiacal subtraction @xcite . the large discrepancy between the ngc 4565 and ngc 5907 fields may arise from the zodiacal subtraction . the ngc 4565 field lies at low ecliptic latitude ( @xmath162 ) where zodiacal residuals are more pronounced . we confirm that a bright star , prominent in the dirbe map and just outside of the ngc 4565 field , does not contribute significant emission to the dirbe pixels we used to estimate the sky brightness . the ngc 5907 residual agrees with residuals reported by several authors : wright & reese ( 2000 ) , @xmath163 @xmath4 @xmath5 @xmath6 at 3.5 @xmath0 m ( cirb detection ) , gorjian , wright & chary ( 2000 ) , @xmath164 @xmath4 @xmath5 @xmath6 at 3.5 @xmath0 m ( tentative cirb detection ) , and dwek & arendt ( 1998 ) , @xmath165 @xmath4 @xmath5 @xmath6 at 3.5 @xmath0 m , and @xmath166 @xmath4 @xmath5 @xmath6 at 4.9 @xmath0 m ( tentative cirb detection and upper limit , respectively , with updated values taken from hauser & dwek 2001 ) . the ngc 5907 residual also agrees with the residual brightness in the dirbe hqb regions , @xmath167 @xmath4 @xmath5 @xmath6 at 3.5 @xmath0 m and @xmath168 @xmath4 @xmath5 @xmath6 at 4.9 @xmath0 m , but hauser et al . 1998 report only upper limits on the cirb . although the ngc 5907 residual adds to the circumstantial evidence for a detection of the cirb , we can not test for isotropy in our small field . we estimate the cirb by accounting for the sources producing the observed surface brightness fluctuations . because we assume a truncated power law for the source counts , we can only constrain sources down to @xmath122 , and thus produce a lower limit on the cirb . the rapidly rising source counts required to explain the histogram are very clearly due to galaxies , as expected from @xmath11-band survey counts . we evaluate the detectable infrared background by integrating over sources down to the simulated 1@xmath41 cutoff flux , @xmath169 @xmath4 @xmath5 , 16.7 times below the point source detection flux , contributing @xmath170 @xmath4 @xmath5 @xmath6 to the cosmic infrared background . we analyze the point source contribution to deep , wide - field 4 @xmath0 m images . detected sources are dominated by stars , and are in general agreement with predictions from the wright & reese ( 2000 ) stellar model . we report a detection of surface brightness fluctuations detected with an ensemble of 5000 pixels . the brightness fluctuations do not show correlated structure and appear to be dominated by galaxies . we estimate the cirb arising from sources that contribute to the fluctuations . sirtf can use sky fluctuations to study the cosmic infrared background in greater detail , probing galaxy counts to fainter limits . shallow sirtf images , distributed over the sky , can remove the stellar foreground from dirbe , allowing a test of the isotropy of the residuals . we wish to thank the staff from the wallops flight facility and white sands missile range for their dedicated support during the two flights of the nite instrument . we also acknowledge the assistance of eric reese in providing us the predictions of the stellar model , and brendan crill for development of the telescope translation mechanism . steven price , giovanni fazio , judith pipher , and william forrest provided essential hardware and expertise for the development of the instrument . the authors thank bill reach , peter eisenhardt , and edward l. wright for several useful discussions . the cobe data sets were developed by the nasa goddard space flight center under the guidance of the cobe science working group and were provided by the nssdc . arendt , r. g. , et al . 1998 , , 508 , 74 bershady , m.a . , lowenthal , j.d . , and koo , d.c . 1998 , , 505 , 50 bock , j. j. , et al . 1998 , spie , 3354 , 1139 bond , j. r. , carr , b.j . , and hogan , c. j. 1986 , , 306 , 428 cambresy , l. , reach , w. t. , beichman , c.a . , and jarrett , t. h. 2001 , , 555 , 563 cole , s. , treyer , m .- a . , and silk , j. 1992 , , 385 , 9 condon , j. j. 1974 , , 188 , 279 dwek , e. , and arendt , r. g. 1998 , , 508 , l9 fabbri , r. 1988 , , 334 , 6 fanson , j. l. et al . 1998 , spie , 3356 , 478 franceschini , a. , toffolatti , l. , mazzei , p. , danese , l. , and de zotti , g. 1991 , a&as , 89 , 285 ganga , k. , cheng e. , meyer s. , and page l. 1993 , , 410 , l57 ganga , k. , ratra , b. , gundersen , j. o. , and sugiyama , n. 1997 , , 484,7 gispert , r. , lagache , g. , and puget , j. l. 2000 , a&a , 360 , 1 gorjian , v. , wright , e. l. , and chary , r. r. 2000 , , 536 , 550 hauser , m. g. , et al . 1998 , , 508 , 25 hauser , m.g . , and dwek , e. 2001 , ara&a , 39 , 249 jarrett , t.h . , van dyk , s. , and chester , t. j. 2001 , aas , 198 , 4911 kashlinsky , a. , and odenwald , s. 2000 , , 528 , 74 kaufman , m. 1976 , , 40 , 369 kelsall , t. , et al . 1998 , , 508 , 44 knox , l. , cooray , a. , eisenstein , d. and haiman , z. 2001 , , 550 , 7 matsumoto , t. 2000 , iaus , 204e , 8 onaka , t. , et al . 2000 , isass , 14 , 281 partridge , r.b . , and peebles , p.j.e . 1967 , , 148 , 377 saracco , p. , dodorico , s. , moorwood , a. , buzzoni , a. , cuby , j .- g . , and lidman , c. 1999 , , 349 , 751 saracco , p. , iovino , a. , garilli , b. , maccagni , d. , and chincarini , g. 1997 , , 114 , 887 stabell , r. , and wesson , p. s. 1980 , , 242 , 443 uemizu , k. , bock , j. j. , kawada , m. , lange , a. e. , matsumoto , t. , watabe , t. , and yost , s. a. 1998 , , 506 , l15 wainscoat , r. j. , cohen , m. , volk , k. , walker , h. j. , and schwartz , d. e. 1992 , , 83 , 111 wright , e.l . , and reese , e.d . 2000 , , 545 , 43 yost , s. a. , bock , j. j. , kawada , m. , lange , a. e. , matsumoto , t. , uemizu , k. , watabe , t. , and wada , t. 2000 , , 535 , 644 zwillinger , d. 1996 , standard mathematical tables and formulae , ( @xmath171 ed . ; crc press , inc . ) dirbe ( zodi sub.)&37.0 @xmath13 ( 1.7)(2.1)&@xmath172&41.5 @xmath13 ( 0.8)(5.9)&32.1 @xmath13 ( 1.3)(2.1)&@xmath172&28.7 @xmath13 ( 0.5)(5.9 ) + ism&0.06&@xmath172&0.09&0.04&@xmath172&0.06 + detected stars&10.5 @xmath13 0.5&6.7 @xmath13 0.3&3.7 @xmath13 0.2&15.7 @xmath13 0.8&10.0 @xmath13 0.5&5.6 @xmath13 0.3 + undetected stars&1.1&@xmath172&0.4&1.7&@xmath172&0.6 + detected galaxies&0.20&0.13&0.07&@xmath172&@xmath172&@xmath172 + galaxy&1.18&0.75&0.41&0.66&0.42&0.23 + residual&24.0 @xmath13 2.7&@xmath172&36.8 @xmath13 6.0&14.0 @xmath13 2.6&@xmath172&22.2 @xmath13 5.9	we present a measurement of faint source confusion in deep , wide - field 4 @xmath0 m images . the @xmath1 images with @xmath2 resolution are centered about the nearby edge - on spiral galaxies ngc 4565 and ngc 5907 . after removing statistical noise and gain fluctuations in the focal plane array , we measure spatial fluctuations in the sky brightness to be @xmath3 @xmath4 @xmath5 @xmath6 , approximately 1% of the diffuse background level observed in a single pixel . the brightness fluctuations are confirmed to be associated with the sky by subtracting sequential images of the same region . an auto - correlation analysis shows the fluctuations are well described by unresolved point sources . we see no evidence for surface brightness fluctuations on larger angular scales ( @xmath7 ) . the statistical distribution of brightness fluctuations allows us to estimate the density of sources below the detection limit . we present a monte - carlo analysis of the undetected sources , which appear to be dominated by galaxies . galaxies producing detected sky fluctuations contribute @xmath8 @xmath4 @xmath5 @xmath6 to the cosmic infrared background , evaluated at @xmath9 @xmath4 @xmath5 . from the fluctuation data we can determine the integrated source counts @xmath10 @xmath6 , evaluated at @xmath9 @xmath4 @xmath5 . the observed fluctuations are consistent with reddened @xmath11-band galaxy number counts . the number counts of extracted point sources with flux @xmath12 @xmath4 @xmath5 are dominated by stars and agree well with the galactic stellar model of wright & reese ( 2000 ) . removing the stellar contribution from dirbe maps with zodiacal subtraction results in a residual brightness of 14.0 @xmath13 2.6 ( 22.2 @xmath13 5.9 ) @xmath4 @xmath5 @xmath6 at 3.5 ( 4.9 ) @xmath0 m for the ngc 5907 field , and 24.0 @xmath13 2.7 ( 36.8 @xmath13 6.0 ) @xmath4 @xmath5 @xmath6 at 3.5 ( 4.9 ) @xmath0 m for the ngc 4565 field . the ngc 5907 residuals are consistent with tentative detections of the infrared background reported by dwek & arendt ( 1998 ) , wright & reese ( 2000 ) , and gorjian , wright & chary ( 2000 ) .
the observation of a d - wave superconducting gap in the high-@xmath6 cuprate superconductors suggests @xcite that strong correlations are responsible for their unconventional properties and superconducting behavior . the two - dimensional ( 2d ) @xmath7 model is one of the simplest effective models proposed @xcite to describe the low energy physics of these materials , @xmath8 the electrons are hopping between nearest neighbor sites of a square lattice leading to a kinetic energy term ( first term of [ htj ] ) as well as an exchange energy due to their spin interaction ( second term ) , where @xmath9 denotes the spin at site @xmath10 : @xmath11 and @xmath12 is the vector of pauli matrices . @xmath13 stands for a pair of nearest neighbors . @xmath14 operates only in the subspace where there are no doubly occupied sites , which can be formally implemented by a gutzwiller projector ( see later ) . in the following we set @xmath15 ( unless specified otherwise ) and we adopt a generic value of @xmath16 throughout the paper . because of the particle - hole symmetry in the square lattice the sign of @xmath17 does not play any role . although this model is formulated in a very simple form , the nature of the quantum correlations makes its physics very rich , and even the ground state of the @xmath7 hamiltonian was not yet characterized for finite doping and large cluster size . however , the @xmath7 model was investigated extensively by unbiased numerical techniques @xcite as well as by mean - field @xcite and variational monte - carlo approaches @xcite . all approaches found a d - wave superconducting phase and a phase diagram which accounts for most of the experimental features of the high-@xmath6 cuprates@xcite . in the limit of vanishing doping ( half - filling ) , such a state can be viewed as an ( insulating ) resonating valence bond ( rvb ) or _ spin liquid _ state . in fact , such a state can also be written ( after a simple gauge transformation ) as a staggered flux state ( sfp ) @xcite , i.e. can be mapped to a problem of free fermions hopping on a square lattice thread by a staggered magnetic field . upon finite doping , although such a degeneracy breaks down , the sfp remains a competitive ( non - superconducting ) candidate with respect to the d - wave rvb superconductor @xcite . in fact , it was proposed by p.a . lee and collaborators @xcite that such a state bears many of the unconventional properties of the pseudo - gap _ normal _ phase of the cuprate superconductors . this simple mapping connecting a free fermion problem on a square lattice under magnetic field @xcite to a correlated wave - function ( see later for details ) also enabled to construct more exotic flux states ( named as _ commensurate flux states _ ) where the fictitious flux could be uniform and commensurate with the particle density @xcite . in this particular case , the unit - cell of the tight - binding problem is directly related to the rational value of the commensurate flux . with an increasing number of materials and novel experimental techniques of constantly improving resolution , novel features in the global phase diagram of high - t@xmath0 cuprate superconductors have emerged . one of the most striking is the observation , in some systems , of a form of local electronic ordering , especially around @xmath5 hole doping . indeed , recent scanning tunnelling microscopy / spectroscopy ( stm / sts ) experiments of underdoped bi@xmath18sr@xmath18cacu@xmath18o@xmath19 ( bsco ) in the pseudogap state have shown evidence of _ energy - independent _ real - space modulations of the low - energy density of states ( dos ) @xcite with a spatial period close to four lattice spacings . a similar spatial variation of the electronic states has also been observed in the pseudogap phase of ca@xmath20na@xmath21cuo@xmath18cl@xmath18 single crystals ( @xmath22 ) by similar stm / sts techniques @xcite . although it is not clear yet whether such phenomena are either generic features or really happening in the bulk of the system , they nevertheless raise important theoretical questions about the stability of such structures in the framework of our microscopic strongly correlated models . in this paper , we analyze the stability and the properties of new inhomogeneous phases ( which may compete in certain conditions with the d - wave superconducting rvb state ) by extending the previous mean - field and variational treatments of the rvb theory . in addition , we shall also consider an extension of the simple @xmath7 model , the @xmath1 model , containing a coulomb repulsion term written as , @xmath23 where @xmath24 is the electron density ( @xmath25 , @xmath26 electrons on a @xmath27-site cluster ) . generically , we assume a screened coulomb potential : @xmath28 where we will consider two typical values @xmath29 and @xmath30 $ ] and where the distance @xmath31 is defined ( to minimize finite size effects ) as the _ periodized _ distance on the torus @xcite . the influence of this extra repulsive term in the competition between the d - wave rvb state and some inhomogeneous phases is quite subtle and will be discussed in the following . to illustrate our future strategy , let us recall in more details the simple basis of the rvb theory . it is based on a mean - field hamiltonian which is of bcs type , @xmath32 where @xmath33 is a constant variational parameter and @xmath34 is a nearest neighbor d - wave pairing ( with opposite signs on the vertical and horizontal bonds ) and @xmath35 is the chemical potential . as a matter of fact , the bcs mean - field hamiltonian can be obtained after a mean - field decoupling of the @xmath7 model , where the decoupled exchange energy leads to the @xmath33 and @xmath34 order parameters . in this respect , we expect that the bcs wave - function is a good starting point to approximate the ground state of the @xmath7 model . however , such a wave - function obviously does not fulfill the constraint of no - doubly occupied site ( as in the @xmath7 model ) . this can be easily achieved , at least at the formal level , by applying the full gutzwiller operator @xcite @xmath36 to the bcs wave - function @xmath37 : @xmath38 the main difficulty to deal with projected wave - functions is to treat correctly the gutzwiller projection @xmath39 . this can be done numerically using a conceptually exact variational monte carlo ( vmc ) technique @xcite on large clusters . it has been shown that the magnetic energy of the variational rvb state at half - filling is very close to the best exact estimate for the heisenberg model . such a scheme also provides , at finite doping , a semi - quantitative understanding of the phase diagram of the cuprate superconductors and of their experimental properties . novel results using a vmc technique associated to inhomogeneous wave - functions will be presented in section [ sec : vmc ] . another route to deal with the gutzwiller projection is to use a _ renormalized mean - field ( mf ) theory _ @xcite in which the kinetic and superexchange energies are renormalized by different doping - dependent factors @xmath40 and @xmath41 respectively . further mean - field treatments of the interaction term can then be accomplished in the particle - particle ( superconducting ) channel . crucial , now well established , experimental observations such as the existence of a pseudo - gap and nodal quasi - particles and the large renormalization of the drude weight are remarkably well explained by this early mf rvb theory @xcite . an extension of this approach @xcite will be followed in section [ sec : mf ] to investigate inhomogeneous structures with checkerboard patterns involving a decoupling in the particle - hole channel . as ( re- ) emphasized recently by anderson and coworkers @xcite , this general procedure , via the effective mf hamiltonian , leads to a slater determinant @xmath42 from which a correlated wave - function @xmath43 can be constructed and investigated by vmc . since the mf approach offers a reliable _ guide _ to construct translational symmetry - breaking projected variational wave - functions , we will present first the mf approach in section [ sec : mf ] before the more involved vmc calculations in section [ sec : vmc ] . we start first with the simplest approach where the action of the gutzwiller projector @xmath39 is approximately taken care of using a gutzwiller approximation scheme @xcite . we generalize the mf approach of ref . , to allow for non - uniform site and bond densities . recently , such a procedure was followed in ref . to determine under which conditions a @xmath4 superstructure might be stable for hole doping close to @xmath5 . we extend this investigation to arbitrary small doping and other kinds of supercells . in particular , we shall also consider 45-degree tilted supercells such as @xmath44 , @xmath45 and @xmath46 . the weakly doped antiferromagnet is described here by the renormalized @xmath7 model hamiltonian , @xmath47 where the local constraints of no doubly occupied sites are replaced by statistical gutzwiller weights @xmath48 and @xmath49 , where @xmath50 is the hole doping . a typical value of @xmath16 is assumed hereafter . decoupling in both particle - hole and ( singlet ) particle - particle channels can be considered simultaneously leading to the following mf hamiltonian , @xmath51 where the previous gutzwiller weights have been expressed in terms of local fugacities @xmath52 ( @xmath53 is the local hole density @xmath54 ) , @xmath55 and @xmath56 , to allow for small non - uniform charge modulations @xcite . the bogoliubov - de gennes self - consistency conditions are implemented as @xmath57 and @xmath58 . in principle , this mf treatment allows a description of modulated phases with _ coexisting _ superconducting order , namely supersolid phases . previous investigations @xcite failed to stabilize such phases in the case of the _ pure _ 2d square lattice ( i.e. defect - free ) . moreover , in this section , we will restrict ourselves to @xmath59 . the case where both @xmath60 and @xmath61 are non - zero is left for a future work , where the effect of a defect , such as for instance a vortex , will be studied . in the case of finite @xmath62 , the on - site terms @xmath63 may vary spatially as @xmath64 , where @xmath35 is the chemical potential and @xmath65 are on - site energies which are self - consistently given by , @xmath66 in that case , a constant @xmath67 has to be added to the mf energy . note that we assume here a _ fixed _ chemical potential @xmath35 . in a recent work @xcite , additional degrees of freedom where assumed ( for @xmath68 ) implementing an unconstrained minimization with respect to the on - site fugacities . however , we believe that the energy gain is too small to be really conclusive ( certainly below the accuracy one can expect from such a simple mf approach ) . we argue that we can safely neglect the spatial variation of @xmath35 in first approximation , and this will be confirmed by the more accurate vmc calculations in section [ sec : vmc ] . incidently , ref . emphasizes a deep connection between the stability of checkerboard structures @xcite and the instability of the sfp due to nesting properties of some parts of its fermi surface @xcite . in principle , the mean - field equations could be solved in real space on large clusters allowing for arbitrary modulations of the self - consistent parameters . in practice , such a procedure is not feasible since the number of degrees of freedom involved is too large . we therefore follow a different strategy . first , we assume fixed ( square shaped ) supercells and a given symmetry within the super - cell ( typically invariance under 90-degrees rotations ) to reduce substantially the number of parameters to optimize . incidently , the assumed periodicity allows us to conveniently rewrite the meanfield equations in fourier space using a reduced brillouin zone with a very small mesh . in this way , we can converge to either an absolute or a local minimum . therefore , in a second step , the mf energies of the various solutions are compared in order to draw an overall phase diagram . unit - cell used in both the mf approach and the variational wave - function . note the existence of 6 independent bonds ( bold bonds ) , that for convenience are labelled from @xmath69 , and of 3 _ a priori _ non - equivalent sites . the center of the dashed plaquette is the center of the ( assumed ) @xmath70 symmetry . other sizes of the same type of structure are considered in the mf case , respectively : @xmath71 , @xmath72 , and @xmath73 unit cells . ] in previous mf calculations @xcite , stability of an inhomogeneous solution with the @xmath4 unit - cell shown in fig . [ cell ] was found around @xmath74 . here , we investigate its stability for arbitrary doping and extend the calculation to another possible competing solution with a twice - larger ( square ) unit - cell containing 32 sites . the general solutions with different phases and/or amplitudes on the independent links will be refered to as bond - order ( bo ) phases @xcite . motivated by experiments @xcite , a @xmath70 symmetry of the inhomogeneous patterns around a central plaquette will again be assumed for both cases . note that such a feature greatly reduces the number of variational parameters and hence speeds up the convergence of the mf equations . starting from a central plaquette , like in fig . [ cell ] , a larger @xmath3 unit - cell ( not shown ) can easily be constructed with 10 non - equivalent bonds ( with both independent real and imaginary parts ) and _ a priori _ 6 non - equivalent sites . note that this new unit - cell is now tilted by 45 degrees . at this point , it is important to realize that patterns with a smaller number of non - equivalent bonds or sites are in fact subsets of the more general modulated structures described above . for example , the sfp is obviously a special case of such patterns , where all the @xmath75 are equal in magnitude with a phase oriented to form staggered currents , and where all the sites are equivalent . this example clearly indicates that the actual structure obtained after full convergence of the mf equations could have _ higher _ symmetry than the one postulated in the initial configuration which assumes a random choice for all independent parameters . in particular , the equilibrium unit - cell could be smaller than the original one and contain a fraction ( @xmath76 or @xmath77 ) of it . this fact is illustrated in fig . [ meanfield_phase ] showing two phase diagrams produced by using different initial conditions , namely @xmath78 ( top ) and @xmath73 ( bottom ) unit - cells . both diagrams show consistently the emergence of the sfp at dopings around @xmath79 and a _ plaquette _ phase ( @xmath71 unit - cell with two types of bonds ) at very small doping @xcite . in addition , a phase with a @xmath80 super - cell is obtained for a specific range of doping and @xmath62 ( see fig . [ meanfield_phase ] on the top ) . interestingly enough , all these bo phases can be seen as _ bond - modulated sfp _ with 2 , 4 , 8 and up to 16 non - equivalent ( staggered ) plaquettes of slightly different amplitudes . this would be consistent with the sfp instability scenario @xcite which suggests that the wavevector of the modulation should vary continuously with the doping . although this picture might hold when @xmath68 , our results show that the system prefers some peculiar spatial periodicities ( like the ones investigated here ) that definitely take place at moderate @xmath62 . lattice ( for @xmath81 ) vs hole doping @xmath50 and repulsion @xmath62 ( in units of @xmath82 ) . top : results obtained assuming a @xmath4 unit cell ; bottom : same with a @xmath83 tilted unit cell . in both cases , a @xmath84 symmetry is assumed ( see text).,title="fig : " ] lattice ( for @xmath81 ) vs hole doping @xmath50 and repulsion @xmath62 ( in units of @xmath82 ) . top : results obtained assuming a @xmath4 unit cell ; bottom : same with a @xmath83 tilted unit cell . in both cases , a @xmath84 symmetry is assumed ( see text).,title="fig : " ] let us now compare the two phase diagrams . we find that , except in some doping regions , the various solutions obtained with the @xmath4 unit - cell are recovered starting from a twice larger unit - cell . note that , due to the larger number of parameters , the minimization procedure starting from a larger unit - cell explores a larger phase space and it is expected to be more likely to converge to the absolute minimum . this is particularly clear ( although not always realized ) at large doping @xmath85 , where we expect an homogeneous fermi liquid ( fl ) phase ( all bonds are real and equal ) , as indeed seen in fig . [ meanfield_phase ] on the bottom . on the contrary , [ meanfield_phase ] on the top reveals , for @xmath86 $ ] , a modulated @xmath80 structure , which is an artefact due to the presence of a local minimum ( see next ) . and for @xmath87 ) obtained by solving the mean - field equations using the _ initial _ @xmath4 unit - cell ( see text ) for a moderate value of @xmath62 . the sfp energy is also shown for comparison . the fl energy has been substracted from all data for clarity . ( b ) comparison of the energies ( for @xmath68 ) using different initial conditions ( see text ) , a @xmath4 or a @xmath3 unit - cell ; due to very small energy differences , the sfp energy is used as a reference for an easier comparison . the different phases specified by arrows and characterized by the number of sites @xmath88 of their actual supercells refer to the ones in fig . [ meanfield_phase ] . for doping @xmath85 , the minimization leads to a solution with small imaginary parts ( of order @xmath89 ) very similar to a fl phase , which we call fl@xmath90 . ] and for @xmath87 ) obtained by solving the mean - field equations using the _ initial _ @xmath4 unit - cell ( see text ) for a moderate value of @xmath62 . the sfp energy is also shown for comparison . the fl energy has been substracted from all data for clarity . ( b ) comparison of the energies ( for @xmath68 ) using different initial conditions ( see text ) , a @xmath4 or a @xmath3 unit - cell ; due to very small energy differences , the sfp energy is used as a reference for an easier comparison . the different phases specified by arrows and characterized by the number of sites @xmath88 of their actual supercells refer to the ones in fig . [ meanfield_phase ] . for doping @xmath85 , the minimization leads to a solution with small imaginary parts ( of order @xmath89 ) very similar to a fl phase , which we call fl@xmath90 . ] since the mf procedure could accidentally give rise to local minima , it is of interest to compare the mf energies obtained by starting with _ random _ values of all independent parameters within the two previously discussed unit - cells . for convenience , we have substracted from all data either the fl ( in fig . [ meanfield_energy](a ) ) or the sfp ( in fig . [ meanfield_energy](b ) ) _ reference _ energy . from figs . [ meanfield_energy](a , b ) we see that we can converge towards a local energy minimum , often modulated in space , which is not the absolute minimum . indeed , over a large doping range , the lowest energy of the all solutions we have found is obtained for homogeneous densities and bond magnitudes . nevertheless , we see that the @xmath4 modulated phase is ( i ) locally stable and ( ii ) is very close in energy to the homogeneous ( sfp ) phase which , often , has a slightly lower energy . note that , around @xmath91 , the states with @xmath80 and @xmath3 supercells are clearly metastable solutions ( and using a larger initial unit - cell is not favorable in the latter case ) . in contrast , in this range of doping , the @xmath4 checkerboard state is very competitive w.r.t . therefore , it makes it a strong candidate to be realized either in the true ground state of the model , or present as very low excited state @xcite . in fact , considering such small energy differences , it is clear that an accurate comparison is beyond the accuracy of the mf approach . we therefore move to the _ approximation - free _ way of implementing the gutzwiller projection with the vmc technique , that allows a detailed comparison between these variational homogeneous and inhomogeneous states . motivated by the previous mean - field results we have carried out extensive variational monte carlo simulations . in this approach , the action of the gutzwiller projection operator is taken care of exactly , although one has to deal with finite clusters . in order to get rid of discontinuities in the d - wave rvb wave - function , we consider ( anti-)periodic boundary conditions along @xmath92 ( @xmath93 ) . as a matter of fact , it is also found that the energy is lower for twisted boundary conditions , hence confirming the relevance of this choice of boundaries . we have considered a @xmath94 square cluster of @xmath95 sites . we also focus on the @xmath5 doping case which corresponds here to @xmath96 electrons on the 256 site cluster . following the previous mf approach , we consider the same generic mean - field hamiltonian , @xmath97 where the complex bond amplitudes @xmath98 can be written as @xmath99 , and @xmath100 is a phase oriented on the bond @xmath101 . the on - site terms @xmath63 allow to control the magnitude of the charge density wave . however , the energy was found to be minimized for all the @xmath63 equal to the same value in the range @xmath102 $ ] and for the two parameters @xmath29 . in fact , we find that strong charge ordered wave - functions are not stabilized in this model @xcite . in this section , we shall restrict ourselves to the @xmath78 unit - cell where all independent variational parameters are to be determined from an energy minimization . this is motivated both by experiments @xcite and by the previous mf results showing the particular stability of such a structure ( see also ref . ) . as mentioned in the previous section , we also impose that the phases and amplitudes respect the @xmath70 symmetry within the unit - cell ( with respect to the center of the middle plaquette , see fig . [ cell ] ) , reducing the numbers of independent links to @xmath103 . to avoid spurious degeneracies of the mf wave - functions related to multiple choices of the filling of the discrete k - vectors in the brillouin zone ( at the fermi surface ) , we add very small random phases and amplitudes ( @xmath104 ) on all the links in the @xmath105 unit cell . let us note that commensurate flux phase ( cfp ) are also candidate for this special @xmath5 doping . in a previous study , a subtle choice of the phases @xmath100 ( corresponding to a gauge choice in the corresponding hofstadter problem @xcite ) was proposed @xcite , which allows to write the @xmath106 flux per plaquette wave - function within the same proposed unit - cell @xcite and is also expected to lead to a better kinetic energy than the landau gauge ( in the landau gauge the unit - cell would be a line of @xmath107 sites ) . however , we have found that the cfp wave - functions turned out _ not _ to be competitive for our set of parameters @xmath62 , due to their quite poor kinetic energy , although they have very good coulomb and exchange energies . we argue that such cfp wave - functions would become relevant in the large coulomb and/or @xmath82 regimes ( see table [ energies ] ) . in order to further improve the energy , we also add a nearest - neighbor spin - independent jastrow @xcite term to the wave - function , @xmath108 where @xmath109 is an additional variational parameter . finally , since the @xmath110 model allows at most one fermion per site , we discard all configurations with doubly occupied sites by applying the complete gutzwiller projector @xmath111 . the wave - function we use as an input to our variational study is therefore given by , @xmath112 in the following , we shall introduce simple notations for denoting the various variational wave - functions , @xmath113 for the bond - order wave function , @xmath114 for the staggered flux phase , @xmath115 for the d - wave rvb superconducting phase , @xmath116 for the simple projected fermi - sea , and we will use the notation @xmath117 ( @xmath118 ) when the jastrow factor is applied on the mean - field wave - function . finally , it is also convenient to compare the energy of the different wave - functions with respect to the energy of the simple projected fermi - sea ( i.e. the correlated wave - function corresponding to the previous fl mean - field phase ) , therefore we define a _ condensation energy _ as @xmath119 . .set of energies per lattice site for @xmath120 and @xmath81 for different wave - functions . the best commensurate flux phase in the landau gauge with flux per plaquette @xmath121 was found for @xmath122 . we also show the energy of the cfp with flux @xmath123 written with another choice of gauge . we show the total energy per site ( @xmath124 ) , the kinetic energy per site ( @xmath125 ) , the exchange energy per site ( @xmath126 ) and the coulomb energy per site ( @xmath127 ) . [ cols="^,^,^,^,^",options="header " , ] nevertheless , we observe that the difference in energy between the bond - order wave - function and the staggered flux phase remains very small . we show in table [ data ] the order parameters measured after the projection for the @xmath128 , @xmath129 and @xmath130 wave - functions . as expected the @xmath128 and the @xmath129 wave - functions are homogenous within the unit - cell . in contrast , the @xmath130 wave - function shows significant modulations ( expected to be measurable experimentally ) of the various bond variables w.r.t their values in the homogeneous sfp . in fig . [ bo_sfp_etot ] we show the small energy difference ( see scale ) between the two wave - functions . interestingly , the difference is increasing with the strength of the potential . we notice that the two wave - functions correspond to two nearby local minima of the energy functional at zero coulomb potential ( see fig . [ barrier ] ) , which are very close in energy ( the @xmath130 wave - function is slightly lower in energy than the @xmath129 ) and are separated by a quite small energy barrier . note that in fig . [ barrier ] we consider the variational bond order parameters and not the projected quantities . when the repulsion is switched on , the height of the energy barrier increases and the @xmath129 wave - function does not correspond anymore to the second local minima . indeed , when @xmath131 the second local energy minima _ shifts _ continuously from the point corresponding to the simple @xmath129 wave - function . the metastable wave - function lying at this second local minima is a weak bond - order ( sfp - like ) wave - function that preserves better the large kinetic energy while still being able to optimize better the coulomb energy than the homogeneous sfp . moreover , to understand better the stabilization of the bo - modulated staggered flux wave - functions w.r.t the homogeneous one , we have plotted in fig . [ sfp_bo ] the difference in the respective kinetic energy , the exchange energy and the coulomb energy of the @xmath129 and @xmath130 wave - functions . we conclude that the two wave - functions , although qualitatively similar ( they both exhibit an underlying staggered flux pattern ) , bear quantitative differences : the staggered flux phase ( slightly ) better optimizes the kinetic energy whereas the bond - order wave - function ( slightly ) better optimizes the coulomb and exchange energies so that a small overall energy gain is in favor of the modulated phase . therefore , we unambiguously conclude that , generically , bond - order modulations should spontaneously appear on top of the staggered flux pattern for moderate doping . finally , we emphasize that the bond - order wave - function is not stabilized by the coulomb repulsion alone ( like for a usual electronic wigner cristal ) exhibiting coexisting bond order and ( small ) charge density wave . moreover , the variational parameters @xmath63 in eq . ( [ mf_hamilt ] ) are found after minimizing the projected energy to be set to equal values on every site of the unit - cell . let us also emphasize that the bond - order wave - function is not superconducting as proposed in some scenarios @xcite . in the actual variational framework , we do not consider bond - order wave - function embedded in a sea of d - wave spin singlet pairs . in fact , we do not expect a bulk d - wave rvb state to be stable at large coulomb repulsion ( because of its very poor coulomb energy ) nor a bulk _ static _ checkerboard sfp at too small coulomb energy . however , for moderate coulomb repulsion for which the d - wave rvb remains globally stable , sizeable regions of checkerboard sfp could be easily nucleated e.g. by defects . this issue will be addressed using renormalized mf theory in a future work . an extension of our vmc study with simultaneous inhomogeneous bond - order and singlet pair order parameters ( as required to treat such a problem ) is difficult and also left for a future work . note also that low - energy _ dynamic fluctuations _ of checkerboard ( and sfp ) characters could also exist within the d - wave rvb state but this is beyond the scope of this paper . the properties of the @xmath130 staggered flux wave - function are summarized in fig . [ snapshots ] showing the real and imaginary parts of the measured hopping term @xmath132 between every nearest neighbor sites of our candidate @xmath130 wave - function . we also present the exchange term on each bonds of the lattice , and the local on - site charge density . we find that the bond - order wave - function has both ( spin - spin ) bond density wave and ( small ) charge density wave components . nonetheless , the charge modulations are very small ( the maximum charge deviation from the mean on - site charge is of the order of @xmath133 ) , and the charge density is a little bit larger in the center of the unit - cell . as expected , the @xmath129 has homogeneous hopping and exchange bonds within the unit - cell . therefore , we conclude that after projection the modulated variational wave - function differs quantitatively from the uniform one : the @xmath130 staggered flux wave - function is quite inhomogeneous ( although with a very small charge modulation ) leading to an increased magnetic energy gain while still preserving a competitive kinetic energy , a characteristic of the homogeneous @xmath129 wave - function . in conclusion , in this work we have investigated the @xmath134 model using both mean - field calculations as well as more involved variational monte - carlo calculations . both approaches have provided strong evidence that bond - order wave - functions ( of underlying staggered flux character ) are stabilized at zero and finite coulomb repulsion for doping close to @xmath5 . in particular , variational monte - carlo calculations show that a bond modulation appears spontaneously on top of the staggered flux phase . this is in agreement with the work of wang et al . @xcite predicting an instability of staggered flux type . we have also shown that the modulated and homogeneous sfp , although nearby in parameter space , are nevertheless separated from each other by a small energy barrier . while both staggered flux wave - functions provide an optimal kinetic energy , the bond - modulated one exhibits a small extra gain of the exchange energy . on the other hand , a short range coulomb repulsion favors both staggered flux wave - function w.r.t the d - wave rvb superconductors and brings them close in energy . finally , it would be interesting to study if the checkerboard pattern could spontaneously appear in the vicinity of a vortex in the mixed phase of the cuprates . such an issue could be addressed by studying the @xmath134 model on a square lattice extending our variational scheme to include _ simultaneously _ nearest neighbor pairing and bond modulated staggered currents . it is expected that , while the pairing is suppressed in the vicinity of the vortex , the checkerboard pattern might be variationally stabilized in this region . lee , n. nagaosa , t.k . ng and x.g . wen , phys . b * 57 * , 6003 ( 1998 ) . wen and p.a . lee , phys . lett . * 76 * , 503 ( 1996 ) . ubbens and p.a . lee , phys . rev . b * 46 * , 8434 ( 1992 ) . hofstadter , phys . b. * 14 * , 2239 ( 1976 ) . anderson , b.s . shastry and d. hristopulos , phys . b. * 40 * , 8939 ( 1989 ) ; d. poilblanc , _ ibid . _ * 40 * , r7376 ( 1989 ) ; p. lederer , d. poilblanc and t.m . rice , phys . lett . * 63 * , 1519 ( 1989 ) ; for related results using slave boson mf techniques see e.g. f. nori , e. abrahams and g.t . zimanyi , phys . b. * 41 * , r7277 ( 1990 ) . m. vershinin , s. misra , s. ono , y. abe , y. ando , and a. yazdani , science * 303 * , 1995 ( 2004 ) . note that the first observation was made around a vortex core in j.e . hoffman , e.w . hudson , k.m . lang , v. madhavan , h. eisaki , s. uchida , and j.c . davis , science * 295 * , 466 ( 2002 ) . note that the _ energy - dependent _ spatial modulations of the tunneling conductance of optimally doped bsco can be understood in terms of elastic scattering of quasiparticles . see j.e . hoffman , k. mcelroy , d.h . lee , k.m . lang , h. eisaki , s. uchida , and j.c . davis science * 297 * 1148 ( 2002 ) . our @xmath4 solution bears some similarities with those obtained within @xmath137/@xmath138 mean field theories ; see m. vojta , phys . b. * 66 * , 104505 ( 2002 ) . note however that the large - n sp(2n ) scheme implies a superconducting state .	motivated by the observation of inhomogeneous patterns in some high - t@xmath0 cuprate compounds , several variational gutzwiller - projected wave - functions with built - in charge and bond order parameters are proposed for the extended @xmath1 model on the square lattice at low doping . first , following a recent gutzwiller - projected mean - field approach by one of us ( phys . rev . b. * 72 * , 060508(r ) ( 2005 ) ) , we investigate , as a function of doping and coulomb repulsion , the stability of the staggered flux phase with respect to small spontaneous modulations of squared unit cells ranging from @xmath2 to @xmath3 . it is found that a @xmath4 bond - order ( bo ) modulation appears spontaneously on top of the staggered flux pattern for hole doping around @xmath5 . a related wave - function is then constructed and optimized accurately and its properties studied extensively using an approximation - free variational monte carlo scheme . finally , the competition of the bo - modulated staggered flux wave - function w.r.t . the d - wave rvb wave - function or the commensurate flux state is investigated . it is found that a short range coulomb repulsion penalizes the d - wave superconductor and that a moderate coulomb repulsion brings them very close in energy . our results are discussed in connection to the stm observations in the under - doped regime of some cuprates .
the cold dark matter ( cdm ) scenario is now the standard model for structure formation . the `` concordance '' @xmath14 model ( e.g. , ostriker et al . 1995 ) is supported by many observations , in particular the large - scale structure of the universe ( e.g. jenkins et al . 1998 ; peacock et al . 2001 ; tegmark et al . 2004 ) and the cosmic microwave background ( e.g. , spergel et al . 2003 ) . giant arcs are formed when background galaxies are distorted into long arc - like shapes by the gravitational shear of intervening clusters of galaxies . they are among the most beautiful images on the sky ( e.g. , see the hubble space telescope [ hst ] images of a2218 , kneib et al . 1996 ) . as clusters of galaxies are the largest bound structures in the universe , they are at the tail of the mass function of bound structures . the formation of giant arcs therefore critically depends on the abundance and density profiles of clusters . observationally , the number of giant arcs was first estimated using the einstein medium - sensitivity survey ( emss ) by gioia & luppino ( 1994 ) . the largest dedicated search for giant arcs in x - ray clusters was performed by luppino et al . ( 1999 ) who found strong lensing in 8 out of 38 clusters . these fractions were confirmed by zaritsky & gonzalez ( 2003 ) using the las campanas distant cluster survey and gladders et al . ( 2003 ) using the red - sequence cluster survey . a recent extensive analysis of hst images found 104 candidate tangential arcs in 128 clusters with length to width ratios exceeding 7 ( sand et al . 2005 ) ratio is similarly defined as in dalal et al . ( 2004 ) where giant arcs are fitted with a rectangle rather than an ellipse , so their @xmath5 ratio is @xmath15 our value . ] . all these recent studies show that giant arcs are quite common in clusters of galaxies . earlier predictions of giant arcs use simple spherical models ( e.g. , wu & hammer 1993 ) . however , such models are clearly inadequate , as the ellipticities and substructures of clusters can enhance the lensing cross - section by a large factor ( bartelmann & weiss 1994 ; bartelmann , steinmetz & weiss 1995 ) . more recently , torri et al . ( 2004 ) studied the importance of mergers for arc statistics . wu & mao ( 1996 ) compared the arc statistics in the einstein - de sitter universe and @xmath0cdm using simple spherical cluster models and found a factor of @xmath16 increase in the @xmath0cdm model . a much more realistic study was performed by bartelmann et al . ( 1998 , hereafter b98 ) who first pointed out that the number of giant arcs in the @xmath14 model appears to be below the observed rate in clusters by a factor of 5 - 10 . meneghetti et al . ( 2000 ) investigated the effects of cluster galaxies on arc statistics and found that cluster galaxies do not introduce perturbations strong enough to significantly change the number of giant arcs . flores , maller & primack ( 2000 ) drew a similar conclusion . meneghetti et al . ( 2003 ) further examined the effect of cd galaxies on the giant arcs and concluded they are insufficient to resolve the discrepancy . oguri et al . ( 2003 ) studied the ability of tri - axial models of dark matter halos to form giant arcs . they concluded that an inner power - law profile of @xmath17 can reproduce the observed giant arcs , while the standard nfw profile ( navarro , frenk , & white 1997 ) with an inner profile of @xmath18 is unable to match the observations . bartelmann et al . ( 2003 ) and meneghetti et al . ( 2005 ) studied the probability of the formation of giant arcs in galaxy clusters in different dark - energy cosmologies . but the effect is still insufficient . in an important paper , wambsganss et al . ( 2004 ) pointed out the simple fact that the probability of high lensing magnifications is very sensitive to the assumed source redshift . for example , a source at @xmath19 has a factor of @xmath20 higher optical depth than a source at redshift @xmath6 . as the redshift distribution for the background source population that forms arcs is not well known , this introduces an uncertainty in the comparison between observations and predictions . in this paper , we use high - resolution simulations to re - examine the predicted number of giant arcs and compare our predictions with observations . the outline of the paper is as follows . in 2 , we discuss the numerical simulations we use and the method we employ to identify giant arcs in simulated clusters . in 3 , we present the main results of the paper and compare them with several previous studies ( bartelmann et al . 1998 ; wambsganss et al . 2004 ; dalal et al . in particular , we confirm the results of wambsganss et al . ( 2004 ; see also dalal et al . 2004 ) that the predicted arc number is sensitive to the assumed source redshift distribution . however , our overall optical depth appears to be lower than these previous studies , sometimes by a large factor . we examine in detail how the differences arise . this reduction makes it somewhat difficult to match the observed high frequency of giant arcs with our predictions . in 4 , we summarize our results and discuss areas for future improvements . the cosmological model considered here is the current popular @xmath0cdm model with the matter - density parameter @xmath21 and the cosmological constant @xmath22 . the shape parameter @xmath23 and the amplitude @xmath10 of the linear density power spectrum are taken to be 0.21 and 0.9 , respectively , where @xmath24 is the hubble constant in units of @xmath25 and we take @xmath26 . a cosmological n - body simulation with a box size @xmath27 , which was generated with our vectorized - parallel @xmath28 code ( jing & suto 2002 ; jing 2002 ) , is used in this paper . the simulation uses @xmath29 particles , so the particle mass @xmath30 is @xmath31 . the gravitational force is softened with the @xmath32 form ( hockney & eastwood 1981 ) with the softening parameter @xmath33 taken to be @xmath34 . the simulation has similar mass and force resolutions as that used by b98 , but our simulation volume is 8 times larger , and hence contains @xmath35 times more clusters . compared with the simulations of wambsganss et al . ( 2004 , hereafter w04 ) , our resolutions are a factor of @xmath36 lower in mass and a factor of 10 lower in the force softening . since strong arcs are produced mostly at radii @xmath37 in the lens plane , the resolutions are sufficient ( see also dalal et al . this is also supported by the comparison of our results with those of w04 which are based on much higher resolution our predicted number of giant arcs is in reasonable agreement with that of w04 if the arcs are identified with their method ( see [ sec : comparisonsw04 ] and fig . [ fig : tau ] ) . dark matter halos are identified with the friends - of - friends method using a linking length equal to 0.2 times the mean particle separation . the halo mass @xmath38 is defined as the virial mass enclosed within the virial radius according to the spherical collapse model ( kitayama & suto 1996 ; bryan & norman 1998 ; jing & suto 2002 ) . for convenience of later discussions , we outline the notations we will use in the paper . we denote the lensing potential as @xmath39 . the mapping from the lens and source plane is given by = - , and the distortion of images is then described by the jacobian = ( cc 1-_11 & -_12 + -_21 & 1-_22 + ) , where @xmath40 are dimension - less source coordinates in the source plane , and @xmath41 are dimension - less image coordinates in the lens plane . the jacobian matrix is given in terms of the lensing potential by the @xmath42 matrix on the right hand side , which involves second - order derivatives of @xmath39 with respect to @xmath43 and @xmath44 . the eigenvalues of the jacobian matrix are denoted as @xmath45 and @xmath46 . without losing generality , we will assume @xmath47 . the ( signed ) magnification is given by = 1 _ 1 _ 2 . for an infinitesimal circular source , the length to width ratio is simply given by = 1/\|_2\| = _ [ eq : lw - mu ] notice that @xmath5 is not equal to @xmath48 if @xmath45 is not unity . the lensing properties of numerical clusters are studied using the ray - tracing technique ( e.g. , b98 ; w04 ) . for the source population , we use redshifts ranging from 0.6 to 7 with a uniform interval of 0.4 . to calculate the optical depth , we use 20 outputs for the simulated box from redshift 0.1 to 2.35 with a step size of @xmath49 . for each redshift , we select the 200 most massive clusters of galaxies as lenses . notice that this cluster sample is more than a factor ten larger than the samples used by bartelmann et al . ( 1998 ) and ho & white ( 2004 ) , although our redshift samplings are sparser than the most recent study of torri et al . ( 2004 ) . for each cluster , we use the particles within a cube with a side - length of two virial radii ( @xmath50 ) . the surface densities are then calculated for three orthogonal projections using the sph smoothing algorithm ( monaghan 1992 ) on a @xmath51 grid . usually , the kernel size is taken to be @xmath34 ( comoving ) . if the particle number within the kernel is fewer than 64 , then we double the kernel size until the particle number is larger than 64 . however , for the high density regions , when the particle number is larger than 400 within the kernel , we only use the nearest 400 particles to estimate the density . in order to obtain the lensing potential , we also use a larger grid ( @xmath52 ) centered on the smaller grid ; the surface densities are padded with zeros for all the pixels outside the inner grid . the lensing potential is then obtained using the fft method ( b98 ) . notice that the larger grid is used to avoid aliasing problems due to periodic boundary conditions . to perform efficient lensing simulations , we first identify regions of interest in the lens plane that have magnifications exceeding @xmath53 and regions with @xmath54 ; other regions are unlikely to produce giant arcs with @xmath5 exceeding 7.5 . following b98 , we assume the sources have an ellipticity , which is equal to one minus the axis ratio , randomly distributed between 0 to 0.5 . each source has an area @xmath55 equal to a circular source with a diameter of @xmath1 . to have sufficient numbers of pixels in one image , the resolution in the regions of interest is increased to @xmath56 ( see below ) so that most images have at least @xmath57 pixels . for the regions identified above , the @xmath51 ( coarse ) grid does not provide sufficient resolution . to remedy this , we obtain the lensing potential on a finer grid with resolution of @xmath56 using cubic spline interpolation of the surrounding @xmath58 coarse grid points . in the same step , we obtain the source position for each grid point , and the corresponding magnification and eigenvalues ( @xmath45 and @xmath46 ) . once we obtain the lens mapping for the regions of interest from the lens plane to the source plane , we locate the smallest rectangle with area @xmath59 that contains all the mapped pixels in the source plane . we then put sources randomly in this rectangle and obtain their imaging properties . the mapped regions in the source plane usually have irregular shapes , so the rectangle will enclose not only the regions of interest but also some regions that do not satisfy our selection criteria ( @xmath60 or @xmath61 ) . to avoid sources that straddle the irregular boundary , we only analyze sources further if they contain at least @xmath62 pixels . in order to sample the regions of arc formation well , a large number of sources are placed randomly inside the regions of interest . the number we generate is given by @xmath63 , where @xmath64 in square arc seconds , and usually we have @xmath65 . for the pixels contained in a source inside the region of interest , we use the lens mapping already derived to obtain the corresponding image positions , and then use the friends - of - friends method to identify giant arcs . to obtain the @xmath5 ratio , we assume giant arcs can be approximated as ellipses , hereafter we refer to this method as ellipse - fitting . numerically we find this assumption to be valid for most of our giant arcs . however , some images are irregular and can not be fitted well by an ellipse . they occur when the source straddles higher - order singularities ( such as beak - to - beak ) or when the source size is comparable to the caustics . but such cases are rare ( @xmath66 ) for clusters that contribute most of the lensing cross - sections and will thus not affect our results significantly . to calculate the @xmath5 ratio , we need to identify the center of a giant arc , @xmath67 , which is first taken to be the center of light . due to the ( curved ) arc shape , the center often falls outside the arc itself . to find a better center , we select one half of the image positions that are closest to the current center and then re - calculate the center of light . the process is repeated until the center does not change significantly . we verified visually that this procedure yields reliable centers of giant arcs . once the center is identified , we then locate the point @xmath43 that is the furthest from @xmath67 , and finally the point @xmath44 that is the furthest from @xmath43 . we fit a circular arc that passes through these three points . the length of the arc is taken to be twice the major axis length of the ellipse , @xmath68 , and the length of the minor axis is taken to be @xmath69 , where @xmath70 is the area covered by the image . the @xmath5 ratio is then simply @xmath71 . this procedure is identical to the ellipse fitting method in b98 . in w04 , the @xmath5 ratio is approximated by the magnification . in order to obtain the arc cross - sections , the magnification patterns need to be calculated on a grid in the source plane . in w04 , each pixel has a size of @xmath72 . however , we find that this is too large for small clusters in our simulations , so we instead used a grid of @xmath56 for all the clusters . we use the newton - raphson method to find the image position and its corresponding magnification and eigenvalues . when there are multiple images , following w04 , we use only the image with the largest ( absolute ) magnification . in the following , we will study the validity of two approximate measures of @xmath5 , namely the magnification , @xmath48 , and the ratio of the two eigenvalues , @xmath73 . under these two approximations , we obtain two corresponding cross - sections , which we will denote as @xmath74 and @xmath75 . we will compare these two cross - sections with that obtained through ellipse - fitting , @xmath76 . the corresponding optical depths can be obtained by integrating the cross - sections along the line of sight ( cf . [ eq : tau ] ) , and we will denote these as @xmath77 , @xmath78 and @xmath79 . the subscript @xmath1 indicates that the sources have an effective diameter of @xmath80 . later we will study lensing cross - sections and optical depths for other effective source diameters , @xmath81 , and their cross - sections and optical depths will be labeled accordingly . for illustrative purposes , fig . [ fig:4p ] shows the lensing properties for the fifth most massive cluster ( @xmath82 ) in our simulation at redshift 0.3 . the source is at redshift 1 . the bottom right panel shows the critical curves and the caustics . the other three panels show the maps of @xmath48 , @xmath73 and @xmath83 in the source plane . recall that we only consider regions where @xmath84 or @xmath85 . in other regions , we set @xmath86 and @xmath87 to unity , which are shown as black in the top left panel . the color bars show the range of the quantities plotted in the maps . pixels with values exceeding the maximum of the color bar are set equal to the maximum . if we approximate the @xmath5 ratio as either @xmath48 or @xmath73 , the top two panels clearly indicate that the cross - section @xmath74 will be much larger than the cross - section @xmath75 for a given length - to - width ratio . this effect is illustrated further in the bottom left panel where we plot the map of @xmath88 . in most cases , especially for the high magnification regions where arcs are expected to form , this ratio is larger than 1 , which indicates that for an infinitesimal circular source , @xmath48 will over - estimate the @xmath5 ratio . as a check of our lensing simulations , fig . [ fig : mu ] shows the magnification probability distribution for the cluster shown in fig . [ fig:4p ] . our results nicely reproduce the asymptotic relation @xmath89 expected from the fold caustics when @xmath90 ( e.g. , schneider , ehlers , & falco 1992 ) . to understand why the approximation @xmath91 appears to over - estimate the arc cross - section , we study a cluster with a generalized nfw profile ( navarro , frenk & white 1997 ; see also moore et al . 1998 ) : ( r)r^-(r+)^-3 + , 0<r < , [ eq : gnfw ] where @xmath92 is the scale radius , and @xmath93 is the virial radius . the cluster is at redshift of 0.3 with a mass of @xmath94 . the power - law index ( @xmath95 ) is taken to be 1.5 and the concentration parameter @xmath96 ( oguri 2002 ) . the source redshift is taken to be unity . [ fig : gnfw ] shows the relation between @xmath97 and @xmath48 for the minimum , saddle and maximum images in the time delay surface . the tangential arcs are primarily formed by the minimum images while the radial arcs are formed by the maximum images . for an infinitesimal circular source , the @xmath5 ratio is equal to @xmath48 multiplied by @xmath98 ( see eq . [ eq : lw - mu ] ) . it is clear that for the minimum image , at high magnifications , @xmath99 , while for the maximum image , @xmath100 . as for high magnifications , the asymptotic cross - section follows the probability distribution @xmath89 ( schneider et al . 1992 ) , the cross - sections for the tangential arcs therefore satisfy @xmath101 . similarly , for the radial arcs , we have @xmath102 . as the tangential arc cross - section usually dominates over the radial arc cross - section , it follows that the @xmath91 assumption will over - estimate the arc cross - sections by a factor of @xmath103 . [ fig : mlens ] shows how the cross - section of giant arcs is contributed by clusters with different masses at redshifts 0.3 and 0.2 . the background source population is at redshift 1.0 . for @xmath104 ( right panel ) , the distribution is peaked around @xmath105 , and it rapidly declines below @xmath106 as the much faster decrease in cross - sections dominates over the rising numbers of low - mass clusters . the cross - section also declines above @xmath107 due the exponential drop in the number of very massive clusters . the left panel is for @xmath108 , which can be directly compared with fig . 4 in dalal et al . ( 2004 ) the two distributions are roughly consistent with each other , not @xmath109 ( dalal 2005 , private communication ) . ] . the ratio of @xmath76 and @xmath74 is shown in fig . [ fig : sim2mu ] as a function of the length - to - width ratio ( @xmath5 ) for clusters at redshift 0.3 and background sources at redshift 1.0 . the thick line is the cross - section weighted average for the 200 most massive clusters while the thick dashed line shows the result for the generalized nfw model discussed above . the four thin lines are the results for four individual clusters , which are the first , fifth , tenth and fifteenth most massive clusters with @xmath110 , and @xmath111 respectively . recall that @xmath76 is the cross - section calculated using ellipse - fitting , while @xmath74 is obtained assuming that the @xmath5 ratio is equal to the magnification . clearly , the assumption that @xmath91 leads to an over - estimate of the cross - sections by a factor of 7 and 10 for @xmath112 and @xmath113 respectively . the ratio of @xmath76 and @xmath75 is shown in fig . [ fig : sim2lambda ] , where @xmath75 is again the cross - section calculated assuming @xmath5 to be equal to the ratio of the two eigenvalues . @xmath114 is in the range of 0.5 - 2 when @xmath5 is in the range of 5 - 20 . thus , @xmath75 offers a better approximation for the arc cross - section than @xmath74 . notice that the more massive a cluster is or the larger the caustics are , the higher @xmath115 and @xmath114 become . this can be understood as follows : as the caustics become larger , the effect of finite source size smoothing decreases . thus , @xmath76 increases and so do its ratios with @xmath74 and @xmath75 . to evaluate the optical depth , we first calculate the average cross - section per unit comoving volume : ( , ) = _ i ( , ) v , where @xmath116 is the average cross - section of the three projections of the @xmath117-th cluster at redshift @xmath118 , @xmath119 is the source redshift , and @xmath120 is the comoving volume of the box adopted in our lensing simulations . the optical depth can then be calculated as : [ eq : tau ] ( ) = 1 _ 0^ dz ( z , ) ( 1+z)^3 dv_p(z)dz , where @xmath121 is the angular diameter distance to the source plane , and @xmath122 is the proper volume of a spherical shell with redshift from @xmath123 to @xmath124 . fig . [ fig : tau ] shows the optical depth as a function of the source redshift , @xmath119 . our results can be well fitted by the simple curve _ 1 = 2.25 10 ^ -61 1+(/3.14)^-3.42 within 20% . [ fig : dtaudz ] shows the differential optical depth as a function of the cluster redshift . for a source at redshift 1 , there is little contribution to the optical depth for clusters beyond redshift 0.7 , because the critical surface density increases when the lens is close to the source , and very few clusters are super - critical . this trend is in excellent agreement with b98 ( see their fig . 1 ) and dalal et al . ( 2004 , see their fig . 10 ) . for a source at redshift 1 , the optimal lensing redshift is around 0.3 , which is why we have chosen this redshift to illustrate many of our results for @xmath6 . for a source at redshift 2 , the peak shifts to about 0.5 . for @xmath8 , the optimal redshift is near 0.6 - 0.7 , and the distributions also become noticeably broader , with a full - width - at - half - maximum of about 0.9 . this may have particular relevance to gladders et al . ( 2003 ) who emphasized the lack of lensing clusters with @xmath125 in their sample . there are likely selection biases in the surveys , but we notice that their four ( approximate ) source redshifts range from 1.7 to 4.9 . if such source redshifts are typical for their background galaxies , then the large cluster lens redshift is fully consistent with our predictions . as can be seen from fig . [ fig : dtaudz ] , for sources above redshift 3 , most lensing cross - sections are due to lenses below redshift 2.4 , and there are wiggles in the optical depth curves . the wiggles around @xmath126 are due to numerical effects as the interval of the lens redshift ( where particle positions are dumped from numerical simulations ) is large and the numerical integration is not exact . the prominent wiggle around @xmath127 is , however , real . it arises due to significant merger events in our simulations . for sources at high redshift , the cross - sections are dominated by relatively few ( most massive ) clusters and merger events in these clusters have substantial impact on the cross - section ( torri et al . 2004 ) . our optical depth turns out to be lower than three previous major studies ( b98 , w04 and dalal et al . 2004 ) , by different amounts . we have traced these differences back primarily to the different values adopted for the power - spectrum normalization ( @xmath10 ) , how the @xmath5 ratio is measured , and whether the numerical simulations have sampled the cluster mass function at the high - mass tail sufficiently . the source size and ellipticity and the matter distribution around clusters also have modest effects on the optical depth . below we discuss issues in detail . in this study , we have followed closely the methodology of b98 . nevertheless , our optical depth ( @xmath128 ) is roughly @xmath129 of the value found by b98 ( @xmath130 ) for a source population at redshift 1 . our lower optical depth is mainly due to the lower @xmath10 parameter adopted in the current study . in b98 , the optical depth was obtained as an average over two sets of simulations , one with @xmath131 and the other with @xmath132 . however , the optical depth is dominated by the former , as the clusters in the high @xmath10 simulation form earlier and are also more centrally concentrated both increase the arc cross - section considerably , leading to a higher optical depth than in the current work . fig . [ fig : massfunc ] shows the mass function of halos at redshift 0.3 predicted by the extended press - schechter formalism ( e.g. sheth & tormen 2002 and references therein ; press & schechter 1976 ) . the cluster abundance for @xmath131 is a factor of @xmath133 higher at @xmath134 than that for @xmath132 . even the moderately larger @xmath10 adopted by w04 ( 0.95 ) leads to a 50% increase in the abundance of clusters at at @xmath134 than that for @xmath132 , which can increase their optical depth relative to ours ( see the end of [ sec : comparisonsw04 ] ) . the method used by w04 is quite different from ours . they adopted a multiple - lens plane approach , modeling the universe with a three - dimensional matter distribution . in our study , we assume that the formation of giant arcs is dominated by individual clusters , and the large - scale structure along the line - of - sight is not important . as mentioned above , another key difference between our study and w04 is how the @xmath5 ratio is measured : w04 assume the @xmath5 ratio to equal the magnification , while we measure the @xmath5 ratio through ellipse - fitting . w04 effectively assumed infinitesimally small circular sources , while we assume the sources to be elliptical and to have a finite size . we will examine below how these different assumptions affect our predictions . if we adopt the same assumption as w04 ( i.e. , @xmath91 ) , then our results are within 25% of theirs ( shown as triangles in fig . [ fig : tau ] ) for a source redshift of @xmath6 , and within @xmath135 for @xmath119 out to @xmath136 . we also qualitatively confirm the important result of w04 that the optical depth is highly sensitive to the source redshift . indeed , the optical depth for a source at @xmath19 is a factor of 5 higher than for a source at @xmath6 . however , there are some quantitative differences , namely , their optical depth increases somewhat faster than ours as a function of the source redshift ; a similar discrepancy with w04 was found by dalal et al . we return to this point at the end of this subsection . as we have discussed earlier in [ sec : caustics ] , the magnification is not a reliable estimator of the @xmath5 ratio . instead , ellipse - fitting of simulated images is needed to determine the @xmath5 value accurately . with this approach , the optical depth is reduced by a factor of @xmath137 for a source at redshift unity ( see the thick solid curve labeled as @xmath77 in fig . [ fig : tau ] ) . the smaller value is a direct result of the smaller cross - sections as we have shown in figs . [ fig : sim2mu]-[fig : sim2lambda ] . in w04 , the sources are assumed to be infinitesimally small circular sources , while in our study the sources are uniformly distributed in ellipticity from 0 to 0.5 , and they occupy an area equal to a circular source with a diameter of @xmath1 . the finite source size effect may be important , particularly for low - mass clusters whose caustics are small . ferguson et al . ( 2004 ) recently used the hst to study the size evolution of galaxies as a function of redshift . they found that for galaxies around redshift 1.4 , the half - light diameters range between @xmath138 to @xmath139 with a peak around @xmath140 . the size becomes smaller as the source redshift increases . the distributions are similar for sources above redshift 2.3 , with a peak around @xmath4 and range from @xmath141 to @xmath142 . our adopted size falls within the range for galaxies from redshift 1.4 to 5 . as the redshift of giant arcs spans a wide range ( from 0.4 to 5.6 , see table 1 in w04 ) , it is important to examine how our results change if we adopt difference source sizes . [ fig : sourcesize ] shows the cumulative cross - section as a function of cluster mass for four source diameters , from @xmath143 , 0.5 , 1.0 and 1.5 arc seconds respectively . in this exercise , we have taken the source population to be at redshift 1.0 , and the clusters are at redshift 0.3 , i.e. approximately at the optimal cluster lensing redshift for @xmath6 ( see fig . [ fig : dtaudz ] ) . the cumulative cross - sections are calculated down to @xmath144 , below which clusters do not contribute substantial cross - sections ( see fig . [ fig : mlens ] ) . for giant arcs with @xmath13 , the cross - section increases by about 50% when the source diameter changes from @xmath1 to @xmath145 . for giant arcs with @xmath146 and @xmath147 , as expected , the finite source effects are even more modest . the optical depths for @xmath148 are shown as functions of the source redshift in fig . [ fig : tau ] . the optical depth increases by about 50% as @xmath149 decreases from @xmath1 to @xmath4 . the finite source size effect appears modest , insufficient to explain our discrepancy with w04 . notice that @xmath150 agrees well with @xmath151 for nearly all source redshifts . furthermore , notice that as the source size decreases from @xmath1 to @xmath4 , the optical depth does not follow @xmath151 any more closely , as one might expect . this arises because , following w04 , we have used only the brightest of multiple images for @xmath75 . in contrast , when calculating @xmath152 , @xmath153 and so on , the cross - section has been multiplied by the number of giant arcs that satisfy the selection criteria ( a similar procedure was used by b98 ) . so in general , even when the source size decreases to zero , these two cross - sections do not overlap with each other . another difference between our study and w04 is that we use elliptical sources while w04 use ( infinitesimally small ) circular sources . the ellipticity distribution we adopt ( uniform between 0 to 0.5 ) is in good agreement with the study of ferguson et al . ( 2004 ) who found a flat distribution between 0.1 to 0.6 ( with a small drop in the number of galaxies with zero ellipticity ) . nevertheless , we want to check how the ellipticity changes the results . [ fig : ellipticity ] shows the ratio of the cumulative cross - sections as a function of cluster mass for elliptical sources and circular sources for giant arcs with @xmath5 larger than 5 , 7.5 and 10 . again all the clusters are at redshift 0.3 and the sources are placed at redshift 1 . in all cases , the ellipticity increases the cross - section by 20% to 55% , implying that if we adopted circular sources , the discrepancy with w04 will become slightly worse . notice that the rate of increase depends both on the source size and the @xmath5 ratio , but the dependence is not monotonic due to the competing effects of finite source size and ellipticity , as was also found by oguri ( 2002 ) . in summary , the finite source size and ellipticity of sources both have only modest effects on the optical depth of giant arcs , insufficient to explain the difference between our study and w04 . as we mentioned above , we reproduce the optical depth of w04 if we approximate the @xmath5 by magnification . however , rigorous ellipse - fitting of simulated images reduce the optical depth by a factor of @xmath9 10 . this means only @xmath154 or so of the images with @xmath11 form giant arcs with @xmath13 . the question thus naturally arises : why do the rest 90% of images fail to make giant arcs @xmath13 ? first , it is well - known that for isothermal spheres ( @xmath155 ) , the @xmath5 ratio is identical to the magnification . but clusters are not isothermal spheres . as we have illustrated using a generalized nfw profile ( a better approximation for clusters ) in [ sec : caustics ] , the magnification is no longer equal to the @xmath5 ratio . second , the top right panel in fig . [ fig:4p ] clearly shows that , for this cluster , most giant arcs form at the position of the cusps while few giant arcs are formed near fold sections of caustics ( which contribute most of the high magnification images ) in the source plane . images with high magnification and low distortion must thus be along caustics but away from cusps . this is also expected from catastrophe theory , because images with high magnifications and low distortions are formed near the so - called `` lips '' or `` beak - to - beak '' caustics ( cf . schneider et al . 1992 ) . however , the question still remains : what are the image configurations of the rest of the highly magnified images ? [ fig : fraction ] provides the answer . it shows that roughly 50% and 80% of the highly magnified images with magnification above 10 form arcs with @xmath147 and @xmath156 respectively ; here the magnification is determined by the ratio of the image area and the source area . the remaining 20% or so images are highly magnified but ( largely ) undistorted ( hmus ) . williams & lewis ( 1998 ) studied such images for cored isothermal and nfw density profiles . they found that for isothermal cored clusters , the ratio between hmus and giant arcs with @xmath13 is of the order of unity . for nfw profiles , the ratio is only about 30% . our corresponding ratio is about @xmath16 and lower for smaller sources . notice that there are comparable number of images with @xmath13 and @xmath157 . this will further decrease the relative number of hmus relative to the observed number of giant arcs with @xmath13 . we do not regard this as a serious discrepancy , as williams & lewis ( 1998 ) used spherically symmetric clusters as lenses . cluster lenses forming large arcs are preferentially merging and highly irregular , so the effect of shear due to substructures may be much more important than that in their studies , which can plausibly explain the differences in the ratio of giant arcs and hmus . a related question concerns the width of giant arcs , which provides valuable information about the inner density profile of clusters . to see the general trend , we use a spherically symmetric toy model for clusters , where the density is a power - law as a function of radius , @xmath158 . for an isothermal sphere , @xmath159 , and the inner slope of an nfw profile is @xmath160 . for an infinitesimally small circular source , when @xmath161 , the width of a tangential arc is equal to @xmath162 times the diameter of the ( circular ) source . for an isothermal sphere , lensing conserves the source width . a point lens corresponds to the limit @xmath163 , in this case the width is equal to one half of the source diameter . in reality , clusters are not spherically symmetric , nevertheless , the width and length of giant arcs carry valuable information about the central density profiles in clusters ( hammer 1991 ; miralda - escud 1993 ) . fig . [ fig : widthlw ] shows the distribution of widths of giant arcs with @xmath146 . for @xmath164 , the distribution peaks around @xmath140 , and rapidly drops off at @xmath142 . for @xmath165 , the distribution peaks around @xmath166 and drops off rapidly around @xmath1 , except for the tenth most massive cluster which seems to have a much broader distribution this is because this cluster has the smallest caustics , and the finite source size is most important in this case . the thick solid histogram shows the observed width distribution in sand et al . ( 2005 ) for giant arc candidates with @xmath146 ( in our definition ) . this comparison does not yet fully account for the ( unknown ) redshift distribution of background sources and other potential biases . nevertheless , it appears very interesting that the size distribution is in better agreement with an effective source diameter of @xmath165 , although the tail indicates that larger sources are also needed . w04 emphasized the effect of matter along the line of sight on the arc statistics . we are unable to model the universe using multiple lens planes , as realistically performed by w04 . we can , however , investigate the effect of matter distribution ( e.g. , filaments ) surrounding the clusters . to do this , we increase the projection depth to @xmath167 mpc ( comoving ) and the lens area to a square with a side - length of @xmath168 . we write the new optical depths as @xmath169 and @xmath170 , in contrast to the values obtained using a cube of side length @xmath50 , denoted by @xmath171 , @xmath172 . [ fig : tau ] shows these optical depths as functions of the source redshift . clearly the matter distribution around the clusters increases the optical depth but the effect appears to be small ( @xmath173 ) . if the enhancement due to the large - scale structure along the line - of - sight is more important ( w04 ) , then the optical depth at higher source redshift will be affected more , which may partially explain why our optical depth increases somewhat more slowly as a function of source redshift than w04 , although hennawi , dalal & ostriker ( 2005 ) showed that multi - plane lensing changes the cross - section only slightly compared with the single - plane lensing ( see their fig . 2 ) . in addition w04 used a power - spectrum normalization @xmath174 , slightly higher than our value ( @xmath132 ) . this will also lead to a slightly higher optical depth because in their simulation , the clusters will form somewhat earlier , which will in turn preferentially boost the optical depth for a source at higher redshift as their optimal lens redshift is higher ( see fig . [ fig : dtaudz ] ) . most of our conclusions agree with those of dalal et al . ( 2004 ) , who used numerically - simulated clusters from the gif collaboration ( kauffmann et al . their mass and spatial resolutions are similar to ours . however , their overall optical depths are larger than ours by a factor of 6 for @xmath175 . the difference is partly due to the different definitions of the @xmath5 ratio . dalal et al . fit the giant arcs by a rectangle , rather than an ellipse as in our case . thus for the same giant arc , the @xmath5 ratio in dalal et al . ( 2004 ) is equal to @xmath15 times our @xmath5 ratio . as a result , their optical depth for giant arcs with @xmath113 should be compared with our value for @xmath176 . for example , for @xmath177 and 2 , our optical depths for giant arcs with @xmath112 ( in our definition ) are @xmath178 and @xmath179 respectively . these should be compared with their corresponding values of @xmath180 , @xmath181 , and @xmath182 . our value is about 70% of their value for a source at redshift 1.5 and 2 . but the discrepancy becomes larger for a source at redshift 1 . at such low redshift , the cross - section is increasingly dominated by the few most massive clusters as the critical surface density increases , so the cosmic variance becomes important , particularly for the gif simulations which have a simulation volume 9.6 times smaller than ours . a detailed comparison between our numerical methods indicate that our cross - sections agree within @xmath183 for the same numerical cluster ( with the same projection ) selected from the gif simulations ( dalal 2005 , private communication ) agree within 10%-30% . ] . we believe that the remaining difference is due to the different mass functions in our simulations and the fact that dalal et al . ( 2004 ) used more than three projections to calculate the cross - sections for the more massive clusters . [ fig : massfunc ] shows that the gif mass function at large mass appears slightly higher than ours . this will increase their optical depths for giant arcs and likely explain the remaining difference . in this paper we used high - resolution numerically - simulated clusters to study the rate of giant arcs in clusters of galaxies . the methodology we used is similar to b98 , but the number of clusters in our simulations is about a factor of @xmath184 higher than previous studies ( e.g. , b98 and dalal et al . 2004 ) , so it should provide better statistics for the most massive clusters . we calculated the cross - section and optical depth as a function of the source redshift using ellipse - fitting and compared our results with those obtained with two approximate measures of the length - to - width ratio ( @xmath5 ) . we also found that the effects of source size and ellipticity on the optical depth are modest . furthermore , the matter distribution around clusters increases the lensing cross - section only slightly , by @xmath173 . our optical depth is about 10% of that found b98 . the difference arises because the two studies used two different @xmath10 values b98 adopted a higher @xmath10 ( 1.12 ) than ours ( 0.9 ) . the higher @xmath10 value leads to both a higher number density and more concentrated clusters ( see fig . [ fig : massfunc ] ) . both effects increase their optical depth . our results qualitatively confirm one important conclusion reached by w04 , namely that the optical depth is sensitive to the source redshift , but the rate of increase in our simulations is somewhat slower than theirs . the large - scale structure , which was included in w04 but ignored by us , will become more important as the source redshift increases , but whether this effect can explain the discrepancy in full is unclear ( see also dalal et al . perhaps more importantly , we find that their assumption , @xmath91 , over - estimates the number of giant arcs by a factor of @xmath137 . better agreements were found between our study and dalal et al . once we account for the difference in the arc modeling ( ellipses vs. rectangles ) , our optical depth for giant arcs with @xmath176 is about 70% of that found by dalal et al . ( 2004 ) for a source at redshift 1.5 and 2.0 . for a source at redshift 1 , our prediction is about 40% of their value . we believe our differences are due to cosmic variance the gif simulation has a cosmic volume that is 9.6 times smaller than ours , and their mass function appears to be somewhat higher than ours ( see fig . [ fig : massfunc ] ) , which may explain their higher optical depth . both w04 and dalal et al . ( 2004 ) concluded that the predicted and observed rates of giant arcs are consistent with each other . however , the comparison study we have performed leaves this question open . it appears to us that our predicted rate may be a factor of a few too low compared with the `` observed '' rate . however , this conclusion is not firm as both observations and theoretical predictions are uncertain . observationally , we need more transparent selection criteria ( see dalal et al . 2004 for excellent discussions ) . furthermore , the @xmath5 determination from ground - based telescopes may be affected by seeing , leading to a likely under - estimate of the @xmath5 ratio as the widths of many arcs are unresolved . the recent extensive search for giant arcs with hst images ( sand et al . 2005 ) is a significant step in the right direction . observers need to report not only the @xmath5 ratio but also individual widths , which can provide important constraints on the inner density profile in clusters ( see [ sec : comparisonsw04 ] ) . our predicted width distribution appears to better match the observed distribution of sand et al . ( 2005 ) when the effective source diameter @xmath4 . however , a reliable prediction of the giant arcs also requires accurate information on the source population , including their redshift , size , magnitude , surface brightness and ellipticity . such information is of course a key objective for studying high - redshift galaxies . as the information is currently lacking , the prediction for the number of giant arcs has yet to enter the high - precision era . theoretically , we need higher resolution simulations in very large boxes so that we can sample the cluster mass function and resolve the internal structure of clusters simultaneously . numerical simulations with baryonic cooling and star formation will also be needed to make more detailed comparisons with observational data . while baryonic cooling is not expected to be very efficient in clusters of galaxies ( most baryons in clusters are still in the hot phase seen as x - rays ) , nevertheless , its effects may not be negligible , particularly for the giant arcs at small radii ( dalal et al . 2004 ; see also oguri 2003 ) . in fact , the recent study by puchwein et al . ( 2005 ) found that the effects of baryons on giant arcs depend on the detailed implementation of viscosity , star formation and feedback processes in simulations . if these effects can be implemented in practice , hydro - dynamical cosmological simulations also offer the possibility of a more direct comparison with observations , at least with the emss , as the x - ray luminosities of clusters of galaxies in these simulations can be predicted ( with some uncertainty ) , and hence we can apply similar selection criteria for clusters of galaxies as in observations and examine the number of giant arcs in these clusters . as many observations converge to the concordance cosmology , it will be interesting to use lensing to constrain parameters such as @xmath10 independently in the @xmath185 flat cosmology . as the cluster mass function and internal structures both sensitively depend on @xmath10 , the arc statistics should provide a stringent limit on @xmath10 , as already illustrated by the difference between the current study and b98 . this parameter is still somewhat uncertain : some studies prefer values as high as 1.1 , while others prefer values as low as 0.7 ( see tegmark 2004 for a recent overview of current results ) . we plan to return to some of these issues in the future . we thank drs . l. gao , w. p. lin and in particular n. dalal for helpful discussions . we are grateful to david sand for the data plotted in fig . [ fig : widthlw ] . we are also indebted to an anonymous referee for an insightful report that improved the paper . ypj is supported in part by nkbrsf ( g19990754 ) and by nsfc . sm and lgl acknowledge the financial support of chinese academy of sciences and the european community s sixth framework marie curie research training network programme , contract no . mrtn - ct-2004 - 505183 angles " . bartelmann , m. , huss , a. colberg , j. m. , jenkins , a. , & pearce , f. r. 1998 , a&a , 330 , 1 ( b98 ) bartelmann , m. , & weiss , a. 1994 , , 284 , 285 bartelmann , m. , steimetz , m. & weiss , a. 1995 , , 297 , 1 bartelmann , m. , meneghetti , m. , perrotta , f. , baccigalupi , c. , & moscardini , l. 2003 , a&a , 409 , 449 bryan , g. l. , & norman , m. l. 1998 , apj , 495 , 80 dalal , n. , holder , g. , & hennawi , j. f. 2004 , , 609 , 50 ferguson , h. c. , et al . 2004 , , 600 , l107 flores , r. a. , maller , a. h. , & primack , j. r. 2000 , apj , 535 , 555 gioia , i. m. , & luppino , g. a. 1994 , apjs , 94 , 583 gladders , m. d. , hoekstra , h. , yee , h. k. c. , hall , patrick , b. , & barrientos , l. f. 2003 , apj , 593 , 48 hammer , f. 1991 , , 383 , 66 hennawi , j. f. , dalal , n. , bode , p. , & ostriker , j. p. 2005 , astro - ph/0506171 ho , s. , & white , m. 2004 , astro - ph/0408245 hockney , r. w. , & eastwood , j. w. 1981 , computer simulation using particles ( mcgraw - hill : new york ) jenkins , a. 1998 , apj , 499 , 20 jing , y. p. 2000 , apj , 535 , 30 jing , y. p. & suto y. 2002 , apj , 574 , 538 jing , y. p. 2002 , 335 , l89 kauffmann g. , colberg , j. m. , diaferio , a. , & white , s. d. m. 1999 , , 303 , 188 kitayama , t. , & suto , y. 1996 , , 280 , 638 kneib , j .- p . , ellis , r. s. , smail , i. , couch , w. j. , & sharples , r. m. 1996 , apj , 471 , 643 luppino , g. a. , gioia , i. m. , hammer , f. , le f@xmath186vre , o. , & annis , j. a. , 1999 , a&as , 136 , 117 meneghetti , m. , bolzonella , m. , bartelmann , m. , moscardini , l. , & tormen , g .. 2000 , , 314 , 338 meneghetti , m. , bartelmann , m. , & moscardini , l. 2003 , , 346 , 67 meneghetti , m. , bartelmann , m. , dolag , k. , moscardini , l. , perrotta , f. , baccigalupi , c. , & tormen , g. 2004 , new astronomy reviews , 49 , 111 miralda - escud , j. 1993 , , 403 , 497 monaghan , j. j. 1992 , ara&a , 30 , 543 moore , b. , governato , f. , quinn , t. , stadel , j. & lake , g. 1998 , apj , 499 , l5 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 ostriker , j. p. , & steinhardt , p. j. 1995 , , 377 , 600 oguri , m. 2002 , apj , 573 , 51 oguri , m. , lee , j. , & suto , y. 2003 , apj , 599 , 7 peacock , j. a. , et al . 2001 , nature , 410 , 169 press , w. h. , & schechter , p. 1974 , , 187 , 425 puchwein , e. , bartelmann , m. , dolag , k. , & meneghetti , m. 2005 , astro - ph/0504206 sand , d. j. , true , t. , ellis , r. s. , & smith , g. p. 2005 , astro - ph/0502528 schneider , p. , ehlers , j. , & falco , e. e. 1992 , gravitational lenses ( springer verlag : new york ) sheth , r. k. , & tormen , g. 2002 , , 349 , 1464 spergel , d.n.s . 2003 , , 148 , 175 stoehr , f. , white , s. d. m. , tormen , g. , & springel , v. 2002 , , 335 , l84 tegmark , m. 2004 , , 606 , 702 tegmark , m. et al . 2004 , phys . d , 69 , 103501 torri , e. , meneghetti , m. , bartelmann , m. , moscardini , l. , rasia , e. , & tormen , g. 2004 , , 349 , 476 wambsganss , j. , bode , p. , & ostriker , j. p. , 2004 , apj , 606 , l93 ( w04 ) willams , l. l. r. , & lewis , g. f. 1998 , , 294 , 299 wu , x .- p . , & hammer , f. 2003 , , 262 , 187 wu , x .- p . , & mao , s. 1996 , , 463 , 404 zaritsky , d. & gonzalez , a. h. , 2003 , apj , 584 , 691	we use high - resolution n - body simulations to study the galaxy - cluster cross - sections and the abundance of giant arcs in the @xmath0cdm model . clusters are selected from the simulations using the friends - of - friends method , and their cross - sections for forming giant arcs are analyzed . the background sources are assumed to follow a uniform ellipticity distribution from 0 to 0.5 and to have an area identical to a circular source with diameter @xmath1 . we find that the optical depth scales as the source redshift approximately as @xmath2 $ ] ( @xmath3 ) . the amplitude is about 50% higher for an effective source diameter of @xmath4 . the optimal lens redshift for giant arcs with the length - to - width ratio ( @xmath5 ) larger than 10 increases from 0.3 for @xmath6 , to 0.5 for @xmath7 , and to 0.7 - 0.8 for @xmath8 . the optical depth is sensitive to the source redshift , in qualitative agreement with wambsganss et al . ( 2004 ) . however , our overall optical depth appears to be only @xmath9 10% to 70% of those from previous studies . the differences can be mostly explained by different power spectrum normalizations ( @xmath10 ) used and different ways of determining the @xmath5 ratio . finite source size and ellipticity have modest effects on the optical depth . we also found that the number of highly magnified ( with magnification @xmath11 ) and `` undistorted '' images ( with @xmath12 ) is comparable to the number of giant arcs with @xmath11 and @xmath13 . we conclude that our predicted rate of giant arcs may be lower than the observed rate , although the precise ` discrepancy ' is still unclear due to uncertainties both in theory and observations .
the sierpinski gasket described in 1915 by w. sierpiski is a classical fractal . suppose @xmath0 is the solid regular triangle with vertexes @xmath1 @xmath2 @xmath3 let @xmath4 be the contracting similitude for @xmath5 then @xmath6and the sierpinski gasket @xmath7 is the self - similar set , which is the unique invariant set @xcite of ifs @xmath8 satisfying@xmath9the sierpinski gasket is important for the study of fractals , e.g. , the sierpinski gasket is a typical example of post - critically finite self - similar fractals on which the dirichlet forms and laplacians can be constructed by kigami @xcite , see also strichartz @xcite . for the word @xmath10 with letters in @xmath11 , i.e. , every letter @xmath12 for all @xmath13 we denote by latexmath:[$\|\sigma @xmath10 and @xmath15 we call @xmath16 a prefix of @xmath17 and denote by @xmath18 if @xmath19 and @xmath20 we also write @xmath21 if @xmath22 or @xmath23 when @xmath24 with @xmath25 we say that @xmath17 is the father of @xmath16 and @xmath16 is a child of @xmath26 given @xmath27 we write @xmath28 which is a solid regular triangle with side length @xmath29 for notational convenience , we write @xmath30 with empty word @xmath31 we also denote @xmath32 if @xmath18 then @xmath33 for solid triangle @xmath34 with word @xmath35 we denote by @xmath36 its boundary consisting of @xmath37 sides , where every side is a line segment with side length @xmath29 complex networks arise from natural and social phenomena , such as the internet , the collaborations in research , and the social relationships . these networks have in common two structural characteristics : the small - world effect and the scale - freeness ( _ power - law _ degree distribution ) , as indicated , respectively , in the seminal papers by watts and strogatz @xcite and by barabsi and albert @xcite . in fact complex networks also exhibit _ self - similarity _ as demonstrated by song , havlin and makse @xcite and fractals possess the feature of _ power law _ in terms of their fractal dimension ( e.g. see falconer03 ) . recently self - similar fractals are used to model evolving networks , for example , in a series of papers , zhang et al . zzfgz07,zzzcg08,gwzzw09 use the sierpinski gasket to construct evolving networks . there are also some complex networks modelled on self - similar fractals , for example , liu and kong @xcite and chen et al . @xcite study koch networks , zhang et al . @xcite investigate the networks constructed from vicsek fractals . see also dai and liu @xcite , sun et al . @xcite and zhou et al . @xcite . in the paper , we introduce a new method to construct evolving networks modelled on sierpinski gasket and study the asymptotic formula for average path length . since @xmath7 is connected , we can construct the network from geometry as follows . fix an integer @xmath38 we consider a network @xmath39 with vertex set @xmath40 where @xmath41 for the edge set of @xmath42 there is a unique edge between distinct words @xmath16 and @xmath17 ( denoted by @xmath43 ) if and only if @xmath44we can illustrate the geodesic paths in figure 2 for @xmath45 . we have @xmath46 since @xmath47 and @xmath48 we also get another geodesic path from @xmath49to @xmath50 @xmath51 since @xmath52 and @xmath53,$ ] the line segment between @xmath54 and @xmath55 we also have some geodesic paths from @xmath56to @xmath50 @xmath57 and @xmath58 also we have @xmath59 but @xmath60 then the geodesic distance between @xmath61 and @xmath62 is @xmath63 in figure 2 , @xmath64 , @xmath65 and @xmath66 . in fact , by observation we have [ c : two]suppose @xmath67 with @xmath68 then @xmath69 if and only if there are at most two letters in @xmath70 in particular , if @xmath67 and @xmath71 then @xmath72 for example , @xmath73 and @xmath74 are neighbors , but @xmath73 and @xmath75 are not . for every @xmath38 we denote @xmath76 the geodesic distance on @xmath77 let@xmath78be the average path length of the complex network @xmath79 we can state our main result as follows . we have the asymptotic formula @xmath80 [ r:6 ] since @xmath81 , theorem @xmath82 implies that the evolving networks @xmath39 have small average path length , namely @xmath83 . the paper is organized as follows . in section 2 we give notations and sketch of proof for theorem 1,consisting of four steps . in sections 3 - 6 , we will provide details for the four steps respectively . our main techniques come from the self - similarity and the renewal theorem . we will illustrate our following four steps needed to prove theorem 1 . given a small solid triangle @xmath84 we can find a maximal solid triangle @xmath85 which contains @xmath86 and their boundaries are touching . translating into the language of words , for a given word @xmath87 we can find a unique shortest word @xmath88 such that @xmath89 and @xmath90 for a word @xmath91 where @xmath92 is the maximal suffix with at most two letters appearing , using claim [ c : two ] we have @xmath93 iterating @xmath94 again and again , we obtain a sequence @xmath95 let @xmath96 in particular , we define @xmath97 for @xmath98 we have @xmath99 @xmath100 , @xmath101and @xmath102 then @xmath103 we will prove in section 3 in fact , this proposition shows that @xmath106 is independent of the choice of @xmath107 whenever @xmath108 in this case , we also write @xmath109 write @xmath110then @xmath111 is independent of @xmath38 in fact , @xmath111 is the minimal number of moves for @xmath112 to touch the boundary of @xmath113 * step 2 . * given @xmath114 we consider the average geodesic distance between the empty word and word of length @xmath115 and set @xmath116and @xmath117 we will obtain the limit property of @xmath118 as @xmath119 in section 4 . as illustrated in figure 3 , proposition [ p : asym ] shows that the _ typical _ geodesic path is the geodesic path between @xmath133 and @xmath134 whose first letters of codings are different . on the other hand , for example the geodesic path between @xmath133 and @xmath135 with the same first letter will give negligible contribution to @xmath136 using @xmath137 with @xmath129 we obtain that @xmath138 , ignoring the terms like @xmath139 we have @xmath140where @xmath141 is the average value of @xmath142 using stolz theorem , we have @xmath143 in fact , suppose @xmath147 @xmath148 or @xmath37 for all @xmath149 is composed of infinite words with letters in @xmath150 then we have a natural mass distribution @xmath151 on @xmath152 such that for any word @xmath153 of length @xmath154 @xmath155for any word @xmath35 let @xmath156 denote the cardinality of letters appearing in word @xmath157 for @xmath151-almost all @xmath158 let @xmath159then @xmath160 for @xmath151-almost all @xmath161 suppose @xmath162 and @xmath163 and there is an infinite sequence @xmath164 of integers such that @xmath165 for all @xmath166 and @xmath167satisfying @xmath168 and @xmath169 for all @xmath170 we then let @xmath171since @xmath172 is uniquely determined by word @xmath173 and @xmath151 is symmetric for letters in @xmath174 we find that @xmath175 is a sequence of positive independent identically distributed random variables with @xmath176 for example , for @xmath177 then @xmath178 and @xmath179 @xmath180 , @xmath181	in this paper , we introduce a new method to construct evolving networks based on the construction of the sierpinski gasket . using self - similarity and renewal theorem , we obtain the asymptotic formula for average path length of our evolving networks .
entanglement and discord describe remarkable features of quantum systems . indeed , they are closely related to the amount of quantum correlations contained in the system , coming either from non - separability or the impossibility of local discrimination @xcite . beyond their specific role in fundamental physics , entanglement and discord has been also recognized as resources for quantum technology , e.g. for the processing of quantum information and for the effective implementation of quantum enhanced protocols . in particular , in the last decade it has been recognized @xcite that separable mixed states may represent a resource , if they show a nonzero quantum discord . in fact , mixed separable states with non - zero discord have been exploited to achieve a speed - up for certain computational tasks compared to classical states @xcite . this is true also for continuous variable systems , where gaussian quantum discord has been introduced @xcite , measured @xcite , and exploited for quantum enhanced protocols @xcite . an essential ingredient to exploit the quantumness of a physical system is the preservation of its coherent time evolution . on the other hand , the unavoidable interaction with its environment usually destroys coherence and quantumness @xcite , and in turn its use for quantum technology . for these reasons , much attention has been devoted to the analysis , characterization and control of the dynamics of quantum correlations in different physical systems @xcite , including quantum optics @xcite , nuclear magnetic resonance @xcite , nanophysics @xcite and biology @xcite . for bipartite open quantum systems interacting with a quantum environment , entanglement and discord may exhibit peculiar features such as sudden death and transitions , revivals , and trapping @xcite . such phenomena have been linked either to direct @xcite or indirect @xcite effective two - qubit interactions . for non - interacting qubits they are due to the non - markovian nature of the environment @xcite , which results in the transfer of correlations back and forth from the two - qubit system to the various parts of the total system . in this framework , it was shown @xcite that for a common bosonic environment the entanglement between two qubits is more robust against decoherence , and its dynamics is faster , than for independent environments with the same spectrum . the effect of the noise spectrum on the dynamics of quantum discord has been also analyzed @xcite . recently , revivals of quantum correlations have been found also for quantum system coupled to classical sources @xcite and have been connected to a quantifier of non - markovianity for the dynamics of a single - qubit . indeed , it was proven that a classical noise can mimic , without loss of generality , a quantum environment not affected by the system or influenced in a way that does not result in back - action @xcite . among the class of open quantum systems interacting with a classical environment , a particular attention has been devoted to systems made of two qubits subjected to a classical source of random telegraph noise ( rtn ) @xcite , namely interacting with a bistable fluctuator , randomly switching between its two states with a given rate @xmath1 . depending upon the ratio between the switching rate and the system - environment coupling , the dynamics of a quantum system may exhibit markovian or non - markovian behavior . in fact , the great interest in rtn is due to the fact that it is able to model environmental fluctuations appearing in many nanodevices based on semiconductors , metals and superconductors @xcite . furthermore , it also represents the basic building block to describe noises of the type @xmath0 , which are responsible for decoherence in quantum solid - state devices @xcite . this kind of noise spectra stem from the collection of random telegraph source with different switching rates and are usually referred to as _ colored _ spectra . the color of the noise depends upon the value of the parameter @xmath2 @xcite . for @xmath3 the so - called pink @xmath4 noise is found , which is obtained from a set of random telegraph fluctuators weighted by the inverse of the switching rate . another interesting case is the @xmath5 spectrum , also called brown noise from its relation to a brownian motion . environments characterized by @xmath0 noise spectra usually arise when a system is coupled to a large number of bistable fluctuators , with a specific distribution of their switching rates . upon considering a collection of fluctuators , the colored noise may be implemented by means of a linear combination of sources of rtn , each characterized by a specific switching rate chosen from a suitable distribution . on the other hand , the same spectrum can be obtained if we consider a single fluctuator with a random switching rate . as already observed by several authors @xcite different microscopic configurations of the environment leading to the same spectra , may correspond to different physical phenomena , e.g. different evolutions for the quantum correlations . in these cases , i.e. when the knowledge of the noise spectrum is not sufficient to describe decoherence phenomena , the noise is referred to be non - gaussian . in particular , we investigate the time evolution of entanglement and discord of two initially entangled non - interacting qubits coupled to a classical environment described either by means of a single random fluctuator or a collection of rtn sources . for the model of a single fluctuator with random switching rate , the correlations decay with a damped oscillating behavior . in the configuration with many bistable fluctuators , pink noise leads to a monotonic decay of entanglement and discord , while the presence of brown noise induces phenomena of sudden death and revivals . we ascribe this discrepancies to the different number of decoherence channels in the two configurations . in other words , the time - evolution of the system is determined not only by the spectrum of the environment , but also by its configurations , i.e. to its microscopic structure . for both the configurations , the dynamics of the two qubits is ruled by a stochastic hamiltonian with time dependent coupling . the average of the time - evolved states over the switching parameters describes the evolution of the two - qubit state under the effect of the noise . upon a suitable choice of stochastic time - dependent terms in the hamiltonian , we are able to describe the effects of both separate and common environments : in the former each qubit is locally coupled to a random external signal , while in the latter both qubits are subject to the effect of a common classical environment . the interest in systems interacting with a common environment arises from recent experiments in cavity qed @xcite and circuit qed @xcite where reservoir - mediated interaction between two qubits is at the heart of entanglement generation . environmental effects due to a collection of bistable fluctuators have been experimentally observed in nanoscale electronic devices , where the single electron tunnelling turns out to be affected by charge fluctuations @xcite . our results may be thus used as a valid guideline to analyze decoherence phenomena in josephson circuits subject to noise stemming from fluctuating background charges and flux @xcite . other systems where the noise is non - gaussian and comes from a collection of two - level fluctuators include silicon @xcite and magnetic ones @xcite , as well as vortex matter @xcite . the paper is organized as follows : in section 2 , we briefly review the definition of negativity as a measure of entanglement , and of quantum discord . in section 3 we present the physical model for two qubits interacting with a classical environment , being represented either by a single random bistable fluctuator or a collection of fluctuators . section 4 reports results for the different configurations , whereas section 5 closes the paper with a discussion and some concluding remarks . in this section we briefly review the concepts of entanglement and quantum discord between two qubits . in particular we evaluate entanglement by means of negativity@xcite , which is given by : @xmath6 where @xmath7 are the negative eigenvalues of the partial transpose of the system density matrix . note that the negativity is bound between zero , for separable states , and one , for maximally entangled states . the concept of negativity has recently been extended to the case of tripartite systems of identical particles @xcite . the quantumness of correlations of bipartite a system may be also quantified by discord @xcite , namely the difference between the total and the classical correlations in a system . the quantum mutual information quantifies the total correlations in a system and is defined as : @xmath8 where @xmath9 is the partial trace of the total bipartite system @xmath10 and @xmath11 is the von neumann entropy . the classical correlations are evaluated by means of the expression @xcite @xmath12 $ ] where @xmath13 are projective measurements on subsystem @xmath14 and @xmath15 , with @xmath16/\text{tr}[\pi_j\rho\pi_j]$ ] is the remaining state of @xmath17 after obtaining outcome @xmath18 on @xmath14 . therefore the quantum discord is the difference between the mutual information and the classical correlation : @xmath19 usually , the evaluation of quantum discord is not an easy task , since it involves an optimization problem . however , for two - qubit systems described by a density matrix of the form ( from now on referred to as states with a x shape ) @xmath20 , where @xmath21 are the three pauli matrices , the optimization may be carried out analytically @xcite , leading to @xmath22\nonumber\\ & -\frac{1-c}{2}\log_2(1-c)-\frac{1+c}{2}\log_2(1+c)\label{discordluo},\end{aligned}\ ] ] where @xmath23 . we address quantum correlations , both entanglement and discord , between two non - interacting qubits subjected to noisy environments . our analysis concerns environments with noise spectra of the form @xmath24 , which are realized by different configurations of bistable fluctuators . the two qubits are initially prepared in the bell state @xmath25 . the interaction between the qubits and the environment can be either local or global , i.e. we examine both the case of independent environments acting locally on each qubit and of a common environment affecting the two qubits . if we set @xmath26 and adopt the spin notation , the two - qubit hamiltonian describing the interaction with a single fluctuator is given by @xmath27 where @xmath28 is the hamiltonian of a single qubit subject to a classical time - dependent noise which affects the transition amplitude parameter @xmath29 : @xmath30 with @xmath31 the qubit energy in absence of noise ( energy degeneracy is assumed ) , @xmath32 the identity matrix for subspace @xmath33 , @xmath34 is the coupling constant between the system and the environment , @xmath35 the pauli matrix . if the time - dependent coefficient @xmath29 can randomly flip between two values @xmath36 with a fixed rate @xmath1 , then eq . describes a qubit subject to a random telegraph noise @xcite . this hamiltonian , extended to describe qudits , has also been used to analyze the time evolution of entanglement of a continuous - time quantum walk of two indistinguishable particles on a one - dimensional lattice @xcite . the hamiltonian is stochastic due to the random nature of the noise parameter @xmath37 . for a specific choice of @xmath37 , the total system evolves according to the evolution operator @xmath38 , with positivity ensured by the very structure of the hamiltonian in eq . ( [ h1qubit ] ) @xcite . by averaging the global state over different realizations of the sequences of @xmath37 , the two - qubit mixed state is obtained . in order to reproduce the @xmath0 spectrum , the single rtn frequency power density must be integrated over the switching rates @xmath1 with a proper distribution : @xmath39 where @xmath40 is the random telegraph noise frequency spectral density with lorentzian form @xmath41 . the integration is performed between a minimum and a maximum value of the switching rates , respectively @xmath42 and @xmath43 . @xmath44 is the switching rate distribution and takes a different form depending on the kind of noise : @xmath45&1<\alpha\leqslant2 \ , . \end{array } \right.\end{aligned}\ ] ] it follows that , in order to simulate a frequency spectrum proportional to @xmath0 , the switching rates must be selected from a distribution proportional to @xmath46 . when the integration in eq . is performed , the spectrum has the requested @xmath0 behavior in a frequency interval , so that every frequency belonging to such interval satisfies the condition @xmath47 . eq . can be obtained either considering a single bistable fluctuator whose switching rate is randomly chosen from a distribution @xmath48 or from a collection of sources of rtn each with a switching rate taken from the same @xmath1-distribution . even if the spectrum is the same , the physical systems described are indeed very different . in the case of a single fluctuator , the noise parameter @xmath37 can only flip between two values @xmath49 . the difference with the rtn case , is that here the switching rate is not known a - priori . this means that the bistable fluctuator is described by a statistical mixture whose elements are chosen from the ensemble @xmath50 . this model describes the physical situation in which each of the two qubits interacts with a single fluctuator , for separate baths , or with the same fluctuator in the case of a common bath . the qubits are affected only by one source of noise , and therefore only one decoherence channel is present . the global system evolves according to the hamiltonian , with a specific choice of both the parameter @xmath37 and of its switching rate . in particular , for each value of the switching rate picked from a distribution @xmath48 , we create a large number of sequences @xmath37 . this means that we are actually introducing some uncertainty on the rate @xmath1 . this uncertainty gives rise to a @xmath0 spectrum . for each selected switching rate , the evolution corresponds to the one of a bistable fluctuator with a characteristic rtn phase shown to be : @xmath51 and characterized by a distribution @xcite : @xmath52 + \frac{\gamma}{\nu}[\theta(\varphi+\nu t)+\theta(\varphi-\nu t)]\nonumber\\ & \times\left[\frac{i_1\left(\gamma t\sqrt{1-(\varphi/\nu t)^2}\right)}{\sqrt{1-(\varphi/\nu t)^2 } } + i_0\left(\gamma t\sqrt{1-(\varphi/\nu t)^2}\right)\right]\bigg\}\end{aligned}\ ] ] where @xmath53 is the modified bessel function and @xmath54 is the heaviside step function . for a given @xmath1 the global system is described by a x - shaped density matrix obtained by averaging over the noise phase the density matrix @xmath55 corresponding to a specific choice of the parameter @xmath37 @xcite : @xmath56\label{rtn}\end{aligned}\ ] ] where @xmath57 is a bell state and @xmath58 and @xmath59 are time - dependent coefficients . the function @xmath60 represents the average of the rtn phase factor i.e. @xmath61 where @xmath62 \\ e^{-\gamma t}\left[\cos { \left(\kappa_{m\nu}t\right ) } + \frac{\gamma}{\kappa_{m\nu}}\sin { \left(\kappa_{m\nu}t\right)}\right ] \end{array}\right . , \ ] ] for @xmath63 and @xmath64 respectively , where @xmath65=@xmath66 with @xmath67 . the two - qubit density matrix is obtained by averaging the density operator in eq . over the switching rates : @xmath68 once the average over @xmath1 is performed , the time - evolved density matrix reads : @xmath69 \label{rho_sing}\end{aligned}\ ] ] where the time - dependent coefficient @xmath70 can be written as : @xmath71 ^ 2 \notag\\ \lambda_{ce}&=\int_{\gamma_1}^{\gamma_2}\ , d_{4\nu } \ , p_{\alpha}(\gamma)\,d\gamma \label{lambdas}\end{aligned}\ ] ] this means that the quantum system is again described by a density matrix with a x form . the @xmath0 noise spectrum can arise from the coupling of a system with a large number of fluctuators , each characterized by a specific switching rate , picked from the distribution @xmath48 @xcite in a range @xmath72 $ ] . in this case the random parameters in eq . describes a linear combination of bistable fluctuators @xmath73 , where @xmath74 is the number of fluctuators and we drop the subscript @xmath33 to simplify the notation . each @xmath75 has a lorentzian power spectrum whose sum gives the power spectrum of the noise : @xmath76 in fact , the sum @xmath77 , with the @xmath78 belonging to the distribution @xmath79 , can be viewed as the monte - carlo sampling of the integral @xmath80 , which is , but for a constant , the one in eq . . therefore , in order to obtain a @xmath0 spectrum , it is necessary that a sufficiently large number of fluctuators is considered , and that the selected @xmath78 are a representative sample of the distribution @xmath79 in the range @xmath72 $ ] . this means that the minimum number of fluctuators we can sum - up depends on the range of integration . in fact , a big number of fluctuators is required to sample the distribution of the switching rates over a large range , while few fluctuators are sufficient in the case of a narrow range . note that we assume that all the fluctuators have the same coupling constant with the environment , that is @xmath81 for @xmath82 . the global evolution operator @xmath83 , for fixed values of the parameters associated to each fluctuator , permits to compute the density matrix of the global system as a function of a total noise phase @xmath84 . following the approach used in @xcite to evaluate the dynamics of two qubits subject to a single rtn , the time - evolved density matrix of the system at time @xmath85 can be expressed as : @xmath86 where @xmath87 is the global noise phase distribution . the density matrix in eq . ( [ rhomedio1 ] ) depends on the average of the phase factor @xmath88 , which can be computed in terms of the rtn coefficient @xmath89 of eq .. note that for the sake of simplicity , we omitted the subscripts @xmath90 in writing the @xmath89 coefficient . + we have @xmath91 where the last equality holds since the rtn phase coefficients are independent . by inserting eq . in eq . , we evaluate the two - qubit density matrix : @xmath92.\end{aligned}\ ] ] the coefficients appearing in the density matrix are @xmath93 note that the fluctuators have fixed switching rates @xmath94 , @xmath82 . + for the sake of completeness , here we report also a third scenario which describes an environment with a @xmath0 spectrum . this is the case of two qubits subject to a collection of fluctuators whose switching rates are known with an uncertainty determined by the @xmath46 distribution in the finite range @xmath72 $ ] . the sources of rtn do not have a fixed @xmath1 , so they are described by an ensemble of many fluctuators . the dynamics is evaluated by averaging the two - qubit density matrix for specific values of the switching rates over the @xmath1 in a range @xmath72 $ ] : @xmath95 where @xmath96 and @xmath97 . the time evolution also in this case preserves the x shape of the density matrix and the evolved state has the same functional expression of eq . , but with coefficients @xmath98^{n_f}\label{newcoeff}\ ] ] where @xmath70 are given in eq . . in this section we present analytical expressions for the negativity and the quantum discord for qubitq interacting with a single random fluctuator and a collection of @xmath74 bistable fluctuators , with either fixed or random switching rates . as we will see in details in the following , negativity and discord show the same qualitative behavior in all the analyzed regimes . the reason for this behavior stays in two features of the physical systems we are addressing : a ) the choice of a maximally entangled state as the initial state of the two qubits ; b ) the sheer dephasing nature of the interaction . overall , these two facts imply that the evolved state is a mixture of bell states . for this class of states discord is a function of negativity only , as it follows straightforwardly from eqs . ( [ negativity ] ) and ( [ discordluo ] ) . in the case of two qubits interacting with a single random fluctuator , the dynamics depends upon the selected range @xmath72 $ ] of the switching rate . following the definition of negativity in eq . , the negativity reads : @xmath99 where the @xmath100s are given in eq . ( [ lambdas ] ) . since the density matrix preserves its x shape during time evolution , the discord can be evaluated using eq . : @xmath101 where @xmath102\,.\ ] ] the dynamics of @xmath103 and @xmath104 is shown in fig [ fig2 ] , for the cases of pink and brown noise . the range of integration is @xmath105/\nu$ ] . negativity and discord have been obtained analytically , except for the integral @xmath70 , which must be computed numerically . ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random bistable fluctuator with spectrum @xmath106 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . bottom panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random fluctuator with spectrum @xmath107 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . , title="fig : " ] ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random bistable fluctuator with spectrum @xmath106 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . bottom panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random fluctuator with spectrum @xmath107 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . , title="fig : " ] + ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random bistable fluctuator with spectrum @xmath106 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . bottom panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random fluctuator with spectrum @xmath107 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . , title="fig : " ] ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random bistable fluctuator with spectrum @xmath106 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . bottom panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with a single random fluctuator with spectrum @xmath107 for different ( solid line ) and common ( dashed line ) environments when @xmath72/\nu=[10^{-4},10^{4}]$ ] . , title="fig : " ] quantum correlations decay with damped oscillations . in the case of a qubit subject to different environments with @xmath106 spectrum , the oscillations have a periodicity of @xmath108 . this periodicity can be explained by analyzing the analytical expressions of the quantum correlations . in particular , we note that in the integral of eq . , the @xmath109 functions exhibit damped oscillations for @xmath110 with periodicity @xmath111 , and for @xmath112 monotonically decay . their weighted superposition leads to an interference effect that can be summarized as follows : the oscillating components result in the formation of alternatively positive and negative peaks spaced by @xmath113 . on the other hand , the monotonic decaying components combine to cancel the negative peaks and to preserve the positive ones . finally we are left with an oscillating function with periodicity of @xmath108 . the same concept applies in the case of a common environment , but now the @xmath114 sum up in an oscillating function with periodicity of @xmath113 . the @xmath107 noise spectrum leads to oscillating functions of time with periodicity @xmath113 and @xmath115 for different and common environments respectively . again this periodicity is related to the fact that with such a distribution , the selected values of @xmath1 accumulate near the lower value of the frequency range , thus leading to a beat phenomenon with constructive interference with the above mentioned periodicity . if different ranges of integration are considered , different time - behavior for the quantum correlations can arise , but we will not analyze this effects in this paper . in the case of two qubits interacting with a collection of fluctuators with fixed switching rates , the dynamics is very different , depending on the spectrum of the noise . the analytical expression for the negativity and the discord is computed starting from the density matrix in eq . . the negativity reads : @xmath116 where the @xmath117 s are given in eq . ( [ gammas ] ) . also in this case the density matrix preserves the x form during time evolution , such that the discord is calculated by applying the luo formula : @xmath118 the negativity is the product of many oscillating coefficients , with various periodicities . in the case of switching rates taken from a @xmath119 distribution , the product of these terms results in a monotonic decay for both entanglement and discord . in fig . [ fig3 ] we report the behavior of such quantities in the case of 20 and 100 fluctuators . as the number of fluctuators is increased the quantum correlations decay faster . we consider 20 sources of rtn as the minimum number of fluctuator needed to obtain both a reliable profile of the frequency spectrum and a representative sample of the @xmath44 distribution . though it is possible to obtain a pink noise spectrum even with a smaller number of fluctuators , this approximation does not describe a sample of @xmath120-distributed switching rates . ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment.,title="fig : " ] ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment.,title="fig : " ] + ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment.,title="fig : " ] ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment.,title="fig : " ] a very different behavior arises when the @xmath1 s are selected from a @xmath121 distribution , see fig . phenomena of sudden death and revivals appear for both entanglement and discord . as the number of fluctuators is increased , the heights of the peaks decrease . the peaks have a periodicity of @xmath113 and @xmath115 for different and common environments respectively . as in the case of the single random fluctuator , this is explained by considering that the selected switching rates have small and very close values . the product of functions with almost the same periodicity gives a periodic behavior with narrow peaks . our results clearly show that the sheer knowledge of the spectrum is not sufficient to determine the dynamical evolution of correlations . indeed , it is also the number of decoherence channels that plays a key role . different physical models of environments can lead to the same spectrum . but , if the two - qubit system interacts with only one decoherence channel , then revivals appear because the system is affected only by one source of classical noise and the information can flow back . if many sources of decoherence are present , then the information can be completely lost , depending on the channel characteristics , that is the distribution of the switching rates . for the sake of completeness , we report also the expressions for negativity and discord in the case of two qubits subject to a collection of bistable fluctuators with stochastic switching rates . from eq . , we can write the negativity and discord as : @xmath122^{n_f}\\ n_{ce}(t)&=\|\gamma'_{ce}(t)\|=\left\|[\lambda_{de(ce)}]^{n_f}\right\|\\ q_{de(ce)}(t)&= h(\gamma'_{de(ce)})\,,\end{aligned}\ ] ] the quantum correlations decay exponentially in the case of @xmath106 noise and with smooth damped oscillations in the case of @xmath107 noise . the latter is the most general case , in which the collection of fluctuators is described by a statistical ensemble . also in this scenario , brown noise leads to non - monotonic decay of negativity and discord , since all the selected @xmath60 terms have similar periodicity and beat effects arise , with constructive interference . ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment . , title="fig : " ] ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment . , title="fig : " ] + ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment . , title="fig : " ] ( color online ) upper panels : time evolution of negativity ( left ) and discord ( right ) for two qubits interacting with two independent environments , consisting in a collection of @xmath74 bistable fluctuators with spectrum @xmath107 when @xmath72/\nu=[10^{-4},10^{4}]$ ] . lower panels : same as before , but with qubits subject to a common environment . , title="fig : " ] we have analyzed in details the dynamics of quantum correlations , entanglement and discord , for a two - qubit system initially prepared in a maximally entangled state and then subjected to classical noise , i.e. coupled with an external environment consisting in a collection of classical bistable flcutuators . in particular , due to its relevance for solid state devices , we have addressed environments characterized by a @xmath0 noise spectrum . we analyzed both the case of independent environments associated to each qubit and of a common environment , and we have taken into account explicitly the structure of the environment . more specifically , we have analyzed distinct configurations of the environment , characterized by the same @xmath0 spectrum , but with a different number of decoherence channels . analytical expressions have been found for the negativity and the discord , showing the same qualitative behavior in all the analyzed regimes . indeed the choice of maximally entangled initial state together with the sheer dephasing nature of the interaction with the environment , makes discord a function of negativity only . in these systems , in fact , the evolved state is a mixture of bell states . in the first configuration we have addressed , the two qubits interact with a single bistable fluctuator , which has a random switching rate leading to an overall @xmath0 spectrum . an oscillating behavior of quantum correlations for both pink and brown noise has been found . the effect of a common environment is that of better preserving the correlations and to double the number of revivals compared to the case of independent baths . in the presence of pink noise , the peaks of negativity and discord decay faster than in the presence of brown noise . in fact , the switching rates are more evenly distributed over the range @xmath72 $ ] than in the case of the @xmath107 noise , thus leading to massive destructive interference . in the second configuration the two qubits interact with a collection of bistable fluctuators , each one with fixed switching rate chosen from a distribution @xmath46 ; also in this case the overall noise spectrum is @xmath0 . in the case of pink noise , entanglement and discord show a monotonic decay , while for brown noise , sudden death and revivals occur . quantum correlations can be written as a product of oscillating and exponential functions . since the switching rates of the fluctuators are selected from a distribution @xmath0 , for the pink noise they lead to destructive interference , while brown noise enhances constructive interference . the action of independent or common environments has different effects on the robustness of correlations and this agrees with previous results obtained with different noise models @xcite . our results clearly show that the behavior of quantum correlations in not influenced only by the spectrum of the environment , but also by the number of decoherence channels , i.e. by the very structure of the environment . with a single decoherence channel , revivals of correlations appear , indicating that a back - flow of information from the environment to the system is possible . when the number of decoherence channels is increased , the information is quickly lost , and no revivals can occur . the characterization and the control of quantum correlations are fundamental for the development of quantum technology . not only quantum correlations constitute a resource to process quantum information , but the deep understanding of their nature will provide a better insight on the nature of quantum states themselves and on the transition between quantum and classical description of physical systems . our results indicate that the microscopic structure of environment , besides its noise spectrum , is relevant for the dynamics of quantum correlations , and may be a valid starting point for the engineering of colored non - gaussian environments . this work has been supported by miur ( firb lichis - rbfr10yq3h ) . mgap thanks sabrina maniscalco and elisabetta paladino for useful discussions . 10 k. modi , a. brodutch , h. cable , t. paterek , v. vedral , rev . phys . * 84 * , 1655 ( 2012 ) . h. ollivier and w. h. zurek , phys . rev . lett . * 88 * , 017901 ( 2001 ) . j. maziero , l. c. cleri , r. m. serra , and v. vedral , phys . rev . a * 80 * , 044102 ( 2009 ) . a. datta and s. gharibian , phys . a * 79 * , 042325 ( 2009 ) . t. werlang , s. souza , f. f. fanchini , and c. j. villas boas , phys . a * 80 * , 024103 ( 2009 ) . a. datta , a. shaji , and c. m. caves , phys . 100 * , 050502 ( 2008 ) . b. p. lanyon , m. barbieri , m. p. almeida , and a. g. white , phys . rev . lett . * 101 * , 200501 ( 2008 ) . p. giorda and m. g. a. paris , phys . lett . * 105 * , 020503 ( 2010 ) . g. adesso and a. datta , phys . lett * 105 * , 030501 ( 2010 ) . lett * 109 * , 180402 ( 2012 ) . m. gu , h. m. chrzanowski , s. m. assad , t. symul , k. modi , t. c. ralph , v. vedral , p. k. lam , nature phys . * 8 * , 671 ( 2012 ) . l. s. madsen , a. berni , m. lassen , u. l. andersen , phys . lett . * 109 * , 030402 ( 2012 ) . d. giulini and _ et al . _ , ( springer , 1996 ) . w. h. zurek , rev . phys . * 75 * , 715 ( 2003 ) . hu and h. fan , ann . phys . * 327 * 851 ( 2012 ) ; m .- hu , h. fan ann . 327(9 ) : 2343 - 2353 ( 2012 ) s. maniscalco , s. olivares , m. g. a. paris , phys . rev . a * 75 * , 062119 ( 2007 ) ; r. vasile , s. olivares , m. g. a. paris , s. maniscalco , phys . rev . a * 80 * , 062324 ( 2009 ) . r. vasile , p. giorda , s. olivares , m. g. a. paris , s.maniscalco , phys . a * 82 * , 012313 ( 2010 ) . a * 83 * , 042321 ( 2011 ) . m. g. genoni , p. giorda , m. g. a. paris , phys . a * 78 * , 032303 ( 2008 ) ; g. brida , i. p. degiovanni , a. florio , m. genovese , p. giorda , a. meda , m. g. a. paris , a. shurupov , phys . * 104 * , 100501 ( 2010 ) . a. ferraro , m. g. a. paris , phys . lett * 108 * , 260403 ( 2012 ) . w. s. warren , science * 277 * , 1688 ( 1997 ) . j. g. filgueiras , t. o. maciel , r. e. auccaise , r. o. vianna , r. s. sarthour , i. s. oliveira , q. inf . proc . * 11 * , 1883 ( 2012 ) . f. buscemi , p. bordone , and a. bertoni , phys . b * 81 * , 045312 ( 2010 ) ; f. buscemi , phys . a * 83 * , 012302 ( 2011 ) ; f. buscemi , p. bordone and a. bertoni , j . phys . : condens . matter * 21 * 305303 ( 2009 ) ; f. buscemi , p. bordone and a. bertoni , eur . j. d 66 , 312 ( 2012 ) . the term colored is a metaphor to suggest that white noise , in analogy with white light , comes from an environment which contains all different frequencies with a flat power spectrum , whereas other colors of noise can be obtained by selecting specific frequency ranges . j. majer , j. m. chow , j. m. gambetta , j. koch , b. r. johnson , j. a. schreier , l. frunzio , d. i. schuster , a. a. houck , a. wallraff , a. blais , m. h. devoret , s. m. girvin and r. j. schoelkopf , nature * 449 * , 443 ( 2007 ) . a. ludviksson , r. kree , and a. schmid , phys . * 52 * , 950 ( 1984 ) . s. m. kogan and k. e. nagaev , solid state comm . , * 49 * , 387 ( 1984 ) . d. a. averin , sol . state comm . * 105 * , 659 ( 1998 ) . l. b. ioffe , v. b. geshkenbein , m. v. feigelman , a. l. fauchre , g. blatter , nature * 398 * , 679 ( 1999 ) . mooij , t. p. orlando , l. levitov , l. tian , c. h. van der wal , s , lloyd , science * 285 * , 1036 ( 1999 ) . g. falci , r. fazio , g. m. palma , j. siewert1 , v. vedral , nature * 407 * , 355 ( 2000 ) . r. e. johanson , m. gunes , s.o . kasap , j. non - cryst . sol . * 266 - 269 * , 242 ( 2000 ) . b. raquet , a. anane , s. wirth , p. xiong , and s. von molnr , phys . * 84 * , 4485 - 4488 ( 2000 ) . g. jung , y. paltiel , e . zeldov , y. myasoedov , m. l. rappaport , m. ocio , s. bhattacharya , m. j. higgins proc . spie * 5112 * , 222 ( 2003 ) . g. vidal and r. f. werner , phys . a * 65 * , 032314 ( 2002 ) .	we address the dynamics of entanglement and quantum discord for two non interacting qubits initially prepared in a maximally entangled state and then subjected to a classical colored noise , i.e. coupled with an external environment characterized by a noise spectrum of the form @xmath0 . more specifically , we address systems where the gaussian approximation fails , i.e. the sole knowledge of the spectrum is not enough to determine the dynamics of quantum correlations . we thus investigate the dynamics for two different configurations of the environment : in the first case the noise spectrum is due to the interaction of each qubit with a single bistable fluctuator with an undetermined switching rate , whereas in the second case we consider a collection of classical fluctuators with fixed switching rates . in both cases we found analytical expressions for the time dependence of entanglement and quantum discord , which may be also extended to a collection of flcutuators with random switching rates . the environmental noise is introduced by means of stochastic time - dependent terms in the hamiltonian and this allows us to describe the effects of both separate and common environments . we show that the non - gaussian character of the noise may lead to significant effects , e.g. environments with the same power spectrum , but different configurations , give raise to opposite behavior for the quantum correlations . in particular , depending on the characteristics of the environmental noise considered , both entanglement and discord display either a monotonic decay or the phenomena of sudden death and revivals . our results show that the microscopic structure of environment , besides its noise spectrum , is relevant for the dynamics of quantum correlations , and may be a valid starting point for the engineering of non - gaussian colored environments .
since the discovery of the large thermoelectric power @xcite and then of superconductivity @xcite in the @xmath10 layered cobaltate family , these systems have attracted a lot of attention . indeed , they present a very rich phases diagram , as a function of the sodium content @xmath6 , and the possible water intercalation . for small @xmath6 ( @xmath11 ) the magnetic susceptibility does not depend on temperature . these systems are thus supposed to be paramagnetic metals , however some authors reinterpreted the data ( using nmr measurements @xcite ) and conclude that these systems should be antiferromagnetic metals . for larger @xmath6 ( @xmath12 ) the compounds are curie - weiss metals . over @xmath13 , the system has be seen as a weakly ferromagnetic metal @xcite , however neutron diffraction measurements @xcite rather sees a - type antiferromagnetism , that is ferromagnetic @xmath9 layers , antiferromagnetically coupled . for special values of the sodium content , @xmath14 @xcite , @xmath15 @xcite and @xmath16 @xcite , the system is insulating . a band insulator for @xmath16 and a charge / spin ordered state for @xmath17 . on the contrary , for @xmath18 the system remains conducting @xcite , exhibiting a fermi liquid behavior , while its half - filling character associated with the large electron correlation of the cobalt @xmath5 let expect a mott insulator . the @xmath10 compounds are composed of @xmath19 layers in the @xmath20 plane , between which the sodium ions are located . the @xmath19 layers are formed by @xmath21 edge - sharing octahedra forming a two - dimensional triangular lattice of cobalt ions . the @xmath21 octahedra present a trigonal distortion along the crystallographic @xmath22 axis corresponding to a compression of the oxygen planes toward the cobalt one . additional distortions take place , according to the sodium ordering . in the pure @xmath19 compound , although tarrascon _ @xcite suggested a complex crystallographic structure , further authors refined the structure with a single cobalt crystallographic position @xcite . this system exhibits a geometry quite characteristic compared to the rest of the @xmath23 family . first , the inter - layer distance is very short . indeed , the inter - layer distance is 4.48 , 4.31 or 4.25 according to the authors , while for @xmath24 it ranges from 5.2 to 5.7 and increases with decreasing @xmath6 . second , the @xmath9 layer is very compressed in the @xmath22 direction . indeed , its thickness is only 1.71 while it ranges between 1.92 and 1.98 for the @xmath10 family . in fact , in this system , the @xmath9 layer is even more compressed than in the superconducting , hydrated compound , for which its thickness is 1.76 . it results in a very large trigonal distortion of the @xmath21 octahedra , with 62.3 degrees between the @xmath22 axis and the co o directions . using the results of reference , this strong trigonal distortion lets us expect a strong @xmath2@xmath3 splitting . third , the co o distances are shorter than in all the other compounds of the family , with only 1.847 . @xmath25 + @xmath26 + in the @xmath27 compound ( p2 structure , subject of the present work ) , the sodiums ordering is commensurate and presents two types of alternated stripes along the @xmath28 direction @xcite . in the first stripes , the sodium atoms are located on - top of ( below ) the cobalts atoms , and in the second type of stripes the sodium ions are located on - top of ( below ) the center of the cobalt triangles ( see figure [ f : struct ] ) . it results two types of cobalt atoms organized in stripes along the @xmath28 direction , namely the co(1 ) atoms ( located on top of a sodium ion ) and the co(2 ) ones . compared to the other systems of the family , the @xmath27 compound does not stick out @xcite . indeed , its @xmath9 layers present a thickness of 1.97 . the average co o distances are 1.887for co(1 ) and 1.900 for co(2 ) . the average trigonal distortions are associated with angles of 58.9 degrees ( between the @xmath22 axis and the co(1)o directions ) , and 59.4 degrees ( co(2)o ) . these angles are weaker than for the superconducting system or the pure @xmath9 one , but still quite larger than the 54.7 degrees of the regular octahedron . one can thus expect a splitting of the @xmath29 cobalt @xmath5 orbitals in two quasi - degenerate @xmath3-type orbitals of low energy and a @xmath2-type one at the fermi level . out of all the @xmath30 systems , the @xmath17 compound presents a very rich but not well understood phases diagram . indeed , the @xmath31 ions order at very large temperature , namely slightly above 300 k @xcite , inducing a small charge order in the cobalt layer . despite this small charge order the system remains conducting . at 88 k a magnetic phase transition occurs , associated with a structural distortion , and the onset of a long range magnetic order . again , despite this transition , the system remains conducting up to 53 k where a small charge gap ( 14 mev ) finally opens @xcite . the understanding of this complex phase diagram is still under process . different hypotheses have been advanced such as successive fermi surface nesting @xcite , polarons formation @xcite , one band versus three bands cross - over @xcite . as already mentioned , the @xmath9 compound remains conducting for all temperatures , opposite to the mott insulator behavior that was expected due to its half - filled character . the low temperature dependence of the magnetic susceptibility lead some authors to suggest that this compound is a weakly correlated metal @xcite . however , as for the other compounds with @xmath32 , these data were reinterpreted @xcite in view of nmr measurements as due to strong antiferromagnetic coupling . in order to understand the low energy physics of these systems it is thus necessary to have accurate determinations of the pertinent degrees of freedom and of their coupling , as a function of the exact crystallographic structure . indeed , all authors agree on the importance of the sodium ordering and the induced cobalt local environment distortions , on the low energy properties . the aim of this paper is to provide such informations using state of the art quantum chemical spectroscopy methods for the @xmath17 and @xmath18 compounds . the next section will shortly recall the methods . section iii will presents our results and finally the last section will sum up our conclusions . the method used in this work ( cas+ddci @xcite ) is a configurations interaction method , that is an exact diagonalization method within a selected set of slater determinants , on embedded crystal fragments . this method has been specifically designed to accurately treat strongly correlated systems , for which there is no single - determinant description . the main point is to treat exactly all correlation effects and exchange effects within a selected set of orbitals ( here the @xmath5 shell of the cobalt atoms ) as well as the excitations responsible for the screening effects on the exchange , repulsion , hopping , etc . integrals . the cas+ddci method has proved very efficient to compute , within experimental accuracy , the local interactions ( orbital energies , atomic excitations , exchange and transfer integrals , coulomb repulsion etc . ) of a large family of strongly correlated systems such as high @xmath33 copper oxides @xcite , vanadium oxides @xcite , nickel and cuprate fluorides @xcite , spin chains and ladders @xcite , etc . the clusters used in this work involve one and two cobalt atoms ( @xmath34 and @xmath35 ) and their oxygen first coordination shell ( see figure [ f : clus ] ) . these fragments are embedded in a bath designed so that to reproduce on them the main effects of the rest of the crystal ; that is the madelung potential and the exclusion effects of the electrons of the other atoms of the crystal on the clusters electrons . the electrostatic potential is reproduced by a set of point charges located at the atomic positions . the charges are renormalized next to the bath borders , using the gell s method , in order to obtain an exponential convergence of the madelung potential @xcite . the convergence accuracy was set in the present work to the mili - electron - volt . the nominal atomic charges used in this work are the formal charges , that is @xmath36 and @xmath37 for the cobalt atoms of the @xmath17 and @xmath18 compounds , @xmath38 for the oxygen atoms and @xmath39 for the sodium atoms . the exclusion effects are treated using total ions pseudo - potentials @xcite ( tip ) on the first shell of atomic sites surrounding the clusters . the calculations were done using the molcas @xcite and casdi @xcite suites of programs . the basis sets used can be found in reference . the structural parameters were taken for the @xmath40 system from the 10 k data of reference and for the @xmath41 system from reference . the results of the ab initio calculations on an embedded fragment of the crystal provide us with energies ( @xmath42 ) and eigenfunctions ( @xmath43 ) for the local states from which the crystal ground and low lying excited states are built . even though these eigenfunctions are expanded over millions of determinants , they are dominated by only few of them , that are usually considered as the implicit basis of the effective models ( @xmath0 or hubbard in the present case ) . from the effective hamiltonian theory ( for instance such as derived in the quasi - degenerate perturbation theory @xcite ) one can extract the main concept necessary to get the effective hamiltonian parameters from the ab initio results . indeed , the effective hamiltonian on the fragment should verify _ at the best _ the following set of equations @xmath44 where @xmath42 and @xmath43 are the ab initio energies and eigenstates , and @xmath45 is the projection operator on the local configurations supporting @xmath46 . for instance , in the present case , these configurations can be the determinants supporting the local singlet and triplet state in the @xmath0 or hubbard model on the @xmath2 orbitals . the @xmath46 parameters are thus obtained from a mean - square fit minimization of equation [ eq : heff ] . these mean - square fit equations can sometime be inversed and the parameters obtained from simple algebraic equations . this is for instance the case for the effective exchange @xmath47 that is the singlet - triplet excitation energy of a @xmath48 fragment . similarly , the effective transfer can be simply expressed ( see equation [ eq : t ] in section [ res : tj ] ) from the same excitation energy and the knowledge of the wave - functions . one should however remember that the values of the effective integrals strongly depend on the nature of the effective model . indeed the processes renormalizing the integrals in one model can be explicit in another and one should take this into account in order to avoid double counting . let us illustrate our purpose on the @xmath29 orbital energy splitting . in a hubbard model based on the cobalt @xmath5 orbitals the @xmath21 fragment low lying states can be expressed as @xmath49 % \psi_1&=&\|e_{g1}^\prime e_{g2}^\prime \overline{e_{g2}^\prime } a_{1 g } \overline{a_{1g}}\rangle \\ \text{with energy } & & 2\varepsilon_{a_{1 g } } + \varepsilon_{e_{g1}^\prime } + 2\varepsilon_{e_{g2}^\prime } + 2u + 8v - 4j_h\\[2ex ] % \psi_2&=&\|e_{g1}^\prime \overline{e_{g1}^\prime } e_{g2}^\prime a_{1 g } \overline{a_{1 g } } \rangle \\ \text{with energy } & & 2\varepsilon_{a_{1 g } } + 2\varepsilon_{e_{g1}^\prime } + \varepsilon_{e_{g2}^\prime } + 2u + 8v - 4j_h\end{aligned}\ ] ] where @xmath50 is the repulsion of two electrons on the same orbital , @xmath51 and @xmath52 the repulsion and exchange of two electrons on different @xmath5 orbitals . it results that if @xmath53 , @xmath54 and @xmath55 are the ab initio energies associated with these states , the ab initio spectrum should be associated with the orbital energies differences @xmath56 on the contrary if one extends the model in order to explicitely treat the differences according to the nature of the @xmath5 orbitals in the two - electrons two - orbitals repulsion and exchange integrals , then the ab initio spectrum does no more correspond to pure orbital energy differences but rather to @xmath57 in these equations the two electrons part is non zero as soon as the @xmath58 orbitals are not pure @xmath29 ones , but rather @xmath29-@xmath4 hybrids , as it is the case as soon as the trigonal distortion sets in ( see reference and section [ res : hyb ] for the present results ) . whether the racah s parameters are explicitely included in the model or the spherical approximation used , one should thus use different @xmath59 effective orbital energy differences in order for the effective hamiltonian to fit the same ab initio spectrum . in the case of @xmath9 layers it means different ordering for the @xmath2 and @xmath58 orbitals ( see reference ) . as usually done , we will use the spherical approximation in this work and renormalize the orbital energies by the effect of the racah s parameters . a strong controversy has been going on in the literature about the cobalt 3d orbital splitting . indeed , authors did not agree on the relative order of the @xmath29 orbitals under the trigonal distortion of the regular octahedron ( @xmath60 ) . the origin of the controversy was the fact that the crystalline field theory @xcite as well as some lda calculations @xcite found the @xmath2 orbital of lower energy than the @xmath3 ones , while other density functional calculations @xcite , as well as quantum chemical calculations @xcite or arpes experimental results @xcite found them of reverse order . we recently showed @xcite that the origin of the controversy was an incorrect treatment of the exchange and correlation integrals within the 3d shell . indeed , only when these effects are taken into account with their whole complexity that is when the orbital dependence of the two - electron two - orbital on - site repulsion and exchange are correctly treated , the relative order between the @xmath2 and @xmath3 orbitals issued from the @xmath29 is correctly predicted . unfortunately this is not the case in the lda approximation ( essentially due to the local exchange ) and thus lda predicts incorrect @xmath2@xmath3 ordering . this splitting is however one of the crucial parameter for the determination of the nature of fermi level states as recently shown by marianetti _ et al _ @xcite . indeed , they showed ( using dmft calculations ) that the existence of the @xmath3 pockets in the fermi surface of the @xmath23 compounds are directly related to the this splitting . one can thus conclude that the discrepancy between lda and arpes fermi surfaces can be directly related to the improper treatment of the b racah parameter in lda . in fact , one can expect that an orbital dependant lda+u , where the whole complexity of the racah parameters would be included , should yield fermi surfaces in agreement with the experimental ones . we thus computed the @xmath29 orbitals splitting in the two compounds , for the different cobalt sites . these values can be extracted from the spectroscopy of the @xmath61 embedded fragments . indeed , the excitation energy between the three cobalt states , where one hole is located on one of the @xmath29 orbitals , yield the effective splittings between the @xmath2 and @xmath3 orbitals . the relative order of the @xmath29 orbitals is displayed on figure [ f : orbsplt ] . one sees immediately that the @xmath2 orbitals are always higher than the @xmath3 ones as can be expected from reference . in the @xmath17 system , both sites exhibit @xmath2 orbitals about 250300 mev higher than the @xmath3 ones . similarly , in our dimer calculations we see that the @xmath62 ion is globally of lower energy than the @xmath63 ion . indeed , our calculations yield a global charge order of about 0.17 electrons in favor of the co(1 ) site . this result is in agreement with the common sense expectations that will favor a smaller valence for the cobalt with the largest number of @xmath31 cations neighbors . indeed , computing the electrostatic potential difference between the two cobalt sites one finds the co(1 ) site about 315 mev lower than the co(2 ) site . this result is however in contradiction with the bond valence sum ( bvs ) model , since the later yields a charge ordering of 0.12 electron in favor of the co(2 ) site @xcite , however it is in agreement with the lda+u calculations @xcite ( @xmath64 in favor of co(1 ) ) . in the pure @xmath9 system , the @xmath2-@xmath3 orbital splitting is very large , much larger than in any of the other @xmath65 systems , even larger than in the superconducting compound . this very large splitting can be easily explained from ( i ) the large trigonal distortion ( the largest observed in this family ) and ( ii ) the very small co o distances . indeed , it was shown in reference that ( i ) the orbital splitting increases with increasing trigonal distortion and ( ii ) that the slope of the splitting as a function of the distortion increases linearly with decreasing co o distances . let us now focus on the cobalt 3d orbital hybridization . these orbitals can hybridize in two ways : with the oxygen ligands 2p orbitals , but also within the cobalt 3d shell between the @xmath29 and @xmath4 sets of the regular octahedron . indeed , it was shown ( see ref . for details ) that not only this hybridization is symmetry allowed by the trigonal distortion of the octahedron , but also that it is responsible for the lowering of the @xmath58 orbitals compared to the @xmath2 one . the @xmath29@xmath4 hybridization angle is reported in table [ t : hybd ] and found to be non negligible , of the same order of magnitude as found in the other @xmath9-based systems . the @xmath29@xmath4 hybridization is larger for @xmath18 than for @xmath17 , as can be expected from the larger trigonal distortion . c\|ddd orbital & & + & & & + @xmath66 & 11.59 & 11.54 & 11.86 + @xmath67 & 8.21 & 10.18 & 11.86 + as far as the cobalt ligand hybridization is concerned , we found it to be weak for the @xmath2 and @xmath3 orbitals , with less than 10% weight on the oxygens . it is however quite large for the empty @xmath4 orbitals with 40 to 45% weight on the neighboring oxygens . let us now focus on the interactions between two first neighbor cobalt atoms . many authors assume that the low energy physics of this compound is determined by the @xmath2 orbitals only . the large @xmath2@xmath3 orbital splitting found in our calculations seem to support this assumption . we thus computed the effective integrals both for a @xmath0 model and a one - band hubbard model . analyzing the low temperature crystallographic structure of @xmath40 ( 10 k data of reference ) , one sees that there are four types of independent @xmath68@xmath68 bonds ( see figure [ f : ab ] ) , the co(1)co(1 ) bond , the co(2)co(2 ) bond and two kinds of co(1)co(2 ) bonds , namely those where the cobalt cobalt interactions are mediated by one o(1 ) and one o(3 ) oxygen ligands and those where the cobalt cobalt interactions are mediated by one o(2 ) and one o(3 ) oxygen ligands ( see figure [ f : ab ] ) . the effective magnetic exchange integrals are @xmath69})= -19\,\text{mev } & j(12^{[23]})= -19\,\text{mev}\end{aligned}\ ] ] where @xmath70 refers to the co(i)co(j ) bond and the @xmath71 $ ] superscript refers to the bridging oxygens . for the pure @xmath9 system the magnetic coupling was computed to be much larger with @xmath72 in our notations , the negative integrals correspond to antiferromagnetic coupling . we thus find all exchange integrals , both in the @xmath18 and @xmath17 compounds , antiferromagnetic , in agreement with the neutron scattering @xcite and nmr @xcite data . one should notice that the exchange integrals are strongly modulated according to the local environment . indeed , they differ by a factor larger than two for the different bond of the @xmath17 system . it results that this system will be described by a strongly inhomogeneous @xmath0 model . in the @xmath18 compound the afm exchange is very large , in agreement with nmr experimental results @xcite . the difference in magnitude between the antiferromagnetic coupling of the @xmath18 , @xmath17 and superconducting system @xcite can easily be explained by the variations in the co o distances . indeed , these distances control for a large part the super - exchange term of the coupling that increases with decreasing distances . it results that following de vaulx _ et al _ @xcite we found that afm fluctuations increase with decreasing @xmath6 . let us now focus on the hopping integrals . one can get them either from the charge fluctuation in the singlet state of the @xmath73@xmath74 fragments , or from the spectroscopy of the @xmath75@xmath74 fragments . a simple analysis shows that the two hopping integrals are somewhat different . in the first case , there is only one spectator electron on the bond , while in the second case , there are two of them ( see figure [ f : hop ] ) . it results in a different evaluation of the hopping integrals according to the number of spectator electrons . @xmath76 where the @xmath77 and @xmath78 labels refers to the arbitrary _ `` left '' _ and _ `` right '' _ atoms in the fragment , and @xmath79 is the single electron hamiltonian part . @xmath80 can be extracted from the @xmath73@xmath81 fragment . indeed , mapping a single band hubbard model on the computed low - energy singlet and triplet states wave functions and energies , one gets after a little algebra @xmath82 assuming the following form for the computed singlet and triplet wave functions @xmath83 % \psi_{tp}&= & \frac{\|d_l\bar d_r\rangle - \|d_r\bar d_l\rangle } { \sqrt{2 } } + \text{small terms } \end{aligned}\ ] ] @xmath84 . @xmath85 is the effective on - site repulsion of a simple hubbard model . let us remind that it accounts for both the average on - site repulsion between the two sites , @xmath86 , and the nearest neighbor effective repulsion between the cobalt atoms , @xmath87 . the ( electron-)hopping integrals between the @xmath2 orbitals thus come as @xmath88 } ) = 115\,\text{mev } & t^{1\bar e}_{00}(12^{[23 ] } ) = 111\,\text{mev}&\end{aligned}\ ] ] for the @xmath17 compound and as @xmath89 for the @xmath18 one . once again , one sees that the effective transfer integrals are strongly dependant on the local environment of the cobalts and that the @xmath40 system is more inhomogeneous than first thought . in the pure compound , the effective transfer between the @xmath2 orbitals of nn atoms is larger than both in the @xmath17 and the superconducting system . the relative variations of the co o and large co co distances between these three systems support the idea that the effective transfer integrals are dominated by the ligand - mediated term rather than the direct hopping term , despite the short metal - metal distances in these edge - sharing compounds . the transfer integrals with two spectator electrons on the bond , @xmath90 , can be extracted from the low - energy states of the @xmath91@xmath74 fragments . there are six low energy states according to the localization of the hole on the @xmath2 and @xmath3 orbitals . the associated effective hamiltonian can be written as @xmath92 where @xmath93 are the atomic effective orbital energies ( see figure [ f : orbsplt ] ) , @xmath94 are the effective transfer integrals ( direct plus mediated by the oxygen ligands ) between the @xmath95 orbital of one cobalt and @xmath96 of the other . @xmath97 are effective intra - atomic transfer integrals between the @xmath98 and @xmath99 orbitals of the same atom . one may be surprised to find such terms , since the direct integrals are zero due to the @xmath100 symmetry , however the coupling with the bridging oxygen @xmath101 orbitals yield in second order perturbation theory an effective intra - atomic transfer term of @xmath102 where @xmath103 and @xmath104 refers to the @xmath5 orbitals of the same cobalt atom , the sum over @xmath105 runs over all the ligand bridging orbitals , @xmath106 is the cobalt @xmath95ligand hopping integrals and @xmath107 is the excitation energy toward the ligand - to - metal charge transfer configuration . for a more detailed description of the underlying mechanism , one can refer to reference . summing the @xmath97 contributions coming from the six oxygens around a cobalt atom , one can show that they exactly cancel out in a symmetric system . in the present system , where the atomic @xmath108 symmetry is not exactly respected , these terms will certainly not exactly cancel out , however , one can expect that their sum will remain very weak . table [ t : t2e ] displays the different effective integrals involved in matrix [ eq : h ] for the @xmath17 system . .effective ( electron-)hopping integrals between @xmath2 and @xmath3 orbitals of two neighboring cobalt atoms ( in mev ) . [ cols="^ , < , < , < , < , < , < " , ] one sees immediately that the dominant transfer integrals are the inter - atomic hopping between two @xmath2 orbitals and two @xmath66 orbitals , in agreement with what we found on the superconducting compound . all other integrals are relatively weak ( @xmath109 ) and can probably be omitted in a simple model . comparing these values with what we found on the superconductor system , one sees that they are of the same order of magnitude , except for the inter - atomic hopping between the @xmath2 orbital of one cobalt and the @xmath67 orbital of the other . indeed , the larger value found in the superconducting systems ( -53 mev ) is now spread over the three @xmath110 , @xmath111 and @xmath112 hopping terms , due to local symmetry breaking . let us now examine what we find for on - site the repulsion @xmath50 from our calculations . indeed , as detailed in equation [ eq : u ] , the value of @xmath85 can be extracted from the singlet and triplet energies and wave functions on the @xmath73@xmath74 fragments . results for a single band model are presented in table [ t : u ] . @xmath113 \hline \rm coo_2 & 3.71 & 15.6 \\[1ex ] \rm [ na_{0.35}coo_2][h_2o]_{1.3}~\cite{thsylvain } & 3.65 & 19.4 \\[1ex ] \rm na_{0.5}coo_2 & & \\ \hfill u(1)-v(11 ) & 2.61 & 31.1 \\ \hfill u(2)-v(22 ) & 2.86 & 20.6 \\ \hfill \overline{u } - v(12^{[13 ] } ) & 2.76 & 24.0 \\ \hfill \overline{u } - v(12^{[23 ] } ) & 2.56 & 23.1 \end{array}\ ] ] one sees immediately that the hubbard on - site repulsion decreases with increasing sodium content , that is with increasing number of electrons in the @xmath9 layers . once again correlating this fact with the co o distances one sees that the larger the co o distances , the weaker the cobalt on - site repulsion . this can be understood by the fact that a larger volume of the coordination shell around the metal atom will allow its magnetic orbitals to spread over some more and this reduce the intra - orbital repulsion . if one now looks at the correlation strength ( @xmath114 ) rather than the on - site repulsion value , one sees that the tendency is reverse . indeed , the larger the sodium content , the larger the correlation strength . it is easy to see that the opposite variation of the correlation strength and the one - site repulsion , as a function of @xmath6 , is again due to the co o distances . indeed , if an increase of the metal coordination sphere induce a decrease of the effective @xmath50 , it induces a much stronger reduction of the effective hopping integral @xmath115 , thus resulting in weaker electronic correlation . this results fit quite well with what was originally supposed ( from experimental observations ) in these systems , that is a decrease of the correlation strength with @xmath6 . however it seems hard to say that these systems are weakly correlated , since the smallest @xmath114 ratio is 15 . we determined the effective on - site and nearest - neighbor coupling parameters for the the @xmath27 and @xmath9 compounds within hubbard and @xmath0 models . the effective integrals and orbital energies were computed using ab - initio quantum chemical methods treating exactly the strong correlation effects within the cobalt @xmath5 shell and the screening effects up to the double - excitations . the hubbard on - site repulsion and the correlation strength @xmath114 were found to vary in opposite ways with the sodium content . indeed , when doping the @xmath116 layer , the @xmath115 value decreases more rapidly than the @xmath50 value and the resulting correlation strength increases . for the @xmath18 limit , however , the system is still strongly correlated with @xmath8 . this result is in disagreement with onoda s conclusion that the @xmath9 compound is a weakly correlated metal @xcite . on the contrary , julien @xcite concludes the system is a strongly correlated metal close to the mott phase transition . our results clearly do agree with the latter hypothesis , indeed , it was shown using dmft calculation that the hubbard model reach the mott transition in a triangular lattice at a critical correlation strength of @xmath117 @xcite . our evaluation of the correlation strength : @xmath118 , clearly situates the @xmath9 compound very close to the mott transition . similarly , our results are in agreement with nmr measurements @xcite that sees an increase of the antiferromagnetic coupling with decreasing sodium content . however , contrarily to kawasaki _ @xcite we do not find the pure @xmath116 compound behaving on a specific way . indeed , we found it to display the largest antiferromagnetic coupling among all the systems we studied . we determined the ligand field splitting for the different crystallographically independent cobalts of the @xmath27 and @xmath116 compounds . as for the superconducting compound , we find that the @xmath2 orbitals are of higher energy than the @xmath58 ones , yielding a @xmath119 cobalt atomic configuration . comparing the @xmath29 splitting in these two compounds with the splitting in the superconducting compound , we see that it follows the behavior predicted in our previous work both as a function of the amplitude of the trigonal distortion and as the size of the cobalt first coordination shell . indeed , in the @xmath9 compound , the average distance between the oxygen and cobalt planes is 0.85 while it is 0.88 in the superconducting system and ranges between 0.95 and 0.98 in the @xmath17 compound . as this distance increases with increasing electron doping of the @xmath9 layer , the splitting between the @xmath58 and @xmath2 orbitals decreases , being maximum for the pure system . comparing this orbital order with the lda or lda+u calculations one sees that it is reverse order . indeed , tight - binding fitting of the lda+u calculations yield a higher hole energy by about 0.01ev when located in the @xmath2 orbital @xcite . why do lda calculations yield a reverse order between the @xmath2 and @xmath3 atomic orbitals than wave - function cluster calculations ? the answer is in the treatment of the coulomb and exchange terms within the 3d shell . indeed , in lda calculations the correlation and exchange contributions are treated using a local approximation . it means that the small differences in energy that should occur in these term according to the nature of the occupied 3d orbitals ( the @xmath120 racah s parameter ) is ignored . this is even worse with lda+u calculations since the hartree term is also replaced and a unique @xmath50 ( or rather @xmath50 , @xmath51 and @xmath52 ) is used while the coulomb and exchange integrals between two different @xmath5 orbitals depend on the nature of the considered orbitals . these integrals differences are quite small and , in most cases , the averaging is valid enough . in the present case , however , it can be shown that the 2-electron contribution to the orbital energy that is responsible for the @xmath2@xmath3 inversion in reference , is a direct function of the racah s @xmath120 integral , and of the coulomb and exchange integrals differences . the effects of these imperfect exchange and correlation treatment within the 3d shell , and thus of the resulting orbital order between the @xmath2 and @xmath3 orbitals is to pop up the band issued from the later in comparison with the bands issued from the former . indeed , it was shown by marianetti _ @xcite ( using dmft calculations ) that when the @xmath2 orbital is supposed of higher energy ( lower hole energy ) than the @xmath3 orbital , the @xmath3 pockets in the fermi surface disappear . more recently bourgeois _ @xcite showed that using orbital splitting of the same sign and order of magnitude as in the present work , dmft fermi surfaces and band structure nicely fit the arpes data ( they studied different systems ranging from @xmath121 to @xmath122 ) . in view of our results and the above analysis we think that the @xmath3 fermi pockets found in lda calculations and not observed in arpes experiments are due to an intrinsic feature of the lda exchange and correlation functional and thus that these pockets are true artefacts of the method . finally , one should remember the strong variation of the effective exchange and hopping integrals as a function of the local distortions . this fact can be of importance in the understanding of the changes in magnetic behavior , as a function of @xmath6 , in the @xmath65 family . it would thus be of interest to see if the local distortions observed for larger @xmath6 values will be able to reverse the sign of the magnetic exchange as suggested by neutron diffraction experiments . the authors thank daniel maynau for providing us with the casdi suite of programs . these calculations where done using the cnrs idris computational facilities under project n@xmath1231842 and the crihan computational facilities under project n@xmath1232007013 . s. p. bayracki , i. mirebeau , p. bourges , y. sidis , m. enderle , j. mesot , d. p. chen , c. t. lin , b. keimer , phys . lett . * 94 * , 157205 ( 2005 ) ; l. m. helme , a. t. boothroyd , r. coldea , d. prabhakaran , d. a. tennant , a. hiess , and j. kulda , phys . * 94 * , 157206 ( 2005 ) . j. miralles , j. p. daudey and r. caballol , chem . . lett . * 198 * , 555 ( 1992 ) ; v. m. garca _ * 238 * , 222 ( 1995 ) ; v. m. garca , m. reguero and r. caballol , theor . chem . acc . * 98 * , 50 ( 1997 ) .	we used quantum chemical ab initio methods to determine the effective parameters of hubbard and @xmath0 models for the @xmath1 compounds ( x=0 and 0.5 ) . as for the superconducting compound we found the @xmath2 cobalt orbitals above the @xmath3 ones by a few hundreds of mev due ti the @xmath3@xmath4 hybridization of the cobalt @xmath5 orbitals . the correlation strength was found to increase with the sodium content @xmath6 while the in - plane afm coupling decreases . the less correlated system was found to be the pure @xmath7 , however it is still strongly correlated and very close = to the mott transition . indeed we found @xmath8 , which is the critical value for the mott transition in a triangular lattice . finally , one finds the magnetic exchanges in the @xmath9 layers , strongly dependant of the weak local structural distortions .
we introduced the concept of @xmath0-schemes in @xcite . in this paper , we will investigate further properties of @xmath0-schemes , mainly focusing on zariski - riemann spaces . first , we will show that there is a normalization of @xmath0-schemes , just as for ordinary schemes . this is important , since we are aiming for an analog of zariski s main theorem . one of the advantage of introducing @xmath0-schemes is that we can simplify the proof of nagata embedding theorem ; it can be proven intuitively , as in the original paper of nagata @xcite . note that the essential part is already proven in corollary 4.4.6 of @xcite . compare with the proof of conrad @xcite , which only uses ordinary schemes , but is long ( approximately 50 pages ) . also , we introduce the notion of approximable @xmath0-schemes : an @xmath0-scheme is approximable if it is a ( filtered ) projective limit of ordinary schemes . this notion is convenient , since locally free sheaves on approximable schemes always come from a pull back of a localy free sheaves on ordinary schemes . at the same time , we see that the zariski - riemann space defined in @xcite is identified with the conventional one , namely the limit space of @xmath1-admissible blow ups along the exceptional locus @xmath2 where @xmath1 is an open subscheme of a scheme @xmath3 . this shows that the conventional zariski - riemann space has the desirable universal property in the category of @xmath0-schemes , not only with schemes . this paper is organized as follows . in section 1 , we quickly review the definitions and properties of @xmath0-schemes , which plays the central role in this paper . in section 2 , we construct the normalization functor of @xmath0-schemes . in section 3 , we give a notion of approximable @xmath0-schemes . this actually determines the complete hull of the category @xmath4 of ordinary schemes in the category @xmath5 of @xmath0-schemes , namely the smallest complete full subcategory of @xmath5 containing @xmath4 . in section 4 , we will give a proof of the original version of nagata embedding , which says that any separated scheme of finite type can be embedded as an open subscheme of a proper scheme . * notation and conventions . * in this paper , the algebraic type @xmath0 is always that of rings when we talk of @xmath0-schemes . when we say ordinary schemes , we treat only coherent schemes and quasi - compact morphisms between them ; to emphasize this assumption and to distinguish ordinary coherent schemes from @xmath0-schemes , we will say @xmath6-schemes instead of coherent schemes . for an @xmath0-scheme @xmath3 , the description @xmath7 stands for the underlying topological space , which is coherent . an @xmath0-scheme @xmath3 will be called _ integral _ if it is irreducible and reduced . this condition is in fact , stronger than assuming any section ring @xmath8 is integral . a morphism of @xmath0-schemes is _ proper _ , if it is separated and universally closed . we _ do not include the condition of finite type"_. an open covering of an @xmath0 scheme @xmath3 is denoted by @xmath9 ; here , @xmath10 is a quasi - compact open covering of @xmath3 , and @xmath11 is a quasi - compact open covering of @xmath12 for each @xmath13 . hence , there are open immersions @xmath14 and @xmath15 for each @xmath16 . moreover , if @xmath3 is a @xmath6-scheme , we usually take @xmath17 and @xmath18 as open affine subschemes of @xmath3 . in this section , we will recall some terminologies and definitions in @xcite . a good reference for general lattice theories is @xcite . a topological space @xmath3 is _ coherent _ , if it is sober , quasi - compact , quasi - separated , and has a quasi - compact open basis . the category @xmath19 of coherent spaces and quasi - compact morphisms is isomorphic to the opposite category @xmath20 of distributive lattices by the functor @xmath21 ; for a coherent space @xmath3 , we may regard @xmath22 as the set of quasi - compact open subsets of @xmath3 , or the set of their complements . a quasi - compact morphism @xmath23 induces a morphism @xmath24 of lattices . therefore , for any algebraic type @xmath0 , we can regard an @xmath0-valued sheaf on a coherent space @xmath3 as a continuous covariant functor @xmath25 . on a coherent space @xmath3 , there is a canonical @xmath26-valued sheaf @xmath27 on @xmath3 , which is defined by @xmath28 for quasi - compact open @xmath1 ; this extends uniquely to the entire zariski site of @xmath3 . we have a functor @xmath29 from the category of commutative rings , which sends a ring @xmath30 to the set of finitely generated ideals of @xmath30 modulo the relation @xmath31 . note that this gives the usual spectrum of rings , when combined with the previous isomorphism @xmath21 . also , we have a natural homomorphism @xmath32 of multiplicative monoids , sending @xmath33 to the principal ideal generated by @xmath34 . this homomorphism commutes with localizations . an @xmath0-scheme is a triple @xmath35 , where a. @xmath7 is a coherent space ( the underlying space " ) , b. @xmath36 is a ring - valued sheaf on @xmath7 ( the structure sheaf " ) , and c. @xmath37 is a morphism of @xmath26-valued sheaves ( the support morphism " ) , which satisfies the following condition : for an inclusion @xmath38 of quasi - compact open subsets of @xmath7 , the restriction map @xmath39 factors through @xmath40 , where @xmath40 is the localization along the multiplicative system @xmath41 where @xmath42 is the complement closed subset of @xmath43 in @xmath1 . by this property , @xmath0-schemes are locally ringed spaces . a morphism of @xmath0-schemes @xmath44 is a morphism of ringed spaces , which commutes with the support morphism : @xmath45^{\alpha_{1}f^{\ # } } \ar[d]_{\beta_{y } } & f_{}\alpha_{1}\scr{o}_{x } \ar[d]^{\beta_{x } } \\ \tau_{y } \ar[r]_{f^{-1 } } & f_{}\tau_{x } } \ ] ] the category @xmath5 of @xmath0-schemes is complete , and co - complete . we have a fully faithful functor @xmath46 , which preserves pull backs and finite patchings by quasi - compact opens . let @xmath47 be a dominant morphism of @xmath0-schemes , where @xmath48 is integral and @xmath49 is a field . the _ zariski - riemann space _ @xmath50 is a proper @xmath0-scheme over @xmath48 , defined as follows : the points of the underlying space @xmath51 corresponds to the set of dominant morphisms @xmath52 of local rings , where @xmath30 s are valuation rings of @xmath49 , and @xmath53 are local rings of @xmath48 . the map @xmath54 is defined naturally . the topology of @xmath51 is generated by the _ domains _ @xmath55 for @xmath56 and the inverse images of the open subsets of @xmath48 . for a quasi - compact open subset @xmath1 of @xmath51 , @xmath57 is the ring which is the intersection of all the valuation rings corresponding to the points in @xmath1 . for arbitrary dominant morphism @xmath58 of integral @xmath0-schemes , @xmath59 is defined by the pushout of @xmath60 , where @xmath61 is the rational function field of @xmath3 . @xmath3 is _ of profinite type _ strongly of profinite type _ ) over @xmath48 , if the canonical map @xmath62 is a scheme - theoretic immersion ( resp . open immersion ) . here , we would like to list up some important properties which is going to be used in the sequel : 1 . a separated , of finite type morphism of @xmath6-schemes is always strongly of profinite type . 2 . the map @xmath63 gives the universal proper @xmath0-scheme of profinite type over @xmath48 for each integral @xmath0-scheme @xmath3 of profinite type over @xmath48 . in this section , we fix an integral base @xmath0-scheme @xmath48 , and any @xmath0-scheme is integral , of profinite type over @xmath48 . we denote by @xmath64 the category of integral @xmath0-schemes of profinite type over @xmath48 and dominant morphisms . an @xmath0-scheme @xmath3 is _ normal _ , if the ring of every stalk @xmath65 is integrally closed . we do not assume noetherian property on normal rings ( or schemes ) in this paper . let @xmath66 be the full subcategory of @xmath64 , consisting of normal schemes , and @xmath67 be the underlying functor . then , @xmath1 has a right adjoint ` @xmath68 ' . moreover , the counit @xmath69 is proper dominant . we will refer to this right adjoint as the _ normalization functor_. the proof is somewhat long , so we will divide it into several steps . the construction of the normalization functor is analogous to that of zariski - riemann spaces , described in detail in @xcite . we will denote by @xmath70 the integral closure of a given integral domain @xmath30 in the sequel . * first , we will construct the underlying space of the normalization of a given integral @xmath0-scheme @xmath3 . let @xmath71 be the set of finite sets of pairs @xmath72 , where a. @xmath1 is a quasi - compact open subset of @xmath3 , and b. @xmath73 . + let @xmath74 , @xmath75 be two elements of @xmath71 . we define two operations @xmath76 , @xmath77 on @xmath71 by @xmath78 for a pair @xmath72 , define @xmath79 $ ] as @xmath80=\{x \in u \mid \rom{$x$ is in the image of $ \spec \scr{o}_{x , x}[\alpha^{-1}]\to \spec \scr{o}_{x , x}$}\}.\ ] ] for two elements @xmath74 , @xmath75 , the relation @xmath81 holds if a. @xmath82 \subset \cup_{j}v_{j}[\beta_{j}]$ ] for any @xmath83 , and b. for any @xmath84 $ ] , set @xmath85\}$ ] . then @xmath86 generates the unit ideal in @xmath87 $ ] . + let @xmath88 be the equivalence relation generated by @xmath89 , and set @xmath90 . the addition and multiplication of @xmath71 descends to @xmath91 , which makes @xmath91 into a distributive lattice . set @xmath92 . this is the underlying space of the normalization @xmath93 . * there is a natural homomorphism @xmath94 of distributive lattices , defined by @xmath95 . this defines a quasi - compact morphism @xmath96 of coherent spaces . * let @xmath97 be a point of @xmath98 , and set @xmath99 . then , @xmath100 becomes a prime ideal of @xmath101 . let @xmath102 be the localization of @xmath101 by @xmath103 . then , @xmath102 dominates @xmath65 . * the structure sheaf @xmath104 is defined by @xmath105 where @xmath49 is the function field of @xmath3 . the support morphism @xmath106 is defined by @xmath107 this defines an @xmath0-scheme @xmath108 . * we have a canonical morphism of sheaves @xmath109 , defined by the identity @xmath110 . this yields a morphism @xmath111 of @xmath0-schemes . it is of profinite type , by the criterion 4.3.3 in @xcite . * let us show that @xmath112 is proper . + we can see from the construction that we have a natural morphism @xmath113 : the morphism @xmath114 of underlying spaces is defined by @xmath115 where @xmath116 , and the morphism between the structure sheaves is canonical . note that @xmath117 is already normal . this shows that @xmath93 is proper over @xmath3 by the valuative criterion . * we will show that the normalization is a functor . let @xmath23 be a dominant morphism of @xmath0-schemes . @xmath118 is defined by @xmath119 the morphism @xmath120 extends canonically to @xmath121 . this gives a functor @xmath122 . * it remains to show that the normalization functor is indeed the right adjoint of the underlying functor . the unit @xmath123 is the identity , since the normalization of a normal @xmath0-scheme is trivial . the counit @xmath124 is given by @xmath112 defined above . we know that the integral closure @xmath125 of a noetherian domain @xmath30 is not noetherian @xcite . therefore , we must drop the ` of finite type ' condition from the definition of properness , if we wish to say that the normalization @xmath126 is proper " . [ lem : normal : local ] let @xmath3 be a normal @xmath0-scheme . then . @xmath8 is normal for any open @xmath1 . let @xmath127 be an element which is integral over @xmath8 , where @xmath49 is the function field of @xmath3 . since @xmath128 is integral over the stalk @xmath65 for any @xmath129 and @xmath65 is integrally closed , we have @xmath130 . hence , @xmath131 . the normalization functor coincides with the usual normalization , when restricted to @xmath6-schemes . first , we will show for affine schemes @xmath132 . the universality of the normalization functor gives a canonical morphism @xmath133 . since @xmath134 is normal , we have a canonical homomorphism @xmath135 . this yields a morphism @xmath136 . it is easy to check that these two morphisms @xmath137 are inverse to each other . it is obvious from the construction that normalization commutes with localizations . this shows that the normalization of any @xmath6-scheme coincides with the usual definition . we fix an integral base @xmath6-scheme @xmath48 in the sequel . the next proposition is pure category - theoretical and easy , so we will omit the proof . let @xmath138 be two categories , with @xmath139 finite complete and @xmath140 small complete . let @xmath141 be a finite continuous functor , namely @xmath142 preserves fiber products . for any object @xmath34 of @xmath140 , the followings are equivalent : a. @xmath34 is isomorphic to a limit of the objects in @xmath143 . b. @xmath34 is isomorphic to a filtered limit of the objects in @xmath143 . let @xmath3 be an @xmath0-scheme , and @xmath144 be a class of @xmath6-schemes . 1 . @xmath3 is _ approximable by @xmath144 _ , if @xmath3 is isomorphic to a filtered limit of some objects of @xmath144 . 2 . @xmath3 is _ approximable _ , if @xmath3 is isomorphic to a filtered limit of some @xmath6-schemes . any approximable @xmath0-scheme is approximable by @xmath6-schemes of finite type . it suffices to show that any @xmath6-scheme is approximable by @xmath6-schemes of finite type . let @xmath3 be any @xmath6-scheme , and @xmath145 be a finite affine covering of @xmath3 . since @xmath14 is quasi - compact , @xmath18 is of finite type over @xmath17 . thus , we have approximations @xmath146 and @xmath147 so that @xmath148 and @xmath149 are of finite type and @xmath150 are open immersions . we may also assume that the above limits are filtered . since filtered limits and finite colimits commute , we have @xmath151 and @xmath152 is a @xmath6-scheme of finite type . let @xmath153 be two integral @xmath0-schemes . a morphism @xmath23 is _ birational _ , if @xmath154 induces an isomorphism @xmath155 between the rational function fields . note that , the morphism being birational does not imply the existence of an open dense subset @xmath1 of @xmath3 such that @xmath156 . let @xmath3 be an approximable @xmath0-scheme , say @xmath157 where @xmath158 s are @xmath6-schemes . 1 . if @xmath3 is reduced , then @xmath3 is approximable by reduced @xmath6-schemes . 2 . if @xmath3 is integral , then @xmath3 is approximable by integral @xmath6-schemes . 3 . further , if the rational function field @xmath159 is finitely generated over an integral base @xmath6-scheme , then @xmath3 is approximable by integral @xmath6-schemes birational to @xmath3 . if @xmath3 is normal , then @xmath3 is approximable by normal @xmath6-schemes . if @xmath3 is proper and approximable by separated @xmath6-schemes , then @xmath3 is approximable by proper ( and of finite type ) @xmath6-schemes . the proofs are all similar , so let us just see ( 1 ) . since @xmath3 is reduced , @xmath160 factors through the reduced @xmath6-scheme @xmath161 . this shows that @xmath162 . let @xmath23 be a morphism of @xmath0-schemes over @xmath48 , with @xmath3 approximable and @xmath163 a @xmath6-scheme , of finite type over @xmath48 . 1 . suppose @xmath3 is a filtered projective limit @xmath164 of @xmath6-schemes . then , @xmath154 factors through @xmath160 for some @xmath165 . 2 . furthermore , if @xmath3 is proper over @xmath48 and approximable by separated @xmath6-schemes , and @xmath163 is separated over @xmath48 , then the above @xmath158 can be chosen to be a proper scheme over @xmath163 . 1 . we may assume that @xmath163 is affine . since @xmath163 is of finite type and @xmath166 is a filtered colimit of @xmath167 , @xmath154 factors through @xmath158 for some @xmath165 . 2 . by the above proposition , we may assume that @xmath158 s are proper over the base scheme @xmath48 . since @xmath163 is separated , the morphism @xmath168 is proper . [ thm : str : sheaf : direct ] let @xmath23 be a proper birational morphism , where @xmath3 is an integral @xmath0-scheme approximable by separated @xmath6-schemes , and @xmath163 a normal @xmath6-scheme separated and of finite type over @xmath48 . then , @xmath169 . the previous proposition shows that @xmath154 factors through proper morphisms @xmath170 , where @xmath157 and @xmath171 is a filtered system of integral @xmath6-schemes , proper birational and of finite type over @xmath163 . since @xmath163 is normal , the usual zariski s main theorem tells that @xmath172 is an isomorphism ( corollary iii . 11.4 of @xcite ) , and @xmath173 coincides with the right hand side , since it is a colimit of @xmath174 s . since of profinite type " morphisms are stable under taking limits , approximable @xmath0-schemes are necessarily of profinite type over @xmath48 . let @xmath3 be a normal @xmath0-scheme , proper and of profinite type over the integral base @xmath6-scheme @xmath48 . assume that the rational function field @xmath159 is finitely generated over @xmath175 . the followings are equivalent : a. @xmath3 is approximable by separated @xmath6-schemes . b. let @xmath176 be any quasi - compact open covering of @xmath3 . then , there exists a refinement @xmath177 of @xmath178 such that @xmath179 are open immersions . ( i)@xmath180(ii ) : @xmath3 can be written as a filtered limit @xmath157 , where @xmath158 s are normal @xmath6-schemes , proper and of finite type over @xmath48 . since the number of @xmath17 s and @xmath18 s are finite , @xmath181 , @xmath182 for some @xmath183 , where @xmath184 s and @xmath185 s are quasi - compact open subsets of @xmath158 . take any refinement @xmath186 of @xmath187 , by affine opens @xmath188 and @xmath189 . set @xmath190 and @xmath191 . since @xmath192 is proper and @xmath189 is normal , of finite type , theorem [ thm : str : sheaf : direct ] implies that @xmath193 . this shows that @xmath179 are open immersions . ( ii)@xmath180(i ) : for any covering @xmath176 of @xmath3 , the refinement @xmath177 gives open immersions @xmath179 which patches up to give a @xmath6-scheme @xmath194 and the canonical morphism @xmath195 . the covering @xmath178 is a pull back of a covering of @xmath194 , and ditto for the elements of @xmath196 s . from this observation , we see that the induced morphism @xmath197 is an isomorphism . it is clear from the construction that @xmath194 is proper . in the sequel , any @xmath0-schemes are integral . let @xmath48 be a @xmath6-scheme , and @xmath3 be a @xmath6-scheme over @xmath48 . we say that @xmath3 is _ compactifiable _ over @xmath48 , if there is an open immersion @xmath198 where @xmath163 is a @xmath6-scheme , proper , of finite type over @xmath48 . ( i)@xmath180(ii ) : there exists an open immersion @xmath198 into a @xmath6-scheme @xmath163 , proper of finite type over @xmath48 . this morphism factors through @xmath59 by the universal property . we will show that for any quasi - compact open subset @xmath1 of @xmath59 , there exists a proper birational morphism @xmath200 , such that @xmath201 for some quasi - compact open subset @xmath43 of @xmath202 , where @xmath203 is the canonical extension of @xmath204 : @xmath205 \ar@{.>}[d ] \ar@{}[dr]\|{\box } & \zr^{f}(x , s ) \ar@{.>}[d]_{g } \ar[rd]^{f } & x \ar[l ] \ar[d ] \\ v \ar@{.>}[r ] & y^{\prime } \ar@{.>}[r ] & y } \ ] ] by the construction of @xmath59 , we may assume @xmath1 is of the form @xmath206 , where @xmath207 is a quasi - compact open subset of @xmath48 and @xmath208 is a finite subset of @xmath209 , and @xmath210 note that @xmath211 is open in @xmath163 , since @xmath199 is an open immersion . suppose @xmath212 , where @xmath213 locally . let @xmath200 be the blow up along @xmath214 . then , either @xmath215 or @xmath216 is in @xmath217 locally , which shows that the domain of @xmath215 is open in @xmath202 . this shows that @xmath1 is the pull back of some @xmath43 by the morphism @xmath203 . hence , @xmath218 becomes a homeomorphism on the underlying space , where @xmath219 is the filtered projective limit of @xmath3-admissible blow - ups of @xmath163 . a similar argument shows that @xmath220 also becomes an isomorphism . note that @xmath221 s are separated over @xmath48 , since we only used blow - ups . ( ii)@xmath180(i ) : the zariski - riemann space @xmath59 can be written as a form @xmath222 , where @xmath221 s are proper , of finite type @xmath6-schemes . since @xmath199 is an open immersion and @xmath3 is quasi - compact , @xmath223 becomes an open immersion for some @xmath165 . in this section , we will prove this theorem for the essential case , namely when @xmath48 and @xmath3 are integral . this restriction is due to the fact that we simply havent established the theorem of zariski - riemann spaces for non - integral schemes . consider @xmath59 . since @xmath3 is separated , of finite type over @xmath48 , the morphism @xmath199 is an open immersion by corollary 4.4.6 of @xcite . let @xmath227 ( resp . @xmath228 ) be the complement of the closure of @xmath229 ( resp . @xmath230 ) in @xmath59 . we can see that @xmath231 , since the interior of the complement of @xmath232 is empty . next , we see that @xmath233 . for this , it suffices to show that @xmath234 . suppose there is a point @xmath97 in @xmath235 . since @xmath236 and @xmath230 are coherent subsets of @xmath59 , @xmath97 must be a specialization of some @xmath237 and @xmath238 by corollary 1.2.8 of @xcite . since @xmath239 is surjective , there are inverse images @xmath240 of @xmath241 such that @xmath242 specializes to @xmath97 . the points in @xmath50 are valuation rings , hence @xmath243 must be the specialization of @xmath244 , or the converse . in either cases , this contradicts to the fact that @xmath245 and @xmath246 has no specialization - generalization relations . this also shows that @xmath227 and @xmath228 are quasi - compact . the morphism @xmath247 induces an isomorphism on @xmath227 , hence @xmath227 is approximable by @xmath6-morphisms of finite type over @xmath48 , ditto for @xmath228 . take any @xmath6-model @xmath248 of @xmath249 ( namely , a morphism @xmath250 where @xmath248 is a @xmath6-scheme ) such that the morphism @xmath251 induces an isomorphism on @xmath252 . then , @xmath253 and @xmath254 can be patched along @xmath255 to obtain a @xmath6-scheme @xmath163 of finite type , and a surjective morphism @xmath256 . this shows that @xmath163 is proper .	we define the notion of normal @xmath0-schemes , and approximable @xmath0-schemes . approximable @xmath0-schemes inherit many good properties of ordinary schemes . as a consequence , we see that the zariski - riemann space can be regarded in two ways either as the limit space of admissible blow ups , or as the universal compactification of a given non - proper scheme . we can prove nagata embedding using zariski - riemann spaces .
the material under consideraion here is build from picene and fluorinated tcnq molecules forming molecular crystals or films with an 1:1 composition which can be synthesized by co - evaporation @xcite . the main building block in the crystal is a dimer like structure of picene and tcnq with flat orientation with respect to each other . therefore , we start our discussion with a short comparison of the electronic structure of the free picene-@xmath0 -dimer and anthraquinone derivate based on density functional theory ( dft ) calculations . electronic structure close to the fermi level as obtained from density functional theory calculations for ( a ) the anthraquinone derivate and ( b ) the picene-@xmath0 -dimer . ] the electronic structure of the anthraquinone derivate as depicted in fig . [ fig : elevels ] a ) has already been discussed in detail elsewhere @xcite . it shows the typical properties of a @xmath3-conjugated molecular material ( semiconductor ) having a homo - lumo gap of about @xmath4 and an almost symmetric level arrangement . the homo and homo-1 levels are delocalized and span the whole molecule including the thiol anchor units . the sprawling side structures guarantee that electron withdrawing or pushing effects of the anchor units modify only slightly the underlying electronic structure of the anthraquinone core . the electronic structure of the picene-@xmath0 system on the other hand is determined by the occurrence of hybrid orbitals which are formed between the @xmath3-conjugated picene and the @xmath0 acceptor . the homo-1 and lumo orbital of picene-@xmath0 are formed from states of the free picene and @xmath0 . this hybridization induces a charge transfer of about @xmath5 from the picene to the @xmath0 @xcite . based on the electronic structure calculations the transport properties have been obtained using the non - equilibrium green function formalism ( negf ) . both molecular systems are sandwiched between two au(111 ) leads . [ negf_iv_curves ] shows the i - v curves without gate voltage ( black lines ) for the a ) anthraquinone and b ) picene-@xmath0 system . both curves show typical details which are expected from the i - v - characteristics of a molecular junction . due to the weak coupling of both molecular systems to the au(111 ) leads both systems show features which can be attributed to distinct molecular orbitals . in a very simplified picture if bias voltage is rising then more orbitals will contribute to the conduction through the junction and the current increases . peaks or regions of negative differential resistance ( ndr ) occur if distinct levels contribute to the conduction on low bias and do not contribute in the case of high bias voltages due to e.g. the lowering of the coupling strength between the molecular orbital and the lead @xcite . such a situation can be seen for example in the anthraquinone system at @xmath6 and @xmath7 bias voltage and for the picene-@xmath0 system at @xmath8 . the effect of increasing bias voltage on the electron transmission spectra of the two molecular junctions is depicted in fig . [ negf_iv_curves ] c ) and d ) respectively . in the zero bias transmission function the picene-@xmath0 -system ( solid line ) has a prominent feature at @xmath9 above the fermi level @xmath10 which can be attributed to the lumo level of the dimer . the features right below @xmath11 are therefore linked to the homo , homo-1 etc . dimer orbitals . above @xmath12 picene-@xmath0 exhibits no features relevant to transport . the main effects of an applied bias voltage to a molecular junction are ( i ) shifting the transmission spectra with respect to @xmath13 , ( ii ) strengthening and dampening of transmission features due to the bias induced changes in the molecule - lead coupling and ( iii ) widening of the energy window in which transmission peaks contribute to the current . therefore if we apply a positive bias to the picene-@xmath0 junction we are shifting transmission features which correspond to occupied homo - n orbitals into the energy window relevant for conduction . for negative bias the transmission peaks originating from the homo and lumo are shifted out of that window and due to the large energy gap between the lumo and lumo+n levels ( see fig . [ fig : elevels]b ) there are no additional levels which can contribute to the conduction . only in the case of very high bias voltages @xmath14 additional transmission features can appear and the current starts to rise also for negative bias . for the anthraquinone junction the situation is completely different . we observe a somewhat lower density of transmission peaks for energies above @xmath11 ( see fig . [ negf_iv_curves]c ) . the changes in the bias voltage do not significantly change the overall density of transmission peaks in the energy window around @xmath11 contributing to the conduction . hence the absolute value of the current through the junction is approximately the same whether we apply a negative or positive bias voltage . our results are in good agreement with other theoretical estimates @xcite . in the picene-@xmath0 junction ( fig . [ negf_iv_curves]b ) the current for the negative bias case stays almost at zero up to an @xmath15 . for anthraquinone junction ( fig . [ negf_iv_curves]a ) the current rises more or less symmetrical in case of forward and reverse bias . this asymmetric behavior gives rise to large current rectification for the picene-@xmath0 system . the rectification ratio can be estimated by @xmath16 @xcite . we show a comparison of @xmath17 for the two systems in the voltage range of interest in fig . [ negf_ampl]b . the highest @xmath17 are achievable for very low values of @xmath18 however for real applications voltages between @xmath19 seem to be more manageable whereas higher bias voltages may lead to rapid degradation in the organic material @xcite . from our calculation we obtain rectification rations for @xmath20 which is much higher than the achievable maximum values for @xmath21 . moreover for the anthraquinone system we find a steep decrease of the rectification up to @xmath22 whereas the picene-@xmath0 systems shows only a weak dependence in the considered bias range . with its large rectification values the picene-@xmath0 -system also outweighs already reported maximum values for other charge transfer and molecular materials by a at least one order of magnitude @xcite . the low variability of @xmath23 over a wide bias range seems more appropriate for real world applications . in consequence we propose the picene-@xmath0 system as an molecular material to fabricate organic diodes due to its advantagoues forward and reverse bias properties . further , a required key element for implementing real molecular circuits is electrical switching . for that reason we compared the response of the two molecular systems to the application of an external gate voltage @xmath24 as schematically drawn in the insets of fig . [ negf_iv_curves ] . the i - v curves corresponding to a @xmath25 for the respective junctions are also shown as red lines in fig . [ negf_iv_curves]a and b. both systems show a rather strong response to the application of an external gate field . for better quantification of the effect the change of the current due to the applied gate voltage @xmath26 normalized to current without gate @xmath27 for both systems is presented in fig . [ negf_ampl]a . for the anthraquinone junction we find a constant increase of the current for @xmath28 between a factor two and three . other authors report very large on / off ratios for anthraquinone junctions of @xmath29 . however these ratios are achieved by chemical modification of the molecule itself and are potentially irreversible in contrast to the electrical switching presented here . additionally the already mentioned theoretical @xcite as well as recently measured data @xcite on anthraquinone systems all show almost symmetrical behavior of the i - v characteristics . the additional gate field in the junction induces two main effects . first the gate voltage shifts the molecular energy levels with respect to the energy window in which molecular orbitals contribute to the conduction ( see @xcite ) . in the present case the positive gate voltage results in higher number of occupied molecular levels contributing and hence the current increases . the second effect of the gate field is the induction of changes of the electronic structure of the molecule itself . dft calculations on the anthraquinone molecule with an applied electric field equal to the gate field in the transport calculations indicate that for example the homo - lumo gap of the molecule is reduced and the molecular level alignment changes as well . in fig . [ negf_ampl]a one can see that the amplification of the current due to the gate field for the anthraquinone junction is almost symmetrical with respect to the bias voltage . the same effects of a gate field also occur in case of the picene junction . however , due to the asymmetric character of the molecular orbitals around the fermi level the impact on the i - v characteristics is much larger . under forward bias the current is barely affected by the gate voltage . in fact the current shows even a small decrease . for the reverse bias case however we see a large increase of the current due to gate voltage . we achieve an maximum current increase by about @xmath30 at @xmath31 . in fig . [ negf_ampl]c we additionally show the dependence of the reverse bias current of the picene-@xmath0 junction as a function of the applied gate voltage . as mentioned before the large energy distance between the lumo and lumo+1 leads to almost no transmission for reverse bias . as the gate field shifts the molecular levels we see of course an current increase . however this level shifting alone does not explain the quantity of the observed effect . here a further mechanism comes into play for the picene-@xmath0 system . as discussed in the beginning the picene-@xmath0 is a weakly bonded dimer with charge transfer of @xmath32 electrons from the picene to @xmath0 . our dft calculations on the picene-@xmath0 dimer with an electric field applied perpendicular to the stacking direction (= transport direction ) reveal that the hybridization of the dimer itself depends strongly on the applied field . the applied electric field allows to tune the hybridization between the dimer components . this allows to lower the lumo - lumo+n distance drastically . the result is , that in the reverse bias case with the gate field switched on , the number of levels which account to the conduction is increased . this explains the very large switching ratios of @xmath30 . in addition we wish to point out , that this behavior corresponds perfectly to the arguments given by ratner and coworkers @xcite for achieving molecular rectification . local weak links in a molecule , given in our case through the hybridization in the picene-@xmath0 dimer , can result in large rectification ratios whereas strong bonding suppresses the effect . by direct modification of the hybrid levels using a gate field one can reach effective current switching . we believe that this mechanism is quite general for charge transfer systems and should be applicable to other dimer systems as well @xcite . to summarize , the negf+dft studies of the anthraquinone - au(111 ) and picene-@xmath0 -au(111 ) systems have shown that the respective systems exhibit fundamentally different i - v characteristics . the anthraquinone system shows an approximately ohmic behavior for low bias voltages and the application of an external gate field results in an increase of the overall current through the junction which is almost symmetric for positive and negative bias voltages . due to hybrid dimer states close to the fermi level the picene-@xmath0 i - v curve is very asymmetric with a pronounced diode - like forward / reverse current behavior . in contrast to the anthraquinone system the effect of an applied gate voltage is about two orders of magnitude larger in the reverse bias than in the forward bias case . further , we have shown that the anthraquinone system can also be seen as a electrically controllable switch . however , in terms of achieving maximum switching ratios the picene-@xmath0 junction shows a clear benefit and can be seen as an molecular transistor in terms of classical circuit elements . the anthraquinone system on the other hand offers almost symmetric and linear i - v characteristics in the @xmath33 range and may be better utilized as operational amplifier . consequently , we propose to use the pure organic interface between picene and @xmath0 as a straightforward way to manufacture a molecular switch with very large switching ratio or a molecular transistor / amplifier . + financial support by the deutsche forschungsgemeinschaft within the forschergruppe for 1154 , project ko1924/5 and by the saxonian cluster of excellence `` structure design of novel high - performance materials via atomic design and defect engineering ( adde ) '' is gratefully acknowledged . we especially thank the zih dresden for providing extensive computational resources and support . the ground state electronic structure of the molecules was investigated using the all - electron density functional theory ( dft ) nrlmol program package which achieves a high level of numerical accuracy ( see @xcite and references therein ) . for the exchange correlation gga / pbe @xcite was used and in all calculations dispersion correction utilizing the dft - d2 method @xcite was included . the geometry of the molecules was optimized using a gradient approach , the relaxation was terminated once all atomic forces were below 0.05 ev/@xmath34 . we applied the nonequilibrium green s function method ( negf ) for self - consistent calculation of the electron transport properties as implemented in the gpaw code @xcite to investigate the i - v characteristics of our model devices . for the transport calculations the electronic structure is obtained by dft calculations using the common approach of constructing a model device where the molecule of interest together with additional electrode atoms ( scattering region ) are sandwiched between two semi - infinite ( metallic ) electrodes . in our case we used at least three additional au(111 ) layers on each side of the molecule to construct the scattering region followed by a further geometry optimization step of the system where the topmost two gold layers together with the attached molecules were allowed to relax . for the scattering region as well as for the leads a localized double-@xmath35 polarized basis set was used . schematic drawings of the used model junctions are shown in the insets if fig . [ negf_iv_curves ] . the whole system can be subject to an external bias and/or gate voltage . the electronic structure of the scattering region and therefore the i - v curves are calculated self - consistently in the presence of such external fields . to support the deductions in this paper we just repeat the key facts of the dft - negf method of use whilst a detailed discussion of the method can be found in the cited literature . the gpaw transport code uses the green s function of the central region defined by @xmath36^{-1}\ ] ] where @xmath37 and @xmath38 are the overlap and hamilton matrix of the scattering region written in the localized basis . @xmath39 are the respective self energies of the leads . after self - consistency in the cycle @xmath40 is reached the transmission function is calculated by @xmath41\ ] ] with @xmath42 . therefore @xmath43 gives the transmission probability of an electron having an energy @xmath44 under an applied bias ( and gate ) voltage @xmath18 . further the current through the junction is obtained by @xmath45 were the electronic chemical potentials @xmath46 are connected to the applied bias voltage via @xmath47 ( @xmath48 elementary charge ) @xcite . de facto the current is calculated by integrating the self - consistent transmission function within the bias - dependent energy window spanned by @xmath46 . 100 aviram , a. and ratner , m. a. `` molecular rectifiers . '' 29 , 277283 ( 1974 ) . mujica , v. , ratner , m. a. and nitzan , a. `` molecular rectification : why is it so rare ? '' chem . 281 , 147150 ( 2002 ) . metzger , r. m. `` unimolecular electronics . '' 18 , 4364 ( 2008 ) . zhou , k .- `` can azulene - like molecules function as substitution - free molecular rectifiers ? '' 13 , 1588290 ( 2011 ) . tsuji , y. , staykov , a. and yoshizawa , k. `` molecular rectifier based on stacked charge transfer complex . '' c 116 , 25752580 ( 2012 ) . mahns , b. et al . `` crystal growth , structure , and transport properties of the charge - transfer salt picene/2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane . '' cryst . growth and design 14 , 1338 - 1346 ( 2014 ) . van dijk , e. h. , myles , d. j. t. , van der veen , m. h. and hummelen , j. c. `` synthesis and properties of an anthraquinone - based redox switch for molecular electronics . '' lett . 8 , 23336 ( 2006 ) . markussen , t. , schitz , j. and thygesen , k. s. `` electrochemical control of quantum interference in anthraquinone - based molecular switches . '' 132 , 224104 ( 2010 ) . seidel , n. et al . `` synthesis and properties of new 9,10-anthraquinone derived compounds for molecular electronics . '' new j. chem . 37 , 601 ( 2013 ) . zhao , p. and liu , d .- s . `` electronic transport properties of an anthraquinone - based molecular switch with carbon nanotube electrodes . '' chinese phys . 29 , 047302 ( 2012 ) . zhao , p. et al . `` first - principles study of the electronic transport properties of the anthraquinone - based molecular switch . '' physica b : condens . matter 406 , 895898 ( 2011 ) . kubatkin , s. et al . `` single - electron transistor of a single organic molecule with access to several redox states . '' nature 425 , 698701 ( 2003 ) . gudon , c. m. et al . `` observation of quantum interference in molecular charge transport . '' nanotechnol . 7 , 3059 ( 2012 ) . valkenier , h. et al . `` cross - conjugation and quantum interference : a general correlation ? '' phys . 16 , 65362 ( 2014 ) . darwish , n. et al . `` single molecular switches : electrochemical gating of a single anthraquinone - based norbornylogous bridge molecule . '' c 116 , 2109321097 ( 2012 ) . karthuser , s. `` control of molecule - based transport for future molecular devices . '' j. phys . condens . matter 23 , 013001 ( 2011 ) . schulze , g. et al . `` resonant electron heating and molecular phonon cooling in single c60 junctions . '' 100 , 136801 ( 2008 ) . bogani , l. and wernsdorfer , w. `` molecular spintronics using single - molecule magnets . '' mater . 7 , 17986 ( 2008 ) . lindner , s. , knupfer , m. , friedrich , r. , hahn , t. and kortus , j. `` hybrid states and charge transfer at a phthalocyanine heterojunction : mnpc / f16copc . '' 109 , 027601 ( 2012 ) . pederson , m. r. , porezag , d. v. , kortus , j. and patton , d. c. `` strategies for massively parallel local - orbital - based electronic structure methods . '' status solidi 217 , 197218 ( 2000 ) . porezag , d. and pederson , m. `` optimization of gaussian basis sets for density - functional calculations . '' a 60 , 28402847 ( 1999 ) . perdew , j. p. , burke , k. and ernzerhof , m. `` generalized gradient approximation made simple . '' 77 , 38653868 ( 1996 ) . grimme , s. `` semiempirical gga - type density functional constructed with a long - range dispersion correction . '' j. comput . 27 , 178799 ( 2006 ) . chen , j. , thygesen , k. s. and jacobsen , k. w. `` ab initio nonequilibrium quantum transport and forces with the real - space projector augmented wave method . '' phys . rev . b 85 , 155140 ( 2012 ) . enkovaara , j. et al . `` electronic structure calculations with gpaw : a real - space implementation of the projector augmented - wave method . '' j. phys . condens . matter 22 , 253202 ( 2010 ) . meir , y. and wingreen , n. `` landauer formula for the current through an interacting electron region . '' 68 , 25122515 ( 1992 ) .	we show that the recently synthesized charge - transfer material picene-@xmath0 can be used as a gate - voltage controlled molecular switch . the picene-@xmath0 system is compared with the extensive characterized anthraquinone - based molecular system , which is known to exhibit large switching ratios due to quantum interference effects . in case of picene-@xmath0 we find switching ratios larger by one order of magnitude . further , our calculations reveal that the picene-@xmath0 system resembles remarkable well the i - v characteristics of a classical diode . the reverse - bias current of this molecular diode can be increased two orders of magnitude by an external gate voltage . based on density - functional theory calculations we show that the hybrid states formed by the picene-@xmath0 system are playing the key role to determine the transport properties . we further conclude that the tuning of quantum transport properties through hybrid states is a general concept which opens a new route towards functional materials for molecular electronics . the basic idea of molecular electronics is that typical functionality needed to build integrated circuits has to be realized by single molecules . this idea was already part of the pioneering theoretical description of a molecular diode given by ratner and coworkers many years ago @xcite . since then it was repeatedly discussed @xcite why it is not easy to actually build a satisfying molecular version of the classical semiconductor diode . in practice we face the problem that in nearly all molecular devices the i - v characteristics is rather symmetrical even when using a highly asymmetric molecule . as a result the attempts to manufacture molecular diodes or transistors resulted in rather poor rectification or switching ratios compared to conventional semiconductor devices @xcite . the criteria for achieving diode - like high rectification ratios in molecular junctions where already suggested by ratner et al . @xcite . their proposal was that organic charge - transfer materials where two molecular subunits each carry an opposite charge should be suitable for the task . however , from this theoretical work it was also concluded that strong coupling between the molecular subunits may completely screen the desired effect . recently this idea was picked up by @xcite and they presented rectification ratios @xmath1 . in this paper we present theoretical results on an the recently synthesized charge transfer material picene-@xmath0 @xcite . we will show that we can reach much higher rectification ratios in the range @xmath2 which is at least one order of magnitude higher than other values reported before . we compare our theoretical results to calculations on an anthraquinone based molecular switch . the anthraquinone system was chosen because it is well characterized and known to perform as an molecular switch @xcite . furthermore theoretical calculations with different contact materials and reliable measurements are available for comparison with our results @xcite . we also investigate the dependence of the i - v characteristics on an external gate voltage . apart from electrochemical gating @xcite or redox active switching @xcite this seems to be the most promising way to manufacture working active electronic devices .
in his book the sensual ( quadratic ) form " conway @xcite described the following topographic " way to visualise the values of a binary quadratic form @xmath4 following conway we define the _ superbase _ of the integer lattice @xmath5 as a triple @xmath6 such that @xmath7 is a basis of the lattice and @xmath8 we construct a binary planar tree starting from this superbase , such that each base is identified with an edge and each superbase is identified with a vertex . it is easy to see that the edges and vertices which have some vector in common form a path , and we can therefore label each face by the common vector in the path which bounds it ( see fig . 1 ) . note that all _ primitive _ lattice vectors , which are not multiples of any other lattice vectors , appear on this tree . by taking values of the form @xmath0 on the vectors of superbase , we get what conway called the _ topograph _ of @xmath0 containing the values of @xmath0 on all primitive lattice vectors . in particular , if @xmath9 we have the values @xmath10 one can construct the topograph of @xmath0 starting from this triple using the following property of any quadratic form , which conway called the _ arithmetic progression rule _ and is known also in geometry as the _ parallelogram rule _ : @xmath11 note that quadratic forms ( in any dimension ) can be characterised as the degree 2 homogeneous functions satisfying the relation ( [ apr ] ) . this relation leads to the following relation on the topograph of @xmath0 ( see fig . 2 ) , which can be used to compute the primitive values of @xmath0 recursively . in particular , for the standard quadratic form @xmath12 we have the tree shown on the left of fig . 3 . on the right side of fig . 3 we show the corresponding _ farey tree _ , when at each vertex we have the fractions @xmath13 and its _ farey mediant _ @xmath14 ( see e.g. @xcite ) . using the farey tree we can identify the infinite paths @xmath15 on a binary tree with real numbers @xmath16 $ ] ( see more details in the next section ) . for example , for the golden ratio @xmath17 we have the fibonacci path shown in bold on both trees of fig . 3 . we would like to study the growth of the values of @xmath0 along the path @xmath18 . more precisely , define _ q()=_n , where @xmath19 with @xmath20 being the @xmath21-th triple on the path @xmath22 for example , for the fibonacci path with @xmath23 we have @xmath24 being every second fibonacci number with the growth @xmath25 we will show that a similar result is true for any positive binary quadratic form @xmath0 , namely that _ q()=2 ( ) , where @xmath26 is the function introduced in @xcite describing the growth of the markov numbers , or , equivalently , the growth of the monoid @xmath27 . the monoid @xmath27 consists of matrices from @xmath28 with non - negative entries . they all can be seen on the edges of the farey tree if we combine two neighbouring fractions @xmath13 into the matrix a= p & q + r & s + sl_2(z ) . the function @xmath26 can be defined as ( ) = _ n , where @xmath29 is attached to the @xmath21-th edge along path @xmath18 and @xmath30 is the _ spectral radius _ of the matrix @xmath31 , defined as the maximum of the modulus of its eigenvalues @xcite . the function @xmath26 can be extended to @xmath32 and has very peculiar properties : it is discontinuous everywhere , @xmath33-invariant and takes all real values from @xmath34 $ ] ( see @xcite ) . and the corresponding farey tree with marked golden " fibonacci path.,title="fig:",height=181 ] and the corresponding farey tree with marked golden " fibonacci path.,title="fig:",height=181 ] the situation is different for indefinite binary quadratic forms . the reason is the existence of what conway @xcite called the _ river _ , which , in the case of the forms not representing zero , is an infinite path on the binary tree separating positive and negative values of the form @xmath0 ( see fig . 4 ) . , height=151 ] let @xmath35 be the two real roots of the corresponding quadratic equation @xmath36 our main result says that for an indefinite form @xmath0 not representing zero _ q()=2 ( ) , _ with @xmath37 we discuss also in more detail the geometry of the corresponding exceptional paths in relation to the continued fraction expansions of the quadratic irrationals @xmath35 . the galois result about pure periodic continued fractions naturally appears in this way . in the case when the indefinite form @xmath0 does represent zero ( which means that its discriminant is total square ) the roots @xmath35 are rational , so @xmath38 and @xmath39 for all @xmath40 in this case . finally in the case of semidefinite forms @xmath0 the growth @xmath41 let us first describe in more detail the correspondence between @xmath42 $ ] and paths @xmath15 in the planar binary rooted tree @xmath43 , which is shown on the right of fig . , the corresponding farey tree has a unique path @xmath18 for any irrational positive @xmath40 such that the limits of the adjacent fractions is @xmath40 ( for rational @xmath40 such a path is finite and leads to the corresponding fraction ) . in terms of the corresponding continued fraction expansion @xmath44 , \quad c_0\geq 0 , c_i>0,\ ] ] the path @xmath18 starts from the root and can be described as @xmath45 left - turns on the tree , followed by @xmath46 right - turns , followed by @xmath47 left - turns , etc . the corresponding matrix @xmath48 is the product of the first @xmath21 matrices along path @xmath15 : a_n()= l^c_0r^c_1l^c_2r^c_3 where @xmath49 the matrices @xmath50 and @xmath51 freely generate the monoid @xmath27 , which is the positive part of the group @xmath28 , acting on the rooted tree @xmath43 by left and right turns respectively . we would like to extend this correspondence to the full binary planar tree @xmath52 , which is known to be the dual tree for the farey tesselation shown on the left of fig . 5 ( see e.g. hatcher @xcite , where we have borrowed this picture from with author s permission ) . , title="fig:",height=226 ] , title="fig:",height=226 ] for this we need to add to @xmath50 and @xmath51 the element @xmath53 acting on the full tree @xmath52 as the reflection . note that since @xmath54 the corresponding action group is actually the modular group @xmath55 since @xmath56 it can be thought of as a kind of ` three - point turn ' on the tree @xcite . the matrix @xmath57 for the reflected path @xmath58 can be written then as @xmath59 or , because of the identities @xmath60 more explicitly as @xmath61 this means that when we move in the negative direction , we need to swap left and right ( see right of fig . 4 ) . to be consistent with this it is natural to consider for negative real numbers @xmath62 with positive @xmath63 , \quad c_0\geq 0 , c_i>0 $ ] the _ negative continued fraction expansions _ = -[c_0 , c_1 , c_2 , c_3 ] = [ -c_0,-c_1,-c_2,-c_3 ] . note that one can always avoid the negative @xmath64 with @xmath65 because of the identity -[c_0 , c_1 , c_2 , c_3 , ] = [ -c_0 - 1 , 1 , c_1 - 1 , c_2 , c_3 , ] , but we are not going to do this . this is related to the following galois result . the famous lagrange s theorem @xcite says that any quadratic irrational @xmath66 with @xmath67 and @xmath68 not total square , has a periodic continued fraction expansion @xmath69,\ ] ] and conversely , any periodic continued fraction represents a quadratic irrational . less known is the following addition due to galois ( see e.g. @xcite ) characterising _ pure periodic continued fraction expansions _ @xmath70.\ ] ] * galois s theorem . * _ a quadratic irrational @xmath71 has a pure periodic continued fraction expansion @xmath72\ ] ] if and only if its conjugate @xmath73 satisfies the inequality @xmath74 moreover , in that case _ @xmath75.\ ] ] as we will see later , geometrically the conjugate @xmath76 corresponds to the path going backwards along the infinite periodic conway river defined by @xmath77 to describe our main result we need an answer to the following natural question . assume that we know the continued fraction expansion of a quadratic irrational @xmath69.\ ] ] what is the continued fraction expansion of its conjugate @xmath76 ? for example , we have @xmath78 , \quad \bar\alpha=\frac{6-\sqrt{2}}{17}=[0,3,1,\overline{2}].\ ] ] what is the general rule here ? we could not find the answer to this question in the literature , except in the galois case , so we present it in this section . note first that we can assume that in ( [ eq : alphaexp ] ) @xmath79 , because otherwise @xmath80 can be rewritten with the period @xmath81 $ ] . [ thm : conj ] let @xmath82 $ ] be the continued fraction expansion of a quadratic irrational with @xmath83 , and @xmath84 then the continued fraction expansion of its conjugate is @xmath85.\ ] ] let @xmath86 , @xmath87 , then @xmath88.\ ] ] here it will be convenient for us to allow the zeros in the continued fraction expansion , which can be always avoided using the identities @xmath89 = \left [ c_0 , \ldots , c_i + c_{i+1 } , c_{i+2 } , c_{i+3 } \ldots \right],\ ] ] @xmath90 = \left [ c_0 , \ldots , c_i , c_{i+1 } , c_{i+2 } , c_{i+3 } \ldots \right].\ ] ] in particular , in the example ( [ ex1 ] ) the formula ( [ conj2 ] ) gives for @xmath91 $ ] @xmath92=[0,3,1,\overline{2}].\ ] ] a more sophisticated example : @xmath93,\ ] ] @xmath94 = [ 1,4,4,\overline{3,2,7,9}].\ ] ] it is enough to consider the case @xmath95 . let us first define @xmath96 , \quad \tilde{\beta } = \left[\overline{b_{l-1 } , b_{l-2 } , \ldots , b_1 , b_l}\right].\ ] ] then , by definition , @xmath97 = a_0 + \cfrac{1}{a_1 + \cfrac{1}{\beta } } , \ ] ] so that @xmath98 using the galois result and identity ( [ negid ] ) we have @xmath99 \\ & = \left[-1 , 1 , b_l - 1 , \overline{b_{l-1 } , \ldots , b_1 , b_l } \right].\end{aligned}\ ] ] we can now directly compute @xmath100 @xmath101 @xmath102 @xmath103 @xmath104.\ ] ] this completes the case @xmath105 in the case @xmath106 we have by , @xmath107,\ ] ] and the result then immediately follows from the previous case with @xmath108 replaced by @xmath109 and @xmath110 replaced by @xmath111 . note that formulae ( [ conj1]),([conj2 ] ) determine involutions since @xmath112 \\ & = \left[a_0 , \ldots , a_{k-1}-1 , 0 , 1 , a_k , \overline{b_1 , \ldots , b_{l-1 } , b_l}\right ] \\ & = \left[a_0 , \ldots , a_{k-1 } , a_k , \overline{b_1 , \ldots , b_{l-1 } , b_l}\right ] = \alpha,\end{aligned}\ ] ] @xmath113 \\ & = \left[a_0 , \ldots , a_{k-1 } , a_k - b_l -1 , 0 , 1 , 0 , \overline{b_{l } , b_1 , \ldots , b_{l-1}}\right ] \\ & = \left[a_0 , \ldots , a_{k-1 } , a_k- b_l , 0 , \overline{b_{l } , b_1 , \ldots , b_{l-1}}\right ] = \alpha.\end{aligned}\ ] ] let us consider now the special case @xmath114 [ thm : riverconj ] let @xmath115 $ ] with @xmath116 . then the conjugate @xmath117 can be given as the negative continued fraction expansion @xmath118.\ ] ] when @xmath119 we have @xmath120.\ ] ] we have @xmath121 = a_0 + \left [ 0 , \overline{b_0 , \ldots , b_n}\right],\ ] ] so by the galois result @xmath122 } & = a_0 - \left[\overline{b_l , \ldots , b_1 } \right ] \\ & = - \left[b_l - a_0 , \overline{b_{l-1 } , \ldots , b_1 , b_l } \right].\end{aligned}\ ] ] in the case when @xmath119 , we can rewrite @xmath123 = \left[a_0 - b_l , 0 , \overline{b_l , b_1 \ldots , b_{l-1 } } \right]\ ] ] and proceed as in the case when @xmath124 to get the formula ( [ conj4 ] ) . note that only in the galois pure periodic case and in the case with @xmath125 do we have to use negative continued fractions . we will see now that these are the only cases when the initial position is on the conway river . let @xmath126 be an indefinite binary quadratic form not representing zero , meaning that @xmath127 for all @xmath128 equivalently , this means that the discriminant of the form d = h^2 - 4ab is positive , but not a total square . in that case conway @xcite showed that on the topograph of @xmath0 positive and negative values are separated by a unique periodic river ( see fig . 4 above ) . we are now going to explain how the continued fractions determine the path from @xmath0 ( identified with the corresponding vertex on the topograph ) to the conway river and the relation with the galois result . let us assume for simplicity that all the values @xmath129 are of the same sign ( say , positive ) , otherwise @xmath0 is already on the river . let us assume also that @xmath130 , so we are going down to the river . let @xmath131 be the real roots of the quadratic equation @xmath132 again for simplicity let us assume that @xmath133 is the dominant ( maximal modulus ) root , so that the conjugate is @xmath134 let @xmath135\ ] ] be the continued fraction expansion of @xmath77 from the general theory of the farey tree and continued fractions we have the following result . the continued fraction expansion ( [ eq : alphaexp ] ) describes the unique path @xmath136 going from @xmath0 to the corresponding conway river , and then along the river to the boundary of the tree determined by @xmath80 . a similar path @xmath137 leading to @xmath76 is described by the propositions 1 and 2 above . the conway river of @xmath0 is an infinite in both directions path from @xmath117 to @xmath80 described by the periodic part @xmath138 $ ] of the expansion . the finite path @xmath139 from @xmath0 to the conway river can be given for @xmath124 by @xmath140 & { \text if } \,\ b_l\\ \left [ a_0 , \ldots , a_{k-1 } -1 \right ] & { \text if } \,\ , a_k < b_l \end{array } \right.\ ] ] and for @xmath141 by @xmath142 this also gives a geometric interpretation of our formulas from the previous section , which is illustrated in fig . 7 in the example of @xmath143 with @xmath144 , \quad \bar\alpha=\frac{6-\sqrt{2}}{17}=[0,3,1,\overline{2}].\ ] ] and @xmath76 and conway river for @xmath143,height=302 ] let us describe now the special ( galois ) case when @xmath80 is a pure periodic continued fraction . let us call the quadratic form @xmath145 _ galois form _ if the continued fraction expansion of the dominant root of @xmath146 is pure periodic . by the galois theorem this is equivalent to the conditions @xmath147 a binary quadratic form @xmath0 is galois if and only if @xmath148 , @xmath149 , @xmath150 satisfy @xmath151 the proof is obvious geometrically : @xmath152 and @xmath153 is true if and only if the conway river travels along the edges separating @xmath154 and @xmath155 from @xmath156 and @xmath157 , see fig . 9 . on the conway river , title="fig : " ] [ fig : basepluslimit ] an example of the galois form is the golden " form @xmath158 corresponding to @xmath159 with @xmath160 , \,\bar\alpha=\frac{1-\sqrt{5}}{2}=-[0,\bar 1].\ ] ] now we are ready to state and prove our main results . let @xmath161 be a binary quadratic form ( definite or indefinite ) . if the form @xmath0 is indefinite , then we assume for the beginning that @xmath0 does not represent zero in the sense that @xmath162 for all @xmath163 in that case the two roots @xmath164 of the quadratic equation @xmath165 are real quadratic irrationals , corresponding to the ends of the conway river . to study the growth of the values of @xmath0 along the path @xmath18 we define the corresponding lyapunov exponent as _ q()=_n , \|q_n()\| = ( \|a_n()\| , \|b_n()\| , \|c_n()\| ) , where @xmath166 are the values of @xmath0 at the @xmath21-th superbase on the path @xmath22 let @xmath26 be the lyapunov exponent of the monoid @xmath27 introduced in @xcite : ( ) = _ n , where @xmath167 are the matrices ( [ mata ] ) and @xmath30 is the spectral radius of the matrix @xmath168 for the definite binary quadratic forms @xmath0 the lyapunov exponent @xmath169 for the indefinite binary quadratic forms @xmath0 not representing 0 we have @xmath170 in other words , the only two exceptional paths with zero growth are those leading to two ends of the conway river of @xmath171 let us introduce first for the form @xmath161 its matrix defined by @xmath172 which , slightly abusing the notation , we also denote by @xmath0 : @xmath173 its determinant @xmath174 is minus a quarter of the discriminant @xmath175 of the quadratic equation @xmath176 the action of the group @xmath28 on @xmath0 is defined in the matrix form by @xmath177 where a= p & q + r & s + sl_2(z ) . this determines a 3-dimensional representation @xmath178 of @xmath28 , when @xmath31 acts on the coefficients @xmath179 of the form @xmath0 by @xmath180 if the eigenvalues of the matrix @xmath31 are @xmath181 and @xmath182 then the eigenvalues of @xmath183 are @xmath184 , @xmath185 and 1 . consider first the case when the coefficients of the form @xmath186 are all positive . in that case conway s climbing lemma @xcite guarantees the permanent growth whichever upward path we choose ( see fig . [ fig : cl ] we claim that the corresponding growth is given by @xmath187 to prove this introduce two norms on binary forms @xmath188 . the first norm is the one used in the definition of @xmath189 : \|q\|=(\|a\|,\|b\|,\|c\| ) , c = a+b+h . the second norm is \|q\|_h=(\|a\|,\|b\|,\|h\| ) . since any two norms in a finite - dimensional space are equivalent , we have for all @xmath0 @xmath190 for some positive constants @xmath191 this means that we can replace @xmath192 by @xmath193 in the definition @xmath194 now let @xmath18 be a path and @xmath195 be the corresponding matrices ( [ mata ] ) , so that the matrices of the forms along the path are @xmath196 or , explicitly a_n + h_n + b_n + = p_n^2 & 2p_nr_n & r_n^2 + p_nq_n & p_nr_n+q_ns_n & r_ns_n + q_n^2 & 2q_ns_n & s_n^2 + a + h + b + . recall also that the function @xmath26 can be alternatively defined as @xmath197 where @xmath198 ( see proposition 1 in @xcite ) . similarly to @xcite assume without loss of generality that @xmath199 $ ] and consider the cases @xmath200 and @xmath201 separately . when @xmath200 we have @xmath202 which determines the quadratic growth of @xmath203 in @xmath21 and as a result @xmath204 if @xmath205 we have @xmath206 and thus from ( [ 3d ] ) @xmath207 @xmath208 on the other hand by assumption the initial @xmath186 are all positive , so , since they are also integer , @xmath209 from ( [ 3d ] ) then it follows that @xmath210 thus we have @xmath211 which implies that @xmath212 this completes the proof in the growing case . assume now that the path goes down , so the values @xmath213 may decrease . consider first the case when the form @xmath0 is positive definite . then all the values are positive , so it is clear that for any path at some point the growth will start again and we can repeat our arguments to get the claim . however , if @xmath0 is indefinite this is no longer true , since @xmath0 can take negative values as well . by conway s result @xcite positive and negative values of @xmath0 are separated by the infinite periodic river . there are three possibilities : either the path does not cross the river , it crosses the river ( or starts on the river and then leaves it ) , or it is after some point stuck on the river forever . if the path does not cross the river the values of @xmath0 will remain positive and thus bounded from below , so at some point we will have growth and we repeat the arguments to prove the claim in this case as well . if the path crosses the river then at some point all the values of @xmath214 will become negative , and we can repeat the arguments for @xmath215 to get the claim . if we start on the river , then after the point of departure we can use the same arguments as before . finally , there are exactly two paths which are stuck on the river , corresponding to @xmath216 in that case we have no growth and @xmath217 note that the corresponding @xmath218 so @xmath219 in that case . let us discuss now the remaining case : indefinite forms representing zero and semidefinite forms with zero discriminant . for indefinite forms @xmath0 representing zero @xmath220 while for the semidefinite forms @xmath221 the topograph of the indefinite forms representing zero ( with discriminant @xmath68 being total square ) is described in conway @xcite . in that case we have two `` lakes '' corresponding to zero values connected by a finite river separating positive and negative values of @xmath171 in the exceptional cases the river disappears and two lakes are adjacent , see the first lecture in @xcite . this picture agrees with the fact that the roots @xmath164 of the corresponding equation @xmath165 are rational . a simple analysis shows that the previous arguments in this case work as well , with the only difference that now both @xmath222 and @xmath223 are zero , and thus equality @xmath39 holds in this case as well . in the semidefinite case equivalent to the case @xmath224 we have zero growth , which simply follows from the conway description @xcite . let us compare this with the growth of the markov triples considered in @xcite . markov triples are the positive solutions of the markov equation x^2+y^2+z^2 - 3xyz=0 . they all can be found from the obvious one @xmath225 by compositions of vieta involutions ( x , y , z ) ( x , y , 3xy - z ) and permutations of @xmath226 the right - hand side of the markov equation @xmath227 is cubic polynomial , which is _ quadratic _ in every variable , implying the vieta involution . it is also symmetric under the permutations of @xmath226 one can consider also a version of the markov equation @xmath228 which has similar properties ( see @xcite ) . we claim that this paper can be interpreted in a similar way for degree 2 polynomial @xmath229 indeed , the transformation @xmath230 is nothing but the vieta involution for the equation a^2+b^2+c^2 - 2ab-2ac-2bc = d . in fact , @xmath231 is nothing other than the discriminant of the binary quadratic form @xmath232 @xmath233 our results say that the lyapunov exponents for the equation ( [ b ] ) with @xmath234 are twice the lyapunov exponents for the markov dynamics . when @xmath235 then the same is true with the exception of conway river paths , for which the lyapunov exponents vanish . note that the equation ( [ b ] ) with @xmath234 determines the two - sheeted hyperboloid , while for @xmath235 this is a one - sheeted hyperboloid . the semidefinite ( degenerate ) forms with @xmath236 correspond to the cone . we would like to mention klein s correspondence @xcite between indefinite binary quadratic forms and geodesics in hyperbolic plane , which we have learnt recently from @xcite . as we have just seen the projectivized vector space of real binary quadratic forms is a real projective plane with the degenerate forms forming a conic section determined by ( [ b ] ) with @xmath237 definite forms correspond to the points inside this conic , which in the klein model represents the points of the hyperbolic plane . the indefinite forms correspond to the points outside the conic , which by polarity are represented by the hyperbolic lines . on the binary tree these lines are nothing other than conway rivers . we are very grateful to alexey bolsinov for numerous helpful discussions and to boris springborn for sending us a preliminary version of his very interesting work @xcite .	we study the growth of the values of binary quadratic form @xmath0 on a binary planar tree as it was described by conway . we show that the corresponding lyapunov exponents @xmath1 as a function of path determined by @xmath2 are twice the values of the corresponding exponents for the growth of markov numbers @xcite , except for the paths corresponding to the conway rivers , when @xmath3 the relation with galois results about continued fraction expansions for quadratic irrationals is explained and interpreted geometrically .
directed current in ratchet systems have received much attention over recent years @xcite . one of the reasons to study these type of systems is motivated by the attempt to understand the physical mechanism of motion of molecular motors in biological systems @xcite and to investigate its role in the design of new material properties @xcite . the dynamics of a particle moving in a periodic potential under the action of an applied time periodic term is rather complex and rich , displaying typically even chaotic behavior @xcite . a feature of particular interest is the emergence of directed currents in the system response to an external time - periodic driving with zero time - averaged value . though most of the works about ratchet systems consider the presence of noise , this phenomenon may also arise in deterministic systems , both in the overdamped @xcite and underdamped @xcite regimes . moreover , the phenomenon of anticipated synchronization occurring in inertial deterministic ratchets have been studied recently @xcite . + in particular , in the overdamped regime , the one - dimensional particle dynamics @xmath0 usually considered is governed by a first order differential equation of the type @xmath1+f(t ) , \label{eq : eom}\ ] ] where the dot and the prime denote time and spatial derivatives , respectively , @xmath2 is a periodic potential with spatial period @xmath3 [ i.e. , @xmath4 , and @xmath5 is a time - periodic driving force with period @xmath6 [ i.e. , @xmath7 . in this type of systems , the current is defined as the average velocity @xmath8 as shown in ref . @xcite co - existing attractors can exist for large driving strengths , which however are not current - carrying . a finite current @xmath9 , possessing an unbounded @xmath0-trajectory , is consequently independent of the initial condition @xmath10 . it can also be shown that a directed current ( @xmath11 ) is only possible if at least one of the following symmetries is broken @xcite : @xmath12 @xmath13 our main interest in this paper is to gain a deeper insight into these type of systems exploring some quantitative and qualitative aspects of its very rich dynamics . specifically , we will consider the same sawtooth potential as in refs . @xcite @xmath14 , \label{eq : uofx}\ ] ] which has a spatial period @xmath15 . in addition , rather than using a sinusoidal driving force @xmath5 , in this work we will consider a multifrequency time - periodic force given by @xmath16 with @xmath17 being a constant . since the potential ( [ eq : uofx ] ) breaks the spatial symmetry ( [ eq : symm:1 ] ) , a directed current is possible , even though the time symmetry ( [ eq : symm:2 ] ) is fulfilled . the paper is organized as follows . in section [ sec : num ] , we perform a detailed numerical study of eq . ( [ eq : eom ] ) for a wide range of the driving parameter values @xmath17 and @xmath18 . in particular , we present a colored phase - diagram @xmath9 vs. @xmath17 and @xmath19 , which is inspired by the figure provided by prof . peter talkner and collaborators in @xcite . in order to explain some features of this diagram we propose in section [ sec : ana ] some simple theoretical considerations . finally , in the last section , we summarize our findings . plots of the instantaneous velocities @xmath20 ( solid line ) and @xmath21 ( dashed line ) as a function of @xmath22 for a driving force with @xmath23 . ( all quantities in dimensionless units).,width=226 ] regardless of the value of the driving frequency , there exist two critical amplitudes @xmath24 and @xmath25 separating three different regimes . for amplitudes @xmath26 $ ] , both potential states , @xmath27 and @xmath28 , possess a periodic array of equilibrium points ( see fig . [ fig : vofx ] ) . since the inertial term @xmath29 is absent , the particle can not cross these equilibrium points , and consequently , it remains trapped between them , leading to a zero current . for @xmath30 $ ] , the potential state @xmath27 retains its equilibrium points , while @xmath28 does not have any , that allowing non - bounding motion in the positive direction finally , when @xmath31 , neither of the potential states possesses equilibrium points , and the particle motion is not bounded in either direction . the " talkner-current phase diagram . the drift velocity as a function of the driving parameters @xmath17 and @xmath19 is represented using a color density plot . black color has been used for velocities less than @xmath32 . ( all quantities in dimensionless units),height=340 ] drift velocity @xmath9 ( solid line ) as a function of the driving frequency @xmath19 for @xmath33 . the dashed line shows the adiabatic value . the arrows depicts the zero - velocity bands given by the theory in the text . ( all quantities in dimensionless units.),height=226 ] the same as in fig . [ fig : velofw:1 ] but for @xmath34.,height=226 ] in order to go further in our analysis , we have resorted to a numerical treatment of our model using the freely available integrator rksuite @xcite . wide regions of parameter space have been explored . in fig . [ fig : diagram ] we have used a color code to represent the drift velocity @xmath9 for a rectangular driving force of varying amplitude @xmath17 and frequency @xmath19 . the regions in black correspond to bounded , time - periodic particle motion with zero current . more precisely , the velocity in those regions is smaller than @xmath32 . the diagram has a rich and interesting structure . regions of zero drift velocity are intermingled with regions with finite values , giving rise to a finger - shaped structure . notice that for the parameter values considered , the drift velocity is always positive or zero , showing the largest magnitude in the intermediate region @xmath30 $ ] . this last feature could be understood by taking into account that in this regime the motion in the positive direction is never compensated by motion in the opposite direction . in fig . [ fig : velofw:1 ] we present a section of the diagram for a driving amplitude in the intermediate regime @xmath35 . a series of peaks corresponding to the fingers in fig . [ fig : diagram ] are observed . figure [ fig : velofw:2 ] shows a representative section for @xmath36 . by contrast with fig . [ fig : velofw:1 ] , it presents an intermediate gap of particle localization . if we further increase the value of @xmath17 , more intermediate gaps appear , as can be seen in the phase diagram . even though the nonlinearity of the system precludes a complete and detailed analytical solution of the problem , it is possible to explain some features of the phase diagram by simple considerations . determination of zero - velocity bands with @xmath37 . solid line depicts @xmath38 , defined in eq . ( [ eq : gofx ] ) , as a function of @xmath22 for the same value of @xmath17 as in fig . [ fig : velofw:2].,height=226 ] due to the truncation implicit in any numerical calculation , the simulations reported above are not able to distinguish between a situation of exact particle localization or a very small drift , lower than the tolerance we chose to determine the black regions in fig.[fig : diagram ] . furthermore , a simple explanation of the existence of such regions in the non - trivial regime @xmath39 would be desirable . let us assume that the particle starts at @xmath40 at the beginning of a driving period . it then moves under a force @xmath20 for half a period until it reaches the position @xmath41 . then the driving force switches sign so the total force on the particle is now @xmath21 , which makes it ( in general ) move backwards up to a position @xmath42 after another @xmath43 . if @xmath44 , the particle has returned to the initial position , and consequently , the process is repeated successively , leading to a drift velocity strictly equal to zero . for this situation to happen the following equations must hold @xmath45 subtracting both equations we arrive to the condition @xmath46 where @xmath47 therefore , we can determine the set of pairs @xmath48 with zero current for a given driving strength @xmath17 by plotting the function @xmath38 vs. @xmath22 ( see fig [ fig : gofx ] ) . the intersection of a horizontal line with @xmath38 in this plot provides the possible values of the pair @xmath48 . the period @xmath6 associated with the pair is then given by eq . ( [ eq : tx0x1:1 ] ) or ( [ eq : tx0x1:2 ] ) . since the drift velocity @xmath9 is independent of the initial conditions @xcite , it would be exactly zero for those driving parameter values @xmath17 and @xmath6 . in fig . [ fig : gofx ] , we plot @xmath38 for the same value of @xmath49 as in fig . [ fig : velofw:2 ] . a horizontal line crosses @xmath38 at the points a , b , c , and d , providing the coordinates @xmath50 , and @xmath51 . if we choose @xmath52 , and @xmath53 as any of the other points , we obtain three pairs of points with driving periods @xmath54 , and @xmath55 . since the integrand in eq . ( [ eq : tx0x1:1 ] ) is always positive , the further away @xmath53 is from @xmath40 , the larger the period , which implies @xmath56 . before we proceed in analyzing the plot in greater detail , let us study the symmetries of @xmath38 . since it is an integral of a space - periodic function [ see eq . ( [ eq : gofx ] ) ] we have @xmath57 where @xmath58 . in addition , taking into account that @xmath59 is an even function , necessarily @xmath60 property ( [ eq : gofxsym:1 ] ) leads to the fact that we only need to vary @xmath40 within a spatial interval of length unity , whereas property ( [ eq : gofxsym:1 ] ) combined with ( [ eq : gofxsym:2 ] ) implies that @xmath38 has an inversion center at @xmath61 , with @xmath62 any integer . the driving period @xmath6 , when viewed as a function of @xmath53 or @xmath40 [ see eq . ( [ eq : tx0x1:1 ] ) ] , also obeys these properties . because of these symmetry properties , choosing the point a as @xmath40 in fig . [ fig : gofx ] is equivalent to choosing any of the points a , a , and a , in the sense that they lead to the same driving periods . let us consider a horizontal line between points a and a. as this line gets closer to a , two of the intersection points approach each other , and consequently , the driving period associated to them tends to zero . therefore , @xmath63 ( @xmath64 ) corresponds to particle localization . moving up the line continuously we obtain a whole band of driving periods starting from zero up to a maximum value given by the pair @xmath65 when the line crosses a. this pair is equivalent to @xmath66 . two more intermediate periods in the band can be obtained from the pairs @xmath67 and @xmath68 in the figure . the next value of the driving period obtained from fig . [ fig : gofx ] is the one given by the pair @xmath69 . this pair marks the start of a second band that ends at the maximum value given by the pair @xmath70 . moving up the line up to a gives all the intermediate values in the band . clearly , the first band is associated with movement within a spatial interval less than unity ( i.e. , @xmath71 ) , whereas the second band is related to displacements two times that distance ( @xmath72 ) . in fig . [ fig : velofw:2 ] we have indicated with arrows the calculated frequency bands . it can be seen that they do not cover the entire region with numerically evaluated zero velocity . this is due to the fact that the above discussed mechanism is just the simplest one leading to zero current . the drift velocity can also vanish because the particle returns to its initial position after two or more driving periods , instead of after the first one . the same analysis can be carried out for a subthreshold driving @xmath73 $ ] . in this case @xmath38 presents singularities at the equilibrium points , which prevents the particle from crossing those points when @xmath74 . this leads to a single band which is near the zero - current region observed in the simulations , as shown in fig . [ fig : velofw:1 ] . drift velocity @xmath9 ( solid line ) as a function of the driving amplitude @xmath17 in the adiabatic limit ( @xmath75 ) . the curve is indistinguishable from the numerical results obtained with @xmath76 . ( all quantities in dimensionless units.),height=226 ] in this section we will obtain an analytical expression for the current in the adiabatic limit @xmath77 @xmath78 specifically , as proved later on , @xmath79 is given by the average value @xmath80,\ ] ] where @xmath81 and @xmath82 are , respectively , the velocities in the presence of the static forces @xmath83 and @xmath84 . obviously , @xmath85 for @xmath86 $ ] and @xmath87 for @xmath88 $ ] , as the particle ends up being trapped by an equilibrium point . for @xmath89 , @xmath90 where @xmath91 is the time taken for the particle to travel a distance equal to @xmath92 ( the spatial period ) in the presence of the static force @xmath83 . analogously , for @xmath93 , @xmath94 where @xmath95 is defined as @xmath96 but replacing @xmath83 by @xmath84 . in order to prove eq . ( [ adiab ] ) , let us consider separately the three regimes @xmath97 $ ] , @xmath98 $ ] , and @xmath99 . for @xmath97 $ ] the result in eq . ( [ adiab ] ) is trivial , since @xmath100 . in the intermediate regime @xmath98 $ ] , let us assume that the particle is initially located at a minimum of the potential @xmath101 ( as we have mentioned before , the drift velocity does not depend on this particular initial condition ) . if we choose the time period of @xmath5 as @xmath102 , with @xmath103 , @xmath104 , @xmath105 , @xmath106 , then , during the first half - period , @xmath107 , the particle stays trapped at the initial location . after the second half - period , @xmath108 , the particle arrives to a new minimum of @xmath101 separated from the initial location by a distance @xmath109 . during the third half - period , @xmath110 , the particle stays trapped at that minimum , and so on . thus , for a given @xmath98 $ ] , all the frequencies @xmath111 lead to the same value of the drift velocity @xmath112=\frac{n}{2 n \tau_{+}(a)}=\frac{1}{2 \tau_{+}(a)}=\frac{v_{+}(a)}{2}.\ ] ] consequently , taking into account that @xmath113 , it follows that @xmath114=\frac{v_{+}(a)}{2}.\ ] ] this proves eq . ( [ adiab ] ) for @xmath98 $ ] , since in this regime @xmath87 . finally , for @xmath115 , it is easy to prove that @xmath116 is a continuous strictly increasing function of @xmath17 which takes values in the interval @xmath117 . let us assume first that for a given @xmath115 the ratio @xmath116 is a rational number @xmath118 , with @xmath119 . then , if we choose the time period of @xmath5 as @xmath120 , with @xmath103 , @xmath104 , @xmath105 , @xmath106 , the particle travels a distance equal to @xmath121 every time period . consequently , all the frequencies @xmath122 lead to the same value of the drift velocity @xmath123=\frac{n(q - p)}{2n q \tau_{+}(a)}=\frac{1}{2\tau_{+}(a)}-\frac{1}{2\tau_{-}(a)}= \frac{v_+(a)+v_-(a)}{2}.\ ] ] taking into account once again that @xmath113 , it follows that @xmath114=\frac{v_{+}(a)+v_{-}(a)}{2}.\ ] ] since any irrational number can be approximated as closely as desired by a sequence of rational numbers , eq . ( [ adiab ] ) is also valid if @xmath116 is an irrational number . in fig . [ fig : velofom ] we represent the adiabatic drift velocity in eq . ( [ adiab ] ) as a function of the driving amplitude @xmath17 . we have also plotted the numerical drift velocity obtained for @xmath76 . both results are indistinguishable within the resolution of the plot . finally , from the above argument , another conclusion follows . as mentioned before , the adiabatic value @xmath124 not only is reached in the limit @xmath75 , but also for the finite frequencies given by eqs . ( [ eq : freq:1 ] ) and ( [ eq : freq:2 ] ) . this result becomes manifest in figs . [ fig : velofw:1 ] and [ fig : velofw:2 ] , as an infinite number of intersections between the horizontal line ( adiabatic limit ) and the drift velocity curve . these frequencies are at the core ( see the red regions in fig . [ fig : diagram ] ) of the above discussed fingers . furthermore , they correspond to exact displacements of @xmath62 spatial periods every driving period @xmath6 . this fact , together with the interpretation provided by our analysis of the zero - current bands , lead us to infer that each finger corresponds to motion in which the particle advances a distance of about @xmath62 spatial periods at those time intervals when @xmath125 . we have checked this fact in the numerical simulations . to summarize , we have studied a one - dimensional deterministic ratchet model in the overdamped regime driven by a rectangular force . we have performed extensive numerical calculations , which allowed us to present a colored phase diagram showing the drift velocity as a function of a wide range of the driving parameter values @xmath17 and @xmath19 . in addition , we have studied the conditions leading to bounded particle motion , obeying asymptotically periodic , bounded motion , @xmath37 . we have shown that the resulting regions cover most of the phase space where there is no directed current . moreover , we have provided an explicit analytical expression for the drift velocity in the adiabatic limit . finally , the theoretical considerations are of great help to rationalize the finger - shaped structure shown in the diagram . + we ( dc , jc - p , aa , mm ) acknowledge the support of the ministerio de educacin y ciencia of spain ( fis2005 - 02884 ) and the junta de andaluca . dc also acknowledges the ministerio de educacin y ciencia of spain for a contract under the juan de la cierva program . the authors wish professor talkner many more enjoyable years of doing trend - setting science and personal happiness . breitestes label p. hnggi and r. bartussek , lect . notes phys . * 476 * , 294 ( 1996 ) . astumian and p. hnggi , physics today * 55 * , ( no . 11 ) , 33 ( 2002 ) . h. linke , appl phys . a * 75 * , 167 ( 2002 ) . p. hnggi , f. marchesoni , and f. nori , ann ( leipzig ) * 14 * , 51 ( 2005 ) . p. reimann and p. hnggi , appl . a * 75 * , 169 ( 2002 ) . f. jlicher , a. adjari , and j. prost , rev . phys . * 69 * , 1269 ( 1997 ) . p. jung , j. g. kissner , and p. hnggi , phys . lett . * 76 * , 3436 ( 1996 ) . j. l. mateos , phys . 84 * , 258 ( 2000 ) . s. sengupta , r. guantes , s. miret - artes , and p. hnggi , physica a * 338 * , 406 ( 2004 ) . f. family , h. a. larrondo , d. g. zarlenga , and c. m. arizmendi , j. phys . condens matter , * 17 * , s3719 ( 2005 ) . r. bartussek , p. hnggi , and j. g. kissner , europhys . lett . * 28 * , 459 ( 1994 ) . a. sarmiento and h. larralde , phys . e * 59 * , 4878 ( 1999 ) . m. n. popescu , c. m. arizmendi , a. l. salas - brito , and f. family , phys . 85 * , 3321 ( 2000 ) .	we investigate a dissipative , deterministic ratchet model in the overdamped regime driven by a _ rectangular force_. extensive numerical calculations are presented in a diagram depicting the drift velocity as a function of a wide range of the driving parameter values . we also present some theoretical considerations which explain some features of the mentioned diagram . in particular , we proof the existence of regions in the driving parameter space with bounded particle motion possessing zero current . moreover , we present an explicit analytical expression for the drift velocity in the adiabatic limit .
four decades have passed now since the _ uhuru _ discovery of the x - ray binary ( henceforth xrb ) cen x-3 was first reported @xcite . what began in the early- to mid-1970s as detailed individual studies of the small number of xrbs known at the time , as also pioneering works on the construction of general models of such binaries @xcite has now blossomed into a mature , rich subject ( for overviews , see , _ e.g. _ , shapiro & teukolsky 1983 ; ghosh 2007 ) , with a large number of xrbs discovered and catalogued today in the milky way , the magellanic clouds , and nearby external galaxies , and the fine details of the behavior of each class of xrbs well - recorded from timing and spectral studies with the aid of several generations of x - ray observatories of ever - increasing sensitivity and resolution . it is thus becoming possible now to do studies of the _ collective _ properties of xrb populations , obtaining statistically meaningful distributions of some of their essential properties @xcite , and exploring the essential physics underlying these distributions , which is rooted in the basic scenarios for the formation , evolution , and dynamics of xrbs ( van den heuvel 1983 , 1991 , 1992 , 2001 ) , which have been constructed gradually over these four decades , and which are widely accepted now . by collective properties , we mean here the distributions of x - ray luminosities and binary periods of these xrbs , the distribution of pulse periods of those binaries which exhibit periodic x - ray pulses , and perhaps also the distribution of their spectral properties at a future date . in this series of papers , we explore the theoretical underpinnings of the observed distribution of the collective properties of xrbs in normal galaxies . we focus here on xrbs _ outside _ the globular clusters of these galaxies ( _ i.e. _ , in the stellar field of the galaxy ) , where the effects of encounters between xrbs and neighboring stars , as well as those between xrbs themselves , are generally thought to be negligible , due to the relatively low stellar density there . this situation is at the opposite limit to that obtaining in dense cores of globular clusters , where such encounters are thought to be _ dominant _ in determining the formation , evolution , and destruction of xrbs , and so their collective properties . we have studied this latter limit of xrbs in globular clusters in an earlier series of papers ( banerjee & ghosh 2006 , 2007 , 2008 ) . the current study may thus be regarded as complementary to the earlier one . we shall confine ourselves here to the distributions of x - ray luminosity and binary orbital period , since the most detailed distributions available are on these properties , and also since these properties are most readily associated with binary evolutionary characteristics . to understand the distributions of properties like pulse periods of x - ray pulsars and spectral parameters of xrbs , one must also take into account the details of the accretion torque on the neutron star , and of the x - ray emission processes . we defer these projects to the future , noting that with @xmath1 140 x - ray pulsars already known , the pulse - period distribution may be amenable to such a future study . the crucial simplification that makes a straightforward approach feasible in the limit that we study here is that the absence of any significant influence on a given xrb of either the stellar background or other xrbs implies that each xrb can be thought of as evolving in _ isolation _ , following the well - studied and now widely - accepted scenarios for the formation and evolution of individual xrbs , starting from primordial binaries generated in normal star - formation activity in a galaxy ( see , _ e.g. _ , van den heuvel 2001 and the references therein ) . this , in turn , implies that we need only have a knowledge of the distribution of the essential parameters of these primordial binaries , and evolve these distributions through the essential processes that occur during the evolution of a primordial binary into a xrb . the task becomes even simpler when only initial and final states matter in the above processes , since we need only keep track of these states to carry out the transformation of the relevant probability distribution , and so connect the primordial distribution to the xrb distribution in a one - to - one correspondence . this is , in fact , the case for massive or high - mass x - ray binaries ( henceforth hmxbs ) . on the other hand , when following one or more of these evolutionary processes requires the knowledge of the entire evolution during a particular process , and not just of the initial and final states , the calculation becomes more laborious ( although still straightforward in principle ) . this is the case for low - mass x - ray binaries ( henceforth lmxbs ) . in this first paper of the series , we focus our attention on hmxbs containing neutron stars . a particular circumstance that helps in this case is that , because of the short evolutionary timescales of the massive companions of the neutron stars in hmxbs , the timescale for the entire evolution from the primordial binary stage to the hmxb stage ( as also the operational lifetime of the hmxb ) is short @xcite compared to the timescale on which the star - formation rate ( which determines the creation rate of primordial binaries ) evolves @xcite . as a consequence , the entire process we study here can be viewed as happening at a constant star - formation rate , so that the star - formation history of the galaxy adds no extra complication . figuratively speaking , it is as if we are taking a `` snapshot '' of the above processes confined entirely to one epoch ( the current one , @xmath2 , for the hmxb populations of the milky way and other local galaxies , and an appropriately earlier epoch for the hmxb populations of spiral galaxies at significant redshifts ) . this argument does not , of course , apply to lmxbs , since their evolutionary timescales are comparable to or longer than @xcite those on which the star - formation rate evolves . accordingly , we have to explicitly take into account the evolution in the star - formation rate when we apply our scheme to lmxbs , as we shall describe in later papers in the series . thus , for example , for describing the lmxb populations of local galaxies , we shall need to use not only the star - formation rate at @xmath2 but also the rates at @xmath3 , _ i.e. _ , the star - formation history of the galaxy . this difference between the evolutionary characteristics of hmxb and lmxb populations is , of course , closely related to the earlier statement about the initial - to - final state correspondence for hmxbs versus the additional role of intermediate states for lmxbs . in the following sections , we first describe the distributions of the essential parameters of the primordial binary populations that we adopt for our study of hmxbs , following well - established scenarios and norms in the literature . in particular , we adopt the standard initial mass function ( henceforth imf ) and the standard log - uniform distribution ( also known as pik s law ) for the orbital separation , as is done in the literature . we then describe how these parameters change in the essential processes that occur as a primordial binary evolves @xcite . the first change occurs when the primary evolves , fills its roche lobe , and transfers its envelope to the secondary , leaving behind its he core . the second change occurs when this he core finishes its further evolution , and explodes in the supernova ( sn ) that creates the neutron star . we give explicit relations connecting the binary parameters in the initial and final states of these two processes . using these relations , we show how the standard rules of probability transformation enable us to derive the distributions of the post - sn binary parameters from the original primordial binary parameters through the jacobian formalism . next , we consider what happens when the massive companion to the neutron star evolves off the main sequence , becoming a giant / supergiant , and driving a strong stellar wind , from which the neutron star accretes matter , thus turning on the hmxb phase @xcite . we adopt standard stellar - wind models from the literature and show how we can derive the distributions of both the x - ray luminosities @xmath4 and the orbital periods @xmath5 of the hmxbs from the distributions we obtained above . we then compare our calculated distributions with the observed distributions of @xmath4 @xcite and @xmath5 @xcite reported in the literature . we show that the main feature of the observed @xmath4-distribution , _ viz . _ , the power - law behavior with a differential slope @xmath6 over a wide range of @xmath4 @xcite is reproduced well in our calculated distribution over the luminosity range covered by neutron - star hmxbs . at the lowest luminosities , below the range covered in the above observational work , our calculated distribution shows a shallower rise in @xmath7 with decreasing @xmath4 . we show that this feature is consistent with recent observations ( shtykovskiy & gilfanov 2005a , b ) of the magellanic clouds . for the @xmath5-distribution , we show that the general features of the observed distribution are in reasonable agreement with the results of our calculations within observational uncertainties . subsequent sections of the paper are arranged as follows . in sec.[sec : hmxboverview ] , we give a brief overview of the formation of neutron - star hmxbs from primordial binaries and their subsequent evolution , and we describe the standard distribution of primordial binary parameters that we adopt . in sec.[sec : paramchange ] , we describe how the binary parameters evolve through the processes of ( a ) the first mass transfer in the system and ( b ) the subsequent supernova of the he - core of the primary . in sec.[sec : jacob ] , we introduce the jacobian formalism for calculating how the distribution of binary parameters transforms as these parameters evolve as described in the previous section . in sec.[sec : psnpdf ] , we discuss the distribution of the essential parameters of the post - sn binary . in sec.[sec : psntoxrb ] , we detail the evolution of the post - sn binary into a hmxb , summarizing the essentials of the stellar and stellar - wind models that provide these details , and indicating the further transformation of distributions needed at this point . in sec.[sec : hmxbdist ] , we discuss our calculated hmxb distributions of luminosity ( _ i.e. _ , the x - ray luminosity function or xlf ) and binary period , compare them with observed distributions , and discuss the effect on our model calculations of varying essential inputs like the initial mass function ( imf ) , stellar wind models , and so on . finally , in sec.[sec : discussion ] , we discuss our results from various angles , including issues about primordial binary parameters , roles of be - star binaries and black holes , physical origins of some essential features of the hmxb distributions , and then summarize our conclusions and future outlook . details of some of our calculational methods are summarized in appendix a. scenarios for the formation and evolution of hmxbs have been studied in detail since the early 1970s , and a `` standard '' picture has now emerged , which is very well - documented @xcite , thus making it unnecessary to recount it here in detail . very briefly , one starts with a primordial binary of two massive stars . the more massive of the two ( _ i.e. _ , the primary ) evolves faster , ends its main sequence life and expands rapidly to become a giant , the whole process occurring on a timescale of @xmath8 years . the giant overflows its roche lobe and starts transferring its hydrogen envelope to the secondary , at which point there are two possibilities , depending on the ratio of the thermal timescales of the two stars , which , in turn , depends on their mass ratio . the first possibilty arises if the mass ratio is not so extreme ( quantitative arguments are given in the next section ) that the above two timescales are not different by more than an order of magnitude , say . the secondary can then accept the entire mass transferred by the primary . this leaves behind only the he - core of the primary , and the secondary becomes so massive after assimilating the h - envelope of the primary that it in fact turns into the more massive component of the system . this is the channel through which hmxbs form . the second posibility arises if the mass ratio is so extreme that the thermal timescales are disparate by more than the above amount , in which case the secondary is unable to accept the mass transferred by the primary , and this mass forms a common enevelope ( ce ) surrounding the secondary and the he - core of the primary . this is the standard channel through which lmxbs are thought to form , which we do not discuss further in this paper . in the hmxb channel , the system now consists of a he - core plus a massive star , which was the secondary earlier , but is now the more massive component . the he - core finishes the rest of its evolution rapidly ( in @xmath9 years ) , and explodes in a supernova , leaving behind a neutron star which will become the main x - ray emitter in the hmxb . the massive companion to the neutron star now finishes its main - sequence evolution in @xmath10 years , expands to become a supergiant , and drives a strong stellar wind . accretion from this wind by the neutron star generates x - rays , and the system thus turns on as a hmxb . this hmxb phase lasts @xmath11 years , after which the massive companion fills its roche lobe . the subsequent mass - transfer is very large and rapid , which leads to considerable mass loss from the system , formation of a common envelope ( ce ) surrounding the neutron star and the he - core of the evolved massive companion , and a total extinction of the x - ray source . the outcome of this ce evolution , _ viz . _ , the formation of a compact binary consisting of the neutron star and the he - core of its former massive companion , which , after the sn of this he - star , ultimately leads to the formation of a double - degenerate system consisting either of an eccentric double - neutron - star binary or of two runaway single neutron stars @xcite , lies outside the scope of this work . in recent years , variations , modifications and additions to the above standard scenario have been considered . for example , it has been suggested that ce evolution of sufficiently wide systems during the first mass transfer may still lead to tight post - ce systems which would contribute to hmxb production ( see @xcite and references therein ) , and that these systems might remain tight even after the sn explosion , turning promptly into hmxbs with unevolved companions . further , under certain other circumstances , ce evolution might lead to tightly - bound hmxbs with he - rich donors @xcite . in this first look at the problem , we shall not consider these fascinating possibilities , but rather confine ourselves to the standard picture , since our aim here is to assess the viability of our approach in the simplest testing ground , before attempting further refinements . we consider now the distribution of the essential properties of the primordial binaries from which hmxbs evolve according to the standard scenario sketched above . the progenitor primordial binary system is described by three essential properties , namely , the mass of primary ( @xmath12 ) , the mass of the secondary ( @xmath13 ) , and the orbital separation ( @xmath14 ) . an equivalent description is in terms of @xmath12 , the mass ratio @xmath15 , and @xmath14 . we use the second description throughout this work , following the custom of recent literature on the subject , and briefly mention the correspondence between the two descriptions in sec.[sec : mdist ] . also , as is often done in this subject , we take the primordial binary orbits to be circular for the purposes of the problem we study here . the constraints imposed by the hmxb problem under consideration here limits the allowed range of the above parameters . if we restrict ourselves to only neutron - star hmxbs , as we do in this paper , then the primary mass @xmath12 is restricted to be between roughly 9@xmath16 and 30@xmath16 @xcite . this means , of course , that those hmxbs which have black holes ( either stellar - mass or intermediate - mass ) as their compact , x - ray emitting components are outside the scope of this work . we return to this point in sec.[sec : blackhole ] . next , the secondary mass @xmath13 is restricted from below by the requirement mentioned above , _ i.e. _ , that its thermal timescale must not be larger than that of the primary by more than one order of magnitude , which implies a rough lower bound @xmath17 , since the thermal timescale goes roughly as the inverse square of the mass in this mass - range . the upper bound on @xmath18 is of course @xmath19 by definition . this means , therefore , that the allowed range of the primordial - binary mass ratio is roughly @xmath20 for hmxb formation . finally , the initial orbital separation @xmath14 is restricted from above by the condition of roche lobe overflow when the primary evolves off the main sequence , without which there would be no mass transfer , and the stipulated hmxb formation scenario will not operate . it is customary to take this upper limit as @xmath21 , which is not very restrictive . consider now the distributions of @xmath22 , and @xmath14 we use in this work , in keeping with the common practice in the literature . the @xmath23-distribution is described in terms of a suitable initial mass function ( imf ) , _ i.e. _ , the probability density @xmath24 of a stellar mass being in the range @xmath25 to @xmath26 . the imf is widely taken to be of a power - law form @xmath27 , the classic salpeter imf @xcite corresponding to @xmath28 , and a more recent suggestion being @xmath29 @xcite . we adopt the salpeter imf for our main calculations and study the effect of varying the imf on the @xmath4-distribution in sec.[sec : imfvary ] . the @xmath30-distribution is generally taken as a power - law in @xmath30 , given by @xmath31 , where the exponent @xmath32 may depend on the type of xrb population being studied . following the usage in recent works on population studies of hmxbs @xcite , we adopt a uniform @xmath30-distribution , @xmath33 , in our hmxb studies in this paper . ( however , we do note that our formalism has the provision for handling non - uniform distributions of @xmath30 , which will indeed be used in our lmxb studies described in subsequent papers in this series . ) the distribution of the orbital separation @xmath14 is almost universally taken in the literature to be a loguniform one ( _ i.e. _ , an equal number of systems in equal intervals of @xmath34 ) , also known as pik s law @xcite . this implies a probabilty density @xmath35 of the separation being in the range @xmath14 to @xmath36 as @xmath37 . we adopt pik s law in our work . in order to construct the total probibility distribution function of primordial binaries , we need only note that the above pieces of the probability density function ( pdf ) are independent of each other , subject to the restrictions in the allowed ranges of values detailed above . thus , subject to these restrictions , the total pdf is given by : @xmath38 here @xmath39 is a normalization parameter . note that , although this pdf is mathematically defined over much larger ranges of parameters , we are interested only in the allowed range for hmxb formation , as explained above . hence , we set @xmath40 over the forbidden range , and so choose the normalization parameter @xmath39 that the integral of the pdf over the allowed range is unity . consequently , although this pdf has no explicit dependence on @xmath30 due to the assumed uniform @xmath30-distribution for this hmxb study , the allowed range of @xmath30 , _ i.e. _ , @xmath41 , does have an effect on the value of @xmath39 . in this section , we summarize the changes in the binary parameters as the system evolves through the first mass - transfer phase , and the subsequent sn of the he - core of the original primary . as mentioned in sec.[sec : intro ] , these changes only require keeping track of the relations between the initial and final states in each of the above two processes , which makes the ensuing transformations relatively simple . in keeping with previous work on the subject , we assume that there is negligible mass loss from the system during this process , _ i.e. _ , the mass transfer is _ conservative _ , so that the h - envelope of the primary is entirely transferred to the secondary , leaving behind its he - core of mass @xmath42 given by : @xmath43 where typical values of @xmath44 and @xmath45 used in the literature are 0.073@xmath16and 0.704 respectively , for a metallicity of @xmath46 . this is a commonly - used analytic approximation to the results of numerical stellar - evolution calculations ( ghosh 2007 and references therein ) . because the mass transfer is conservative , the binary parmeters @xmath47 , @xmath48 , and @xmath49 at the end of it are related to the primordial binary parameters as : @xmath50 and @xmath51 ^ 2~. \label{eq : piachange}\ ] ] the post - mass - transfer binary is detached . the he - core described above evolves further and explodes as a supernova ( sn ) , which leaves behind a neutron star of a typical mass of 1.4 @xmath16 and blows off the rest of the he - star . this sudden mass loss from the system alters the orbital separation and makes the orbit eccentric . we assume in this work that the neutron star has the above mass in all cases . neglecting any effects of the expanding supernova ejecta on the secondary , _ e.g. _ , ablation , the secondary remains completely unchanged in this process . since we assume that the mass of the neutron star is always fixed to 1.4 @xmath16 , it can no longer be used formally as a free parameter in the description of the post - sn system . therefore , we choose the orbital eccentricity @xmath52 as the third parameter , so that the post - sn system is described by the companion mass ( @xmath53 ) , the semimajor axis ( @xmath54 ) and the eccentricity ( @xmath52 ) . natal supernova kicks introduce additional changes in the post - sn parameters . it is generally believed today that the sn explosion need not have exact spherical symmetry . many suggestions have been given for the physical origin of this asymmetry , _ e.g. _ , density inhomogeneity in the pre - collapse core , anisotropic neutrino emission , unequal momentum fluxes in the jet and anti - jet directions , and so on @xcite . it has also been pointed out that a seed anisotropy , once introduced by these mechanisms , will be enhanced during the hydrodynamic evolution of the explosion and may impart a large kick to the neutron star . proper motion measurement of radio pulsars is used as a diagnostic of these natal kicks . such studies have shown that the distribution of the pulsar velocities can be described by a 3d isotropic maxwellian , in which the kick speed @xmath55 is distributed as @xmath56 @xcite , with @xmath57 km / s @xcite , and the direction of the kick velocity is distributed isotropically . earlier works had suggested that the distribution might actually be bimodal , represented by two maxwellians , so that the overall distribution would be the weighted sum of these @xcite . recent works have argued that electron - capture supernovae ( ecsn ) are likely to produce much lower kicks with @xmath58 km / s or lower @xcite , while the larger value given in the previous paragraph is appropriate for iron - core - collapse supernovae ( iccsn ) . we note here that studies of the proper motions of pulsars involve single pulsars , some of which would have come from a single star collapse . for the rest , which were obtained by disruption of binary systems , there would clearly be an observational bias towards higher kick velocities , since it is these systems which became preferentially unbound to yield the single pulsars whose proper motions are studied observationally . however , such effects are difficult to account for at this stage of understanding of the problem , and it is customary in the literature to apply the inferred distribution of kick velocities directly to the x - ray binary progenitors , as we have done here . effects of isotropic maxwellian sn - kicks in progenitors of x - ray binaries were studied in a pioneering work by kalogera ( 1996 ) . for our study here , we have devised a method related to that described in the above work , but designed specifically for our purposes here . the method works as follows . if we define our co - ordinate system such that , just before the sn explosion , the line joining the two stars is along the x - axis and the neutron star is moving in the positive z - direction with a keplerian orbital velocity @xmath59 , then a kick of magnitude @xmath60 in the direction @xmath61 would produce a change in orbital parameters given by : @xmath62 & = & m_f^ta_f\left(1-e^2\right ) \label{eqn : postsn}\end{aligned}\ ] ] here the angles @xmath63 and @xmath64 are defined in the usual way , @xmath65 denotes the total mass of the system , and the subscripts @xmath66 and @xmath67 denote the initial and final values ( _ i.e. _ , _ pre - sn _ and _ post - sn _ values ) . all quantities here are in solar units , with @xmath55 in the units of @xmath68 . due to the maxwell - distributed random sn - kicks given to the neutron star , the one - to - one correspondence between the pre - sn and post - sn binary parameteres , which would have existed in the absence of such kicks , is broken now . instead , we need to calculate a suitable _ average _ effect of sn - kicks on the transformation between the pre - sn and post - sn parameters , which takes into account the underlying distribution of the kicks . to this end , we average eqs.[eqn : postsn ] over the distribution of @xmath60 , upon which the linear terms in @xmath69 vanish due to isotropy , and the distribution - averaged transformation equations become : @xmath70 here , @xmath71 is @xmath72 averaged over the kick - distribution . for a maxwellian distribution , it is given by : @xmath73 those properties of @xmath71 for a maxwellian which we need for our work are detailed in appendix a. we assume here a combination of two maxwellians with values of @xmath74 appropriate for iccsn and ecsn , as described above . the way for determining the appropriate proportions of these two components is described below in sec.[invtrans ] . for each of the two components , the parameter transformations are also given in appendix a. probability theory provides us with a method of transforming the pdf of a set of variable to the pdf of another set of variables , if we know the relations between the these two sets of variables . let us assume that a system is described by a set of variables denoted by a vector @xmath75 . a given specific process transforms this set of parameters to another set of ( the same number of ) parameters denoted by @xmath76 . both the forward transformation @xmath77 and the inverse transformation @xmath78 are mathematically defined . the initial pdf as a function of @xmath75 , @xmath79 then transforms as follows : @xmath80 here , @xmath81 is called the jacobian determinant of the inverse transformation . the above theorem can now be applied to our problem in order to transform the pdf of the primordial binaries to the post - sn pdf . in order to apply the above formalism to our problem , we need to invert the above parameter transformation relations , which is straightforward . the jacobian of this transformation ( @xmath82 ) can also be readily calculated . in the two regions described in appendix a , these transformations can be obtained and the jacobian ( @xmath83 ) can be calculated explicitly , as detailed in appendix a. working in either region , the post - sn pdf is given by : @xmath84 the pdf of the parameters of the post - sn binary , @xmath85 , no longer factorizes into individual pdfs for these three parameters , unlike the situation for the pdf of primordial binaries . this is as expected , since the parameter transformation laws given above mix the parameters . an immediate consequence of this is that the pdf for any one of the above post - sn binary parameters needs to be obtained by integrating the pdf of eqn.[eqn : psnpdf ] over the other two parameters . as described above , we use two values of the dispersion in the kick distribution , @xmath74 , appropriate for iccsn and ecsn . we have also described above the method of calculating the distribution in each case . the actual distribution of the parameters with both types of sn present is the weighted sum of these two pdfs for the two values of @xmath74 , and this weighting comes from the following considerations . for the range of @xmath12 which leads to ecsn events , we have adopted here the value advocated in recent works @xcite , which is 8 - 11 @xmath16 . the relative contribution of the ecsn and iccsn is then given by @xmath86 , which depends upon the imf . here , @xmath87 define the allowed range of primary masses given earlier , and @xmath88 is the primary mass for transition from ecsn to iccsn . for a salpeter imf , this ratio is 0.42 for @xmath89 and 0.67 for @xmath90 . we take this ratio as 0.42 wherever it occurs in our calculations here . in order to appreciate the nature of the post - sn distribution , we calculate and plot the individual pdfs for each of these parameters , obtained by integrating @xmath91 over the other two . we now discuss each of these distributions . the distribution of the companion mass shows a slow early rise and then broken power - law behaviour with the break occuring at @xmath92 , as shown in fig.[fig : mcdf ] . the power - law slope is @xmath93 , in the mass range @xmath94 , whereas above the break point the @xmath53-distribution becomes steeper , with a slope of @xmath95 . the kink seen at the breakpoint of 30@xmath16 becomes a _ fold _ in the bivariate distribution as a function of @xmath53 and @xmath54 ( see fig . [ fig : mcadf ] ) . this kink appears at the mass which corresponds to the upper limit of the primary mass in the primordial binary . the allowed range of the primary mass is @xmath96 : this constraint was imposed to ensure the formation of neutron star as an end product of the stellar evolution of the primary . it can be easily seen from the relation given above between the mass of the primary and that of its he - core that the mass range obtained for @xmath53 if we take @xmath97 , corresponding to the entire allowed range of @xmath13 , is @xmath98 . thus for @xmath99 , a large range of @xmath12 can contribute , whereas for @xmath100 the allowed range of @xmath12 is rapidly cut off as @xmath53 increases , which causes a rapid reduction in the number of systems possible in this region . thus , the effect of the allowed phase - space region is reflected in the transition in the power - law index of the @xmath53-distribution . in the region where large parts of the phase space of @xmath12 are allowed to contribute , the slope of the pdf is close to the input imf slope of @xmath101 . for masses above the kink , only small parts can contribute , resulting in a steeper fall . for similar reasons of forbidden phase space , a downturn is expected at very low values of @xmath53 , in the mass range @xmath102 . upon inclusion of sn - kicks , the above mass distribution is essentially unchanged in the mid-@xmath53 range , except that the kink at @xmath92 becomes less prominent . but the distribution is cut off more sharply at the lowest and highest masses , the former being immediately understandable since kicks tend to preferentially unbind systems with lower binding energy . the bivariate @xmath103 distribution shown in fig.[fig : mcadf ] reveal the complexity introduced by the kicks . although the surface shows similarities to its no - kick counterpart in general features , several additional features appear due to maxwellian - averaged kicks corresponding to the two different values of @xmath74 given above for the two types of sn , namely , ecsn and iccsn . fig.[fig : ladf ] shows the pdf of @xmath104 , the post - sn semimajor axis on a logarithmic scale . the pdf without considering the effect of sn - kicks is flat ( _ i.e. _ , loguniform in @xmath54 ) in the mid-@xmath54 range , with smooth rise and fall at low and high values of @xmath54 respectively . this is clearly a direct consequence of the assumed flat , piks - law distribution of primordial binaries , which remains unchanged at the intermediate values of @xmath54 . the pdf with the sn - kicks included clearly shows two additional features . first , there is a general shift of the distribution towards wider orbits , entirely as expected , since sn - kicks generally tend to widen orbits . second , the pdf is not exactly loguniform , _ i.e. _ , flat in the mid-@xmath54 range now : a slow fall with increasing @xmath54 is observed in this range , before the pdf falls off rapidly at large @xmath54 , as before . this second feature is a consequence of the fact that sn - kicks steadily reduce the probability of survival of wider binaries even in the mid-@xmath54 range , before destroying them altogether at large @xmath54-values . in the absence of kicks , the eccentricity introduced by the sn explosion is equal to the fractional mass loss form the system . as this fraction is small ( @xmath105% or less ) in case of hmxbs , only small eccentricities are introduced by the sn in the no - kick scenario . figure [ fig : edf ] , which shows the post - sn eccentricity pdf , confirms this . however , introduction of the sn - kicks changes this result completely , making the post - sn systems much more eccentric , as expected . further , the results for iccsn and ecsn are quite different in terms of post - sn eccentricity , also as expected because of the large difference between @xmath74 in the two cases ( see above ) . due to the small kicks imparted in case of ecsn with @xmath71 independent of the details of the pre - sn system ( see sec.[sec : sn ] and appendix a ) , the eccentricities are generally smaller than in the iccsn case ( although larger than in the no - kick case ) , and the e - distribution is spread over a large range . by contrast , for iccsn the kicks are much larger , leading to large eccentricities , as shown in fig.[fig : edf ] . further , in this case @xmath71 is almost a fixed fraction of the disruption velocity of the pre - sn system for maxwellian ( and possibly also for other similar ) distributions , as explained in appendix a , so that the e - distribution is narrowly peaked around a value determined by that fraction . it is likely that more detailed calculations would widen this peak somewhat , but no qualitative changes are expected . the composite pdf of eccentricity for ecsn and iccsn naturally shows a double - peaked structure , as in fig.[fig : edf ] . of course , the e - distribution calculated above refers to the immediate post - sn systems , which are not observable as xrbs . these systems subsequently undergo tidal circularization rapidly , so that most of them become circular or nearly so by the time mass transfer begins and the system turns on as a hmxb . since eccentricity is an irrelevant parameter at the hmxb stage , we integrate @xmath91 over the eccentricity to obtain only the bivariate pdf as a function of @xmath53 and @xmath54 , which is shown in fig.[fig : mcadf ] . post - sn systems evolve over the main - sequence lives of their companions as detached systems on a short timescale ( @xmath10 years ) , during which the orbit is tidally circularized . therefore , a bivariate post - sn pdf adequately describes the hmxb system , being given by : @xmath106 observed collective properties of hmxbs are usually given as distributions of their luminosities ( @xmath107 ) ( _ i.e. _ , the x - ray luminosity function or xlf ) and orbital periods ( @xmath108 ) , instead of @xmath53 and @xmath54 , which we have worked with upto this point . in order to compare our results with observations , a further transformation @xmath109 is therefore required now . @xmath108 can be calculated for given values of @xmath53 and @xmath54 using kepler s third law . if all the masses and distances are expressed in the units of the solar values , then @xmath108 in hours is given by : @xmath110 here @xmath111 is the mass of the neutron star . to calculate the luminosity as a function of @xmath53 and @xmath54 , prescriptions of the mass - loss rate from the companion and the capture mechanism by the neutron star are required . the companion expands into a giant / supergiant after completing its main - sequence life , and loses mass by driving a strong stellar wind at a rate @xmath112 . the neutron star captures a fraction of this lost mass . the accretion rate onto the neutron star is then given by @xmath113 ( capture fraction ) . in this work , we assume that hmxbs are entirely _ wind - fed giant / supergiant _ systems , thus ignoring those systems in which a main - sequence companion close to its roche lobe may drive a ( weak ) _ atmospheric _ roche - lobe overflow . this is certainly justified at higher x - ray luminosities , since systems fed only by the latter mechanism would be found at the lowest end of the xlf . in a similar vein , we also do not include hmxbs with be - star companions ( see sec.[sec : bestar ] ) , since the transient nature of these systems with low duty cycles implies a low _ time - averaged _ luminosity , which would put them , again , at the low end of a time - averaged xlf , which is generally difficult to determine . the x - ray luminosity @xmath107 is then given by the standard stellar - wind model , according to which the massive companion drives a wind mass - loss at a rate @xmath112 , and with a terminal wind speed of @xmath114 . of this , a fraction @xmath115 ^ 2 $ ] ( the capture fraction introduced above ) is captured and accreted by the neutron star , and the gravitational energy release from this accretion generates the x - rays . in order to connect @xmath107 to the stellar and orbital parameters , we first note that @xmath107 is the rate of release of the gravitational energy of the accreted matter and hence is directly proportional to accretion rate @xmath116as @xmath117 which we can express numerically as @xmath118 , where @xmath119 is @xmath107 in units of @xmath120 ergs / s and @xmath121is @xmath116 in units of @xmath122 @xmath16/yr . next , we note that the accretion rate @xmath116is related to the wind mass - loss rate @xmath112 and the above capture fraction as @xmath123 ^ 2~. \label{eqn : accrate}\ ] ] to proceed further , we need models of stellar winds , which we consider next . models of stellar wind relate the mass loss rate and the terminal velocity of the wind to stellar parameters like mass , radius and luminosity . in our case , these parameters are the mass of the companion , the radius of the companion @xmath124 and the luminosity of the companion @xmath125 . some well - known stellar - wind models which we consider here are those due to castor _ _ , kudritzi - reimers , and vink _ et al . _ @xcite , which span the range from the classic cak model @xcite of the 1970s to the recent vink _ et al . _ model @xcite . the rates of mass loss in these models are given by : @xmath126 the constants @xmath127 , @xmath128 , and @xmath129 are given in the original works . we use the kr model in our main work here , and summarize the effects of varying the stellar - wind model in sec.[sec : windvary ] . results of numerical calculations of stellar evolution give stellar parameters like radius and luminosity as functions of the mass , metallicity and the age of the star . for using the above stellar - wind models in our calculations , we need these parameters for a given mass and metallicity of the wind - driving companion . in the spirit of our semi - analytic approach , we use the parameters given by hurley , pols and tout ( 2000 , henceforth hpt ) in their comprehensive work on the construction of analytic approximations to the standard numerical results of stellar - evolutionary codes . the range of @xmath53 shown in fig.[fig : mcdf ] and discussed in sec.[sec : compmass ] leads to the conclusion that , except for a small region at the lower end of this range , the massive companions ignite he in the core while in the hertzsprung gap ( hg ) , while those in that small region do so at the top of the giant branch ( gb ) . the companion s luminosity remains nearly constant in the hg , as discussed in hpt . relevant fitting formulae for the luminosity and radius in the core helium burning ( cheb ) phase are given in sec.5 of hpt , particularly secs.5.1 and 5.3 . we have used a metallicity @xmath130 in our main calculations , varying it later to test its effect on hmxb distribution ( see sec.[sec : othervary ] ) . the terminal velocity @xmath114 which appears in eqn.[eqn : accrate ] is of the order of escape velocity from the surface of the companion and hence can be taken as a function of @xmath53 , so that the accretion rate can be expressed in the form : @xmath131 where the function @xmath132 contains the physics of wind mass - loss . equation [ eqn : accr ] gives the accretion rate as a function of @xmath53 and @xmath54 and so , along with eqn.[eqn : kepler ] , provides the transformation from @xmath133 to @xmath134 , inverse transformations for eqns . [ eqn : accr ] and [ eqn : kepler ] need to be obtained to transform the pdf . we first carry out the transformation from @xmath133 to @xmath134 . we note here that the pdf as a function of @xmath116 and @xmath107 will be identical up to a numerical factor since the two are linearly related to each other . the inverse transformation can become complicated because @xmath135 can be a very complicated function and hence difficult to invert analytically . our goal would , of course , be to construct a procedure that works with a general @xmath135 , so that the formalism can be applied to any model without changing the procedure . we first write the accretion rate as @xmath136 the numerical factor is so chosen that @xmath137 ; for other models the numerical constants can be appropriately adjusted in a similar way . with the aid of eqns.[eqn : kepler ] and [ eqn : accrate ] , we can eliminate @xmath54 and write @xmath138 where @xmath139 this equation can be numerically solved to obtain @xmath53 as a function of @xmath140 , which is the first inverse transformation equation . after computing @xmath53 numerically , @xmath54 can be computed using @xmath141 where @xmath142 using these inverse transformations , we calculate the jacobian . the partial derivatives take the following forms in terms of the functions @xmath143 , @xmath144 , @xmath145 and @xmath146 : @xmath147 from eqs.([eqn : htr1 ] ) , ( [ eqn : htr2 ] ) & ( [ eqn : partials ] ) , we can calculate the jacobian as a function of @xmath108 and @xmath121 as : @xmath148 ^ 3\ ; \left\{3\:\frac{df}{dm_c}-2\:\frac{f(m_c)}{m_c+m_{ns}}\right\}^{-1}\ ; \left[1.84\times10^{-5}\:{\ensuremath{\dot{m}_{-10}}}^3p_b^5\right ] \label{eqn : hjacob}\ ] ] the distribution of hmxbs in @xmath108 and @xmath121 can now be obtained with the aid of eqs.([eqn : jacob ] ) and ( [ eqn : psnbi ] ) . it is : @xmath149 and @xmath150 and @xmath107 are of course proportional to each other , as explained above . the resultant bivariate pdf @xmath151 is displayed as a surface in fig.[fig : pbldf ] , and the individual pdfs for the @xmath107- and @xmath108-distributions are shown in figs.[fig : hxlf ] and [ fig : pdist ] respectively . for obtaining the individual distribution of each variable , the bivariate distribution is integrated over the other variable , as explained earlier , the range of integration extending over only the allowed range of the concerned variable . in fig.[fig : pblrange ] , we show the allowed region in the @xmath152 plane , which shows how the allowed region in the @xmath103 plane shown in fig.[fig : mcarange ] is mapped onto the plane of the new variables . the effect of the above transformation is a rotation of the allowed zone in @xmath153 plane . the simple bounds on parameters of primordial binaries given earlier transformed into a nearly rectangular allowed region in the parameter space @xmath133 for post - sn binaries ( see fig.[fig : mcarange ] ) . after transformation to the _ hmxb parameters _ , however , this allowed zone rotates into an inclined band , as shown in fig.[fig : pblrange ] . the boundaries given in this figure serve as integration limits for the calculation of the individual pdfs from the bivariate distribution in terms of @xmath107 and @xmath108 . we now discuss the nature of the distributions of the luminosities and binary periods of hmxbs calculated above , and compare them with the current state of observational knowledge of these distributions . we emphasize that we are not attempting a detailed fit to the data at this stage , but focusing instead on a comparison between the general trends in calculated and observed distributions . the purpose of such a comparison is of course to determine if a more detailed computational scheme , _ e.g. _ , a numerical population synthesis , is worthwhile in the future for a giving a more detailed account of the observations . a major virtue of an approach like ours is that it is capable , at least in principle , of assessing in a transparent way the relative importance of various components ( _ e.g. _ , the role of the primordial binary distributions vis - a - vis that of the evolutionary processes like the first mass transfer and the sn ) in determining the final hmxb distribution . such an assessment helps greatly in planning the strategy of subsequent computational studies . the pdf of hmxbs as a function of @xmath108 is obtained immediately by integrating @xmath154 given in eqn.[eqn : hmxbpdf ] over the accretion rate . the computed @xmath108 distribution without sn - kicks ( see fig . [ fig : pdist ] ) is rather similar to the corresponding post - sn @xmath54 distribution , _ i.e. _ , a flat pik s law in the mid - region ( @xmath155 hrs ) , with a gradual rise to this flatness at shorter periods and a slightly sharper fall - off from it at longer periods . the distribution with the inclusion of sn - kicks also resembles the post - sn @xmath54 distribution with sn - kicks : the rise now is more gradual , reaching a maximum at @xmath156 hrs . instead of a flat top , a slow fall is observed mid-@xmath108 range , followed by a sharper fall - off at long periods , as before . the major and obvious difference between the two distributions is an overall shift towards wider orbits and longer periods when sn - kicks are included , entirely as expected and as seen earlier in the post - sn @xmath54-distribution . we compare our theoretically obtained @xmath108-distribution with observations in sec.[sec : observe ] . the xlf of hmxbs can be obtained by integrating eqn.[eqn : hmxbpdf ] over @xmath108 , using the limits described previously and carrying out a simple linear transformation for @xmath121 to @xmath119 . the xlf is given by : @xmath157 where @xmath158 . figure [ fig : hxlf ] shows the numerically computed xlf . a broken power - law can be fitted to this computed xlf , with a cut - off at the neutron - star eddington limit . equation [ eqn : hxlf ] gives the power - law exponents in the low and high luminosity regimes , obtained by least - squares fits to the computed xlf : @xmath159 the xlf calculated without sn - kicks is also given for reference . it overlaps with the xlf with sn - kicks in the low - luminosity regime , and shows a slightly shallower power - law of exponent is @xmath160 in the high - luminosity regime , the cross - over point being at @xmath161 erg / s . we now examine the effects of the various model parameters , _ e.g. _ , stellar - wind model , imf slope , metallicity , and so on on our theoretical xlf . we first compare the calculated @xmath107-distribution of fig . [ fig : hxlf ] with the observed distribution . the major feature of the observed @xmath107-distribution is a power law with a differential slope @xmath162 over a wide range of luminosities @xmath163 . this is the so - called `` universal '' x - ray luminosity function ( xlf ) of hmxbs , obtained in the following way . whereas the observed xlfs of various nearby early - type galaxies ( _ e.g. _ , the milky way , smc , m82 and m83 , the antennae , ngc 4736 , and so on ) follow this trend , their normalizations are not the same . however , when the xlf of a given galaxy is normalized by the current star - formation rate ( sfr ) in that galaxy , xlfs for all these galaxies fall essentially on top of each other , yielding the above `` universal '' xlf @xcite . this is due to the well - known fact that the strength of the hmxb population and their x - ray output is proportional to the current sfr , and is closely related to the discussion given in earlier on how the rapid hmxb evolution leads to a distribution of hmxbs which is , in effect , a `` snapshot '' of the galaxy taken at its current sfr . in comparing calculated and observed xlfs , we note first that , since we confine ourselves to only neutron - star hmxbs in this work , our calculated xlf applies only to luminosities not exceeding the eddington luminosity @xmath164 for a @xmath1 1.4@xmath16 neutron star , _ i.e. _ , @xmath165 . the observed xlf extends upto almost 2 decades of luminosity above this , and the hmxbs at these higher luminosities are believed to be black - hole systems , with the possibility of both stellar - mass black holes and intermediate - mass black holes ( at the highest luminosities ) being present . we return to this question in more detail in sec.[sec : blackhole ] . with this caveat in mind , we note that there is a remarkable agreement between the calculated and observed xlfs in the luminosity range @xmath166 , above which the calculated xlf cuts off at the eddington limit for neutron stars , as expected , so that no comparison with observations is possible at higher @xmath107 . below @xmath167 , there was no data when the works referred to at the beginning of this subsection were published . however , more recent @xmath168 observations of the magellanic clouds have extended the xlf below this lower limit , down to about @xmath169 or slightly lower , for smc and lmc @xcite . the results indeed suggest a flattening of the xlf at low luminosities , and the best - fit differential slope of the observed xlf at these luminosities for smc , which is given in the above reference as @xmath170 , is in fact almost consistent with our calculated slope at low luminosities , given in fig.[fig : hxlf ] and eq.([eqn : hxlf ] ) . however , we must be cautious with our calculated xlf in the low - luminosity regime , since we have neglected ( a ) main - sequence companions undergoing atmospheric roche - lobe overflow , and ( b ) be - star companions in our calculations here , as explained above . observations of hmxb orbital periods are available in substantial numbers only for our galaxy and the magellanic clouds . measurements of orbital periods have been performed for @xmath171 systems @xcite . figure[fig : pbobs ] shows the distribution of orbital periods constructed with this data . the distribution shows a rising part in the range @xmath172 hrs . with the ( large ) error bars , the distribution in the region @xmath173 hrs is consistent with either a uniform trend or a slow rise / decay with a peak around @xmath174 hrs . it can be seen that our theoretical @xmath108-distribution is generally consistent with the observed one within the error bars , except at very long orbital periods , where the observed distribution shows a sharp cut - off beyond orbital periods @xmath175 year . such an apparent cut - off is expected for two observational reasons . first , it is very difficult to follow systems with such long orbital periods for sufficient times to establish reliable orbital periods . second , such wide binary systems would generally have low luminosities , which would put them at the faint end of the xlf , and add further to the difficulties of a successful observation . thus strong selection effects work against observation of hmxbs with long orbital periods . by contrast , the theoretical distribution extends upto the widest orbits that can survive the basic processes involved , particularly the sn explosion . the fact that these very wide binaries do not show up in the observed distribution does not imply that they do not exist , but simply that they are unobservable in practice . we now examine the effects of the various model parameters , _ e.g. _ , stellar - wind model , imf , metallicity , and so on on our theoretical xlf . we consider the two alternative models by cak and vink _ et al . _ , which were introduced in sec.[sec : windmodel ] . xlfs computed for these models are shown in fig.[fig : xlfmod ] . it can be seen that the xlfs overlap in the low - luminosity regime for all models , which means that the xlf is independent of the exact form of @xmath135 ( see above ) in this region . the xlf s differential slope is close to -1 in this regime . the position of the cross - over luminosity @xmath176 is different for the three models with similar power - law exponents above respective @xmath176 . the cak model cuts off rather abruptly at luminosities considerably below the eddington luminosity of neutron stars and so appears unable to account for the observations . barring the difference in the @xmath176 , the kudritzki - reimers model and the vink model give similar results , in general agreement with observations . the imf is typically given as a single power - law in the mass range that is relevant to the problem at hand . the power - law index of the imf ( @xmath177 ) is considered to be within the range @xmath178 , @xmath179 being the standard value given by salpeter and widely used in this mass range . we study the effect of varying @xmath177 on xlf in the range 2.0 - 3.0 . the effect is shown in the fig.[fig : xlfmod ] . the xlf at low luminosities below the kink shows hardly any variation . we therefore display the xlf only for luminosities above the kink . even above the kink , the xlf slope varies by a small amount ( from -1.47 to -1.66 between the lower and upper limits to @xmath177 considered here ) , showing relative insensitivity of the xlf to the imf slope . for an exact matching with the observed xlf , a slightly steeper imf seems to be preferred , if other parameters remain unchanged . among other parameters in the problem , we consider the stellar - wind velocity @xmath114 and the metallicity @xmath180 . the terminal velocity of the wind is an important factor in determining the capture fraction . it is typically of the order of the escape velocity at the surface of the companion and generally thought to be @xmath181 km / s . we varied @xmath114 around this canonical value to test the effect on the xlf . the changes in both the xlf slope and the high - luminosity cutoff were insignificant . consider next the metallicity @xmath180 , which affects the stellar parameters and the wind mass loss rate . we considered a large range of @xmath180 from @xmath182 to @xmath183 , and used the model of vink _ et al . _ ( 2001 ) to study the effects of varying @xmath180 on our final results . the results are shown in fig.[fig : xlfmod1 ] . there is no significant difference between the xlfs for @xmath184 and @xmath185 : both have a slope @xmath0 in the luminosity range @xmath186 erg s@xmath187 . the xlf for @xmath188 also has nearly the same slope , but a somewhat different crossover luminosity . finally , the xlf for @xmath189 is markedly different , but such low metallicities are not realistic for hmxbs , as they would be expected only in old stellar systems . in this work , we have developed a strightforward , first - principles scheme for understanding collective properties of hmxb populations , starting from standard , well - known collective properties of primordial binaries which are the progenitors of hmxbs , and following the transformations of the probability distributions through the evolutionary processes that lead to the formation of hmxbs . our purpose , of course , was to assess if the standard picture of primordial binaries and the standard evolutionary scenario for hmxbs together can account for the observed collective properties of hmxbs in a basic , simple way . the fact that we find that such an account can indeed be given is most encouraging , and it constitutes , in our view , an essential step towards attempts at elaborate population synthesis schemes designed for understanding further details of these collective properties . since our procedure is transparent and readily understandable at each step , it is easy in our scheme to follow the role of each ingredient in shaping the final hmxb distribution . in this section , we discuss various issues of principle and procedure which are relevant to this line of approach , and conclude with plans for the future . we pointed out in sec.[sec : primordial ] that distributions of the masses @xmath190 of the primary and secondary in the primordial binary can be described either in terms of these masses themselves , _ i.e. _ , the pair ( @xmath190 ) , or alternatively in terms of the primary mass and the mass ratio @xmath191 , _ i.e. _ , the pair ( @xmath192 ) . both have been done in the literature @xcite , and we have chosen here the second description for its closer correspondence with essentially all recent work . the relation between these two approaches has been discussed thoroughly by tout ( 1991 ) . massive companions in hmxbs are of two types in general , _ viz . _ , ob giants / supergiants and be stars , the latter being characterized by ( a ) strong , broad emission lines that supply evidence for rapid stellar rotation , and ( b ) somewhat lower masses and wider orbits . since we have confined our detailed calculations in this work to the former , as stated in sec.[sec : psntoxrb ] , we now discuss the expected role of be - star binaries in hmxb xlf . it is believed that be stars are often surrounded by an outflowing disk of matter expelled by centrifugal forces from the fast - roating equatorial regions of the star , in addition to the usual stellar wind emitted from all over the stellar surface . accretion by the neutron star from this outflow material generally follows the basic description from stellar winds given in sec.[sec : windmodel ] , with appropriate values of @xmath114 for the fast wind and the slowly outflowing disk , with one major caveat . since the matter in the rapidly - rotating outflow disk has much angular momentum , it may form an _ accretion _ disk around the neutron star , which would then drain on the neutron star a slow , viscous timescale . this complicates the description considerably , as the outflow disk is generally expected to be tilted with respect to the orbital plane @xcite , so that the orbiting neutron star would `` crash '' through this outflow disk ( twice per orbit in general ) , possibly acquire an accretion disk , and accrete it slowly over the rest of the orbit . this would naturally lead to outbursts of x - ray emission , which are indeed observed in be - star hmxbs . the question for our purposes here is : how does all this affect the xlf of hmxbs ? first consider the observational situation . the point to note here is that be - star hmxbs are basically transient systems with low duty cycles , so that a long - term average of the luminosity of a be - system is much lower than that of an ob supergiant - system . thus , in an xlf constructed from a long - term monitoring of the x - ray sky with an all - sky monitor , one would expect the high-@xmath107 parts to be dominated by ob - systems , while the low-@xmath107 parts may have considerable contributions from be - systems . indeed , the observed xlfs cited earlier in this paper have been constructed from recent work with x - ray observatories in the following way . the xlf for the milky way has be constructed from @xmath193 years of observation with the all - sky monitor on @xmath194 . thus , the above argument certainly applies to this case . indeed , since the typical long - term average expected from be - systems would be at or below the lower end of the luminosity - range over which the xlf is actually reported ( and this applies to basically all observed xlfs except those obtained from @xmath168 observations of smc and lmc ; see sec.[sec : compxlf ] ) , we would expect little contribution to the reported xlf from be - star systems . however , note that the xlf for other nearby galaxies reported in these references have been constructed essentially from one `` snapshot '' ( _ i.e. _ , a single exposure ) taken by @xmath195 and @xmath168 @xcite . given this , it is remarkable that the xlfs of all these galaxies ( suitably normalized by their sfrs ) are essentially coincident with one another over the same range of @xmath107 , as explained earlier . considering the fact that , because of their low duty cycles , only a small fraction of the be - systems would be present in these `` snapshot '' xlfs , we would still expect some contribution from them in a range of @xmath107 typical of the outbursts of be - systems . the fact that these xlfs appear very similar to the above long - term average xlf of the milky way over the canonical luminosity range @xmath196 is therefore most noteworthy , and may imply that , for reasons which are not clear at present , be - systems may not have made a substantial contribution to these observed xlfs . now consider the inclusion of be - systems in calculational schemes like ours . to the extent that the wind - accretion formalism can be applied to accretion from both the fast , low - density wind from the stellar surface and the slow , high - density outflow in the equatorial disk @xcite , these systems are already in our scheme , at least in principle . however , while a quantitative description of the periodic acquisition and drainage of accretion disks described above has been done for individual binary systems @xcite , its inclusion in a study of collective properties of hmxb populations is more complex and outside the scope of this work . as mentioned in sec.[sec : compxlf ] , the observed `` universal '' xlf of hmxbs extends to about two decades of luminosity above the eddington luminosity for canonical 1.4@xmath16 neutron stars . while we focus on neutron - star hmxbs in this study , it is interesting to consider this brightest end of the xlf briefly . if the x - ray sources here are accretion - powered , they can only contain accreting black holes ( some sources upto luminosities @xmath197 erg s@xmath187 can of course be close juxtapositions of several neutron - star sources , as has sometimes been suggested ) , either stellar - mass ones ( @xmath198 ) , or even the so - called intermediate - mass black holes ( imbh ) , with masses @xmath199 , the x - ray sources corresponding to the latter objects being often called ultra - luminous x - ray sources ( ulx ) . a truly remarkable feature of the universal hmxb xlf is that a single , smooth power law gives an excellent account of x - ray binaries containing neutron stars , stellar - mass black holes , and imbhs over the entire luminosity range @xmath200 , _ i.e. _ , above the `` kink '' in the xlf @xcite . while it is not difficult to extend the formation and evolution scenario outlined in the earlier sections to include higher primary masses which would produce stellar - mass black holes , and then imagine that the other systematics would go through in such a way as to extend the power - law xlf , this argument does not automatically include ulxs and imbhs , whose formation scenario has to be different and more exotic , _ e.g. _ , black - hole mergers in dense stellar clusters . and yet , as gilfanov ( 2004 ) has noted , these `` rare '' and `` exotic '' objects appear to form a smooth extension of the `` ordinary '' hmxb population . whether this is really true or not can possibly be probed with future observations in a way suggested by this author : if we adopt the alternative hypothesis that the apparent cutoff in the currently observed xlf at @xmath201 really corresponds to the maximum possible luminosity of what we may call `` ordinary '' , stellar - mass black holes referred to above , then these `` exotic '' imbhs may show up beyond this cutoff if we can observe regions with extremely high star - formation rates , since the merger scenario for the formation of imbhs implies that they would occur in very dense regions with very high star - formation rates . however , since such imbhs would necessarily be rarer than stellar - mass black holes , we should expect a `` step down '' at the presently observed cutoff , beyond which the xlf would continue at a lower strength . this is a fascinating possibility . understanding the xlf shape is an important step towards extracting information about various processes which determine the collective properties of hmxbs . two main features of the xlf calculated using our scheme are ( 1 ) a kink at @xmath202 erg / s , and ( 2 ) a power - law behavior with the differential slope of -1.6 above @xmath176 . we will discuss the possible origin of these two features in this section . the origin of the kink in the hmxb xlf can be understood in terms of allowed zones in the parameter space . this is demonstrated in a clear way by overplotting contours of constant @xmath107 on the allowed zone in the @xmath133 plane , as shown in fig.[fig : mcarange ] . let us first examine a simple `` toy '' model for the xlf . in this model , we approximate the post - sn pdf as a function of @xmath53 and @xmath54 as @xmath203 ( as discussed earlier , this `` toy '' is not too bad an approximation over much of the parameter range in case of no sn - kicks ) . we now note that @xmath107 can be written in a schematic way as @xmath204 . next we transform the approximate form of the post - sn pdf from @xmath133 to @xmath205 . a bit of simple algebra shows that such a transformation gives a pdf of the form @xmath206 . integrating this pdf over @xmath53 would yield @xmath207 , if the limits of integration were independent of @xmath107 . figure [ fig : mcarange ] shows that this is indeed the case for @xmath208 , since the @xmath53-limits are clearly seen to be almost independent of @xmath107 when @xmath107 lies in this range . by contrast , as we move to higher @xmath107-values , the contours become shorter by cutting off low-@xmath53 regions . in other words , the lower limit of integration becomes a function of @xmath107 . shortening of the contour length results in a power - law xlf with the slope steeper than -1 . eventually , near the neutron - star eddington luminosity , the contour passes out of the allowed region altogether , and the xlf is cut off . at very low luminosities we expect a similar mechanism to cause a turnover in the xlf ( which would cut it off eventually at extremely low luminosities ) , as is clear from fig.[fig : mcarange ] , since higher-@xmath53 regions are progressively cut off as @xmath107 decreases in this range . however , this effect is not expected to be very important for two reasons . first , cutting of the highest - mass regime in @xmath53 is not a severe problem , since the @xmath53-distribution drops steeply in that regime anyway . therefore , one needs to go to very low luminosities ( @xmath209 erg / s ) to observe this effect . second , with current observational sensitivity , we can track the xlf behavior below @xmath176 only for a few nearby galaxies , and the very low luminosities indicated above are not even approached . it appears , therefore , that this xlf turnover is unlikely to be amenable to observation in the near future , and accordingly we do not consider it any further . the xlf of hmxbs obtained observationally does not reach luminosities below @xmath176 for most of the galaxies ( see sec.[sec : observe ] ) . therefore , only the power - law regime above @xmath176 can be compared with observations in most cases . physical origins of this power - law index are an important aspect of our understanding of the collective properties of hmxbs . note first that our calculation of the xlf is a numerical one , and the power - law result given in sec . [ sec : hxlf ] is only an analytic approximation to it . however , simple qualitative arguments may serve to illustrate the basic physics underlying such calculations , and we consider such arguments below . in a pioneering argument of this type , postnov ( 2003 ; also see postnov & kuranov 2005 ) proceeded as follows . expressing the xlf as : @xmath210 one can use a suitable estimate of the mass - function @xmath211 of the companion for these arguments . the estimation of @xmath212 is more involved , and the original postnov method was to estimate @xmath213 as follows , and use its reciprocal . using @xmath214 ( see sec.[sec : psntoxrb ] ) , this author obtained a power - law form @xmath215 , with the aid of a simple stellar - wind model @xmath216 , into which the following scalings for massive stars were inserted : @xmath217 , @xmath218 . this gave @xmath219 . finally , assuming a power - law form for the companion mass distribution , _ i.e. _ , @xmath220 , straightforward algebra with eq.([eqn : postnov ] ) leads to an xlf slope of @xmath221 . postnov ( 2003 ) assumed @xmath179 ( _ i.e. _ , the exponent of the salpeter imf ) for the @xmath222-distribution , which led to an xlf slope of @xmath223 . in revisiting the above argument , we note first that eq.([eqn : postnov ] ) is incomplete , since n is a function of both @xmath53 and @xmath54 , whose form we have calculated explicitly in the earlier sections . thus , the complete equation is @xmath224 next , we note that the scalings of @xmath125 and @xmath124 with @xmath53 used in the postnov ( 2003 ) work apply to massive _ main - sequence _ stars , but not to the evolved massive stars of interest here . also , the wind mass - loss prescription used in that work is similar to that in the cak model of the 1970s , which our calculations have shown to be inadequate . the rough scalings for @xmath125 and @xmath124 for evolved massive companions of interest here can be obtained from the appropriate formulae in sec.5 of hpt , and are @xmath225 , @xmath226 . with these scalings , and our @xmath53-distribution shown in fig.[fig : mcdf ] , which we can approximate with the power - law of exponent @xmath227 that applies to its principal part , we can evaluate the first term on the right - hand side of eq.([eqn : bhadghosh ] ) . for the kudritzki - reimers model , it can be easily shown that @xmath228 , so that this first term gives an xlf slope @xmath229 . the other stellar model gives an essentially identical final result . the rest of the contribution comes from the second term on the right - hand side of eq.([eqn : bhadghosh ] ) , which can be easily calculated and which leads to our overall xlf slope @xmath230 . we have gone through this argument in detail because it addresses an interesting observation made by gilfanov ( 2004 ) that , in hmxb systems powered by stellar - wind accretion , the distribution of @xmath116 and therefore @xmath107 should be governed by the properties of the massive companion , in particular the distribution of @xmath53 and @xmath125 . our work here shows that this is largely , but not completely , true . the properies of the massive companion are contained in the first term on the right - hand side of eq.([eqn : bhadghosh ] ) , and keeping only this term , as postnov ( 2003 ) did , amounts to neglecting altogether the binary orbital properties which are contained in the second term on the right - hand side of eq.([eqn : bhadghosh ] ) , and which also influence the xlf . the estimates summarized above give a measure of the relative sizes of the effects of the companion and the orbit , and demonstrate that , while the latter are certainly smaller , they are by no means negligible . in this work , we have described a method for obtaining the distributions of some of the essential collective properties of hmxb populations in the stellar fields of normal / starburst galaxies , wherein we start from accepted distributions of primordial binaries which are progenitors of such hmxbs , and follow the transformation of these distributions with the aid of a jacobian formalism as the primordial binary population evolves into the hmxb population . our method , which is semi - analytic , traces in a transparent way the effects of various processes in the course of this evolution , and so assesses with ease which physical processes dominate in determining which distribution . for example , the distribution of the properties of the massive companions seems to have the dominant effect on the xlf , although the distribution of the orbital parameters does have a significant effect , as we demonstrated in sec.[sec : xlfslope ] . but the distribution of hmxb orbital periods appears to be strongly influenced by both the primordial orbital distribution and the sn - kick properties . the agreement between our calculated xlf and binary - period distribution and the observed hmxb distributions is most encouraging , and it justifies a future monte carlo population synthesis scheme for a more detailed undrstanding of how hmxb populations are built in the stellar fields of normal galaxies . however , we must first extend our present method to the more complex problem of following the formation and evolution of lmxb populations from their corresponding primordial binaries , as mentioned in sec.[sec : intro ] . this will occupy us in the next papers in this series . it is a pleasure to thank m. gilfanov , e. p. j.van den heuvel , v. kalogera , l. stella , and r. a. sunyaev for stimulating discussions , and an anonymous referee for valuable comments which greatly improved the paper . in this appendix we describe the method of averaging the effects of sn - kicks for a maxwellian kick - distribution . it has been widely assumed that the distribution of kicks will be isotropic and recent observations by have supported it @xcite . therefore the angle dependence in eq.[eqn : postsn ] is averaged in a straightforward way , which leads to eq.[eqn : postsnav ] . two points are to be noted here . first , the only surviving kick - term in the averaged transformation equations is @xmath71 which is to be obtained by averaging @xmath72 over the maxwellian distribution . second , it must be remembered that the process of averaging gives the average values of post - sn quantities . these therefore should be compared with the average behaviour of a number of sample observational sets . a further , most crucial , point to be noted at this stage is that in our study here ( and in all similar studies ) , we are interested in _ only _ those post - sn systems which remain bound as xrbs , so that they can eventually produce hmxbs whose distribution we are interested in . accordingly , we must exclude at this point all systems which become unbound in the sn . the way we do so is as follows . we note that , in the @xmath72-averaging process , we would make an error if we carried out the integrations over the maxwellians in eq.[eqn : vsqav ] upto infinitely large values of @xmath55 , since this would include all the unbound systems that are to be excluded . rather , we must _ truncate _ the integration at a suitable upper limit @xmath231 which corresponds to the point at which the post - sn system becomes just unbound . this point is readily obtained from the first of eqns.[eqn : postsnav ] by setting @xmath232 , and @xmath231 given by : @xmath231 is well - defined for each set of values of pre - sn parameters . for algebraic convenience , we can express it as @xmath235 for a given maxwellian distribution with dispersion @xmath74 . this simplifies the expression for @xmath71 , which can be written as @xmath236 , where @xmath237 is given by : fig.[fig : fhf ] shows @xmath237 as a function of @xmath67 . two regimes are clearly demarcated , with a small transition region between the two . the low-@xmath67 region can be described by a power - law given by , @xmath239 , whereas in the high-@xmath67 region , @xmath240 , which is its asymptotic value . the crossover point of the two regimes is at @xmath241 . this clear division in two regions give us a straightforward relation for @xmath71 . for @xmath242 , @xmath71 is independent of @xmath74 , and is given by @xmath243 . on the other hand , for @xmath244 @xmath71 is independent of @xmath231 , and is given by @xmath245 . this behavior of @xmath237 can be understood as follows . for @xmath242 , @xmath231 is small and hence only the initial rising part of the maxwellian is relevant . thus the distribution is essentially given by @xmath246 , which is independent of @xmath74 . on the contrary , for @xmath244 , @xmath231 is large and almost the entire maxwellian profile is included , excluding only a small tail . detailed calculations given above show that the transition region between these two regimes is small , so that , as a first approximation , we can use the following simple prescription : the above prescription can be substituted appropriately in eqn . [ eqn : postsnav ] to obtain the post - sn parameter transformation or the inverse transformation , which is required for the jacobian formalism . we describe in the next section the relevant transformation equations and the jacobian for the two cases . in this region , @xmath71 is independent of the parameters of the binary system . we can simplify eqn.[eqn : postsnav ] by using following ratios . let @xmath248 , @xmath249 , and @xmath250 . also , we define orbit size change factor as @xmath251 . equation [ eqn : postsnav ] can then be used to write the inverse transformations : these equations are indirect transformations which give @xmath253 and @xmath254 on substituting the definitions of @xmath255 and @xmath256 in above equations . since the companion mass is unchanged in the sn explosion , one can readily calculate @xmath42 from @xmath254 for known @xmath53 . the jacobian of this transformation can be calculated as follows . note first that the transformation is only in two variables _ i.e. _ @xmath257 , since the third parameter _ @xmath53 is unchanged . for the same reason , one can also write @xmath258 . all the necessary partial derivatives can be written in terms of the derivatives of @xmath255 and @xmath256 as follows : where @xmath255 is the orbital - size factor defined similarly as @xmath261 . the jacobian in this case is given by an even simpler relation . one can easily see that @xmath254 is independent of @xmath262 and depends only on e. therefore @xmath263 . the jacobian in that case is given by @xmath264 . straightforward algebra shows that	we introduce a method for obtaining the x - ray luminosity function ( xlf ) and the binary - period distribution of populations of high - mass x - ray binaries ( hmxbs ) in the stellar fields ( _ i.e. _ , outside globular clusters ) of normal galaxies . we start from standard distributions of the parameters of those primordial binaries which are the progenitors of hmxbs , and follow the transformation of these ditributions with the aid of a jacobian formalism as the former evolve into the latter through the processes of the first mass transfer and the supernova ( sn ) that follows . we discuss the distributions of the post - sn binaries and the hmxbs . we show that our calculated model xlf has a differential slope @xmath0 with a flattening at low luminosities , in excellent agreement with observations . the calculated binary - period distribution , which basically has a slightly sloping plateau - like character at intermediate periods , with a rise to this plateau at shorter periods and fall - off from it at longer periods , is in agreement with the observed distribution within observational uncertainties . we discuss the physical origin of these distributions . we demonstrate that , while the effects of both ( a ) the distribution of the properties of the massive companion in the hmxbs , and ( b ) the primordial orbital distribution and the sn dynamics are important , the former appear to be dominant in determining the xlf , and the latter in determining the hmxb binary - period distribution . we discuss the possible roles of stellar - mass black holes and ultra - luminous x - ray sources ( ulx ) in the observed `` universal '' xlf of hmxbs .
double - diffusive instabilities commonly occur in any astrophysical fluid that is stable according to the ledoux criterion , as long as the entropy and chemical stratifications have opposing contributions to the dynamical stability of the system . they drive weak forms of convection described below , and can cause substantial heat and compositional mixing in circumstances reviewed in this paper . two cases can be distinguished . in _ fingering convection _ , entropy is stably stratified ( @xmath0 , but chemical composition is unstably stratified @xmath1 ; it is often referred to as _ thermohaline _ convection by analogy with the oceanographic context in which the instability was first discovered . in _ oscillatory double - diffusive convection _ ( oddc ) , entropy is unstably stratified ( @xmath2 , but chemical composition is stably stratified @xmath3 ; it is related to semiconvection , but can occur even when the opacity is independent of composition . fingering convection can naturally occur at late stages of stellar evolution , notably in giants , but also in main sequence stars that have been polluted by planetary infall ( as first proposed by vauclair , @xcite ) , or by material transferred from a more evolved companion star . oddc on the other hand is naturally found in stars in the vicinity of convective nuclear - burning regions , including high - mass core - burning main sequence or red clump stars , and shell - burning rgb and agb stars . it is also thought to be common in the interior of giant planets that have been formed through the core - accretion scenario . beyond competing entropy and compositional gradients , a necessary condition for double - diffusive instabilities to occur is @xmath4 ( @xmath5 and @xmath6 are the microscopic compositional and thermal diffusivities ) . this is usually the case in astrophysical fluids , where @xmath7 is typically _ much _ smaller than one owing to the added contribution of photon and electron transport to the thermal diffusivity . the somewhat counter - intuitive manner in which a high thermal diffusivity can be destabilizing is illustrated in figure [ fig : instab ] . in the fingering case , a small @xmath7 ensures that any small displaced fluid element rapidly adjusts to the ambient temperature of its new position , while retaining its original composition . an element displaced downward thus finds itself denser than the surrounding fluid and continues to sink ; the opposite occurs for an element displaced upward . in the case of oddc , thermal diffusion can progressively amplify any internal gravity wave passing through , by heating a fluid element at the lowest position of its displacement and cooling it near the highest . in both cases , the efficient development of the instability is conditional on the fluid element being small enough for thermal diffusion to take place . double - diffusive convection is therefore a process driven on very small scales , usually orders of magnitude smaller than a pressure scaleheight . consequently , a common way of studying fingering and odd convection is by a _ local _ linear stability analysis , in which the background gradients of entropy ( related to @xmath8 ) and composition @xmath9 are approximated as being constant ( baines & gill , @xcite ) . the governing equations in the boussinesq approximation are then : @xmath10 where @xmath11 , @xmath12 , @xmath13 and @xmath14 are the non - dimensional velocity field , pressure , temperature and mean molecular weight perturbations of the fluid around the background state , pr@xmath15 is the prandtl number ( and @xmath16 is the viscosity ) , and @xmath17 is called the _ density ratio_. here the unit lengthscale used is the typical horizontal scale @xmath18 of the basic instability , the unit time is @xmath19 , the unit temperature is @xmath20 and the unit compositional perturbation is @xmath21 . the @xmath22 sign in the temperature and composition equations should be used to model fingering convection , while the @xmath23 sign should be used to model oddc . mathematically speaking , this sign change is the only difference between the two processes . assuming perturbations have a spatio - temporal structure of the form @xmath24 where @xmath25 is either one of the dependent variables , @xmath26 is the wavenumber of the perturbation and @xmath27 its growth rate ( which could be complex ) , @xmath27 satisfies a cubic equation : @xmath28 \lambda + \left [ k^6 { \rm pr } \tau \pm l^2 { \rm pr } ( \tau - r_0^{-1 } ) \right ] = 0 \mbox { , } \nonumber \label{eq : cubic}\end{aligned}\ ] ] where @xmath29 again refers to @xmath22 for fingering convection and @xmath23 for oddc , @xmath30 and @xmath31 is the norm of the horizontal component of @xmath26 . @xmath27 is real in the case of fingering convection but complex in the case of oddc , as expected from the physical description of the mechanism driving the instability . the fastest growing mode in both cases is vertically invariant . its growth rate @xmath32 and horizontal wavenumber @xmath33 can be obtained by maximizing @xmath34 over all possible @xmath26 . finally , setting @xmath35 identifies marginal stability , and reveals the parameter range for double - diffusive instabilities to be:@xmath36 note that @xmath37 in both cases corresponds to the ledoux criterion . while linear theory is useful to identify _ when _ double - diffusive convection occurs , nonlinear calculations are needed to determine how the latter saturates , and how much mixing it causes . vertical mixing is often measured via non - dimensional vertical fluxes , called nusselt numbers . the temperature and compositional nusselt numbers are defined here as @xmath38 where @xmath39 and @xmath40 are the dimensional temperature and compositional turbulent fluxes . to reconstruct the dimensional _ total _ fluxes of heat and composition @xmath41 and @xmath42 , we have ( wood , @xcite ) @xmath43 where @xmath44 is the thermal conductivity , and @xmath45 is the specific heat at constant pressure . it is worth noting that @xmath46 can also be interpreted as the ratio of the effective to microscopic compositional diffusivities . direct numerical simulations , which solve the fully nonlinear set of equations ( [ eq : goveqs ] ) for given parameter values pr , @xmath7 and @xmath47 from the onset of instability onward , can in principle be run to estimate the functions @xmath48 and nu@xmath49 . however , the actual nonlinear behavior of double - diffusive systems reveals a number of surprises , that must be adequately studied before a complete theory for mixing can be put forward . it has long been known in oceanography that double - diffusive convection has a tendency to drive the growth of structures on scales much larger than that of the basic instability ( cf . stern , @xcite ) . this tendency was recently confirmed in the astrophysical context as well ( rosenblum , @xcite ; brown , @xcite ) . these structures either take the form of large - scale internal gravity waves or thermo - compositional staircases , as shown in figure [ fig : large - scale ] . , @xmath50 . the basic instability first saturates into a near - homogeneous state of turbulence , but later develops large - scale gravity waves . bottom : oddc simulation for pr = @xmath51 , @xmath52 . the basic instability first saturates into a near - homogeneous state , but later develops into a thermo - compositional staircase , whose steps gradually merge until only one is left . the mean nusselt numbers increase somewhat in the presence of waves in the fingering case , and quite significantly when the staircase forms , and at each merger , in the oddc case . , scaledwidth=90.0% ] for density ratios close to one , fingering convection tends to excite large - scale gravity waves , through a process called the _ collective instability _ first discovered by stern ( @xcite ) . these waves grow to significant amplitudes , and enhance mixing by fingering convection when they break . the same is true for oddc , but the latter can sometimes also form thermo - compositional staircases excited by a process called the @xmath53_instability _ ( radko , @xcite ) . the staircases spontaneously emerge from the homogeneously turbulent state , and appear as a stack of fully convective , well - mixed regions ( the layers ) separated by thin strongly stratified interfaces . the layers have a tendency to merge rather rapidly after they form . vertical mixing increases significantly when layers form , and with each merger . for these reasons , quantifying transport by double - diffusive convection requires understanding not only how and at what amplitude the basic small - scale instabilities saturates , but also under which circumstances large - scale structures may emerge and how the latter affect mixing . given their ubiquity in fingering and odd convection , it is natural to seek a unified explanation for the emergence of large - scale structures that is applicable to both regimes . mean - field hydrodynamics is a natural way to proceed , as it can capitalize on the separation of scales between the primary instability and the gravity waves or staircases . to understand how mean - field instabilities can be triggered , first note that the intensity of vertical mixing in double - diffusive convection is naturally smaller if the system is closer to being stable , and vice - versa . if a homogeneously turbulent state is spatially modulated by large - scale ( but small amplitude ) perturbations in temperature or chemical composition , then vertical mixing will be more efficient in regions where the _ local _ density ratio is closer to one , and smaller in regions where it is further from one . the spatial convergence or divergence of these turbulent fluxes can , under the right conditions , enhance the initial perturbations in a positive feedback loop , in which case a mean - field instability occurs . first discussed separately in oceanography , the collective and @xmath54-instabilities were later discovered to be different unstable modes of the same mean - field equations by traxler ( @xcite ) , in the context of fingering convection . their work has successfully been extended to explain the emergence of thermo - compositional layers in oddc in astrophysical systems by mirouh ( @xcite ) . a formal stability analysis of the mean - field equations shows that they are unstable to the @xmath53instability ( the layering instability ) whenever the _ flux ratio _ @xmath55 is a _ decreasing _ function of @xmath47 ( radko , @xcite ) . similarly , a necessary condition for the collective instability was given by stern ( @xcite ) , who argued that large - scale gravity waves can develop whenever @xmath56 where @xmath57 . this criterion is often much less restrictive than the one for the development of the @xmath53instability . note that @xmath58 and @xmath46 in ( [ eq : gamma ] ) and ( [ eq : a ] ) are the nusselt numbers associated with the small - scale turbulence present _ before _ any large - scale structure has emerged . traxler ( @xcite ) were the first to run a systematic sweep of parameter space to study fingering convection in astrophysics , and to measure @xmath48 and @xmath59 in 3d numerical experiments . however , they were not able to achieve very low values of pr and @xmath7 . brown ( @xcite ) later presented new simulations with pr and @xmath7 as low as @xmath60 , but this is still orders of magnitude larger than in stellar interiors , where pr and @xmath7 typically range from @xmath61 to @xmath62 . to bridge the gap between numerical experiments and stellar conditions , brown ( @xcite ) derived a compelling semi - analytical prescription for transport by small - scale fingering convection , that reproduces their numerical results and can be extrapolated to much lower pr and @xmath7 . their model attributes the saturation of the fingering instability to the development of shearing instabilities between adjacent fingers ( see also denissenkov , @xcite and radko & smith , @xcite ) . for a given set of governing parameters pr , @xmath7 and @xmath47 , the growth rate and horizontal wavenumber of the fastest - growing fingers can be calculated from linear theory ( see section [ sec : linear ] ) . meanwhile , the growth rate @xmath63 of shearing instabilities developing between neighboring fingers is proportional to the velocity of the fluid within the finger times its wavenumber ( a result that naturally emerges from dimensional analysis , but can also be shown formally using floquet theory ) . stating that shearing instabilities can disrupt the continued growth of fingers requires @xmath63 and @xmath32 to be of the same order . this sets the velocity within the finger to be @xmath64 where @xmath65 is a universal constant of order one . meanwhile , linear stability theory also relates the temperature and compositional fluctuations @xmath13 and @xmath14 within a finger to @xmath66 . the turbulent fluxes can thus be estimated _ only using linear theory _ : @xmath67 comparison of these formula with the data helps calibrate @xmath65 . brown ( @xcite ) found that using @xmath68 can very satisfactorily reproduce most of their data within a factor of order one or better , except when pr @xmath69 ( which is rarely the case in stellar interiors anyway ) . equation ( [ eq : nusfinger ] ) implies that for low pr and @xmath7 , turbulent heat transport is negligible , while turbulent compositional transport is significant only when @xmath47 is close to one . however , the values of @xmath46 obtained by brown ( @xcite ) are still not large enough to account for the mixing rates required by charbonnel & zahn ( @xcite ) to explain surface abundances in giants . such large values of @xmath70 might on the other hand be achieved if mean - field instabilities take place . as discussed in section [ sec : large ] , one simply needs to estimate @xmath54 and @xmath71 in order to determine in which parameter regime mean - field instabilities can occur . using ( [ eq : gamma ] ) and ( [ eq : nusfinger ] ) it can be shown that @xmath54 is always an _ increasing _ function of @xmath47 at low pr and @xmath7 . this implies that fingering convection is stable to the @xmath53instability , and therefore not likely to transition _ spontaneously _ to a state of layered convection in astrophysics . the simulations of brown ( @xcite ) generally confirm this statement , except in a few exceptional cases discussed below . by contrast , fingering convection does appear to be prone to the collective instability ( as shown in figure [ fig : large - scale ] ) for sufficiently low @xmath47 . by calculating @xmath71 for a typical stellar fluid with pr @xmath72 and @xmath73 , we find for instance that gravity waves should emerge when @xmath74 or so . in this regime , we expect transport to be somewhat larger than for small - scale fingering convection alone , although probably not by more than a factor of 10 ( see figure [ fig : large - scale ] ) . nevertheless , a first - principles theory for the vertical mixing rate in fingering convection , in the presence of internal gravity waves , remains to be derived . finally , as first hypothesized by stern ( @xcite ) and found in preliminary work by brown ( @xcite ) , it is possible that these large - scale gravity waves could break on a global scale and mechanically drive the formation of layers . if this is indeed confirmed , transport could be much larger than estimated in ( [ eq : nusfinger ] ) in the region of parameter space for which @xmath75 . this , however , remains to be confirmed . until we gain a better understanding of the various effects of gravity waves described above , ( [ eq : nusfinger ] ) is our best current estimate for transport by fingering convection in astrophysical objects . an example of the numerical implementation of the model by brown ( @xcite ) is now available in mesa , and consists of the following steps . ( 1 ) to estimate the local properties of the star , and calculate all governing parameters / diffusivities . ( 2 ) to estimate the properties of the fastest - growing fingering modes using linear theory ( see section [ sec : linear ] ) and ( 3 ) to apply ( [ eq : nusfinger ] ) to calculate @xmath46 , and then ( [ eq : totalfluxes ] ) to calculate @xmath42 . turbulent heat transport is negligible , so @xmath76 . 3d numerical simulations of oddc were first presented by rosenblum ( @xcite ) and mirouh ( @xcite ) . both explored parameter space to measure , as in the case of fingering convection , the functions @xmath48 and @xmath59 after saturation of the basic odd instability . the values of pr and @xmath7 achieved , however , were not very low , and models are needed once again to extrapolate these results to parameters relevant for stellar interiors . mirouh ( @xcite ) proposed an empirical formula for nu@xmath77 ( and nu@xmath49 , via @xmath54 ) , whose parameters were fitted to the experimental data . however , a theory based on first principles is more desirable . we have recently succeeded in applying a very similar method to the one used by brown ( @xcite ) to model transport by small - scale oddc . as described in moll ( @xcite ) , a simple approximate estimate for temperature and compositional nusselt numbers can also be derived from the linear theory for the fastest - growing mode ( see section [ sec : linear ] ) , this time in the form of @xmath78 where @xmath79 , @xmath80 and @xmath81 . the constants @xmath82 and @xmath83 must again be fitted to the existing data ; preliminary results suggest that @xmath84 and @xmath85 . by contrast with fingering convection , oddc is subject to both @xmath53 and collective instabilities . both modify the vertical heat and compositional fluxes quite significantly so ( [ eq : nusoddc ] ) should _ not _ be used _ as is _ to model mixing by oddc . it is used , on the other hand , to determine when mean - field instabilities occur . the region of parameter space unstable to layering can again be determined by calculating @xmath54 ( using [ eq : nusoddc ] this time ) , and checking when @xmath86 . moll ( @xcite ) ( see also mirouh , @xcite ) showed that layering is always possible for pr and @xmath7 below one , provided @xmath87 $ ] . the critical value @xmath88 is fairly close to zero in the parameter regime appropriate for stellar interiors , as @xmath89 in the limit @xmath90 . this implies that the region of parameter space unstable to layer formation spans nearly the entire oddc range . as discussed in section [ sec : large ] , the vertical heat and compositional fluxes increase dramatically when a staircase first emerges , and then again at each layer merger . this was studied by wood ( @xcite ) , who ran and analyzed simulations for @xmath91 $ ] , and argued that their results are consistent with the following empirical transport laws for layered convection : @xmath92 where @xmath93 and @xmath94 are slowly varying functions of @xmath47 and @xmath7 , and where the rayleigh number is @xmath95 , where @xmath96 is the mean step height in the staircase . for numerically achievable pr and @xmath7 , wood ( @xcite ) estimated that @xmath97 and @xmath98 . equation ( [ eq : nuslayers ] ) has two important consequences . the first is that turbulent heat transport can be significant in layered convection , provided @xmath96 is large enough . secondly , both nusselt numbers are ( roughly ) proportional to @xmath99 , but nothing so far has enabled us to determine what @xmath96 may be in stellar interiors . indeed , in _ all _ existing simulations of layered oddc to date , layers were seen to merge fairly rapidly until a single one was left . whether staircases in stellar interiors evolve in the same way , or eventually reach a stationary state with a layer height smaller than the thickness of the unstable region itself , is difficult to determine without further modeling , and remains the subject of current investigations . odd systems which do not transition into staircases ( @xmath100 ) also usually evolve further with time after saturation of the basic instability , with the small - scale wave - turbulence gradually giving way to larger - scale gravity waves . whether the latter are always excited by the collective instability , or could be promoted by other types of nonlinear interactions between modes that transfer energy to larger scales , remains to be determined . in all cases , these large - scale waves have significant amplitudes and regularly break . this enhances transport , as it did in the case of fingering convection . moll ( @xcite ) found that the resulting nusselt numbers are between 1.2 and 2 across most of the unstable range , regardless of pr or @xmath7 . these results are still quite preliminary , however , and their dependence on the domain size ( which sets the scale of the longest waves ) remains to be determined . based on the results obtained so far , and summarized above , a plausible mixing prescription for oddc can be obtained by applying the following steps . ( 1 ) to estimate the local properties of the star , and calculate all governing parameters / diffusivities . ( 2 ) to estimate the properties of the fastest - growing modes using linear theory ( see section [ sec : linear ] ) ( 3 ) to determine whether layers are expected to form or not by calculating @xmath54 ( using [ eq : nusoddc ] ) for neighboring values of @xmath47 , and evaluating @xmath101 . ( 4 ) if the system is layered , then _ assume a layer height _ ( for instance , some small fraction of a pressure scaleheight ) , and calculate the heat and compositional fluxes using ( [ eq : totalfluxes ] ) with ( [ eq : nuslayers ] ) . if the system is not expected to form layers , then calculate these fluxes using ( [ eq : totalfluxes ] ) and @xmath102 instead . the unknown layer height is the only remaining free parameter of this model , and will hopefully be constrained in the future by comparison of the model predictions with asteroseismic results .	much progress has recently been made in understanding and quantifying vertical mixing induced by double - diffusive instabilities such as fingering convection ( usually called thermohaline convection ) and oscillatory double - diffusive convection ( a process closely related to semiconvection ) . this was prompted in parts by advances in supercomputing , which allow us to run direct numerical simulations of these processes at parameter values approaching those relevant in stellar interiors , and in parts by recent theoretical developments in oceanography where such instabilities also occur . in this paper i summarize these recent findings , and propose new mixing parametrizations for both processes that can easily be implemented in stellar evolution codes .
most quantum error - correction codes ( qecc ) have been constructed for quantum systems that are composed of two - dimensional subsystems quantum bits or short qubits . at first glance , the size of the alphabet used to encode the information to be processed seems to be irrelevant . but already in the context of classical information processing and communication this is not true . for example , the entropy of a message depends on the alphabet used for the encoding , and increasing the size of the alphabet allows the construction of better error - correcting codes @xcite . in the context of quantum information , a quantum system consisting of two three - dimensional subsystems qutrits shows new features when compared to a two - qubit system . for example , bound entanglement exists only for the former @xcite , and for two qutrits there is even `` nonlocality without entanglement '' @xcite . in this paper , we consider quantum systems which have subsystems of dimension @xmath4 , where @xmath5 is prime . as a shorthand , we will use the term `` qudit '' . quantum codes for qudit systems have been studied , e.g. , in @xcite . the question of encoding and decoding these codes , however , was not explicitly addressed . here we present efficient ( classical ) algorithms to compute efficient quantum networks for the encoding process . the method applies to qubit codes as well , for which encoding algorithms have been discussed , e.g. , in @xcite . in the following section , we present the mathematical framework of quantum information processing using composed quantum systems . in section [ sec : qecc ] we recall the basic concepts and constructions of quantum error - correcting codes . a first encoding algorithm for the class of css codes is presented in section [ sec : css ] . the main result , an encoding algorithm for stabilizer codes over higher dimensional quantum - systems , is derived in section [ sec : jacobi ] and illustrated in section [ sec : example ] . in the context of classical information processing , information is encoded using words over some finite alphabet @xmath6 , most often @xmath7 . the elementary `` symbols '' of quantum information processing are states of a finite dimensional quantum system . those can be modeled by normalized vectors in a complex hilbert space @xmath8 . again , the simplest case of @xmath9 is considered most often . then the orthonormal basis states of such a quantum bit , or short qubit , are written as @xmath10 and @xmath11 . ( the notation of quantum states as `` ket vectors '' @xmath12 is attributed to dirac @xcite ) . a general state of a qubit is given by @xmath13 if both @xmath14 and @xmath15 are non - zero , the state @xmath16 is a so - called superposition of @xmath10 and @xmath11 with amplitudes @xmath14 and @xmath15 . for a @xmath17-dimensional system , we label the orthonormal basis states by the elements of some alphabet of size @xmath17 , e.g. , the numbers @xmath18 or the elements of a finite field , if @xmath17 is a prime power . the general state of a qudit is given by @xmath19 combining several qudits , we obtain a quantum register . the canonical basis states of a quantum register of length @xmath1 are tensor products of the basis states of the single qudits . hence we can label them by words @xmath20 of length @xmath1 . for the basis states of a quantum register we use the following notations : @xmath21 a general state of a quantum register of length @xmath1 is a normalized vector in the exponentially large hilbert space @xmath22 , given by @xmath23 when writing states of quantum registers , normalization factors may be omitted . the elements of the dual space @xmath24 will be denoted by `` bra vectors '' @xmath25 . then the inner product of two vectors @xmath26 and @xmath27 reads @xmath28 , and linear operators @xmath29 can be written in the form @xmath30 as the operations have to preserve the normalization of the vectors , the admissible operations are unitary operators , which we will discuss next . in the following we will consider qudit systems where each qudit corresponds to a @xmath31-dimensional hilbert space where @xmath32 is a prime power . [ def : elemgates ] let @xmath31 be a prime power , i.e. , @xmath32 where @xmath5 is prime . by @xmath33 we denote a primitive complex @xmath5-th root of unity , i.e. , @xmath34 . furthermore , let @xmath35 denote the trace of an element @xmath36 which is defined as @xmath37 . then we define the following operations : @xmath38 for @xmath39[def : x ] @xmath40 for @xmath41 @xmath42 for @xmath43[def : mult ] @xmath44 [ def : dft ] @xmath45[def : add ] @xmath46 here , when writing @xmath47 , we identify @xmath48 and @xmath49 , the integers modulo @xmath5 . then the function @xmath50 is an additive character of @xmath51 . different values of @xmath15 yield the @xmath31 different additive characters of @xmath51 ( see , e.g. , @xcite ) . 1.2 ( 1,3)@xmath52 ( 1,3)@xmath53 ( 1,3)@xmath54 ( 1,3)@xmath55 ( 1,2,3 ) ( 3,3 ) a graphical representation of these elementary operations so - called quantum gates is given in figure [ fig : elem_gates ] . each horizontal line corresponds to one qudit . the first three operations operate on single qudits . a superscript in brackets indicates on which subsystem the transformation acts , e.g. , @xmath56 . the operations @xmath52 and @xmath54 correspond to the addition of a fixed element @xmath39 and the multiplication with a fixed element @xmath57 , respectively . the operation @xmath53 has no direct classical analogue , it changes the phases of the basis states . the fourier transformation @xmath55 can be used to transform the state @xmath10 to a superposition of all basis states with equal amplitudes , i.e. , @xmath58 starting from this superposition , a quantum computation can , e.g. , evaluate a function in parallel for all possible inputs . arbitrary classical functions over @xmath51 can be implemented using the gates @xmath59 and @xmath60 . the former corresponds to the reversible implementation of the addition of two elements . the first qudit is called `` control '' , the second `` target '' . the latter is a universal reversible gate over @xmath51 , as any function over @xmath51 corresponds to a polynomial which can be evaluated using the horner scheme . it is the generalization of the so - called toffoli gate @xcite for qubits , which is universal gate for reversible boolean functions , as well as the fredkin gate @xcite . in the following , we will recall the basic properties and constructions of quantum error - correcting codes . in order to construct an error - correcting code , one has to specify an error model . the error model can be specified by a ( finite ) set @xmath61 of error operators . in @xcite , the following characterization of error - correcting codes is given . let @xmath62 be a subspace of the hilbert space @xmath63 with orthonormal basis @xmath64 . then @xmath62 is a quantum error - correcting code for the error - operators @xmath65 if and only if there exists @xmath66 such that for all @xmath67 , @xmath68 and @xmath69 @xmath70 most quantum error - correcting codes are designed to correct local errors , i.e. , errors that effect only some subsystems . the error acting on a single subsystem can be any linear transformation . it is sufficient that condition ( [ eq : knilllaflamme ] ) holds for a basis of the linear space of error operators . for qudit systems of prime power dimension @xmath31 , we consider the following set of unitary operators : @xmath71 it is not hard to show that those @xmath72 operators are an orthogonal basis with respect to the inner product @xmath73 . furthermore , they generate an error group @xmath74 of size @xmath75 with center @xmath76 ( see @xcite ) . any element of @xmath74 can uniquely be written as @xmath77 where @xmath78 and @xmath79 . the commutation relations of two elements are derived from @xmath80 hence commuting two elements results in a phase factor , i.e. , @xmath81 for an @xmath1-qudit system , the error basis and the error group are the @xmath1-fold tensor products @xmath82 and @xmath83 , respectively . the weight of an error - operator @xmath84 is the number of tensor factors that are different from identity . if a @xmath85-dimensional subspace @xmath62 of @xmath86 can correct all errors of weight no greater than @xmath87 , @xmath62 is a @xmath87-error - correcting code with minimum distance @xmath88 , denoted by @xmath89 . a particular class of quantum error - correcting codes are so - called stabilizer codes ( see @xcite ) . the basic idea is to consider an abelian subgroup @xmath90 of the error group @xmath91 such that its intersection with the center of @xmath92 is trivial . the stabilizer code @xmath62 is defined as the common eigenspace of the operators in @xmath90 . stabilizer codes can be described in terms of certain classical codes over finite fields . any element @xmath93 of the error group @xmath91 can uniquely be written as @xmath94 where @xmath78 and @xmath95 . the weight of an element @xmath96 is the number of indices @xmath97 for which not both @xmath98 and @xmath99 are zero . from the commutation relation ( [ eq : commute ] ) , it follows that for @xmath100 @xmath101 where the inner product @xmath102 is defined by @xmath103 this shows that the group @xmath104 is isomorphic to @xmath105 . furthermore , two elements @xmath96 and @xmath106 commute if and only if @xmath107 . hence an abelian subgroup @xmath108 of @xmath91 corresponds to a subspace @xmath109 of @xmath105 that is contained in its dual @xmath110 with respect to the inner product ( [ eq : traceip ] ) . [ def : stab_mat ] let @xmath90 be an abelian subgroup of @xmath91 which has trivial intersection with the center of @xmath91 . furthermore , let @xmath111 where @xmath112 with @xmath113 and @xmath114 be a minimal set of generators for @xmath90 . then a stabilizer matrix of the corresponding stabilizer code @xmath62 is a generator matrix of the ( classical ) linear code @xmath115 . we will write this matrix in the form @xmath116 any error operator @xmath93 that does not commute with all elements @xmath117 will change the eigenvalue of an eigenstate @xmath118 of @xmath119 which can be detected by a measurement . recall that @xmath120 does not commute with all @xmath117 if and only if @xmath121 . finally , the minimum distance of a stabilizer code is the minimum weight of the vectors @xmath122 , since the errors corresponding to @xmath110 are those that can not be detected , and the operators corresponding to @xmath109 have no effect on the code . a special class of stabilizer codes are so - called css codes named after calderbank , shor @xcite and steane @xcite . originally , they have been designed for qubit systems , but they can easily generalized to any dimension ( see , e.g. , @xcite ) . given two linear codes @xmath123_q$ ] and @xmath124_q$ ] over @xmath51 with @xmath125 , the basis states of the corresponding css code are @xmath126 two states @xmath127 and @xmath128 are identical if and only if @xmath129 , otherwise they are orthogonal . as the dimension of @xmath130 is @xmath131 , there are @xmath132 different cosets where @xmath133 . hence the dimension of the css code is @xmath132 . its minimum distance is @xmath134 . we illustrate the encoding of a css code for the code @xmath135\!]_8 $ ] over seven @xmath136-dimensional quantum systems . let @xmath137 be a primitive element of @xmath138 with minimal polynomial @xmath139 . the code @xmath140_8 $ ] with generator matrix @xmath141 is contained in its dual @xmath142 $ ] with generator matrix @xmath143 the first two rows of @xmath144 are equal to @xmath145 . the cosets of @xmath146 are given by linear combinations of the last three rows . denoting the rows of @xmath144 by @xmath147 , we can rewrite equation ( [ eq : cssstate ] ) as @xmath148 where @xmath149 . applying a fourier transformation to the first and second qudit of the initial state @xmath150 , we obtain @xmath151 now we sequentially add the corresponding multiple of the rows of @xmath144 in reverse order , i.e. , starting with @xmath152 . as @xmath144 is in row echelon form , this corresponds to a sequence of modified @xmath153 gates instead of adding the first qudit to the second , we have to add a multiple of it . the resulting encoding circuit is shown in figure [ fig : cssencoding ] . 0.7 ( # 1,#2,#3)-5(#1,#2,#3)-5 ( 7 ) ( 1,1)@xmath154 ( 5,7,7)(1,1)@xmath155 ( 1,1)@xmath156 ( 4,7,7)(1,1)@xmath157 ( 1,1)@xmath156 ( 3,7,7)(1,1)@xmath157 ( 1,1)@xmath154 ( 2,7,7)(1,1)@xmath158 ( 1,1)@xmath159 ( 1,7,7)(1,1)@xmath14 ( 7 ) for each row @xmath160 of @xmath144 , we have a sequence of gates @xmath161 corresponding to the non - zero entries @xmath162 for @xmath163 . the gate @xmath161 is conjugated by a multiplication gate @xmath54 where @xmath164 . note that we have not combined adjacent multiplication gates . in general , @xmath145 is a generator matrix of @xmath125 , and @xmath144 is its completion to a generator matrix of @xmath165 . using the construction illustrated above , we get let @xmath123_q$ ] and @xmath124_q$ ] be linear codes over @xmath51 with @xmath125 . then there exists a quantum circuit to encode the resulting css code @xmath166\!]_q$ ] over qudits of dimension @xmath31 , where @xmath133 , using @xmath131 @xmath55 gates , at most @xmath167 @xmath153 gates , and at most @xmath168 multiplication gates . as the rle of @xmath165 and @xmath169 in the construction of css codes is completely symmetric , one has the freedom to choose the matrices @xmath145 and @xmath144 to be generator matrices of either @xmath130 and @xmath165 , respectively @xmath170 and @xmath169 . this may reduce the number of gates by a constant factor , as well as optimizing the particular generator matrices . in this section we derive an encoding algorithm for general stabilizer codes over qudit systems of prime power dimension @xmath32 . the main idea is to transform the abelian stabilizer group @xmath171 of the stabilizer code @xmath172\!]_q$ ] into a stabilizer group @xmath173 for which encoding is particularly easy . up to a permutation @xmath174 of the qudits which we may ignore in the sequel , that group is @xmath175 . the corresponding stabilizer code @xmath176 can not correct single qudit errors at arbitrary positions as the common eigenstates of @xmath177 are tensor products of the fixed @xmath178-qudit state @xmath179 and the unencoded state @xmath180 , i.e. , they are of the form @xmath181 as the stabilizer groups @xmath90 and @xmath173 are conjugated to each other there exists a transformation @xmath182 such that @xmath183 where @xmath184 denotes equality up to the permutation @xmath185 of the qudits . combining ( [ eq : code0 ] ) and ( [ eq : conjugation ] ) , it follows that the states @xmath186 are common eigenstates of @xmath90 . hence the transformation @xmath187 can be used to encode that states of @xmath62 . in the following , we show that the transformation @xmath182 can be efficiently decomposed into single qudit transformations and @xmath153-gates . first we consider the operation of the fourier transformation over the additive group of the field @xmath51 ( see definition [ def : elemgates ] [ def : dft ] ) on the error group modulo the center . the action on @xmath53 is given by @xmath188 similarly , @xmath189 as any element of @xmath190 corresponds to a row vector @xmath191 , we can describe the action of @xmath55 on @xmath190 as the linear transformation @xmath192 . next we investigate the action of the matrix @xmath54 corresponding to the multiplication with @xmath193 , @xmath194 ( see definition [ def : elemgates ] [ def : dft ] ) . we compute @xmath195 hence , @xmath54 acts on @xmath191 as @xmath196 . for the next operation , we have to distinguish the cases whether @xmath31 is odd or @xmath31 is even . if @xmath31 is odd , we define the operator @xmath197 which commutes with @xmath53 . furthermore , @xmath52 acts on @xmath198 as follows : @xmath199 equivalently , we get @xmath200 hence @xmath201 acts on @xmath191 as @xmath202 . if @xmath31 is even , we have to use another definition for the operator @xmath201 as we can not divide by two . for this , we fix an arbitrary self - dual basis @xmath203 of @xmath204 over @xmath205 ( see e.g. , @xcite ) . hence , by definition , @xmath206 . any element @xmath39 can uniquely be written as @xmath207 where @xmath208 . the coefficients @xmath98 are given by @xmath209 since @xmath210 let @xmath211 and let @xmath203 be an arbitrary self - dual basis of @xmath51 over @xmath205 . furthermore , we define an integer - valued function on @xmath51 as @xmath212 then the following holds for all @xmath213 : @xmath214 where @xmath215 with @xmath216 . now we are ready to define an operator @xmath217 as @xmath218 again , @xmath217 commutes with all matrices @xmath53 . the action on @xmath52 is derived from @xmath219 hence @xmath220 finally , for any @xmath193 , @xmath194 , we define @xmath201 as @xmath221 ( note that @xmath222 is uniquely defined as @xmath223 is an automorphism of @xmath204 . ) the matrix @xmath54 is a permutation matrix , hence @xmath201 is diagonal and commutes with @xmath53 . the action of @xmath201 on @xmath52 is given by @xmath224 summarizing , the operators @xmath201 ( defined by ( [ def : p_odd ] ) for @xmath31 odd and ( [ def : p_even ] ) for @xmath31 even ) acts on @xmath191 always as @xmath202 . altogether , we have shown [ satz : sl2 ] the group @xmath225 acts on @xmath190 , the error group @xmath74 modulo the center , as @xmath226 . as @xmath226 acts transitively on the non - zero vectors ( see , e.g. , @xcite ) , we obtain [ cor : trans ] the group @xmath227 acts transitively on the non - trivial elements of @xmath190 . so far , we have only considered the action of single qudit operations on the error group @xmath91 . our goal is to transform an arbitrary abelian stabilizer group @xmath228 into @xmath177 which corresponds to the code @xmath229\!]$ ] . as @xmath176 can not correct errors and single qudit operations do not change the error - correcting properties of a stabilizer code , we need additional transformations which are @xmath153-gates acting on pairs of qudits . combining the definitions of @xmath52 and @xmath59 ( see definition [ def : elemgates ] [ def : x ] and [ def : add ] ) , we can rewrite the @xmath153-gate as @xmath230 using ( [ eq : rewriteadd ] ) , it is easy to show that @xmath59 commutes with all matrices of the form @xmath231 . for elements of the form @xmath232 we get @xmath233 this proofs [ lemma : add ] the transformation @xmath59 acts on @xmath234 as @xmath235 i.e. , @xmath236 is added to @xmath237 and @xmath238 is subtracted from @xmath239 . with this preparation , we can formulate our algorithm to efficiently compute efficient quantum circuits for the encoding of stabilizer codes . ' '' '' [ cols= " < , < " , ] we illustrate the algorithm of figure [ fig : encoding ] using a quantum code over qutrits . a stabilizer matrix of the code @xmath240\!]_3 $ ] is given by @xmath241 in the first step , we transform each pair @xmath242 of the first row that is non - zero to @xmath243 . this is achieved by the transformation @xmath244 the resulting stabilizer matrix is @xmath245 the first non - zero column is the first one . so using the transformation @xmath246 we obtain @xmath247 note that the first column of the @xmath248-matrix is zero . this follows from the fact that the corresponding stabilizer elements commute with the @xmath249 which corresponds to the first row . in the next step , we use the transformations @xmath250 and @xmath251 to obtain @xmath252 using the transformations @xmath253 and @xmath254 yields @xmath255 for the last row , we use the transformations @xmath256 and @xmath257 and get @xmath258 note that after the first three steps , we have @xmath259 . hence not the third , but the fourth qudit is the control qudit for the @xmath153 gates of the last row . equivalently , we could have subtracted the third row from the fourth to obtain @xmath260 . the stabilizer group would not have changed as the addition of one row to another corresponds to multiplying one generator by another . finally , a fourier transformation on the first four positions transforms the stabilizer group into the form @xmath261 . 0.7 ( 9 ) ( 1,1)@xmath262(1,4)@xmath263(1,9,9)(1,1)@xmath263(1,1)@xmath55 ( 2,9,9)(1,1)@xmath217(1,1)@xmath263 ( 3,9,9)(1,1)@xmath217(1,1)@xmath263 ( 4,9,9)(1,6)@xmath55(9 ) the complete decoding circuit is shown in figure [ fig : dec_circuit ] . reversing the order of the gates and replacing each gate by its inverse yields a quantum circuit for encoding . the quantum circuits for both css codes and general stabilizer codes over qudit systems of prime power dimension resulting from our algorithms have a similar structure . they consist of an alternating sequence of single qudit gates and @xmath153-gates with the same control qudit . the total number of gates is always at most quadratic in the number of qudits . for css codes , the single qudit gates are either fourier transformations or multiplication gates , which are trivial for the case of qubits . furthermore , some optimizations are possible by interchanging the rle of the two classical codes involved . the algorithm of @xcite , respectively the modified version of @xcite , to compute quantum circuits for the encoding of qubit stabilizer codes can also be generalized to non - qubit stabilizer codes . it turns out that the resulting quantum circuits have again the same structure and complexity as those presented in this paper . the approach presented here , however , has the advantage that the transformation @xmath182 used to conjugate the stabilizer group @xmath90 to @xmath173 can also be used to compute `` encoded gates '' which preserve the code @xmath62 . conjugating gates from the error - group which preserve the code @xmath176 by @xmath182 results in transformations that are also in the error - group and preserve the code @xmath62 . as those transformations consist only of single qudit operations , they can be used for fault - tolerant quantum computation . markus grassl and martin rtteler acknowledge the hospitality of the mathematical sciences research institute , berkeley , usa . part of this work was supported by _ deutsche forschungsgemeinschaft ( dfg ) , schwerpunktprogramm quanten - informationsverarbeitung ( spp 1078 ) , projekt aqua ( be 887/13 ) _ and the ist / fet project q - acta ( ist-1999 - 10596 ) funded by the european commission . , _ fault - tolerant quantum computation with constant error _ , in proceedings of the 29th annual acm symposium on theory of computing , association for computing machinery , 1997 , pp . see also lanl preprint quant - ph/9611025 . height 2pt depth -1.6pt width 23pt , _ fault - tolerant quantum computation with higher - dimensional systems _ , in quantum computing & quantum communications ; first nasa international conference ; selected papers , qcqc98 , c. p. williams , ed . 1509 of lecture notes in computer science , palm springs , february 1720 1998 , springer , pp . 302313 . , _ quantum reed - solomon codes _ , in proceedings applied algebra , algebraic algorithms and error - correcting codes ( aaecc-13 ) , m. fossorier , h. imai , s. lin , and a. poli , eds . 1719 of lecture notes in computer science , honolulu , hawaii , november 1519 1999 , springer , pp . see also lanl preprint quant - ph/9910059 . , _ reversible computing _ , in automata , languages and programming , j. w. de bakker and j. van leeuwen , eds . , vol . 85 of lecture notes in computer science , noordweijkerhout , the netherlands , 1418 july 1980 , springer , pp .	we present two methods for the construction of quantum circuits for quantum error - correcting codes ( qecc ) . the underlying quantum systems are tensor products of subsystems ( qudits ) of equal dimension which is a prime power . for a qecc encoding @xmath0 qudits into @xmath1 qudits , the resulting quantum circuit has @xmath2 gates . the running time of the classical algorithm to compute the quantum circuit is @xmath3 . # 1 # 1 # 1\|#1 # 1#1\| # 1#2#1\|#2 markus grassl , martin rtteler , and thomas beth
the two most characteristic ensembles is , on one side , the one resulting in exponential ( boltzmann - gibbs - ( bg ) ) distributions , @xmath5 , and , on the other side , the one with power distributions , @xmath6 . both are encountered in the realm of the high energy multiparticle production processes investigated in hadronic and nuclear collisions . they are connected there with , respectively , nonperturbative _ soft _ dynamics operating at small @xmath7 s and described by exponential distributions , @xmath8 ) , and with perturbative _ hard _ dynamics , responsible for large @xmath7 s and described by power distributions , @xmath9 . these two types of dynamics are investigated separately . it is usually assumed that they operate in distinct parts of phase space of @xmath7 , separated by @xmath10 . however , it was found recently that the new high energy data covering the whole of phase space ( cf . , for example , @xcite ) are best fitted by a simple , quasi - power law formula extrapolating between both ensembles @xcite : @xmath11 this formula coincides with the so called tsallis nonextensive distribution @xcite for @xmath12 , @xmath13^{\frac{1}{1-q } } \stackrel{def}{= } c_q\exp_q\left(-\frac{x}{x_0}\right ) \stackrel{q \rightarrow 1}{\longrightarrow } c_1\exp \left(-\frac{x}{x_0}\right).\label{eq : tsallis}\ ] ] this is the distribution we shall shall concentrate on and discuss . in section [ sec : examples ] we shall discuss some examples of processes leading to such distributions . it depends on the nonextensivity parameter @xmath0 and this can be different depending on whether it arises from a tsallis distribution ( @xmath2 ) or from the nonextensive tsallis entropy ( @xmath3 ) . both are connected by @xmath14 and this relation seems to be confirmed experimentally . this is presented in section [ sec : duality ] . in section [ sec : shannon ] we shall discuss necessary conditions for obtaining a tsallis distribution from shannon information entropy . section [ sec : logosc ] demonstrates that a tsallis distribution can also accommodate the log - periodic oscillations apparently observed in high energy data . our conclusions and a summary are presented in section [ sec : summary ] . in many practical applications , a tsallis distribution is derived from tsallis statistics based on his nonextensive entropy , @xmath15 on the other hand , there are even more numerous examples of physical situations not based on @xmath16 and still leading to quasi - power distributions in the tsallis form . in what follows , we shall present some examples of such mechanisms , concentrating on those which allow for an interpretation of the parameter @xmath0 . the first example is _ superstatistics _ @xcite ( cf , also @xcite ) based on the property that a gamma - like fluctuation of the scale parameter in exponential distribution results in the @xmath0-exponential tsallis distribution with @xmath17 ( cf . ( [ eq : tsallis ] ) ) . the parameter @xmath0 defines the strength of such fluctuations , @xmath18 . from the thermal perspective , it corresponds to a situation in which the heat bath is not homogeneous , but has different temperatures in different parts , which are fluctuating around some mean temperature @xmath19 . it must be therefore described by two parameters : a mean temperature @xmath19 and the mean strength of fluctuations , given by @xmath0 . as shown in @xcite , this allows for further generalization to cases where one also has an energy transfer to / from heat bath . the scale @xmath20 in the tsallis distribution becomes then @xmath0-dependent : @xmath21 here the parameter @xmath22 depends on the type of energy transfer , cf . @xcite for illustrative examples from , respectively , nuclear collisions and cosmic ray physics . the second example is the _ preferential attachment approach _ ( used in stochastic networks @xcite ) . here the system under consideration exhibits correlations of the preferential attachment type ( like , for example , `` rich - get - richer '' phenomenon in networks ) and the scale parameter depends on the variable under consideration . if @xmath23 then the probability distribution function , @xmath24 , is given by an equation the solution of which is a tsallis distribution ( again , with @xmath17 ) : @xmath25^{\frac{1}{1-q}}. \label{eq : nets}\ ] ] for @xmath26 one again gets the usual exponential distribution . consider now a _ tsallis distribution from multiplicative noise _ @xcite . we start from the langevin equation @xcite , @xmath27 where @xmath28 and @xmath29 denote stochastic processes corresponding to , respectively , multiplicative and additive noises . this results in the following fokker - planck equation for the distribution function @xmath30 , @xmath31 stationary @xmath30 satisfies @xmath32 with @xmath33 in the case of no correlation between noises and no drift term due to additive noise ( i.e. , for @xmath34 @xcite ) its solution is a tsallis distribution for @xmath35 , @xmath36^{\frac{q}{1-q}}~~{\rm with}~~ t = \frac{2var(\xi)}{\langle \xi\rangle};~~q = 1 + \frac{2var(\gamma)}{\langle \gamma\rangle}. \label{eq : solutionpp}\ ] ] however , if we insist on a solution in the form of @xmath37^n,~~n = \frac{1}{q-1 } , \label{eq : singlep}\ ] ] eq . ( [ eq : k1vk2 ] ) has to be replaced by @xmath38 . \label{eq : k1vk2p}\ ] ] one then gets @xmath39 in the form of a tsallis distribution , ( [ eq : singlep ] ) , but with @xmath40 and with @xmath0-dependent @xmath20 ( reminiscent of @xmath41 from eq . ( [ eq : teff ] ) discussed before , cf . @xcite ) : @xmath42\quad{\rm with}~~ t_0=\frac{cov(\xi,\gamma)}{\langle \gamma\rangle},~~t_1 = \frac{\langle \xi\rangle}{2\langle \gamma\rangle}. \label{eq : teff}\ ] ] let us now remember that the usual situation in statistical physics is that out of three variables considered , energy @xmath43 , multiplicity @xmath44 and temperature @xmath20 , two are fixed and one fluctuates . fluctuations are then given by gamma distributions @xcite ( in the case of multiplicity distributions where @xmath44 are integers , they become poisson distributions ) and only in the thermodynamic limit ( @xmath45 ) does one get them in the form of gaussian distributions , usually discussed in textbooks . in @xcite we discussed in detail situations when two or all three variables fluctuate . if all are fixed we have a distribution of the type of @xmath46 this is nothing else but a tsallis distribution with @xmath47 randomly chosen independent points @xmath48 breaks segment @xmath49 into @xmath44 parts , length of which is distributed according to eq . ( [ eq : constn ] ) . the length of the @xmath50 such part corresponds to the value of energy @xmath51 ( for ordered @xmath52 ) . one could think of some analogy in physics to the case of random breaks of string in @xmath53 points in the energy space . notice that induced partition differs from _ successive sampling _ from the uniform distribution , @xmath54 $ ] , which results in @xmath55 @xcite . ] . for the _ constrained systems _ one gets @xmath56 . for example , if we have @xmath57 independent energies , @xmath58 , then each of them is distributed according to the boltzman distribution , @xmath59 ( and their sum , @xmath60 , is distributed according to gamma distribution , @xmath61(e/\lambda)^{n-1 } \exp(-e/\lambda)$ ] ) . however , if the available energy is limited , @xmath62 , then the resulting _ conditional probability _ becomes a tsallis distribution with @xmath63 : only if the scale parameter @xmath64 would fluctuate in the same way as in in the case of superstatistics , see @xcite . ] @xmath65^{\frac{1}{1 - q } } , \label{eq : constraints}\\ & & q = \frac{n-3 } { n-2 } < 1 , \qquad \qquad \lambda = \frac{\alpha n}{n-1}. \label{eq : c1}\end{aligned}\ ] ] we end this part by a reminder of how tsallis distribution with @xmath66 arises from _ statistical physics considerations_. consider an isolated system with energy @xmath67 and with @xmath68 degrees of freedom ( @xmath57 particles ) . choose single degree of freedom with energy @xmath69 ( i.e. , the remaining , or reservoir , energy is @xmath70 ) . if this degree of freedom is in a single , well defined , state then the number of states of the whole system is @xmath71 and probability that the energy of the chosen degree of freedom is @xmath69 is @xmath72 . expanding ( slowly varying ) @xmath73 around @xmath43 , @xmath74 and ( because @xmath75 ) keeping only the two first terms one gets @xmath76 i.e. , a boltzmann distribution ( or @xmath77 ) . on the other hand , because one usually expects that @xmath78 ( where @xmath79 are of the order of unity and we put @xmath80 and , to account for diminishing the number of states in the reservoir by one , @xmath81 ) @xcite , one can write @xmath82 and write the full series for probability of choosing energy @xmath69 : @xmath83=\nonumber\\ & = & c\left(1 - \frac{1}{\nu - 2}\beta e\right)^{(\nu - 2 ) } = \nonumber\\ & = & \beta(2-q)[1 - ( 1-q)\beta e]^{\frac{1}{1-q } } ; \label{eq : statres}\end{aligned}\ ] ] ( where we have used the equality @xmath84 $ ] ) . this result , with @xmath85 , coincides with results from conditional probability and the induced partition . constant values of @xmath43 , @xmath44 and @xmath20 result in eq . ( [ eq : constn ] ) and @xmath56 . to get larger values of @xmath0 , one has to allow for fluctuations of one of the variables : @xmath43 , @xmath44 or @xmath20 . in superstatistics @xcite it was @xmath20 that was fluctuaing , let us now consider the example of fluctuating @xmath44 . it means that , whereas for fixed @xmath44 ( to simplify notation we changed @xmath86 to @xmath44 here ) @xmath87 for @xmath44 fluctuating according to some @xmath88 the resulting distribution is @xmath89 the most characteristic for our purposes are situations provided by the , respectively , _ binomial distribution _ ( bd ) , eq.([eq : bd ] ) , _ poissonian distribution _ ( pd ) , eq.([eq : pd ] ) and by the _ negative binomial distributions _ ( nbd ) , eq.([eq : nbd ] ) ( cf . @xcite ) : @xmath90 they lead , respectively , to tsallis distribution with @xmath91 , eq.([eq : bdf ] ) , to exponential boltzmann distribution , eq.([eq : pdf ] ) with @xmath77 , and to tsallis distribution with @xmath92 , eq.([eq : nbdf ] ) ( @xmath93 ) : @xmath94 note that in all three cases @xmath95 it is natural that for bd where @xmath96 one has @xmath63 , for pd with @xmath97 also @xmath77 and for nbd where @xmath98 one has @xmath17 ( cf . , also case @xmath67 and @xmath99 considered in @xcite ) . fluctuations of @xmath44 can be translated into fluctuations of @xmath20 . notice first that @xcite ( cf . also @xcite ) nbd , @xmath100\cdot\gamma^k(1+\gamma)^{-k - n}$ ] , arises also if in the poisson multiplicity distribution , @xmath101 , one fluctuates the mean multiplicity @xmath102 using gamma distribution @xmath103 with @xmath104 means average value in a given event whereas @xmath105 denotes averages over events ( or ensembles ) . ] . now , identifying fluctuations of mean @xmath102 with fluctuations of @xmath20 , one can express the above observation via fluctuations of temperature . noticing that @xmath106 ( i.e. , that @xmath107 and @xmath108 $ ] ) one can rewrite @xmath109 as @xmath110 and this is just a gamma distribution describing fluctuations of @xmath111 discussed in @xcite . there is more to the physical meaning of @xmath0 . because @xmath112 the heat capacity @xmath113 can be written as @xmath114 . however , because in our case @xmath67 , i.e. , @xmath115 ( or @xmath116 and @xmath117 ) , the formula @xmath118 obtained before coincides with a similar formula obtained in @xcite : @xmath119 this can be confronted with @xmath120 from @xcite , which , because of the form of @xmath121 in eq . ( [ eq : ntot ] ) , can be rewritten approximately as @xmath122 [ h ] or @xmath123 . @xmath124 distributions of @xmath125 for particles inside the jets with different values of @xmath126 . distributions are fitted using a tsallis distribution ( [ eq : tsallis ] ) with @xmath127 . @xmath128 @xmath129 as function of multiplicity @xmath44 in jets with different values of @xmath130 as presented in @xmath124 . depending on the range of phase space covered , the corresponding values of @xmath0 are @xmath131 for @xmath132 gev and @xmath133 for @xmath134 gev . data are from @xcite and from the durham hepdata project ; http://hepdata.cedar.ac.uk/view/irn9136932 . , title="fig : " ] or @xmath123 . @xmath124 distributions of @xmath125 for particles inside the jets with different values of @xmath126 . distributions are fitted using a tsallis distribution ( [ eq : tsallis ] ) with @xmath127 . @xmath128 @xmath129 as function of multiplicity @xmath44 in jets with different values of @xmath130 as presented in @xmath124 . depending on the range of phase space covered , the corresponding values of @xmath0 are @xmath131 for @xmath132 gev and @xmath133 for @xmath134 gev . data are from @xcite and from the durham hepdata project ; http://hepdata.cedar.ac.uk/view/irn9136932 . , title="fig : " ] tsallis distribution is usually derived either from tsallis entropy ( via the maxent variational approach , not discussed here ) or from some dynamical considerations , some examples of which are presented in this paper . however , it turns out that it also emerges in a quite natural ( in the same maxent approach ) way from shannon entropy , provided one imposes the right constraints . in fact , as shown in @xcite , one can establish a transformation between these two variational problems and shown that they contain the same information . this means that the two approaches seem to be equivalent , one can either use tsallis entropy with relatively simple constraints , or the shannon entropy with rather complicated ones ( cf . , for example , a list of possible distributions one can get in this way @xcite ) . in general , shannon entropy for some probability density @xmath24 , @xmath135 $ ] , supplied with constraint @xmath136 , where @xmath137 is some function of @xmath138 , subjected to the usual maxent variational procedure , results in the following form of @xmath24 : @xmath139 , \label{eq : tfroms}\ ] ] with constants @xmath140 and @xmath64 calculated from the normalization of @xmath24 and from the constraint equation . it is now straightforward to check that @xmath141 \label{eq : constrt}\ ] ] results in @xmath142 which translates to ( remembering that @xmath143 ) a tsallis distribution @xmath144^{\frac{1}{1 - q}}. \label{eq : tfs}\ ] ] the parameter @xmath19 can be deduced from the additional condition which must be imposed , namely from the assumed knowledge of the @xmath145 ( notice that in the case of bg distribution this would be the only condition ) . so far the physical significance of the constraint ( [ eq : constrt ] ) is not fully understood . its form can be deduced from the idea of varying scale parameter in the form of the preferential attachment , eq . ( [ eq : nets ] ) , which in present notation means @xmath146 . as shown in ( [ eq : nets ] ) it results in tsallis distribution ( [ eq : tfs ] ) . this suggest the use of @xmath147 $ ] constrained as in eq . ( [ eq : constrt ] ) . in such approach @xmath148\ln [ t(e ) ] + [ ( 2-q)/(q-1)]\ln \left ( t_0\right)$ ] and , because @xmath149 , therefore @xmath150 for tsallis distribution becoming @xmath151 for boltzmann - gibbs ( bg ) distribution ( @xmath77 ) . it is interesting that the constraint ( [ eq : constrt ] ) seems to be natural a for multiplicative noise described by the langevine equation : @xmath152 , with traditional multiplicative noise @xmath153 and additive noise ( stochastic processes ) @xmath29 ) ( see @xcite for details ) . in fact , there is a connection between the kind of noise in this process and the condition imposed in the maxent approach . for processes described by an additive noise , @xmath154 , the natural condition is that imposed on the arithmetic mean , @xmath155 , and it results in the exponential distributions . for the multiplicative noise , @xmath156 , the natural condition is that imposed on the geometric mean , @xmath157 , which results in a power law distribution @xcite . one has to start with some explanatory remarks . the tsallis distribution can be also obtained via maxent procedure from tsallis entropy , @xmath159 where @xmath160 and @xmath161 . now , depending on the condition imposed one gets from @xmath16 @xmath162^{-\frac{1}{1-q}}\quad { \rm for}\quad\langle x\rangle_1 \label{eq : cond1}\\ { \rm or}\quad f(x ) & = & ( 2-q)[1+(q-1)x]^\frac{1}{1-q}\quad { \rm for}\quad\langle x\rangle_q . \label{eq : condq}\end{aligned}\ ] ] however , after replacement of @xmath0 by @xmath163 , the distribution ( [ eq : cond1 ] ) becomes the usual tsallis distribution ( [ eq : condq ] ) . therefore one encounters an apparent puzzle , namely the @xmath2 of tsallis distribution does not coincides with the @xmath3 of corresponding tsallis entropy , instead they are connected by relation @xmath158 . the natural question therefore arises : is such a relation seen in data ? as shown in @xcite that seems really be the case , at least quantitatively . this is seen when comparing @xmath164 obtained from data on @xmath165 _ distributions _ [ duality]@xmath124 ) to @xmath166 obtained from data on _ multiplicities _ in @xmath167 collisions assuming that entropy is proportional to the number of particles produced ( cf . [ duality ] @xmath128 ) . whereas @xmath2 is deduced from a tsallis distribution taken in one of the forms discussed above , @xmath3 is deduced directly from the corresponding entropy @xmath16 of the @xmath167 collision . assume that such collision can be adequately described by a superposition model in which the main ingredients are @xmath68 nucleons which have interacted at least once @xcite . assume further that they are identical and independent and produce @xmath168 secondaries of each other . as a result a @xmath169 are produced in one collision and the mean multiplicity is @xmath170 ( where @xmath171 is the mean number of nucleons participating in the collision and @xmath172 the mean multiplicity in @xmath173 elementary collision . the corresponding entropy @xmath174 of such process will then be @xmath0-sum of @xmath68 entropies @xmath175 of individual collisions and is given by : [ t ] energy dependencies of the parameters @xmath0 obtained from , respectively : multiplicity distributions @xmath88 @xcite ( squares ) , from different analysis of transverse momenta distributions @xmath176 in @xmath177 data ( @xcite - circles , full symbols ) and from data on @xmath176 from pb+pb collisions ( @xcite - half filled circles ) . @xmath128 energy dependence of the charged multiplicity for nucleus - nucleus collisions divided by the superposition of multiplicities from proton - proton collisions fitted to data on multiplicity taken from @xcite ( na49 ) and from compilation @xcite.,title="fig : " ] energy dependencies of the parameters @xmath0 obtained from , respectively : multiplicity distributions @xmath88 @xcite ( squares ) , from different analysis of transverse momenta distributions @xmath176 in @xmath177 data ( @xcite - circles , full symbols ) and from data on @xmath176 from pb+pb collisions ( @xcite - half filled circles ) . @xmath128 energy dependence of the charged multiplicity for nucleus - nucleus collisions divided by the superposition of multiplicities from proton - proton collisions fitted to data on multiplicity taken from @xcite ( na49 ) and from compilation @xcite.,title="fig : " ] @xmath178^k = \frac{\left [ 1 + ( 1-q)s^{(1)}_q\right]^{\nu } - 1}{1 - q}. \label{eq : sqnu}\ ] ] notice that @xmath179 = \nu \ln \left [ 1 + ( 1 - q ) s^{(1)}_q\right]$ ] and @xmath180 . for @xmath56 entropy @xmath174 is nonextensive because @xmath181 . for @xmath182 one has @xmath183 only for @xmath184 and @xmath185 , i.e. , entropy is extensive , @xmath186 . because @xmath187 , the nonextensivity parameter obtained from the corresponding entropies must be smaller than unity , @xmath188 . on the other hand , all estimates of nonextensivity parameter from tsallis distributions lead to @xmath189 . the pure power - like distributions are known to be in many cases decorated by specific log - periodic oscillations ( i.e , multiplied by some dressing factor @xmath190 ) @xcite . they suggest some hierarchical fine - structure existing in the system under consideration and are usually regarded as possibly indicating some kind of multifractality in the system . closer inspection of recent data from lhc @xcite reveals that , for large transverse momenta @xmath165 , one observes a similar effect , cf . [ summary ] . so far , the prevailing opinion is that this is just an apparatus induced artifact with no meaning . however , its persistence in the type of experiment considered , energy or type of collision process ( provided that the range of @xmath165 covered is large enough ) calls for some explanation . such an explanation was considered in @xcite . in @xcite , the only possibility investigated was to attribute such oscillations to complex values of the power index @xmath57 in eq . ( [ eq : h ] ) . as shown in @xcite and also below , it also works in the case of quasi - power like tsallis distributions . however , because one now also has a scale parameter @xmath20 and a constant term , one can also offer another explanation : real @xmath57 but log - periodically oscillating scale parameter @xmath20 . this was discussed in detail in @xcite . we shall present both possibilities here . [ h ] @xmath191 for the highest energy @xmath192 tev , the tsallis behavior is evident . only cms data are shown @xcite , others behave essentially in the identical manner . @xmath128 log - periodic oscillations showing up in different experimental data like @xcite or atlas @xcite taken at @xmath192 tev . @xmath193 results from cms @xcite for different energies . @xmath194 results for different systems ( @xmath177 collisions compared with @xmath195 taken for @xmath196 % centrality @xcite . results from alice are very similar @xcite.,title="fig : " ] @xmath191 for the highest energy @xmath192 tev , the tsallis behavior is evident . only cms data are shown @xcite , others behave essentially in the identical manner . @xmath128 log - periodic oscillations showing up in different experimental data like @xcite or atlas @xcite taken at @xmath192 tev . @xmath193 results from cms @xcite for different energies . @xmath194 results for different systems ( @xmath177 collisions compared with @xmath195 taken for @xmath196 % centrality @xcite . results from alice are very similar @xcite.,title="fig : " ] @xmath191 for the highest energy @xmath192 tev , the tsallis behavior is evident . only cms data are shown @xcite , others behave essentially in the identical manner . @xmath128 log - periodic oscillations showing up in different experimental data like @xcite or atlas @xcite taken at @xmath192 tev . @xmath193 results from cms @xcite for different energies . @xmath194 results for different systems ( @xmath177 collisions compared with @xmath195 taken for @xmath196 % centrality @xcite . results from alice are very similar @xcite.,title="fig : " ] @xmath191 for the highest energy @xmath192 tev , the tsallis behavior is evident . only cms data are shown @xcite , others behave essentially in the identical manner . @xmath128 log - periodic oscillations showing up in different experimental data like @xcite or atlas @xcite taken at @xmath192 tev . @xmath193 results from cms @xcite for different energies . @xmath194 results for different systems ( @xmath177 collisions compared with @xmath195 taken for @xmath196 % centrality @xcite . results from alice are very similar @xcite.,title="fig : " ] for simple power laws , one has some function @xmath197 which is scale invariant , @xmath198 and @xmath199 with @xmath200 . however , this can be written as @xmath201 , where @xmath202 is an arbitrary integer . we then have not a single power @xmath203 but rather a whole family of complex powers , @xmath204 , with @xmath205 . their imaginary part signals a hierarchy of scales leading to log - periodic oscillations . this means that , in fact , @xmath206 $ ] ( where @xmath207 are coefficients of the expansion ) . this the origin of the usual dressing factor appearing in @xcite and used to describe data : @xmath208 \label{eq : r}\ ] ] ( only @xmath209 and @xmath210 terms are kept ) . it turns out that a similar scaling solution can also be obtained in case of a tsallis quasi - power like distribution . to this end one must start from stochastic network approach , section [ sec : sn ] and eq . ( [ eq : nets ] ) , in which tsallis distribution is obtained by introducing a scale parameter depending on the variable considered . in our case it is @xmath211 resulting in @xmath212 in final difference form ( with change in notation : @xmath19 replaced by @xmath20 ) @xmath213 we consider a situation in which @xmath214 . it depends now on the new scale parameter @xmath215 ( @xmath216 in order to keep changes of @xmath217 to be of the order of @xmath20 ) and can be very small but always remains finite . it can now be shown that @xmath218 = ( 1 - \alpha n)f(e)$ ] which , when expressed in the new variable @xmath219 , corresponds formally to the following scale invariant relation : @xmath220 = ( 1 - \alpha n)g(x ) . \label{eq : gscaling}\ ] ] following the same procedure used to obtain dressed solutions of scale invariant functions discussed at the beginning of this section , one arrives at the dressed tsallis distribution ( we keep , as before , only the two lowest terms , @xmath221 and @xmath222 ) ) @xmath223 . however , @xmath57 and @xmath20 are both unknown _ a priori _ parameters . therefore , for fitting purposes , where we have to use two ( and not three ) parameter tsallis distribution , we use @xmath224^{-m_0}$ ] with fitting parameters @xmath225 and @xmath226 . in terms of @xmath20 , @xmath57 and @xmath215 we have @xmath227 . ] : @xmath228\right\}. \label{eq : approx}\ ] ] with @xmath229 . in addition to the scale parameter @xmath215 one has two more parameters occurring in the dressing factor @xmath190 , @xmath230 and @xmath209 . the other parameters occurring in eq . ( [ eq : r ] ) are expressed by the original parameters in the following way : @xmath231 , @xmath232 , @xmath233 and @xmath234 . one can , however , consider a more involved evolution process , with @xmath235 sequential cascades ; in this case the additional parameter @xmath235 changes parameter @xmath236 in ( [ eq : r ] ) , @xmath237 . it does not affect the slope parameter @xmath225 but changes the frequency of oscillations which now decrease as @xmath238 . comparison with data requires @xmath239 ( cf . @xcite for details ) . as mentioned before , one can translate a dressed tsallis distribution into a normal one but with a log - periodically oscillating in @xmath165 scale factor @xmath20 , cf . [ figlpot]a . the formula used there to fit the obtained results resembles that for dressing factor ( [ eq : r ] ) , @xmath240 . \label{eq : t}\ ] ] in fit shown in fig . [ figlpot]a parameters ( generally energy dependent ) are @xmath241 . [ t ] oscillations of scale parameter @xmath20 leading to identical dressed tsallis distribution as shown in fig . [ summary]b ( obtained for cms data at @xmath192 tev and fitted using eq . ( [ eq : t ] ) ) . @xmath128 dependencies of @xmath242 from eq . ( [ eq : taue ] ) and @xmath243 from eq . ( [ eq : st6 ] ) resulting in oscillations of @xmath20 shown in panel @xmath124 . , title="fig : " ] oscillations of scale parameter @xmath20 leading to identical dressed tsallis distribution as shown in fig . [ summary]b ( obtained for cms data at @xmath192 tev and fitted using eq . ( [ eq : t ] ) ) . @xmath128 dependencies of @xmath242 from eq . ( [ eq : taue ] ) and @xmath243 from eq . ( [ eq : st6 ] ) resulting in oscillations of @xmath20 shown in panel @xmath124 . , title="fig : " ] to explain eq . ( [ eq : t ] ) one uses a stochastic equation for the temperature evolution @xcite written in langevin formulation with energy dependent noise , @xmath244 , and allowing for time dependent @xmath245 but results are for transverse momenta @xmath165 here . however , they are taken at the midrapidity , i.e. , for @xmath246 , and for large transverse momenta , @xmath247 , and in this region one has @xmath248 . ] : @xmath249 assuming now a scenario of _ preferential attachment _ ( cf . section [ sec : sn ] above ) known from the growth of networks @xcite ) one has @xmath250 and eq . ( [ eq : st2 ] ) has now the form : @xmath251 after straightforward manipulations ( cf . @xcite for details ) one gets , for large @xmath69 ( i.e. , neglecting terms @xmath252 ) : @xmath253\frac{dt}{d(\ln e ) } + t\frac{d\xi(t , e)}{d(\ln e ) } = 0.\label{eq : st5}\ ] ] assuming now that noise @xmath244 increases logarithmically with energy , @xmath254 in this case eq . ( [ eq : st5 ] ) becomes an equation for the damped hadronic oscillator with solution in the form of log - periodic oscillation of temperature with frequency @xmath255 and depending on initial conditions phase shift parameter @xmath256 : @xmath257\ln e\right\}\cdot \sin(\omega\ln e + \phi ) . \label{eq : st7}\ ] ] averaging the noise fluctuations over time @xmath258 and taking into account that the noise term can not on average change the temperature , @xmath259 , one arrives at @xmath260 this should now be compared with the parametrization of @xmath261 given by eq . ( [ eq : t ] ) and used to fit data in fig.[figlpot ] , of the order of @xmath262 , emerges from the stochastic process with energy dependent noise ; the main contribution comes from the usual energy - independent gaussian white noise . ] . we close with the remark that , instead of using energy dependent noise @xmath244 given by eq . ( [ eq : st6 ] ) and keeping the relaxation time @xmath263 constant . we could equivalently keep the energy independent white noise , @xmath264 , but allow for the energy dependent relaxation time , for example in the form of @xmath265 in this case the temperature evolution has the form @xmath266 e^{-t\omega^2/n } \exp\left(-\frac{t}{\tau_0}\right ) , \label{eq : ttau}\ ] ] and @xmath20 gradually approaches its equilibrium value @xmath267 . actually , for @xmath268 , as in our case , this approach towards equilibrium is faster for large @xmath69 . this is because , in addition to the usual exponential relaxation characteristic for @xmath269 case , we have an additional factor @xmath270 . we presented examples of possible mechanisms resulting in quasi - power distributions exemplified by tsallis distribution , eq . ( [ eq : tsallis ] ) . our presentation had to be limited , therefore we did not touch thermodynamic connections of this distribution @xcite or the possible connection of tsallis distributions with qcd calculations discussed recently @xcite . * statistical physics consideration , as well as `` induced partition process '' , results in eq . ( [ eq : constn ] ) , i.e. , in tsallis distribution with @xmath271 . fluctuations of the multiplicity @xmath44 modify the parameter @xmath0 which is now equal to @xmath272 , cf . ( [ eq : fluctn ] ) . notice that conditional probability for the bg distribution again results in eq . ( [ eq : constn ] ) . * fluctuations of the multiplicity @xmath44 are equivalent to results of an application of superstatistics , where the convolution @xmath273 becomes a tsallis distribution , eq . ( [ eq : tsallis ] ) , for @xmath274 * differentiating eq . ( [ eq : sst ] ) one gets @xmath275 this is nothing else than a `` preferential attachment '' case , again resulting in a tsallis distribution which for @xmath276 becomes a bg distribution , cf . eq . ( [ eq : nets ] ) . * replacing in eq . ( [ eq : diffsst ] ) differentials by finite differences , cf . eq . ( [ eq : deltae ] ) , one gets for @xmath277 the scale invariant relation , eq . ( [ eq : gscaling ] ) , which results in log - periodic oscillations in tsallis distributions . of such well known distributions as the snedecor distribution ( with @xmath278 with integer @xmath68 , for @xmath279 it becomes an exponential distribution ) , can be extended to complex nonextensivity parameter . ] in addition to this line of reasoning , we have also brought in the problem of the apparent duality between the nonextensive parameters obtained from the whole phase space measurements of multiplicity and more local measurements of transverse momenta . this point deserves an experimental and phenomenological scrutiny . finally , we tentatively suggested that , by choosing the right constraints , which account for additive or multiplicative processes considered , one can also get a tsallis distribution directly from the shannon information entropy . c. tsallis , j. stat . phys . * 52 * ( 1988 ) 479 and eur . j. a * 40 * ( 2009 ) 257 ; cf . also c. tsallis , _ introduction to nonextensive statistical mechanics _ ( berlin 2009 : springer ) . for an updated bibliography on this subject , see http://tsallis.cat.cbpf.br/biblio.htm . o. j. e. maroney , phys . rev e * 80 * ( 2009 ) 061141 ; t. s. bir , k. rmssy and z. schram , j. phys . g * 37 * ( 2010 ) 094027 ; t. s. bir and p. vn , phys . e * 83 * ( 2011 ) 061147 ; t. s. bir and z. schram , eur . . j. web conf . * 13 * , ( 2011 ) 05004 ; t. s. bir , _ is there a temperature ? conceptual challenges at high energy , acceleration and complexity _ ( springer , new york dordrecht heidelberg london , 2011 ) ; p. vn , g. g. barnafldi , t. s. bir and k. rmssy , j. phys . : conf . * 394 * ( 2012 ) 012002 . g. wilk , z. wodarczyk , acta phys . b * 35 * ( 2004 ) 871 and 2141 ; c. tsallis , eur . j. st * 161 * ( 2008 ) 175 ; d. j. b. soares , c. tsallis , a. m. mariz , l. r. da silva , europhys . * 70 * ( 2008 ) 70 . t. s. bir , g. g. barnafldi , p. van , physica a * 417 * ( 2015 ) 215 ; t. s. bir , p. van , g. g. barnafldi , k. rmssy , _ statisytical power law due to reservoir fluctuations and the universal thermostat independence principle _ , arxiv:1409.5975[cond - mat.stat - mech ] , to be published in entropy . y. huang , h. saleur , c. sammis , d. sornette , europhys . lett . * 41 * ( 1998 ) 43 ; h. saleur , c.g . sammis , d. sornette , j. geophys . ( 1996 ) 17661 ; a. krawiecki , k. kacperski , s. matyjaskiewicz , j. a. holyst , chaos solitons fractals * 18 * ( 2003 ) 89 ; j. bernasconi , w. r. schneider , j. stat . 30 * ( 1983 ) 355 ; d. stauffer , d. sornette , physica a * 252 * ( 1998 ) 271 ; d. stauffer , physica a * 266 * ( 1999 ) 35 . c. y. wong , g. wilk , acta phys . b * 43 * ( 2012 ) 2047 ; phys . rev . d * 87 * ( 2013 ) 114007 and _ relativistic hard - scattering and tsallis fits to @xmath165 spectra in @xmath281 collisions at the lhc _ , arxiv:1309.7330[hep - ph ] , to be published in the open nuclear and particle physics journal ( 2014 ) . l. j. l. cirto , c. tsallis , c .- y . wong , g. wilk , _ the transverse - momenta distributions in high - energy @xmath281 collisions - a statistical - mechanical approach _ arxiv:1409.3278[hep - ph ] and c .- y . wong , g. wilk , l.j . l. cirto , c. tsallis , _ possible implication of a single nonextensive @xmath165 distribution for hadron production in high - energy pp collisions _ ; arxiv:1412.0474[hep - ph ] .	quasi - power law ensembles are discussed from the perspective of nonextensive tsallis distributions characterized by a nonextensive parameter @xmath0 . a number of possible sources of such distributions are presented in more detail . it is further demonstrated that data suggest that nonextensive parameters deduced from tsallis distributions functions @xmath1 , @xmath2 , and from multiplicity distributions ( connected with tsallis entropy ) , @xmath3 , are not identical and that they are connected via @xmath4 . it is also shown that tsallis distributions can be obtained directly from shannon information entropy , provided some special constraints are imposed . they are connected with the type of dynamical processes under consideration ( additive or multiplicative ) . finally , it is shown how a tsallis distribution can accommodate the log - oscillating behavior apparently seen in some multiparticle data .
the dynamic stability of steady - state homogeneous sliding between two macroscopic bodies in frictional contact is a basic problem of interest in various scientific disciplines such as physics , solid mechanics , materials science and geophysics . the emergence of instabilities may give rise to rich dynamics and play a dominant role in a broad range of frictional phenomena @xcite . the response of a frictional system to spatiotemporal perturbations , and the accompanying instabilities , are governed by several physical properties and processes . generally speaking , one can roughly distinguish between bulk effects ( i.e. the constitutive behavior and properties of the bodies of interest , their geometry and the external loadings applied to them ) and interfacial effects related to the frictional constitutive behavior . the ultimate goal of a theory in this respect is to identify the relevant physical processes at play , to quantify the interplay between them through properly defined dimensionless parameters and to derive the stability phase diagram in terms of these parameters , together with the growth rate of various unstable modes . as a background and motivation for what will follow , we would like first to briefly discuss the various players affecting the stability of frictional sliding , along with stating some relevant results available in the literature . focusing first on bulk effects , it has been recognized that when considering isotropic linear elastic bodies , there is a qualitative difference between sliding along interfaces separating bodies made of identical materials and interfaces separating dissimilar materials . in the former case , there is no coupling between interfacial slip and normal stress variations , while in the latter case due to broken symmetry such coupling exists @xcite . this coupling may lead to a reduction in the normal stress at the interface and consequently to a reduction in the frictional resistance . hence , bulk material contrast ( i.e. the existence of a bi - material interface ) potentially plays an important destabilizing role in the stability of frictional sliding . another class of bulk effects is related to the finite geometry of any realistic sliding bodies and the type of loading applied to them ( e.g. velocity or stress boundary conditions ) . to the best of our knowledge , the latter effects are significantly less explored in the literature ( but see @xcite ) . in relation to interfacial effects , it has been established that sliding along a bi - material interface described by the classical coulomb friction law , @xmath5 ( @xmath6 is the local friction stress , @xmath7 is the local compressive normal stress and @xmath8 is a constant friction coefficient ) , is unstable against perturbations at all wavelengths and irrespective of the value of the friction coefficient , when the bi - material contrast is such that the generalized rayleigh wave exists @xcite . the latter is an interfacial wave that propagates along frictionless bi - material interfaces , constrained not to feature opening @xcite . it is termed the generalized rayleigh wave because it coincides with the ordinary rayleigh wave when the materials are identical and it exists when the bi - material contrast is not too large . in fact , the response to perturbations in this case is mathematically ill - posed @xcite . ill - posedness , which is a stronger condition than instability ( i.e. all perturbation modes can be unstable , yet a problem can be mathematically well - posed ) , will be discussed later . it has been then shown that replacing coulomb friction by a friction law in which the friction stress @xmath6 does not respond instantaneously to variations in the normal stress @xmath7 , but rather approaches @xmath5 over a finite time scale , can regularize the problem , making it mathematically well - posed @xcite . subsequently , the problem has been addressed within the constitutive framework of rate - and - state friction models , where the friction stress depends both on the slip velocity and the structural state of the interface . within this framework @xcite , the simplest version of the friction stress takes the form @xmath9 , where @xmath10 is the difference between the local interfacial slip velocities of the two sliding bodies and @xmath11 is a dynamic coarse - grained state variable . under steady - state sliding conditions the state variable @xmath11 attains a steady - state value determined by @xmath10 , @xmath12 . within such a constitutive framework , the most relevant physical quantities for the question of stability , which will be extensively discussed below , are the instantaneous response to variations in the slip velocity , @xmath13 ( the so - called `` direct effect '' ) , and the variation of the steady - state frictional strength with the slip velocity , @xmath14 @xcite . note that here and below we use the following shorthand notation : @xmath15 and @xmath16 . previous studies have argued that an instantaneous strengthening response , i.e. the experimentally well - established positive direct effect @xmath17 ( which is associated with thermally activated rheology @xcite ) , is sufficient to give rise to the existence of a quasi - static range of response to perturbations at sufficiently small slip velocities @xcite . the existence of such a quasi - static regime is non - trivial ( e.g. it does not exist for coulomb friction ) ; it implies that when very small slip velocities are of interest , one can reliably address the stability problem in the framework of quasi - static elasticity ( i.e. omitting inertial terms to begin with ) , rather than considering the full and more difficult elastodynamic problem and then take the quasi - static limit . within such a quasi - static framework , it has been shown that @xmath17 can lead to stable response against sufficiently short wavelength perturbations , even if the interface is velocity - weakening in steady - state , @xmath18 @xcite . furthermore , it has been shown that sufficiently strong velocity - strengthening , @xmath19 , can overcome the destabilizing bi - material effect , leading to the stability of perturbations at all wavelengths in the quasi - static limit @xcite . despite this progress , several important questions remain open . first , to the best of our knowledge the fully elastodynamic stability analysis of bi - material interfaces in the framework of rate - and - state friction models has not been performed . this is important since the quasi - static regime when it exists is expected to be valid only for very small slip velocities ( as was argued in @xcite and will be explicitly shown below ) . second , a very recent compilation of a large set of experimental data for a broad range of materials @xcite has revealed that dry frictional interfaces generically become velocity - strengthening over some range of slip velocities @xcite . in other cases , frictional interfaces are intrinsically velocity - strengthening @xcite . as velocity - strengthening friction is expected to play a stabilizing role in the stability of frictional sliding , there emerges a basic question about the interplay between the stabilizing velocity - strengthening friction effect and the destabilizing bi - material effect , when elastodynamics is fully taken into account . finally , in almost all of the previous studies we are aware of , the sliding bodies were assumed to be infinite ( but see , for example , @xcite ) . yet , realistic sliding bodies are of finite extent and the interaction with the boundaries may be of importance . to address these issues we analyze in this paper the linear stability of a deformable solid of height @xmath0 steadily sliding on top of a rigid solid within a generic rate - and - state friction constitutive framework , fully taking into account elastodynamic effects . the rate - and - state friction constitutive framework includes a single state variable @xmath11 , but is otherwise general in the sense that no special properties of @xmath20 are being specified and rather generic dynamics of @xmath11 are considered . nevertheless , we will be mostly interested in the physically relevant case in which the interface exhibits a positive instantaneous response to velocity changes , @xmath17 ( positive direct effect ) @xcite , and is steady - state velocity - strengthening , @xmath21 , over some range of slip velocities @xcite . in addition , we will consider two variants of the constitutive model , each of which incorporates a regularized response to normal stress variations @xcite . while our analysis remains rather general , the main simplification we adopt is that we consider the limit of strong material contrast , i.e. we take one of the solids to be non - deformable . the motivation for this is two - fold . first , we know that the bi - material effect that emerges from the coupling between interfacial slip and normal stress variations becomes stronger as the material contrast increases @xcite . we are interested here in exploring the ultimate range of stability and consequently we focus on strongly dissimilar materials , which will allow us to extract upper bounds on the stability of bi - material frictional interfaces . second , the strong dissimilar materials limit somewhat reduces the mathematical complexity involved and makes the problem more amenable to analytic progress , as will be shown below . we suspect that this simplification does not imply qualitative differences compared to the finite material contrast case , though interesting quantitative differences may emerge and will be explored elsewhere . the finite material contrast case can be studied along the same lines , though it is more technically involved . the structure of this paper is as follows ; in sect . [ sec : executivesummary ] the main results of the paper are listed . in sect . [ sec : math ] the basic equations and the constitutive framework are introduced . in sect . [ sec : spectrum ] the linear stability spectrum for finite height @xmath0 systems is derived , with a focus on standard rate - and - state friction . in sect . [ sec : finiteh ] the linear stability spectrum is analyzed in the small wave - number @xmath1 limit , demonstrating the existence of a universal instability ( independent of the details of the friction law ) with a fastest growing mode characterized by a wave - number satisfying @xmath2 and a growth rate that scales with @xmath3 . in sect . [ sec : infiniteh ] the linear stability spectrum is analyzed in the large systems limit , @xmath4 . we derive the stability phase diagram in terms of a relevant set of dimensionless parameters that quantify the competing physical effects involved . we show that the dominant instability mode travels at nearly the dilatational wave - speed ( super - shear ) in the opposite direction to the sliding motion . in a certain parameter range this instability is manifested through unstable modes at all wave - numbers , yet the frictional response is shown to be mathematically well - posed . instability modes which travel at nearly the shear wave - speed in the sliding direction are shown to exist in a relatively small region of the parameter space . finally , previous results obtained in the quasi - static regime @xcite are shown to be relevant within a narrow region of the parameter space . in sect . [ sec : modified - models ] a finite - time regularized response to normal stress variations , within the framework of generalized rate - and - state friction models , is studied . we show that this regularized response tends to promote stability . section [ sec : summary ] offers a brief discussion and some concluding remarks . the analysis to be presented below is rather extensive and somewhat mathematically involved . yet , we believe that it gives rise to a number of physically significant and non - trivial results . in order to highlight the logical structure of the analysis and its major outcomes , we list here the main results to be derived in detail later on : * the stability of a deformable body of finite height @xmath0 steadily sliding on top of a rigid solid is studied . the linear stability spectrum is derived in the constitutive framework of velocity - strengthening rate - and - state friction models and an instantaneous response to normal stress variations . * the spectrum takes the form of a complex - variable equation implicitly relating the real wave - number @xmath1 and the complex growth rate @xmath22 of interfacial perturbations . physically , it represents the balance between the perturbation of the elastodynamic shear stress at the interface and the perturbation of the friction stress ( the latter includes the elastodynamic bi - material effect and constitutive effects ) . the spectrum can feature several distinct classes of solutions . * the linear stability spectrum is first analyzed in the small @xmath23 limit and the existence a universal instability ( independent of the details of the friction law ) is analytically demonstrated . the instability is shown to be related to waveguide propagating modes , which are strongly coupled to the height @xmath0 . they are characterized by a wave - number satisfying @xmath2 and a growth rate that scales with @xmath3 . * as the growth rate of the waveguide - like instability vanishes in the @xmath24 limit , the linear stability spectrum is analyzed also in the large @xmath23 limit , where additional instabilities are sought for . two classes of new instabilities , qualitatively different from the waveguide - like instability , are found : ( i ) a dynamic instability which is mediated by modes propagating at nearly the dilatational wave - speed ( super - shear ) in the opposite direction to the sliding motion and features a vanishingly small wave - number at threshold ( ii ) a dynamic instability which is mediated by modes propagating at nearly the shear wave - speed in the direction of sliding motion and features a finite wave - number at threshold . * in addition , a third type of instability which was previously discussed in the literature exists in the quasi - static regime . * in all cases , even when all wave - numbers become unstable in a certain parameter range , the frictional response is shown to be mathematically well - posed . * a comprehensive stability phase diagram in the large @xmath23 limit is derived , presented and physically rationalized . the stability phase diagram is expressed in terms of relevant set of dimensionless parameters that quantify the competing physical effects involved . * a regularized , finite - time , response of the friction stress to normal stress variations is analyzed in detail and is shown to promote stability . that is , the instabilities mentioned above still exist , but their appearance is delayed and the range of unstable wave - numbers is reduced as compared to the case of an instantaneous response to normal stress variations . * the results may have implications for understanding the failure / rupture dynamics of a large class of bi - material frictional interfaces . the problem we study involves an isotropic linear elastic solid , of infinite extent in the @xmath25-direction and height @xmath0 in the @xmath26-direction , homogeneously sliding at a velocity @xmath27 in the @xmath25-direction on top of a rigid ( non - deformable and stationary ) half space . two - dimensional plane - strain deformation conditions are assumed and the geometry of the problem is sketched in fig . [ fig1 ] . in fig . [ fig : gaps ] we present numerical solutions of eqs . and for the very same set of parameters . it is observed that while each of these solutions is discontinuous in itself , superposing the two solutions generates two _ continuous _ functions , each of which is composed of two segments from different solutions . we thus propose that the gap observed in fig . [ fig : comparison ] is related to the solution in eq . , which features a non - physical exponential divergence as @xmath28 . in a real system , with a finite height @xmath0 ( however large ) , increasing solutions always exist and will naturally lead to the regularization of the discontinuities that emerge in the strict @xmath24 limit . ampuero , j .- p . , ben - zion , y. , 2008 . cracks , pulses and macroscopic asymmetry of dynamic rupture on a bimaterial interface with velocity - weakening friction . geophysical journal international 173 ( 2 ) , 674692 . ben - zion , y. , 2008 . collective behavior of earthquakes and faults : continuum - discrete transitions , progressive evolutionary changes , and different dynamic regimes . reviews of geophysics 46 ( 4 ) , rg4006 . hawthorne , j. c. , rubin , a. m. , 2013 . tidal modulation and back - propagating fronts in slow slip events simulated with a velocity - weakening to velocity - strengthening friction law . journal of geophysical research : solid earth 118 ( 3 ) , 12161239 . martins , j. , simes , f. m. f. , 1995 . on some sources of instability / ill - posedness in elasticity problems with coulomb s friction . in : raous , m. , jean , m. , moreau , j. ( eds . ) , contact mechanics . springer us , pp . 95106 . prakash , v. , clifton , r. , 1993 . time resolved dynamic friction measurememts in pressure - shear . in : ramesh , k. ( ed . ) , experimental techniques in the dynamics of deformable solids . american society of mechanical engineers . applied mechanics division , pp . prakash , v. , clifton , r. j. , 1992 . pressure - shear plate impact measurement of dynamic friction for high speed machining applications . in : proceedings of the 7th international congress on experimental mechanics . . 811 . rice , j. r. , lapusta , n. , ranjith , k. , 2001 . rate and state dependent friction and the stability of sliding between elastically deformable solids . journal of the mechanics and physics of solids 49 ( 9 ) , 18651898 . teufel , l. w. , logan , j. m. , 1978 . effect of displacement rate on the real area of contact and temperatures generated during frictional sliding of tennessee sandstone . pure and applied geophysics 116 ( 4 - 5 ) , 840865 . tonazzi , d. , massi , f. , culla , a. , baillet , l. , fregolent , a. , berthier , y. , 2013 . instability scenarios between elastic media under frictional contact . mechanical systems and signal processing 40 ( 2 ) , 754766 .	understanding the dynamic stability of bodies in frictional contact steadily sliding one over the other is of basic interest in various disciplines such as physics , solid mechanics , materials science and geophysics . here we report on a two - dimensional linear stability analysis of a deformable solid of a finite height @xmath0 , steadily sliding on top of a rigid solid within a generic rate - and - state friction type constitutive framework , fully accounting for elastodynamic effects . we derive the linear stability spectrum , quantifying the interplay between stabilization related to the frictional constitutive law and destabilization related both to the elastodynamic bi - material coupling between normal stress variations and interfacial slip , and to finite size effects . the stabilizing effects related to the frictional constitutive law include velocity - strengthening friction ( i.e. an increase in frictional resistance with increasing slip velocity , both instantaneous and under steady - state conditions ) and a regularized response to normal stress variations . we first consider the small wave - number @xmath1 limit and demonstrate that homogeneous sliding in this case is universally unstable , independently of the details of the friction law . this universal instability is mediated by propagating waveguide - like modes , whose fastest growing mode is characterized by a wave - number satisfying @xmath2 and by a growth rate that scales with @xmath3 . we then consider the limit @xmath4 and derive the stability phase diagram in this case . we show that the dominant instability mode travels at nearly the dilatational wave - speed in the opposite direction to the sliding direction . in a certain parameter range this instability is manifested through unstable modes at all wave - numbers , yet the frictional response is shown to be mathematically well - posed . instability modes which travel at nearly the shear wave - speed in the sliding direction also exist in some range of physical parameters . previous results obtained in the quasi - static regime appear relevant only within a narrow region of the parameter space . finally , we show that a finite - time regularized response to normal stress variations , within the framework of generalized rate - and - state friction models , tends to promote stability . the relevance of our results to the rupture of bi - material interfaces is briefly discussed . friction , bi - material interfaces , dynamics instabilities , rupture , elastodynamics
stephan s quintet ( hereafter sq ) includes ngc 7317 ( e ) , binary galaxies ngc 7318a ( e ) and ngc 7318b ( sbc pec ) , sy2 galaxy ngc 7319 ( sbc pec ) , and the foreground galaxy ngc 7320 ( sd ) . two very long parallel optical tidal tails ( @xmath11 kpc ; arp & kormendy 1972 ) extend from the south end of ngc 7319 toward another galaxy ngc 7320c on the east of sq . the hi observations ( shostak et al . 1984 ; williams et al . 2002 , hereafter w02 ) show hi tails following the optical tails . a large scale shock front ( @xmath1240 kpc ) in the intragroup medium ( igm ) between ngc 7319 and ngc 7318b was first discovered by allen & hartsuiker ( 1972 ) as a radio emission ridge , then confirmed by high resolution x - ray maps ( trinchieri et al . moles et al . ( 1997 , hereafter m97 ) suggested a two - intruders scenario for the history of sq : an old intruder ( ngc 7320c ) stripped most of the gas from group members , and a new intruder ( ngc 7318b ) is currently colliding with this gas and triggered the large scale shock . there are still many uncertainties about this scenario . these include the question whether both optical tails , one of age @xmath13 yrs ( young tail ) and another of age @xmath14 @xmath15 yrs ( old tail ) , are triggered by two passages of the same galaxy ngc 7320c over ngc 7319 ( m97 ) . also , based on new hi maps which show that the hi in the region occupied by ngc 7318b is clearly separated into two clumps with distinctively different velocities , w02 challenged the conclusion that ngc 7318b was _ not _ affected by interaction until its collision with sq starting about @xmath16 yrs ago ( s01 ) . -0.3 cm -0.8 cm the star formation activity in sq is apparently very much influenced by the interactions . the most spectacular star formation region in sq is the igm starburst sq - a , which is associated with a bright mir ( 15@xmath17 ) source ( xu et al . 1999 , hearafter x99 ) just beyond the northern tip of the shock front . it is triggered by the same ngc 7318b / igm collision that triggered the shock ( xu et al . 2003 , hearafter x03 ) . the young tail , the brighter one among the two optical tails , is also active in star formation . hunsberger et al . ( 1996 ) found 13 tidal dwarf galaxy candidates along this tail . a bright hii region ( sq - b ) is detect in both h@xmath5 ( arp 1973 ) and mir ( x99 ) . recently , mendes de oliveira et al . ( 2004 , hereafter mdo04 ) discovered four intergalactic hii regions in a region north of the young tail , possibly associated with the stripped ism of ngc 7319 ( s01 ) . the h@xmath5 observations ( arp 1973 ; vilchez & iglesias - pramo 1998 ; plana et al . 1999 ; s01 ) reveal numerous huge hii regions along several arms of ngc 7318b . hunsberger et al . ( 1996 ) , iglesias - pramo & vilchez ( 2001 ) , and mendes de oliveira et al . ( 2001 ) classified them as tidal dwarf galaxy candidates , though s01 argued that the crossing time of ngc 7318b ( @xmath18 yrs ) is too short for any tidal effects . there has been no estimate of the total star formation rate in sq in the literature . therefore it is not clear whether the overall star formation activity is enhanced by the interactions . also , because a clear picture for the interaction history of sq is still missing , most results in the literature linking the star formation activity to interactions in sq were at best suggestive . in this letter we present the first uv images of sq obtained using galex ( martin et al . the galex observations are sensitive to very low levels of star formation averaged over @xmath13 yrs , the time scale of the tidal effects . this enables the first quantitative study on the overall star formation and its distribution in sq . comparisons between the uv , optical , hi , and the emission in other wavebands put new constraints on the star formation history and on the relation between star formation and galaxy - galaxy interaction . sq was observed by galex on 2003 - 08 - 23 , 2003 - 09 - 05 and 2003 - 09 - 06 in 5 orbits with total exposure time of 3327s . data reduction was done using standard galex pipeline ( ir0.2 calibration , morrissey et al . the r.m.s noise is 27.65 mag arcsec@xmath19 and 28.11 mag arcsec@xmath19 in the fuv ( 1530 ) and nuv ( 2310 ) , respectively . the uv magnitudes are in the ab system . the fwhm of the fuv beam is [email protected] , and that of the nuv beam [email protected] . -0.5 cm the galex image ( fig.1a ) looks very different from the optical image ( fig.1b ) . the foreground sd galaxy ngc 7320 is the most prominent uv source , but as ngc 7320 is not a member of sq we do not discuss it further . a striking feature of the uv emission is that there are large regions that do not coincide with significant optical light from the galaxies or optically detected tidal features . in the merger which gives rise to the antennae ( hibbard et al . 2004 ) the uv emission is seen to follow the tidal tails and merger body closely , while being of @xmath21 , of age simlar to the distributed uv emission in sq . there is an extended uv disk ( outlined by the red ellipse in the uv image ) centered between the nuclei of ngc 7318a and ngc 7318b . it is most likely to be associated with ngc 7318b and the contribution from the ngc 7318a is insignificant . this is because ngc 7318a has the optical apparence of an ordinary elliptical galaxy and no hii region or hi gas in this area is likely to be associated with it ( m97 ; s01 ) . the uv disk , with a physical size of @xmath9 kpc , is about 1@xmath22 larger ( along the major axis ) than the disk found by s01 on the b - r image . the emission peaks in the south side of the uv disk are in good agreement with those in the hi gas in this region which has the same redshift of ngc 7318b ( w02 ) . this agrees with the results of thilker et al . ( 2004 ) who found that the uv emission regions in outer disk of m83 are typically associated with hi structures . the north tip of the disk goes beyond the hi gas boundary ( fig.1c ) . this discrepancy between the uv and the cold gas may well be due to the shock . sq - a is also located within this disk . the sy2 nucleus of ngc 7319 is faint in the uv , indicating very high extinction within the agn , consistent with the sy2 classification . both the young tail and the old tail are detected . the young tail looks more like a loop in the uv image . compared to the optical tail , it extends further in the northeast direction and includes all the 4 intergalactic hii regions found by mdo04 . the uv emission within the ngc 7319 disk shows a disturbed morphology . it is interesting to note that no hi gas has been detected in ngc 7319 disk ( fig.1c ; w02 ) . the uv emission in the northeast part of the disk is along two strong optical arms . it _ does not _ coincide with the h@xmath5 emission , nor with the co emission ( fig.1c ) . the elliptical galaxy ngc 7317 is faint in the uv , and there is no indication that it has been involved in any recent interactions with other members of sq . the uv data are analyzed quantitatively to study the star formation activity in different regions in sq . the results are presented in table 1 . for both the fuv and nuv magnitudes , a calibration uncertainty of 0.1 mag is assumed ( morrissey et al . 2004 ) . the fuv magnitude , the fuv - nuv color and the b - v color in table 1 are corrected for the foreground galactic extinction ( e(b - v)=0.079 and the extinction curve of cardelli et al . 1989 ) . for regions detected in isocam 15@xmath17 map ( x99 ) , the internal extinction is determined using the luminosity ratio @xmath23 according to the following formulae : @xmath24 @xmath25 the relation @xmath26 is taken from chary & elbaz ( 2001 ) . ( 2 ) is taken from buat et al . other regions are either undetected or outside the isocam 15@xmath17 map . for them the internal extinction is estimated using the average hi column density ( w02 ) and the dust - to - gas ratio of the solar neighborhood ( @xmath27 atoms@xmath3 @xmath28 , savage and mathis 1979 ) . the foreground screen model and a modified calzetti attenuation curve ( buat et al . 2004 ) are assumed . for the extinction estimated using the hi , the error is assumed to be 100% . estimates of the internal extinction of reg . vi ( sq - a ) using the mir - to - fuv ratio and using the hi column density , respectively , are found to be consistent with each other . star formation rate is estimated using the extinction corrected fuv luminosity according to the formula of kennicutt ( 1998 ) . the age of the stellar population responsible for the uv emission is determined using the extinction corrected fuv - nuv color and the single population synthesis model ( i.e. the burst model ) of fioc & rocca - volmerange ( 1997 , pegase ) . this is compared with the age of the stellar population responsible for the optical emission , determined using the extinction corrected b - v color and the same synthesis model . * sfr of sq - a ( reg . vi ) * : the sfr of 1.34@xmath4 m@xmath2 yr@xmath3 estimated from the extinction corrected fuv luminosity is in excellent agreement with that from the extinction corrected h@xmath5 luminosity ( 1.45 m@xmath2 yr@xmath3 ) by x03 . on the other hand , the extinction estimated from the mir - to - fuv ratio , a@xmath29 , is significantly lower than that indicated by the balmer decrement ( a@xmath30 0.6 2 ) . this suggests that o stars and b stars which are responsible for the ionizing and non - ionizing uv radiation , respectively , have different distributions relative to the dust . the extinction corrected b - v color suggests an underlying stellar population of @xmath31 myr , consistent with the results of gallagher et al . ( 2001 ) . * star formation in regions related to ngc 7319 * : these include reg . i , ii , iii , and iv . among the net sfr of 1.98@xmath6 m@xmath2 yr@xmath3 , 68% is contributed by the disk ( reg . iv ) , and 15% from the young tail ( reg . ii ) . there is a disk / tidal - tail gradient among the undereddened uv colors of reg . iv , iii and ii ( fig.2b ) . however , the extinction corrected uv colors are consistent with each other ( fig.2c ) , indicating that the color gradient is mainly due to the dust reddening , and there is no significant difference in the age of the uv population in the disk and in the tail . the extinction corrected uv and optical colors of reg . i ( old tail ) are consistent with a single population of age @xmath32 yrs , somewhat younger than that estimated by s01 from dynamic arguments ( 612 @xmath33 yrs ) . ii includes 11 subregions . iia , iib and iic correspond to the optical tail . the age of the uv population in iia and iib is @xmath13 yrs while that of the optical population is @xmath34 @xmath35 yrs , indicating an old underlying population possibly stripped from the disk of ngc 7319 in the encounter . the bright h@xmath5/mir source sq - b is in the east end of iia , signaling active current star formation . iic has a much younger uv population ( @xmath18 yrs ) . it is outside our b and v image . the extragalactic hii regions found by mdo04 are also in reg . iii ( c and d in mdo04 ) shows very young ages in both uv color and optical color ( @xmath36 yrs ) , consistent with the ew(h@xmath5 ) result of mdo04 . iij , which is north of iii and is not in the list of mdo04 , shows similar properties as iii . the uv population of iie ( b in md004 ) has an age of @xmath37 , older than that estimated by mdo04 ( 5.6 myr ) from ew(h@xmath5 ) . in the uv image these subregions , together with other subregions which are discovered for the first time , form a loop - like structure . given that many of these subregions were also detected in other bands ( e.g. optical and h@xmath5 ) and available redshift data ( s01 ; mdo04 ) are consistent with they all having the same redshift , this structure is very likely real . m97 suggested that the young tail is triggered by a high velocity ( @xmath38 km / sec ) passage of ngc 7320c through ngc 7319 1.5 @xmath35 yrs ago . however , the recently measured redshift of ngc 7320c is almost identical to that of ngc 7319 , indicating instead a slow passage ( s01 ) . therefore in order for ngc 7320c to move to its current position , the ngc 7319/n7320c encounter must have occured @xmath39 yrs ago . this is close to the age of the old tail , but older than that of the young tail ( m97 ) . an alternative scenario is that the young tail is triggered by a close encounter between ngc 7319 and the elliptical galaxy ngc 7318a . since ngc 7318a is about 3 times closer to ngc 7319 than ngc 7320c , the time argument is in favor of this new scenario . also the uv loop , on the other side of ngc 7319 with regard to ngc 7318a , looks very similar to the counter tidal loop found frequently in the dynamic simulations of galaxy - galaxy interactions ( see , e.g. , fig.2 of toomre & toomre 1972 , t=2 10@xmath40 yr ) . * star formation in regions related to ngc 7318b * : these include reg . v , vii , and viii . the peaks of the uv emission in reg . v correspond to a chain of hii regions in one of star forming arms of ngc 7318b ( s01 ; gallagher et al . therefore most of the uv emission is likely associated with the star formation rather than with the shock . the total sfr is 3.37@xmath10 m@xmath2 yr@xmath3 . this is very close to the sfr of the milky way ( @xmath41 m@xmath2 yr@xmath3 , mezger 1988 ) , a normal sbc galaxy . viii includes 4 subregions , viiia and viiib are in the south end of the uv disk , and viiic and viiid in the north end . they are 30 40 kpc away from the nucleus of ngc 7318b . these are even larger than the largest distance among the uv emission regions in the extreme outer disk of m83 , studied by thilker et al . whether the remarkable size of this uv disk , the peculiar morphology of the hi gas , and the long and open arms ( tidal tails ? ) of ngc 7318b are due to a head - on collision with ngc 7318a about 10@xmath40 yrs ago ( x03 ) , is an interesting subject for future studies . the uv and optical colors of viiia are consistent with a single population of @xmath13 yrs . * summary remarks * : every time when sq is observed in a waveband for the first time , it reveals new , spectecular phenomona related to its complex interaction history . the long term fate of the star forming regions in the igm starburst sq - a , in the remarkably extended uv loop associated with the young tail , and in the very large uv disk of ngc 7318b is of great interest . this can have far - reaching inference to the interaction induced star formation in multi - galaxy systems which may play important roles in the early universe . _ acknowledgments _ : 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 of france and the korean ministry of science and technology . allen , r.j . , hartsuiker , j.w . 1972 , nature , 239 , 324 . arp , h. , kormendy , j. 1972 , apj , 178 , l101 . arp , h. 1973 , apj , 185 , 797 . buat , v. , et al . 2004 , present volume . cardelli , j.a . , clayton , g.c . , mathis , j.s . 1989 , apj , 345 , 245 . chary , r. , elbaz , d. 2001 , apj , 556 , 562 . fioc , m. , rocca - volmerange , b. 1997 , a&a , 326 , 950 . gao , y. , xu , c. 2000 , apjl , 542 , l83 . gallagher , s. , charlton , j.s . , hunsberger , s.d . , zaritsky , d. , whitmore , b.c . 2001 , aj , 122 , 163 . hibbard , j. , et al . 2004 , present volume . hunsberger , s.d . , charlton , j.c . , zaritsky , d. 1996 , apj , 462 , 90 . iglesias - pramo , j. vilchez . j.m . 2001 , apj , 550 , 204 . kennicutt , r.c . 1998 , ara&a , 36 , 189 . lisenfeld , u. , braine , j. , duc , p .- a . , leon , s. , charmandaris , v. , brinks , e. 2002 , a&a , 394 , 823 . martin , d.c . , 2004 , present volume . mendes de oliveira , c. , cypriano , e.s . , sodr , jr . , l. , , balkowski , c. 2004 , apjl , 605 , l17 ( mdo04 ) . mendes de oliveira , c. , plana , h. , amram , p. , balkowski , c. , bolte , m. 2001 , aj , 121 , 2524 . mezger , p.g . 1988 , in _ galactic and extragalactic star formation _ , eds . pudrutz & m. fich ( dordrecht : kluwer ) , 227 . moles , m. , sulentic , j.w . , mrquez , i. 1997 , , 485 , 69 ( m97 ) . morrissey , p. , et al . 2004 , present volume . arp , h. , sulentic , j.w . 1997 , , 322 , 89 . plana , h. , mendes de oliveira , c. , amram , p. , bolte , m. , balkowski , c. , boulesteix , j. 1999 , apjl , 516 , l69 . savage , b.d . , mathis , j.s . 1979 , ara&a , 17 , 73 . shostak , g.s . , sullivan iii , w.t . , allen , r.j . 1984 , , 139 , 15 . sulentic , j.w . , rosado , m. , dultzin - hacyan , d. , verdes - montenegro , l. , trinchieri , g. , xu . c. , pietsch , w. 2001 , aj , 122 , 2993 ( s01 ) . thilker , d. , et al . 2004 , present volume . toomre , a. , toomre , j. 1972 , , 178 , 623 . trinchieri , g. , sulentic , j.w . , breitschwerdt , d. , pietsch , w. 2003 , a&a , 401 , 173 . , iglesias - pramo , j. 1998 , apjs , 117 , 1 . williams , b.a . , yun , m.s . , verdes - montenegro , l. 2002 , aj , 123 , 2417 ( w02 ) . xu , c. , sulentic , j.w . , tuffs , r. 1999 , apj , 512 , 178 ( x99 ) . xu , c.k . , lu , n.y . , condon , j.j . , dopita , m. , tuffs , r. 2003 , apj , 595 , 665 ( x03 ) . llllllllll old tail & i & 19.74@xmath42 & 8.49@xmath43 & 0.060@xmath44 & 0.44@xmath4 & 8.4@xmath45 & 0.28@xmath46 & 8.5@xmath47 & 0.17@xmath48 + young tail & ii & 18.14@xmath49 & 9.18@xmath50 & 0.296@xmath51 & 0.21@xmath46 & 7.9@xmath52 & 0.45@xmath53 & 8.7@xmath52 & 0.31@xmath53 + tail / disk overlap & iii & 19.34@xmath42 & 9.17@xmath54 & 0.285@xmath55 & 0.43@xmath4 & 7.9@xmath52 & 0.69@xmath53 & 8.9@xmath52 & 1.47@xmath49 + ngc 7319 disk & iv & 18.19@xmath49 & 9.84@xmath46 & 1.341@xmath56 & 0.77@xmath46 & 8.3@xmath52 & 0.77@xmath53 & 9.0@xmath52 & 2.00@xmath57 + shock front & v & 16.87@xmath58 & 9.87@xmath59 & 1.450@xmath60 & 0.31@xmath53 & 7.9@xmath52 & 0.55@xmath53 & 8.8@xmath61 & 0.76@xmath54 + sq - a & vi & 17.32@xmath58 & 9.84@xmath59 & 1.335@xmath62 & 0.27@xmath53 & 7.4@xmath63 & 0.44@xmath53 & 8.7@xmath52 & 1.12@xmath50 + ngc 7318b inner disk & vii & 16.76@xmath58 & 9.91@xmath59 & 1.561@xmath64 & 0.39@xmath53 & 8.1@xmath52 & 0.86@xmath53 & 9.3@xmath52 & 0.74@xmath54 + ngc 7318b outer disk & viii & 17.72@xmath49 & 9.33@xmath54 & 0.360@xmath65 & 0.02@xmath46 & 7.4@xmath66 & 0.38@xmath53 & 8.7@xmath45 & 0.10@xmath58 + total & & 15.36@xmath58 & 10.54@xmath59 & 6.687@xmath67 & 0.33@xmath53 & ... & ... & ... & ... + ... & & & & & & & & +	we present the first _ galex _ uv images of the well known interacting group of galaxies , stephan s quintet ( sq ) . we detect widespread uv emission throughout the group . however , there is no consistent coincidence between uv structure and emission in the optical , @xmath0 , or hi . excluding the foreground galaxy ngc 7320 ( sd ) , most of the uv emission is found in regions associated with the two spiral members of the group , ngc 7319 and ngc 7318b , and the intragroup medium starburst sq - a . the extinction corrected uv data are analyzed to investigate the overall star formation activity in sq . it is found that the total star formation rate ( sfr ) of sq is 6.69@xmath1 m@xmath2 yr@xmath3 . among this , 1.34@xmath4 m@xmath2 yr@xmath3 is due to sq - a . this is in excellent agreement with that derived from extinction corrected h@xmath5 luminosity of sq - a . the sfr in regions related to ngc 7319 is 1.98@xmath6 m@xmath2 yr@xmath3 , most of which ( 68% ) is contributed by the disk . the contribution from the young tail is only 15% . in the uv , the young tail is more extended ( @xmath7 kpc ) and shows a loop - like structure , including the optical tail , the extragalactic hii regions recently discovered in h@xmath5 , and other uv emission regions discovered for the first time . the uv and optical colors of the old tail are consistent with a single stellar population of age @xmath8 yrs . the uv emission associated with ngc 7318b is found in a very large ( @xmath9 kpc ) disk , with a net sfr of 3.37@xmath10 m@xmath2 yr@xmath3 . several large uv emission regions are 30 40 kpc away from the nucleus of ngc 7318b . although both ngc 7319 and ngc 7318b show peculiar uv morphology , their sfr is consistent with that of normal sbc galaxies , indicating that the strength of star formation activity is not enhenced by interactions .
in a parton model picture , parton distribution functions ( pdfs ) describe combinations of number densities of quarks and gluons in a fast - moving hadron and are a crucial ingredient for our understandings of high - energy experiments involving hadrons . at leading twist , the quark structure of spin - half hadrons is described by three pdfs : the unpolarized distribution @xmath2 , the longitudinal polarization ( helicity ) distribution @xmath3 , and the transverse polarization ( transversity ) distribution @xmath4 . from the phenomenological point of view , @xmath4 is the least known one because , being chiral - odd , it can be measured only in processes with two hadrons in the initial state ( _ e.g. _ proton - proton collision ) or one hadron in the initial state and at least one hadron in the final state ( _ e.g. _ semi - inclusive dis - sidis ) . combining data on polarized single - hadron sidis , together with data on almost back - to - back emission of two hadrons in @xmath0 annihilations , the transversity distribution was extracted for the first time by the torino group @xcite . the main difficulty of such analysis lies in the factorization framework used to interpret the data , since it involves transverse momentum dependent pdfs ( tmds ) . qcd evolution of tmds must be included to analyze sidis and @xmath0 data obtained at very different scales , but the computation of its effects is still under active debate . recently , the impact of evolution on @xmath4 was estimated in the so - called css framework @xcite . alternatively , the transversity distribution can be extracted in the standard framework of collinear factorization using data on sidis with two hadrons detected in the final state . in fact , @xmath4 is multiplied with a specific chiral - odd di - hadron fragmentation function ( diff ) @xcite , which can be extracted from the corresponding @xmath0 annihilation process leading to two back - to - back pion pairs @xcite . the collinear framework allows to keep under control the evolution equations of diffs @xcite . using two - hadron sidis data on proton and deuteron targets from hermes @xcite and compass @xcite , as well as belle data for the process @xmath5 @xcite , the transversity @xmath4 was extracted for the first time in the collinear framework @xcite and the valence components of up and down quark were separated @xcite . in this contribution to the proceedings , we summarize the parametrization and the error analysis both for @xmath4 and for the diffs @xcite . we also comment on re - analyzing the @xmath5 process by including structures violating parity invariance ( p - odd structures ) . p - odd diffs arise when the fragmenting quark crosses local ( in space and time ) domains where strong interactions break parity invariance . these p - odd bubbles can be induced by topologically nontrivial solutions of qcd equations of motion , which suggest that the physical vacuum is a superposition of degenerate states , the so - called @xmath1 vacuum @xcite , and that topological fluctuations in this vacuum structure can be equivalently represented by a local p - odd term in the qcd lagrangian @xcite . p - odd diffs generate new azimuthally asymmetric terms in the cross section for the @xmath5 process , similarly to what happens for the @xmath6 one @xcite but in a collinear framework . here , we write the full structure of the leading twist cross section and we try to estimate the size of the p - odd contributions . we consider the process @xmath7 , where @xmath8 denotes the beam lepton , @xmath9 the nucleon target with mass @xmath10 and polarization @xmath11 , @xmath12 and @xmath13 the produced unpolarized hadrons with masses @xmath14 and @xmath15 , respectively ( the four - momenta are given in parentheses ) . we define the total @xmath16 and relative @xmath17 momenta of the pair , with @xmath18 and @xmath19 the momentum transferred . we define the azimuthal angles @xmath20 and @xmath21 as the angles of @xmath22 and @xmath23 , respectively , around the virtual photon direction @xmath24 . we also define the polar angle @xmath1 which is the angle between the direction of the back - to - back emission in the center - of - mass ( cm ) frame of the two hadrons , and the direction of @xmath25 in the photon - proton cm frame . then , @xmath26 and @xmath27 is a function of the invariant mass only @xcite . finally , we usually define the sidis invariants @xmath28 to leading twist , the differential cross section for the two - hadron sidis off a transversely polarized nucleon target becomes @xcite @xmath29 where @xmath30 is the fine structure constant , @xmath31 , @xmath32 , and @xmath33 with @xmath34 the fractional charge of a parton with flavor @xmath35 . the @xmath36 is the diff describing the hadronization of an unpolarized parton with flavor @xmath35 into an unpolarized hadron pair . the @xmath37 is its chiral - odd partner describing the same fragmentation but for a transversely polarized parton @xcite . diffs can be expanded in legendre polynomials in @xmath38 @xcite . after averaging over @xmath38 , only the term corresponding to the unpolarized pair being created in a relative @xmath39 state survives in the @xmath40 expansion , while the interference in @xmath41 survives for @xmath42 @xcite . without ambiguity , the two terms will be identified with @xmath40 and @xmath42 , respectively . inserting the structure functions of eq . ( [ structfunct ] ) into the cross section ( [ crosssidis ] ) , we get the single - spin asymmetry ( ssa ) @xcite @xmath43 for the specific case of production of @xmath44 pairs , isospin symmetry and charge conjugation suggest @xmath45 and @xmath46 , with @xmath47 , with also @xmath48 @xcite . moreover , from eq . ( [ sidisssa ] ) the @xmath49-dependence of transversity is more conveniently studied by integrating the @xmath50- and @xmath51-dependences of diffs . so , the actual combinations of transversity used in the sidis analysis are , for the proton target @xcite , @xmath52 and , for the deuteron target , @xmath53 \ , x f_1^{q+\bar{q}}(x ; q^2 ) \ ; , \nonumber \\ & & \label{xh1d}\end{aligned}\ ] ] where @xmath54 , @xmath55 , @xmath56 if @xmath47 , respectively ( _ i.e. _ it reflects isospin symmetry of strong interactions inside the deuteron ) , and @xmath57 the latter quantities can be determined by extracting diffs from the @xmath5 process . in fact , after the annihilation a quark and an antiquark are emitted back - to - back , each one fragmenting into a residual jet and a @xmath58 pair . the leading - twist cross section in collinear factorization , namely by integrating upon all transverse momenta but @xmath22 and @xmath59 , can be written as @xcite @xmath60 where @xmath61 is the symmetric term corresponding to the production of unpolarized quark and antiquark , and the azimuthal angles @xmath20 and @xmath62 give the orientation of the planes containing the momenta of the pion pairs with respect to the lepton plane ( see fig.1 of ref . for more details ) . the azimuthally asymmetric term is generated from the production of the quark and antiquark being transversely polarized . we define the so - called artru - collins asymmetry @xcite @xmath63 before the measurement of the artru - collins asymmetry by the belle collaboration @xcite , the only information available on diffs were coming from model calculations in the context of the spectator approximation @xcite , which have produced solid predictions successfully compared with data for the asymmetry ( [ sidisssa ] ) recently released by the compass collaboration @xcite . as a function of @xmath50 at @xmath64 gev@xmath65 and @xmath66 gev for various flavors . right panel : @xmath67 as a function of @xmath51 at @xmath64 gev@xmath65 for three different @xmath68 . [ d1],title="fig:",width=238 ] as a function of @xmath50 at @xmath64 gev@xmath65 and @xmath66 gev for various flavors . right panel : @xmath67 as a function of @xmath51 at @xmath64 gev@xmath65 for three different @xmath68 . [ d1],title="fig:",width=238 ] since a measurement of the unpolarized @xmath0 cross section @xmath61 is still missing , the unpolarized diff @xmath40 was parametrized to reproduce the two - pion yield of the pythia event generator tuned to the belle kinematics @xcite . the fitting expression at the starting scale @xmath64 gev@xmath65 was inspired by the above mentioned spectator model calculations @xcite and it contains three resonant channels ( pion pair produced by @xmath69 , @xmath70 , and @xmath71 decays ) and a continuum . for each channel and for each flavor @xmath72 , a grid of data in @xmath73 was produced using pythia for a total amount of approximately 32000 bins . each grid was separately fitted using the corresponding parametrization of @xmath40 and evolving it to the belle scale of @xmath74 gev@xmath65 . an average @xmath75 per degree of freedom ( @xmath75/dof ) of 1.62 was reached using in total 79 parameters ( see ref . for more details ) . in fig . [ d1 ] , the @xmath36 for @xmath76 flavors at @xmath64 gev@xmath65 and @xmath66 gev is reported as a function of @xmath50 ( left panel ) . then , the chiral - odd diff @xmath42 was extracted from the artru - collins asymmetry by using the above mentioned isospin symmetry and charge conjugation of diffs and by integrating upon the hemisphere of the antiquark jet @xcite . the experimental data for @xmath77 are organized in a @xmath73 grid of 64 bins @xcite . they were fitted starting from an expression for @xmath78 at @xmath64 gev@xmath65 with 9 parameters , and then evolving it to the belle scale . the final @xmath75/dof was 0.57 @xcite . in the right panel of fig . [ d1 ] , the ratio @xmath79 at @xmath64 gev@xmath65 is reported as a function of @xmath51 for three different @xmath68 ( for a more detailed discussion also of errors , see ref . ) . the knowledge of diffs in eq . ( [ xh1p ] ) allowed us to get a glimpse of the combination @xmath80 directly from the hermes data for @xmath81 @xcite . recently , the compass collaboration has released new data for @xmath81 on a proton target and for @xmath82 on a deuteron target @xcite . thus , the combination of eqs . ( [ xh1p ] ) and ( [ xh1d ] ) made it possible to separately parametrize for the first time each valence flavor of the transversity distribution @xcite . here , we summarize the fitting procedure and error analysis , and comment some significant results . the main theoretical constraint on transversity is soffer s inequality @xcite . if the soffer bound is fulfilled at some initial @xmath83 , it will hold also at higher @xmath84 . we impose this condition by multiplying the functional form by the corresponding soffer bound @xmath85 at the starting scale @xmath64 gev@xmath65 . the soffer bound is built from the mstw08 set @xcite for @xmath86 , and from the dssv parameterization @xcite for @xmath87 . our analysis was carried out at lo in @xmath88 ; the explicit form of sb@xmath89 is reported in appendix of ref . . the functional form for the valence transversity distributions at @xmath90 gev@xmath65 reads @xmath91\ , x \ , \bigl [ \mbox{\small sb}^q(x ; q_0 ^ 2)+ \mbox{\small sb}^{\bar q}(x ; q_0 ^ 2)\bigr ] \ , . \label{funct_form}\ ] ] the hyperbolic tangent is such that the soffer bound is always fulfilled . the low-@xmath49 behavior is determined by the @xmath92 term , which is imposed by hand to grant the integrability of eq . ( [ funct_form ] ) and a finite tensor charge . present fixed - target data do not allow to constrain it . the functional form is very flexible and can contain up to three nodes . here , we show the results employing all the parameters , the so - called _ extra - flexible scenario _ ( for results with other choices , see refs . ) . the fit and the error analysis were carried out in two different ways : using the standard hessian method and using a monte carlo approach . the latter was introduced because it does not rely on the assumptions of a quadratic dependence of @xmath75 and a linear expansion of the theoretical quantity around the minimum , which are prerequisites for the hessian method . this freedom is essential whenever the minimization pushes the theoretical function towards its upper or lower bounds . the monte carlo approach is inspired to the work of the nnpdf collaboration @xcite , although our results are not based on a neural - network fit . the approach consists in creating @xmath9 replicas of the data points by shifting them by a gaussian noise with the same variance as the measurement . each replica , therefore , represents a possible outcome of an independent measurement . then , the standard minimization procedure is applied to each replica separately ( for details , see ref . ) . the number of replicas is chosen so that the mean and standard deviation of the set of replicas accurately reproduces the original data points . in our case , it turns out to be 100 . the @xmath9 theoretical outcomes can have any distribution , not necessarily gaussian . in which case , the @xmath93 confidence interval is in general different from the 68% interval . the latter can be easily obtained by rejecting the largest and lowest 16% of replicas for each experimental point . contribution to the transversity in the _ extra - flexible scenario_. the standard hessian method gives the uncertainty band with solid boundaries , the central thick solid line visualizing the central value . the band with dashed boundaries is the outcome when adopting the monte carlo approach . the band with short - dashed boundaries is the parametrization obtained in ref . 1 from the collins effect . the dark thick solid lines indicate the soffer bound . all results at the scale of @xmath94 gev@xmath65 . right panel : same notations for the @xmath95 contribution . [ xh1],title="fig:",width=226 ] contribution to the transversity in the _ extra - flexible scenario_. the standard hessian method gives the uncertainty band with solid boundaries , the central thick solid line visualizing the central value . the band with dashed boundaries is the outcome when adopting the monte carlo approach . the band with short - dashed boundaries is the parametrization obtained in ref . 1 from the collins effect . the dark thick solid lines indicate the soffer bound . all results at the scale of @xmath94 gev@xmath65 . right panel : same notations for the @xmath95 contribution . [ xh1],title="fig:",width=226 ] in fig . [ xh1 ] , the left panel displays the @xmath96 contribution in eq . ( [ funct_form ] ) for the _ extra - flexible scenario _ , while @xmath97 is in the right one . for each panel , the outcome with the standard hessian method is represented by the uncertainty band with solid boundaries , the central thick solid line visualizing the central value . the partially overlapping band with dashed boundaries is the outcome when adopting the monte carlo approach , where the band width corresponds to the @xmath98 of all the 100 replicas , produced by rejecting the largest and lowest @xmath99 among the replicas values in each @xmath49 point . for sake of comparison , each panel displays also the corresponding results in the framework of the collins effect @xcite , depicted as a band with short - dashed boundaries . since the latter was extracted at the scale @xmath100 gev@xmath65 , our results are properly evolved at the same scale . finally , the dark thick solid lines indicate the soffer bound , also evolved at the same scale . the uncertainty bands in the standard and monte carlo approaches are quite similar within the @xmath49-range where data exist , namely @xmath101 , and are in agreement with the other extraction based on the collins effect . outside this range , the standard approach tends to saturate the soffer bound , and the boundaries of the band can occasionally cross it because the assumed quadratic dependence of @xmath75 on the parameters around its minimum is not reliable . on the contrary , in the monte carlo approach each replica is built such that it never violates the soffer bound . for @xmath102 , the replicas entirely fill the area between the upper and lower soffer bounds , giving an explicit visualization of the realistic degree of uncertainty where there are no experimental data points . in the right panel , for @xmath103 our results tend to saturate the lower limit of the soffer bound because they are driven by the compass deuteron data , in particular by the bins number 7 and 8 . no such trend is evident in the parametrization corresponding to the single - hadron measurement of the collins effect . as a matter of fact , this is the only source of significant discrepancy between the two extractions . this statement remains valid also for the other _ scenarios _ , indicating that this is not an artifact of the chosen functional form . the tensor charge , a fundamental quantity of hadrons at the same level as the vector , axial , and scalar charges , is related to the transversity by @xmath104 the region of validity of our fit is restricted to the experimental data range . we can therefore give a reliable estimate for the tensor charge by truncating the integral to the interval @xmath105 . for the _ extra - flexible scenario _ , we find at @xmath64 gev@xmath65 @xmath106 these results are compatible with other _ scenarios _ and , within errors , with the extraction in the framework of the single - hadron collins effect @xcite . we tried also to extend the range of integration outside the experimental data to @xmath107 . the result is heavily influenced by the adopted functional form , in particular by the low-@xmath49 exponent , and this fact reflects again the uncertainty because of missing data at very low @xmath49 . qcd equations of motion possess topologically nontrivial solutions which suggest that the physical vacuum is a superposition of degenerate states differing by their topological charge . this vacuum structure is conventionally addressed as the @xmath1 vacuum @xcite . it can be reflected in the qcd lagrangian by introducing a term proportional to the parameter @xmath1 that globally breaks p and cp symmetries of qcd . experimental findings indicate that @xmath1 is very small . nevertheless , topological fluctuations in this vacuum structure can be equivalently represented by making @xmath1 varying in space and time , therefore locally breaking p and cp @xcite . consequently , during fragmentation a quark can cross local domains ( p - odd bubbles ) where p - invariance of strong interactions is broken . the most general lorentz - invariant decomposition of the leading - twist quark - quark correlator for fragmentation in a pair of pseudoscalar mesons , depicted in ref . , can then be generalized to include terms violating parity invariance . in the collinear case , it reads @xmath108 \frac{1}{4 } \gamma^+ \nonumber \\ & & + \frac{1}{4\pi } \left [ \tilde{g}_1 ( z , \zeta , { \bf r}_t ) \gamma_5 + \gamma_5 \frac{{\kern 0.2 em r \kern -0.56 em \raisebox{0.3ex}{/}}_t}{m_h } \tilde{h}_1^\sphericalangle ( z , \zeta , { \bf r}_t ) \right ] \frac{1}{4 } \gamma^+ \ ; , \label{deltapv}\end{aligned}\ ] ] where @xmath109 . @xmath110 and @xmath111 are new p - odd diffs . similarly to what happens in the single - hadron fragmentation @xcite , they are responsible for a new azimuthally asymmetric term in the cross section for the @xmath5 process , but in a collinear framework . in fact , after averaging over @xmath112 the leading - twist cross section for the semi - inclusive production of collinear pairs reads @xmath113 where each diff now depends on @xmath114 or @xmath115 , and @xmath116 is the angle between the annihilation direction and the @xmath117 axis ( see fig.1 of ref . ) . in a model calculation based on the spectator approximation , both @xmath118 and @xmath111 have the same size and arise from the interference of pion pairs being produced with relative partial waves with @xmath119 . but the former is related to the imaginary part of this interference , while the latter is given by the real part . as such , @xmath111 displays a node at @xmath120 , the mass of the @xmath69 resonance . in eq . ( [ e+e - totxsectlm ] ) , a new azimuthally asymmetric term appears that linearly depends on the p - odd @xmath111 ( or its antiquark partner ) . the size of the effect should be proportional to @xmath121 . the ratio of p - odd diffs is of the same size as that of the p - even partners @xmath122 , as mentioned above , and it turns out to be of the order 0.4 @xcite . from ref . , @xmath123 is proportional to the average gradient of the @xmath1 field . using the instanton vacuum model , kang and kharzeev estimate it to be approximately equal to 10 mev . then , we get @xmath124 . we acknowledge many useful discussions with d. boer , who inspired the last section on p - odd fragmentation , and a. bianconi , who took part in diff studies at several steps . this work is partially supported by the joint research activity `` study of strongly interacting matter '' ( acronym hadronphysics3 , grant agreement no . 283286 ) under the 7th framework programme of the european community .	we summarize the latest achievements about the extraction of the transversity parton distribution and proton tensor charge based on an analysis of pion - pair production in deep - inelastic scattering off transversely polarized targets . recently released data for proton and deuteron targets by hermes and compass allow for a flavor separation of the valence components of transversity . at variance with the collins effect , this extraction is performed in the framework of collinear factorization and relies on di - hadron fragmentation functions . the latter have been taken from the first recent analysis of the semi - inclusive production of two pion pairs in back - to - back jets in @xmath0 annihilation . we also comment on the possibility of isolating new azimuthally asymmetric correlations of opposite pion pairs , which could arise when a fragmenting quark crosses parity - odd domains localized in minkowski space - time and induced by the topologically nontrivial qcd background ( the so - called @xmath1 vacuum ) .
[ [ section ] ] let @xmath1 be a simple algebraic group scheme over a field @xmath11 of prime characteristic @xmath12 which is defined and split over the prime field @xmath13 , and @xmath14 denote the frobenius map . the fixed points of the @xmath15th iterate of the frobenius map , denoted @xmath16 , is a finite chevalley group where @xmath17 denotes the finite field with @xmath18 elements . an elusive problem of major interest has been to determine the cohomology ring @xmath19 . until recently , aside from small rank cases , it was not even known in which degree the first non - trivial cohomology class occurs . this present paper is a sequel to @xcite where we began investigating three related problems of increasing levels of difficulty : .25 cm ( 1.1.1 ) determining vanishing ranges : finding @xmath20 such that the cohomology group @xmath21 for @xmath22 . .15 cm ( 1.1.2 ) locating the first non - trivial cohomology class : finding a @xmath23 satisfying ( 1.1.1 ) such that @xmath24 a @xmath23 satisfying this property will be called a _ sharp bound_. .15 cm ( 1.1.3 ) determining the least non - trivial cohomology : for a sharp @xmath23 as in ( 1.1.2 ) compute @xmath25 . .25 cm vanishing ranges ( 1.1.1 ) were found in earlier work of quillen @xcite , friedlander @xcite and hiller @xcite . sharp bounds ( 1.1.2 ) were later found by friedlander and parshall for the borel subgroup @xmath26 of the @xmath27 , and conjectured for the general linear group by barbu @xcite . a more detailed discussion of these results can be found in ( * ? ? ? * section 1.1 ) . in @xcite , for simple , simply connected @xmath1 and primes @xmath12 larger than the coxeter number @xmath28 , we proved that @xmath29 for @xmath30 . this provided an answer to ( 1.1.1 ) and improved on hiller s bounds @xcite . for a group with underlying root system of type @xmath31 , we demonstrated that @xmath32 is in fact a sharp bound , answering ( 1.1.2 ) . the first non - vanishing cohomology , as in ( 1.1.3 ) , was also determined . for type @xmath33 , questions ( 1.1.2 ) and ( 1.1.3 ) were also answered , where the @xmath34 cases required the prime to be at least twice the coxeter number . our methods also yielded a proof of barbu s conjecture ( * ? ? ? * conjecture 4.11 ) . in this paper , we continue these investigations in two directions . first we consider the case that @xmath1 is a group of adjoint type ( as opposed to simply connected ) . for such @xmath1 with @xmath35 , when the root system is simply laced , one obtains a uniform sharp bound of @xmath36 answering ( 1.1.2 ) ( cf . corollary [ c : simplylacedadjoint ] ) . the same uniform bound also holds for the adjoint versions of the twisted groups of types @xmath37 , @xmath23 , and @xmath38 when @xmath39 . we then consider the remaining types in the simply connected case . for @xmath1 being simple , simply connected and having root system of type @xmath40 with @xmath35 , ( 1.1.2 ) and ( 1.1.3 ) are answered ( cf . theorem [ t : dsummary2 ] ) . for type @xmath41 , ( 1.1.2 ) is answered for all primes @xmath39 ( with the exceptions of @xmath42 for type @xmath38 ) ; cf . theorems [ t : e6q - results ] , [ t : e7summary - r ] , and [ t : e8summary ] . the calculations for the non - simply - laced groups are considerably more complicated . for type @xmath43 we answer ( 1.1.2 ) when @xmath44 and @xmath45 , see theorem [ t : bsumm ] . some discussion of the situation for types @xmath46 and @xmath47 is given in sections 7 and 8 respectively . for @xmath48 and @xmath35 , we find improved answers to ( 1.1.1 ) ; cf . theorem [ t : g2summary ] and theorem [ t : f4van ] . finding an answer to ( 1.1.2 ) and ( 1.1.3 ) continues to be elusive in these types although some further information towards answering these questions is obtained . throughout this paper , we will follow the notation and conventions given in the standard reference @xcite . @xmath1 will denote a simple , simply connected algebraic group scheme which is defined and split over the finite field @xmath49 with @xmath12 elements ( except in section [ s : adjoint ] where @xmath1 is assumed to be of adjoint type rather than simply connected ) . throughout the paper let @xmath11 be an algebraically closed field of characteristic @xmath12 . for @xmath50 , let @xmath51 be the @xmath15th frobenius kernel of @xmath1 and @xmath16 be the associated finite chevalley group . let @xmath52 be a maximal split torus and @xmath53 be the root system associated to @xmath54 . the positive ( resp . negative ) roots are @xmath55 ( resp . @xmath56 ) , and @xmath57 is the set of simple roots . let @xmath43 be a borel subgroup containing @xmath52 corresponding to the negative roots and @xmath58 be the unipotent radical of @xmath43 . for a given root system of rank @xmath59 , the simple roots will be denoted by @xmath60 ( via the the bourbaki ordering of simple roots ) . for type @xmath61 , @xmath62 denotes the unique short simple root and for type @xmath31 , @xmath62 denotes the unique long simple root . the highest ( positive ) root will be denoted @xmath63 , and for root systems with multiple root lengths , the highest short root will be denoted @xmath64 . let @xmath65 denote the weyl group associated to @xmath53 , and , for @xmath66 , let @xmath67 denote the length of the element @xmath68 ( i.e. , number of elements in a reduced expression for @xmath68 ) . let @xmath69 be the euclidean space associated with @xmath53 , and the inner product on @xmath70 will be denoted by @xmath71 . let @xmath72 be the coroot corresponding to @xmath73 . the fundamental weights ( basis dual to @xmath74 ) will be denoted by @xmath75 , @xmath76 , , @xmath77 . let @xmath78 be the integral weight lattice spanned by these fundamental weights . the set of dominant integral weights is denoted by @xmath79 . for a weight @xmath80 , set @xmath81 where @xmath82 is the longest word in the weyl group @xmath65 . by @xmath83 we mean the dot " action of @xmath65 on @xmath78 , with @xmath84 being the half - sum of the positive roots . for @xmath85 , @xmath86 denotes the reflection in the hyperplane determined by @xmath87 . for a @xmath1-module @xmath88 , let @xmath89 be the module obtained by composing the underlying representation for @xmath88 with @xmath90 . moreover , let @xmath91 denote the dual module . for @xmath92 , let @xmath93 be the induced module and @xmath94 be the weyl module of highest weight @xmath95 . [ [ section-1 ] ] we will employ the basic strategy used in @xcite in addressing ( 1.1.1)-(1.1.3 ) which uses effective techniques developed by the authors which relate @xmath96 to extensions over @xmath1 via an induction functor @xmath97 . when @xmath97 is applied to the trivial module @xmath11 , @xmath98 has a filtration with factors of the form @xmath99 ( * ? ? ? * proposition 2.4.1 ) . the @xmath1-cohomology of these factors can be analyzed by using the lyndon - hochschild - serre ( lhs ) spectral sequence involving the frobenius kernel @xmath100 ( cf . * section 3 ) ) . in particular for @xmath44 , we can apply the results of kumar - lauritzen - thomsen @xcite to determine the dimension of a cohomology group @xmath101 , which can in turn be used to determine @xmath102 . the dimension formula involves the combinatorics of the well - studied kostant partition function . this reduces the question of the vanishing of the finite group cohomology to a question involving the combinatorics of the underlying root system @xmath53 . for root systems of types @xmath37 and @xmath103 the relevant root system combinatorics was analyzed in ( * ? ? ? * sections 5 - 6 ) . in the cases of the other root systems ( @xmath43 , @xmath23 , @xmath104 , @xmath105 , @xmath1 ) the combinatorics is much more involved and we handle these remaining cases in sections 4 - 8 . in this section , for the convenience of the reader , we state the key results from @xcite which will be used throughout this paper . [ [ observ ] ] we first record here a formula for @xmath106 that will be used at various times in the exposition ( * ? ? ? * observation 2.1 ) : [ o : uniquedecomp ] if @xmath107 admits a reduced expression @xmath108 with @xmath109 and @xmath110 , then @xmath111 moreover , this is the unique way in which @xmath106 can be expressed as a sum of distinct positive roots . let @xmath112 . the functor @xmath97 is exact and one can use frobenius reciprocity to relate extensions over @xmath1 with extensions over @xmath16 ( * ? ? ? * proposition 2.2 ) . [ p : extcomparison ] let @xmath113 be rational @xmath1-modules . then , for all @xmath114 , @xmath115 in order to make the desired computations of cohomology groups , we will make use of proposition [ p : extcomparison ] ( with @xmath116 ) . in addition , we will use a special filtration on @xmath117 ( cf . * proposition 2.4.1 ) ) . [ p : sections ] the induced module @xmath117 has a filtration with factors of the form @xmath118 with multiplicity one for each @xmath92 . the filtration from proposition [ p : sections ] allows one to obtain a condition on @xmath1-cohomology which leads to vanishing of @xmath119-cohomology ( cf . * corollary 2.6.1 ) ) . [ p : vanishing ] let @xmath120 be the least positive integer such that there exists @xmath121 with @xmath122 . then @xmath123 for @xmath124 . while the identification of an @xmath120 satisfying proposition [ p : vanishing ] gives a vanishing range as in ( 1.1.1 ) , it does not a priori follow that @xmath125 . the following theorem provides conditions which assist with addressing ( 1.1.2 ) or ( 1.1.3 ) ( * ? ? ? * theorem 2.8.1 ) . [ t : nonvanishing ] let @xmath120 be the least positive integer such that there exists @xmath126 with @xmath127 . let @xmath92 be such that @xmath128 . suppose @xmath129 for all @xmath130 that are linked to @xmath131 . then * @xmath132 for @xmath124 ; * @xmath125 ; * if , in addition , @xmath133 for all @xmath134 with @xmath135 , then @xmath136 from the filtration on @xmath117 in proposition [ p : sections ] , @xmath137 can be decomposed as a direct sum over linkage classes of dominant weights . for a fixed linkage class @xmath138 , let @xmath120 be the least positive integer such that there exists @xmath139 with @xmath127 . let @xmath140 be such that @xmath128 . suppose @xmath129 for all @xmath130 in @xmath138 . then analogous to theorem [ t : nonvanishing ] , it follows that @xmath125 . see ( * ? ? ? * theorem 2.8.2 ) . from sections [ ss : vanishing ] and [ ss : nonvanishing ] , the key to understanding the vanishing of @xmath102 is to understand @xmath142 for all dominant weights @xmath131 . for @xmath48 , these groups can be related to @xmath141-cohomology groups ( cf . * lemma 3.1 ) ) . [ l : g1cohom ] suppose @xmath35 and let @xmath143 . then for all @xmath144 @xmath145 we remark that the aforementioned lemma would hold for arbitrary @xmath15th - twists if it was known that the cohomology group @xmath146 admits a good filtration , which is a long - standing conjecture of donkin . for @xmath35 , this is known for @xmath48 by results of andersen - jantzen @xcite and kumar - lauritzen - thomsen @xcite . in that case , the lemma is only needed when @xmath147 . for arbitrary @xmath15 we can often work inductively from the @xmath48 case . this requires slightly more general @xmath148-computations and the possibility that @xmath149 . from lemma [ l : g1cohom ] , for @xmath126 , the cohomology group @xmath150 can be identified with @xmath151 . it is well - known that , from block considerations , @xmath152 unless @xmath153 for @xmath107 and @xmath154 . for @xmath35 , from @xcite and @xcite , we have @xmath155 where @xmath156 . note also that , since @xmath35 and @xmath157 is dominant , @xmath158 must also be dominant . for a dominant weight @xmath159 , observe that , from lemma [ l : g1cohom ] and ( [ ind ] ) , we have @xmath160 hence , if @xmath161 , then @xmath162 must be a sum of @xmath163 positive roots . for a weight @xmath157 and @xmath164 , let @xmath165 denote the dimension of the @xmath157-weight space of @xmath166 . equivalently , for @xmath167 , @xmath165 denotes the number of times that @xmath157 can be expressed as a sum of exactly @xmath59 positive roots , while @xmath168 . the function @xmath169 is often referred to as _ kostant s partition function_. by using @xcite , ( * ? ? ? * thm 2 ) , lemma [ l : g1cohom ] , and ( [ ind ] ) , we can give an explicit formula for the dimension of @xmath170 ( cf . * proposition 3.2.1 , corollary 3.5.1 ) ) . [ p : kostantpartcohom ] let @xmath39 and @xmath171 . then @xmath172 the following gives a fundamental constraint on non - zero @xmath173 such that @xmath174 it is stated in a more general @xmath148-context as it will also be used in some inductive arguments for @xmath34 ( cf . * proposition 3.4.1 ) ) . [ p : degreebound1 ] let @xmath35 with @xmath175 , both non - zero , such that @xmath176 with @xmath177 and @xmath178 for @xmath179 . assume @xmath180 . * let @xmath181 . if @xmath53 is of type @xmath46 , assume that @xmath182 is a long root . then @xmath183 * if @xmath63 denotes the longest root in @xmath184 , then @xmath185 equality requires that @xmath186 and @xmath187 . * if @xmath188 , then @xmath189 . proposition [ p : degreebound1 ] can be generalized to the following ( cf . * proposition 4.3.1 ) ) . [ p : degreebound2 ] let @xmath35 , @xmath190 and @xmath114 . if @xmath191 , then there exists a sequence of non - zero weights @xmath192 such that @xmath193 for some @xmath194 and nonzero @xmath177 . moreover , for each @xmath195 , there exists a nonnegative integer @xmath196 with @xmath197 and @xmath198 . furthermore , @xmath199 equality requires that @xmath200 and that @xmath201 for all @xmath202 . note that the assumption @xmath203 in the proposition can be replaced by @xmath204 where @xmath205 . in that case one arrives at the same conclusions with @xmath206 . in this section we obtain some general vanishing information for those cases when the root system @xmath53 is simply laced . in such cases , the longest root @xmath63 and the longest short root @xmath64 coincide . following the discussion in section 2 , we want to consider when @xmath207 for @xmath208 and @xmath92 . [ [ section-2 ] ] to gain information on such cohomology groups , we will use lemma [ l : g1cohom ] and ( [ ind ] ) . the following proposition will aid us in showing that certain cohomology groups are non - zero . [ p : homcalculation ] let @xmath209 denote the longest root of @xmath53 and @xmath210 be a nonnegative integer . then @xmath211 the claim follows from the diagram below and the fact that all modules in the commutative diagram below have a one - dimensional highest weight space with weight @xmath212 . the first embedding is a consequence of the fact that the module @xmath213 has a weyl module filtration . the weyl module @xmath214 is isomorphic to the dual of the adjoint representation , @xmath215 . clearly , @xmath215 maps onto @xmath216 and @xmath217 maps onto @xmath218 ( of weight @xmath63 ) as @xmath43-modules . hence , we obtain the two @xmath43-surjections in the first line of the diagram . the remaining maps and the commutativity of the diagram arise via the universal property of the induction functor . ( 340,60)(0,0 ) ( 5,50)@xmath219 ( 63,50)@xmath220 ( 80,50)@xmath221 ( 190,50)@xmath222 ( 170,50)@xmath223 ( 272,50)@xmath224 ( 290,50)@xmath225 ( 280,5)@xmath226 ( 310,30)@xmath227 ( 170 , 35)(4,-1)100 [ [ section-3 ] ] for a @xmath1-module @xmath228 and a dominant weight @xmath229 let @xmath230_{\gamma}$ ] denote the unique maximal summand of @xmath228 whose composition factors have highest weights linked to @xmath231 . [ l : simplylacedcohom ] assume that the root system @xmath53 of @xmath1 is simply laced . let @xmath63 denote the longest root and define @xmath232 then * for any non - zero dominant weight @xmath158 linked to zero we have + @xmath233 whenever @xmath234 ; * for any non - zero dominant weight @xmath158 linked to zero we have + @xmath235 whenever @xmath236 ; * @xmath237_0 \cong \begin{cases } h^0({\lambda } ) & \mbox { if } i = 2p-3\\ 0 & \mbox { if } 0 < i < 2p-3 . \end{cases}$ ] we apply proposition [ p : degreebound2 ] . note that @xmath158 being linked to zero forces all the weights @xmath238 of proposition [ p : degreebound2 ] to be in the root lattice . this forces @xmath239 . parts ( a ) and ( b ) now follow from equation ( [ rbound ] ) and the remark in section [ ss : degreebounds ] . for part ( c ) we make use of proposition [ p : homcalculation ] with @xmath240 and conclude that @xmath241 note that in the simply laced case @xmath242 which combined with ( [ ind ] ) yields @xmath243 the weight @xmath131 is the smallest non - zero dominant weight in the zero linkage class . any other non - zero weight @xmath158 in the linkage class will be of the form @xmath244 , where @xmath245 is a non - zero sum of positive roots . clearly @xmath158 can not be a weight of @xmath246 whenever @xmath247 hence , @xmath248 for all @xmath249 since @xmath250 has a good filtration one obtains @xmath251_0 \cong h^0({\lambda}).\ ] ] part ( b ) now implies that @xmath252_0=0 $ ] whenever @xmath253 [ t : firstnontrivial ] assume that the root system of @xmath1 is simply laced . then @xmath254 let @xmath158 be a weight in the zero linkage class . from lemma [ l : simplylacedcohom](a ) , we know that @xmath255 for @xmath256 . let @xmath257 . we next show by induction on @xmath15 that @xmath258 . for @xmath48 , this follows from lemma [ l : g1cohom ] and lemma [ l : simplylacedcohom](c ) . for @xmath34 , we look at the lyndon - hochschild - serre spectral sequence @xmath259 lemma [ l : simplylacedcohom](b ) implies that the @xmath260 for @xmath236 . note that @xmath261 lemma [ l : simplylacedcohom](c ) implies that @xmath262 for @xmath263 , and , moreover , that @xmath264 is a summand of @xmath265 . hence , ( cf . * lemma 5.4 ) ) , @xmath266 has a summand isomorphic to @xmath267 . by induction , this ext - group is non - zero , and hence @xmath268 and transgresses to the @xmath269-page , which implies that @xmath270 since @xmath131 is the lowest non - zero dominant weight in the zero linkage class , the claim now follows by applying the argument given in section [ ss : nonvanishing ] to the weight @xmath131 and the zero linkage class . in this section we assume that @xmath1 is simply laced and of adjoint type . the fixed points of the @xmath15th iterated frobenius map on @xmath1 will again be denoted by @xmath16 . for example , if @xmath1 is the adjoint group of type @xmath37 then @xmath16 is the projective linear group with entries in the field with @xmath271 elements . propositions [ p : extcomparison ] and [ p : sections ] can also be applied to groups of adjoint type . note that the root lattice and the weight lattice coincide in this case . therefore all dominant weights of the form @xmath272 are automatically in the zero linkage class . from lemma [ l : simplylacedcohom ] , theorem [ t : firstnontrivial ] , and proposition [ p : vanishing ] , one obtains the following corollary . [ c : simplylacedadjoint ] assume that the root system @xmath53 of @xmath1 is simply laced and that @xmath1 is of adjoint type . then * @xmath273 for @xmath274 ; * @xmath275 . note that , for type @xmath33 and @xmath276 and @xmath277 being relatively prime , the adjoint and the universal types of the finite groups coincide . in these cases the above claim was already observed in theorems 6.5.1 and 6.14.1 of @xcite . let @xmath1 be of type @xmath37 , @xmath23 , @xmath38 and of adjoint type . let @xmath245 denote an automorphism of the dynkin diagram of @xmath1 . then @xmath245 induces a group automorphism of @xmath1 that commutes with the frobenius morphism , which we will also denote by @xmath245 . let @xmath278 be the finite group consisting of the fixed points of @xmath245 composed with @xmath105 . note that @xmath245 fixes the maximal root @xmath209 . therefore the discussion in this section also applies to the twisted groups @xmath278 of adjoint type . in particular , corollary [ c : simplylacedadjoint ] holds for these groups as well . assume throughout this section that @xmath53 is of type @xmath40 , @xmath280 , and that @xmath281 . following section 2 , our goal is to find the least @xmath282 such that @xmath283 for some @xmath131 . suppose that @xmath284 for some @xmath282 and @xmath285 with @xmath286 and @xmath107 . from proposition [ p : degreebound1](c ) , @xmath287 . for a fundamental dominant weight @xmath288 , @xmath289 therefore , if @xmath290 , we will have @xmath291 and @xmath292 . this reduces us to analyzing the cases when @xmath293 , we consider first the case that @xmath294 and obtain the following restrictions . [ l : omega1 ] suppose @xmath53 is of type @xmath40 with @xmath279 and @xmath295 . suppose @xmath296 with @xmath107 . then * @xmath297 for @xmath298 ; * if @xmath299 , then @xmath300 . following the discussion in section [ ss : g1analysis ] , @xmath301 must be a weight of @xmath302 for @xmath303 . recall that @xmath304 . consider the decomposition of @xmath106 into a sum of @xmath67 distinct positive roots ( see observation [ o : uniquedecomp ] ) . write @xmath305 where @xmath306 is the number of positive roots in this decomposition that contain @xmath307 and @xmath308 is the number of roots in this decomposition that do not contain @xmath307 . then @xmath309 contains @xmath310 copies of @xmath307 . since any root contains at most one copy of @xmath307 , we have @xmath311 replacing @xmath67 by @xmath312 and simplifying gives @xmath313 the total number of positive roots containing an @xmath307 is @xmath314 . since we necessarily then have @xmath315 , we can rewrite the above as @xmath316 since @xmath317 . this proves part ( a ) . furthermore , we see that @xmath318 if and only if @xmath319 and @xmath320 . in other words , when @xmath106 is expressed as a sum of distinct positive roots , it consists precisely of all @xmath321 roots which contain an @xmath307 . that is , @xmath322 , which gives part ( b ) . we will show in proposition [ p : omega1summary ] that @xmath323 for @xmath324 . to do this , we will make use of proposition [ p : kostantpartcohom ] . we first make some observations about relevant partition functions . note that @xmath325 for @xmath53 of type @xmath40 , with @xmath279 , and integers @xmath326 , we set @xmath327 note that @xmath328,\ ] ] when @xmath329 . [ l : omega1combinatorics ] suppose @xmath53 is of type @xmath40 with @xmath279 and @xmath330 . * @xmath331 whenever @xmath332 . * @xmath333 * @xmath334 , for @xmath335 * @xmath336 * @xmath337 , for @xmath338 * @xmath339 for @xmath340 and @xmath341 even . \(a ) the weight @xmath342 , written as a sum of simple roots , contains one copy of @xmath343 . assume that @xmath344 and @xmath279 . recall that @xmath345 . note that @xmath346 is a sum of positive roots if and only if @xmath347 therefore , @xmath348 . if @xmath349 then @xmath350 in addition @xmath351 , written as a sum of simple roots , contains no @xmath343 . hence , @xmath352 , written as a sum of simple roots , contains exactly @xmath120 copies of @xmath343 . each positive root of @xmath53 contains at most one copy of @xmath343 . therefore at least @xmath120 positive roots are needed to sum up to @xmath352 . one concludes that @xmath353 for @xmath354 . + + ( b ) again @xmath355 is a sum of positive roots only if @xmath356 . therefore , @xmath357 ( c ) for the expression @xmath358 to be a sum of positive roots one needs either @xmath359 or @xmath360 and @xmath361 . set @xmath362 then @xmath363 + + ( d ) we make use of the fact that @xmath364 is a minuscule weight and obtain : @xmath365\\ & = & [ { \operatorname{ch}}h^0(g / b , s^k({\mathfrak{u}}^)\otimes - \epsilon_1)):{\operatorname{ch}}h^0(m\epsilon_1)]\\ & = & [ { \operatorname{ch}}h^0(g / b , s^{k-2n+2}({\mathfrak{u}}^)\otimes \epsilon_1)):{\operatorname{ch}}h^0(m\epsilon_1 ) ] \qquad ( \text{by \cite[lemma 6]{klt}})\\ & = & \sum_{u \in w } ( -1)^{\ell ( u ) } p_{k-2n-2}(u\cdot m \epsilon_1- \epsilon_1 ) \qquad ( \text{by \cite[3.8]{aj}})\\ & = & p(m-1,k-2n+2,n ) \qquad ( \text{by ( b)}).\end{aligned}\ ] ] + + part ( e ) now follows directly from ( c ) and ( d ) . + + ( f ) if @xmath366 , it follows from ( e ) and ( a ) that @xmath367 so the claim holds if it holds for @xmath368 . if @xmath369 and @xmath370 , then ( e ) has to be replaced by @xmath371 note that the both sides of the equation are zero unless @xmath120 is even . for even @xmath120 , a direct computation similar to that in the proof of ( * lemma 6.11 ) shows that @xmath372 [ p : omega1summary ] suppose @xmath53 is of type @xmath40 with @xmath279 . assume that @xmath373 . let @xmath374 then @xmath375 from the previous discussion we know that @xmath376 is of the form @xmath377 with @xmath378 . set @xmath379 . from proposition [ p : kostantpartcohom ] and lemma [ l : omega1combinatorics](b ) , one concludes @xmath380= p(p-2n+1,k , n).\ ] ] by lemma [ l : omega1combinatorics](a ) , this expression is zero unless @xmath381 and lemma [ l : omega1combinatorics](d ) it follows that @xmath382 . replacing @xmath11 by @xmath383 and solving for @xmath173 yields the claim . we now consider the case that @xmath385 or @xmath386 for @xmath107 with @xmath387 . [ l : omegan]suppose @xmath53 is of type @xmath40 with @xmath279 and @xmath295 . suppose @xmath388 or @xmath389 with @xmath107 , and @xmath284 for @xmath390 . then * @xmath391 ; * if @xmath366 , then @xmath392 . we consider the case of @xmath77 . by symmetry , the case of @xmath384 can be dealt with in a similar manner . following the discussion in section [ ss : g1analysis ] , @xmath393 must be a weight of @xmath302 for @xmath303 . recall that @xmath394 consider the decomposition of @xmath106 into a sum of distinct positive roots ( cf . observation [ o : uniquedecomp ] ) . write @xmath305 where @xmath306 is the number of positive roots in this decomposition which contain @xmath62 and @xmath308 is the number of roots in this decomposition which do not contain @xmath62 . then @xmath395 contains @xmath396 copies of @xmath62 . since any root contains at most one copy of @xmath62 , we have @xmath397 substituting @xmath305 , rewriting , and simplifying , we get @xmath398 the total number of positive roots containing @xmath62 is @xmath399 . since we necessarily have @xmath400 and @xmath317 , we get @xmath401 which gives part ( a ) . for part ( b ) , assume that @xmath366 . we want to show that @xmath402 this is equivalent to showing that @xmath403 . consider the left hand side : @xmath404 hence the problem is reduced to showing that @xmath405 or @xmath406 . since @xmath407 , we have @xmath408 since @xmath366 . part ( b ) follows . note that if @xmath368 , @xmath409 the following two theorems summarize our findings when the root system is of type @xmath5 . [ t : dsummary1 ] suppose @xmath53 is of type @xmath40 with @xmath279 . assume that @xmath373 . then * @xmath410 for @xmath411 ; * @xmath412 part ( a ) follows from section [ ss : initialrestrictions ] , lemma [ l : omega1](a ) , lemma [ l : omegan ] , and proposition [ p : vanishing ] . for part ( b ) , when @xmath366 , it follows from section [ ss : initialrestrictions ] , lemma [ l : omega1 ] , proposition [ p : omega1summary ] and lemma [ l : omegan ] that @xmath413 is the only dominant weight with @xmath414 . since @xmath131 is the lowest weight in its linkage class , the claim follows from theorem [ t : nonvanishing ] . for @xmath369 the symmetry of the root system yields @xmath415 for the weights @xmath416 , @xmath417 and @xmath418 , and those are the only weights with non - zero @xmath1-cohomology in degree @xmath419 . each weight is minimal in its own linkage class . the claim follows . working inductively from the @xmath48 case , we can obtain sharp vanishing bounds for arbitrary @xmath15 . [ t : dsummary2 ] suppose @xmath53 is of type @xmath40 with @xmath279 . assume that @xmath373 . then * @xmath420 for @xmath421 ; * @xmath422 for part ( a ) , we need to show that @xmath423 for @xmath424 and @xmath92 . if that is true , then the claim follows from proposition [ p : vanishing ] . for part ( b ) , we require precise information on those dominant weights @xmath131 for which @xmath425 . the root lattice @xmath426 has four cosets within @xmath78 : @xmath426 , @xmath427 , @xmath428 , and @xmath429 . if @xmath131 is a weight in the root lattice claim ( a ) follows from lemma [ l : simplylacedcohom](a ) . furthermore , no such weights can contribute to cohomology in degree @xmath430 . assume that @xmath191 for some @xmath431 and apply proposition [ p : degreebound2 ] . suppose first that @xmath432 with @xmath433 . for each @xmath202 , in order to have @xmath434 , then @xmath435 must be a weight in @xmath436 . this implies that @xmath437 must lie in the positive root lattice . since @xmath438 , we necessarily have @xmath439 . since @xmath440 , it then follows that @xmath441 . inductively one concludes that @xmath442 for all @xmath144 . for a weight @xmath229 , when expressed as a sum of simple roots , let @xmath443 denote the number of copies of @xmath307 that appear . since a positive root contains at most one copy of @xmath307 , we have @xmath444 from observation [ o : uniquedecomp ] , we know that @xmath445 can be expressed uniquely as @xmath446 distinct positive roots . write @xmath447 where @xmath448 denotes the number of roots containing an @xmath307 and @xmath449 the number that do not . then @xmath450 , and rewriting the above gives @xmath451 hence , summing over @xmath144 gives @xmath452 recall that the @xmath238 are non - zero dominant weights . by the assumption on @xmath238 , @xmath453 . the total number of positive roots containing an @xmath307 is @xmath321 . hence , @xmath454 . with this , we get @xmath455 since @xmath456 . this gives the necessary condition for part ( a ) for the coset @xmath457 . before considering the remaining two cosets , towards addressing part ( b ) , we consider when equality can hold in ( [ d.5cond ] ) . as in section [ ss : caseomega1 ] we see that equality holds in ( [ d.5cond ] ) if and only if @xmath458 and @xmath459 which forces @xmath460 for all @xmath144 . moreover , one obtains from the discussion above and proposition [ p : omega1summary ] for @xmath461 that @xmath462_{{\lambda } } \cong \begin{cases } h^0({\lambda } ) & \mbox { if } i = 2p-2n\\ 0 & \mbox { if } 0 < i < 2p-2n . \end{cases}\ ] ] using the spectral sequence argument in the proof of theorem [ t : firstnontrivial ] ( see also the proof of ( * ? ? ? * lemma 5.4 ) ) , we can show that @xmath463 . we prove this by induction on @xmath15 with the @xmath48 case being theorem [ t : dsummary1 ] . consider the lyndon - hochschild - serre spectral sequence @xmath464 from the remarks in section [ ss : degreebounds ] and the discussion above , @xmath262 for @xmath465 . furthermore , from ( [ d2p2n ] ) , @xmath262 for @xmath466 . finally , if @xmath467 and @xmath468 , then , from the above conclusion that @xmath469 for each @xmath144 , we must have @xmath470 . hence , the @xmath471-term survives to @xmath269 and is the only term to contribute in degree @xmath430 . hence , by ( [ d2p2n ] ) and our inductive hypothesis , @xmath472 to complete the proof of ( a ) we need to consider the case that @xmath473 with @xmath474 . as above , @xmath475 must lie in the positive root lattice . since @xmath476 does , this implies that @xmath437 must lie in the positive root lattice . when expressed as a sum of simple roots @xmath477 ( as does @xmath77 ) . whereas , for @xmath478 , @xmath479 . since @xmath480 , for @xmath481 to lie in the positive root lattice , when expressed as a sum of fundamental weights , @xmath482 must contain an odd number of copies of @xmath384 and @xmath77 in total . since @xmath12 is odd , this also holds for @xmath483 . inductively , every @xmath238 has this property . we may assume therefore that each @xmath238 contains at least one copy of @xmath77 or one copy of @xmath384 . proceed as above , but let @xmath484 and @xmath485 denote the number of copies of @xmath62 and @xmath486 , respectively , appearing in @xmath229 . set @xmath487 . note that , for the weights @xmath229 that appear in what follows , both @xmath488 and @xmath485 are nonnegative . again , a positive root contains at most one copy of @xmath62 or @xmath489 . just as above , we get @xmath490 and the corresponding dual statement for @xmath491 by choosing the appropriate root we obtain @xmath492 or @xmath493 either one will result in @xmath494 from earlier arguments we know that @xmath495 . hence @xmath496 summing over @xmath144 , one obtains @xmath497.\ ] ] clearly , @xmath498 moreover , we can say that @xmath499 . substituting this gives @xmath500 where the last inequality follows as in the proof of lemma [ l : omegan ] . thus part ( a ) follows . for @xmath366 , the last inequality is strict . hence , @xmath324 is the only dominant weight for which @xmath501 . as in the proof of theorem [ t : dsummary1 ] , since @xmath131 is minimal in its linkage class , part ( b ) follows . similarly , for @xmath368 , by symmetry , part ( b ) follows . assume for this subsection that @xmath53 is of type @xmath38 with @xmath502 ( so @xmath503 ) . the only dominant weights @xmath158 with @xmath504 are @xmath75 and @xmath505 . one concludes from proposition [ p : degreebound1 ] and proposition [ p : vanishing ] that @xmath506 for all @xmath507 unless there exists a weight @xmath131 of the form @xmath508 or of the form @xmath509 with @xmath510 for some @xmath511 [ l : e6vanishing ] suppose @xmath53 is of type @xmath38 , @xmath503 and @xmath92 is of the form @xmath508 or @xmath512 with @xmath107 . assume in addition that @xmath513 , @xmath514 . then @xmath515 for all @xmath516 we prove the assertion for @xmath517 let @xmath518 denote the number of times that @xmath343 appears in @xmath519 when written as a sum of simple roots . note that all positive roots of @xmath53 contain the simple root @xmath343 at most once . this implies that @xmath520 . moreover , there are exactly @xmath521 distinct positive roots containing @xmath343 . hence , @xmath522 . using @xmath523 , we note that @xmath524 , written as a sum of simple roots contains at least @xmath525 copies of @xmath343 . from section [ ss : g1analysis ] we know that @xmath526 and @xmath527 imply that @xmath524 is a sum of @xmath528 many positive roots . note that this can only happen if @xmath529 is divisible by @xmath530 . again using the fact that @xmath343 appears at most once in each positive root , one obtains the inequality : @xmath531 solving for @xmath173 yields @xmath532 note that equality holds if and only if @xmath533 . for the proof of part ( b ) , set @xmath537 . part ( i ) follows from the proof of lemma [ l : e6vanishing ] . let @xmath546 denote the subgroup of @xmath65 generated by the simple reflections @xmath547 and let @xmath68 denote the distinguished representative of the left coset @xmath548 . then @xmath549 and @xmath550 equals the sum of all positive roots in @xmath53 that contain @xmath343 , which equals the weight @xmath551 . let @xmath552 clearly , @xmath553 since @xmath131 is the smallest dominant weight in its linkage class the assertion follows from the remarks in section [ ss : nonvanishing ] . for @xmath554 , part ( a)(ii ) follows from theorem [ t : firstnontrivial ] . it remains to show part ( a)(i ) . if @xmath555 for @xmath556 and @xmath121 , then @xmath557 or @xmath558 for some @xmath107 . from the proof of lemma [ l : e6vanishing ] , one can see that @xmath559 . consider the case that @xmath560 . the @xmath505 case is dual and analogous . one can explicitly , with the aid of magma , identify all @xmath107 such that @xmath121 and @xmath561 lies in the positive root lattice . by considering the number of copies of @xmath307 appearing in @xmath309 ( as in the proof of lemma [ l : e6vanishing ] ) , one can identify the least @xmath11 such that @xmath309 is a weight in @xmath562 , and hence the least possible value of @xmath173 . the three weights which can give a value of @xmath563 are listed in the following table along with the minimum possible value of @xmath11 . as seen in section [ ss : g1analysis ] , @xmath564 must be a weight of @xmath565 , and hence @xmath173 is congruent to @xmath67 mod 2 . it follows that some of the above degree bounds are even higher . for example , consider @xmath566 . since @xmath567 but @xmath568 is odd , the least value @xmath173 could take would be @xmath569 . a similar situation holds for @xmath570 and @xmath571 . similarly , for the other weights in the above list , if the cohomology vanishes in the degree @xmath173 listed , then the next possible non - vanishing degree is @xmath572 . we summarize this in the following lemma . in principle , one could use proposition [ p : kostantpartcohom ] to compute the dimension of @xmath585 in terms of partition functions for the weights in lemma [ l : f4nonvan ] . for small @xmath12 , one can use magma to make this computation . for @xmath586 or @xmath514 , one finds that the two candidates in degree @xmath568 have zero cohomology . they do give cohomology in degree @xmath587 . and in degree @xmath569 , the only one weight ( of the three ) which has cohomology is @xmath588 . we make the following if part ( a ) of the conjecture holds , then @xmath593 for @xmath590 thus improving upon theorem [ t : f4van ] . however , even if part ( b ) of the conjecture also holds , it does not necessarily follow that @xmath594 . analogous to the situation for type @xmath46 ( cf . section [ ss : gvan ] ) , cohomology in degree @xmath595 from the weight @xmath596 could cancel out the cohomology in degree @xmath597 from the weight @xmath598 . conjecture [ c : f4conj1 ] is a special case of a more general conjecture on partition functions ( known to hold for small values of @xmath120 ) . conjecture [ c : f4conj1](a ) would follow from parts ( a ) and ( b ) while conjecture [ c : f4conj1](b ) would follow from part ( c ) .	the computation of the cohomology for finite groups of lie type in the describing characteristic is a challenging and difficult problem . in @xcite , the authors constructed an induction functor which takes modules over the finite group of lie type , @xmath0 , to modules for the ambient algebraic group @xmath1 . in particular this functor when applied to the trivial module yields a natural @xmath1-filtration . this filtration was utilized in @xcite to determine the first non - trivial cohomology class when the underlying root system is of type @xmath2 or @xmath3 . in this paper the authors extend these results toward locating the first non - trivial cohomology classes for the remaining finite groups of lie type ( i.e. , the underlying root system is of type @xmath4 , @xmath3 , @xmath5 , @xmath6 , @xmath7 , @xmath8 , @xmath9 , and @xmath10 ) when the prime is larger than the coxeter number . [ multiblock footnote omitted ]
the theory of mechanism design has been developed and applied to many branches of economics for decades . nash implementation is a cornerstone of the mechanism design theory . the maskin s theorem @xcite provides an almost complete characterization of social choice rules ( scrs ) that are nash implementable : when the number of agents is at least three , the sufficient conditions for nash implementation are monotonicity and no - veto , and the necessary condition is monotonicity . note that an scr is specified by a designer , a desired outcome from the designer s perspective may not be desirable for the agents ( see table 1 in section 3.1 ) . the maskin mechanism ( page 394 , @xcite ) constructed in the proof of maskin s sufficiency theorem is an abstract mechanism . people seldom consider how the designer actually receives messages from agents . roughly speaking , there are two distinct manners : direct and indirect manner . in the former manner , agents report their messages to the designer directly ( _ e.g. _ , speak face to face , hand over , _ etc _ ) , thereby the designer can know exactly that a message is reported by an agent himself , not by any other device . in the latter manner , agents report messages to the designer through channels ( _ e.g. _ , internet , cable _ etc _ ) . thereby , when the designer receives a message from a channel , he can not know what has happened on the other side of the channel : whether the message is reported by an agent himself , or generated by some device authorized by an agent . traditionally , nobody notice the difference between the two manners in the maskin mechanism . however , in this paper , we will point out that traditional sufficient conditions on nash implementation may fail if agents report messages to the designer in an indirect manner . the rest of the paper is organized as follows : section 2 recalls preliminaries of the mechanism design theory given by serrano @xcite ; section 3 is the main part of this paper , where we will propose a self - enforcing agreement to help agents break through the restriction of maskin s sufficiency theorem . section 4 draws the conclusion . let @xmath0 be a finite set of _ agents _ with @xmath1 , @xmath2 be a finite set of social _ outcomes_. the information held by the agents is summarized in the concept of a _ state_. the true state is not verifiable by the designer . we denote by @xmath3 a typical state and by @xmath4 the domain of possible states . at state @xmath5 , each agent @xmath6 is assumed to have a complete and transitive _ preference relation _ @xmath7 over the set @xmath8 . we denote by @xmath9 the profile of preferences in state @xmath3 , and denote by @xmath10 the strict preference part of @xmath7 . fix a state @xmath3 , we refer to the collection @xmath11 as an _ environment_. let @xmath12 be the class of possible environments . a _ social choice rule _ ( scr ) @xmath13 is a mapping @xmath14 . a _ mechanism _ @xmath15 describes a message or strategy set @xmath16 for agent @xmath17 , and an outcome function @xmath18 . @xmath16 is unlimited except that if a mechanism is direct , _ i.e. _ , @xmath19 . an scr @xmath13 satisfies _ no - veto _ if , whenever @xmath20 for all @xmath21 and for every agent @xmath17 but perhaps one @xmath22 , then @xmath23 . an scr @xmath13 is _ monotonic _ if for every pair of environments @xmath24 and @xmath25 , and for every @xmath23 , whenever @xmath20 implies that @xmath26 , there holds @xmath27 . we assume that there is _ complete information _ among the agents , _ i.e. _ , the true state @xmath3 is common knowledge among them . given a mechanism @xmath28 played in state @xmath3 , a _ nash equilibrium _ of @xmath29 in state @xmath3 is a strategy profile @xmath30 such that : @xmath31 . let @xmath32 denote the set of nash equilibria of the game induced by @xmath29 in state @xmath3 , and @xmath33 denote the corresponding set of nash equilibrium outcomes . an scr @xmath13 is _ nash implementable _ if there exists a mechanism @xmath15 such that for every @xmath34 , @xmath35 . maskin @xcite provided an almost complete characterization of scrs that were nash implementable . the main results of ref . @xcite are two theorems : 1 ) ( _ necessity _ ) if an scr is nash implementable , then it is monotonic . 2 ) ( _ sufficiency _ ) let @xmath36 , if an scr is monotonic and satisfies no - veto , then it is nash implementable . in order to facilitate the following investigation , we briefly recall the maskin mechanism given by serrano @xcite as follows : consider a mechanism @xmath15 , where agent @xmath17 s message set is @xmath37 , @xmath38 is the set of non - negative integers . a typical message sent by agent @xmath17 is described as @xmath39 . the outcome function @xmath40 is defined in the following three rules : ( 1 ) if for every agent @xmath6 , @xmath41 and @xmath42 , then @xmath43 . ( 2 ) if @xmath44 agents @xmath45 send @xmath41 and @xmath42 , but agent @xmath22 sends @xmath46 , then @xmath43 if @xmath47 , and @xmath48 otherwise . ( 3 ) in all other cases , @xmath49 , where @xmath50 is the outcome chosen by the agent with the lowest index among those who announce the highest integer . this section is the main part of this paper . in the beginning , we will show an example of scr which satisfies monotonicity and no - veto . it is nash implementable although all agents dislike it . then , we will propose a self - enforcing agreement using complex numbers , by which the agents may break through the maskin s sufficiency theorem and make the scr not nash implementable . let @xmath51 , @xmath52 , @xmath53 . in each state @xmath5 , the preference relations @xmath54 over the outcome set @xmath8 and the corresponding scr @xmath13 are given in table 1 . the scr @xmath13 is _ pareto - inefficient _ from the agents perspectives because in state @xmath55 , all agents unanimously prefer a pareto - optimal outcome @xmath56 : for each agent @xmath6 , @xmath57 . _ table 1 : an scr satisfying monotonicity and no - veto is pareto - inefficient from the agents perspectives . _ + [ cols="^,^,^,^,^,^ " , ] suppose the true state is @xmath55 . at first sight , @xmath58 might be a unanimous @xmath59 for each agent @xmath17 , because by doing so @xmath56 would be generated by rule 1 of the maskin mechanism . however , @xmath60 has an incentive to unilaterally deviate from @xmath58 to @xmath61 in order to trigger rule 2 ( where @xmath62 stands for any legal value ) , since @xmath63 , @xmath64 ; @xmath65 also has an incentive to unilaterally deviate from @xmath58 to @xmath66 , since @xmath67 , @xmath68 . note that either @xmath60 or @xmath65 can certainly obtain her expected outcome only if just one of them deviates from @xmath58 ( if this case happened , rule 2 would be triggered ) . but this condition is unreasonable , because all agents are rational , nobody is willing to give up and let the others benefit . therefore , both @xmath60 and @xmath65 will deviate from @xmath58 . as a result , rule 3 will be triggered . since @xmath60 and @xmath65 both have a chance to win the integer game , the winner is uncertain and the final outcome is also uncertain between @xmath69 and @xmath70 . to sum up , although every agent prefers @xmath56 to @xmath71 in state @xmath55 , @xmath56 can not be yielded in nash equilibrium . indeed , the maskin mechanism makes the pareto - inefficient outcome @xmath71 be nash implementable in state @xmath55 . can the agents find a way to break through the maskin s sufficiency theorem and let the pareto - efficient outcome @xmath56 be nash implementable in state @xmath55 ? interestingly , we will show that the answer may be `` yes '' if agents report messages to the designer through channels ( _ e.g. _ , internet ) . in what follows , first we will define some matrices with complex numbers , then we will propose a self - enforcing agreement to help agents break through the maskin s sufficiency theorem . * definition 1 * : let @xmath72 be two @xmath73 matrices , and @xmath74 be two basis vectors : @xmath75 hence , @xmath76 , @xmath77 ; @xmath78 , @xmath79 . * definition 2 * : for @xmath80 agents , suppose each agent @xmath6 possesses a basis vector . @xmath81 is defined as the tensor product of @xmath82 basis vectors @xmath83 : @xmath84 @xmath85 contains @xmath82 basis vectors @xmath83 and @xmath86 elements . @xmath85 is also denoted as @xmath87 . similarly , @xmath88 obviously , there are @xmath86 possible vectors : @xmath89 @xmath90 . * definition 3 * : @xmath91 , _ i.e. _ , @xmath92 where the symbol @xmath93 denotes an imaginary number , and @xmath94 is the conjugate transpose of @xmath95 . * definition 4 * : @xmath96 * definition 5 * : for @xmath97 $ ] , @xmath98 $ ] , @xmath99 @xmath100,\phi\in[0,\pi/2]\}$ ] . hence , @xmath101 . * definition 6 * : for @xmath102 , @xmath103 $ ] , @xmath104 $ ] , let @xmath105 , @xmath106\overrightarrow{\psi}_{1}.\ ] ] the dimension of @xmath107 is @xmath108 . since only two elements in @xmath109 are non - zero , it is not necessary to calculate the whole @xmath110 matrix to yield @xmath111 . indeed , we only need to calculate the leftmost and rightmost column of @xmath107 to derive @xmath111 . * definition 7 * : @xmath112 . suppose @xmath113^{t}$ ] , let latexmath:[$\delta=[\|\eta_{1}\|^{2 } , \cdots , @xmath95 , @xmath115 ( @xmath116 ) and @xmath94 are all unitary matrices . hence , @xmath117 . thus , @xmath118 can be viewed as a probability distribution , each element of which represents the probability that we randomly choose a vector from the set of all @xmath86 possible vectors @xmath119 @xmath120 . * definition 8 * : condition @xmath121 contains five parts . the first three parts are defined as follows : + @xmath122 : given an scr @xmath13 , there exist two states @xmath123 , @xmath124 , @xmath125 such that @xmath126 ( for each @xmath6 , @xmath127 , @xmath128 ) with strict relation for some agent ; and the number of agents that encounter a preference change around @xmath129 in going from state @xmath123 to @xmath130 is at least two . denote by @xmath131 the set of these agents and @xmath132 the number of these agents . + @xmath133 : consider the state @xmath130 specified in condition @xmath122 , if there exists another @xmath134 , @xmath135 that satisfies @xmath122 , then @xmath136 ( for each @xmath6 , @xmath127 , @xmath137 ) with strict relation for some agent . + @xmath138 : consider the outcome @xmath129 specified in condition @xmath122 , for any state @xmath139 , @xmath129 is top ranked for each agent @xmath140 . as specified before , in this paper we will discuss how the traditional results on nash implementation will be changed when agents report messages to the designer through channels . we assume that : \1 ) each agent has a unique channel connected with the designer . the agents report messages to the designer through these channels . \2 ) after the designer claims the outcome function @xmath40 , if all agents anticipate the scr @xmath13 appeared in rule 1 of @xmath40 satisfies condition @xmath121 ( given in definition 8 and 11 ) , agents can negotiate and construct an agreement _ complexmessage _ as shown in fig . the algorithm _ messagecomputing _ is given in definition 9 . in the initial configuration , the computer controls all channels . \3 ) each agent can freely decide whether to leave his channel to the computer and let the computer send a message to the designer , or to take back his channel and send his message to the designer by himself . if any agent deviates from the initial configuration and chooses the latter option , then this deviation is _ observable _ to the rest agents and all of them will choose the latter option too , _ i.e. _ , all agents will take back their channels and send their messages to the designer by themselves . if all agents leave their channels to the computer , the algorithm _ messagecomputing _ works , _ i.e. _ , calculates @xmath141 and sends them to the designer through channels . * definition 9 * : the algorithm _ messagecomputing _ is defined as follows : + * input * : @xmath142\times [ 0,\pi]\times a\times \mathcal { t } \times \mathbb{z}_{+}$ ] , @xmath143 . + * output * : @xmath144 , @xmath143 . + step 1 : reading @xmath145 from each agent @xmath6 . + step 2 : computing the leftmost and rightmost columns of @xmath107 . + step 3 : computing @xmath146\overrightarrow{\psi}_{1}$ ] , @xmath147 , and the probability distribution @xmath118 . + step 4 : randomly choosing a vector from the set of all @xmath86 possible vectors @xmath119 @xmath120 according to the probability distribution @xmath118 . + step 5 : for each agent @xmath6 , let @xmath148 ( or @xmath39 ) if the @xmath17-th basis vector of the chosen vector is @xmath83 ( or @xmath149 ) , where @xmath150 are specified in condition @xmath122 . + step 6 : sending @xmath151 to the designer through channels @xmath152 . * definition 10 * : suppose @xmath122 and @xmath133 are satisfied and @xmath151 is computed by _ messagecomputing_. suppose the true state is @xmath130 specified in condition @xmath122 . @xmath153 , @xmath154 , @xmath155 and @xmath156 are defined as the payoffs to the @xmath82-th agent in state @xmath130 when the chosen vector in step 4 of _ messagecomputing _ is @xmath87 , @xmath157 , @xmath158 or @xmath159 respectively . * definition 11 * : suppose conditions @xmath122 , @xmath133 and @xmath138 are satisfied and the true state is @xmath130 , consider each message @xmath160 , where @xmath161 is top - ranked for each agent @xmath17 . the rest two parts of condition @xmath121 are defined as : + @xmath162 : for each agent @xmath6 , let him be the @xmath82-th agent , @xmath163 . + @xmath164 : for each agent @xmath165 , let him be the @xmath82-th agent , @xmath166 . + it can be seen from fig . 1 that after the agreement _ complexmessage _ is constructed , each agent @xmath6 independently faces two options : + @xmath167 @xmath168 : leaving his channel to the computer , and submitting @xmath169 to the algorithm _ messagecomputing_. + @xmath167 @xmath170 : taking back his channel , and submitting @xmath171 to the designer by himself . to sum up , if agents report messages to the designer through channels and all agents anticipate the scr @xmath13 appeared in rule 1 of @xmath40 satisfies condition @xmath121 , they can construct the agreement _ complexmessage _ after the designer claims the outcome function @xmath40 . the timing steps of the maskin mechanism are updated as follows : + time 1 : the designer claims the outcome function @xmath40 to all agents ; + time 2 : the agents construct the agreement _ complexmessage _ ; + time 3 : each agent @xmath6 chooses an option between @xmath168 and @xmath170 , and @xmath172 are sent through channels . + time 4 : the designer receives @xmath151 from @xmath82 channels ; + time 5 : the designer announces the outcome @xmath173 . * proposition 1 * : for @xmath80 , suppose agents send messages to the designer through channels ( _ e.g. _ , internet ) . consider an scr @xmath13 that satisfies monotonicity and no - veto . if condition @xmath121 is satisfied , then in state @xmath130 the agents can construct a self - enforcing agreement _ complexmessage _ to make the pareto - inefficient outcome @xmath174 not be implemented in nash equilibrium . * proof * : since @xmath122 and @xmath133 are satisfied , then there exist two states @xmath123 , @xmath124 , @xmath125 such that @xmath126 ( for each @xmath6 , @xmath127 , @xmath128 ) with strict relation for some agent , and @xmath131 contains the agents that encounter a preference change around @xmath129 in going from state @xmath123 to @xmath130 ( @xmath175 ) . suppose the true state is @xmath130 , now let us check what will happen after the agents construct the agreement _ complexmessage _ in time 2 . from the viewpoints of agents , after constructing _ complexmessage _ , there are two possible cases in time 3 : + 1 ) suppose every agent @xmath6 chooses @xmath168 , then the algorithm _ messagecomputing _ works . consider the following strategy profile : each agent @xmath140 submits @xmath176 , @xmath177 ; each agent @xmath178 submits @xmath179 . according to lemma 1 ( see appendix ) , this strategy profile is a nash equilibrium of @xmath29 in state @xmath130 . in step 4 of _ messagecomputing _ , the chosen vector is @xmath180 . in step 5 of _ messagecomputing _ , @xmath148 for each @xmath6 . therefore , in time 5 , @xmath181 . each agent @xmath17 s payoff is @xmath153 . + 2 ) suppose some agent @xmath6 chooses @xmath170 , _ i.e. _ , takes back his channel and reports @xmath59 to the designer by himself . then according to assumption 3 ( see section 3.3 ) , all of the rest agents will observe this deviation , thereby take back their channels and submit messages to the designer by themselves . according the maskin mechanism , in time 5 , the final outcome will be either @xmath174 by rule 1 , or uncertain by rule 3 ( _ i.e. _ , each agent @xmath17 s payoff is @xmath156 ) . according to conditions @xmath122 and @xmath162 , it is not profitable for any agent @xmath17 to choose @xmath170 , _ i.e. _ , unilaterally take back his channel and send a message to the designer by himself . as telser pointed out in page 28 , line 2 @xcite : `` _ _ a party to a self - enforcing agreement calculates whether his gain from violating the agreement is greater or less than the loss of future net benefits that he would incur as a result of detection of his violation and the consequent termination of the agreement by the other party ... hence both parties continue to adhere to an agreement if and only if each gains more from adherence to , than from violations of , its terms . _ _ '' therefore , it can be seen that _ complexmessage _ is a self - enforcing agreement among the agents . put differently , although the agents collaborate to construct _ complexmessage _ in time 2 , they do not require a third - party to enforce it after then and the game is still non - cooperative . to sum up , in state @xmath130 , the agents can construct a self - enforcing agreement _ complexmessage _ to make the pareto - inefficient outcome @xmath174 not be implemented in nash equilibrium.@xmath182 let us reconsider table 1 . let @xmath183 , @xmath184 , @xmath185 , @xmath186 . suppose the true state is @xmath55 . since both @xmath60 and @xmath65 encounter a preference change around @xmath56 in going from state @xmath187 to @xmath55 , condition @xmath122 is satisfied . obviously , @xmath133 and @xmath138 are also satisfied . consider the strategy profile as follows : @xmath188 let @xmath189 be the first agent , and for any agent @xmath190 , let her be the last agent . consider the payoff to the third agent , suppose @xmath191 ( the corresponding outcome is @xmath56 ) , @xmath192 ( the corresponding outcome is @xmath70 if @xmath193 , and @xmath69 if @xmath194 ) , @xmath195 ( the corresponding outcome is @xmath69 if @xmath193 , and @xmath70 if @xmath194 ) , @xmath196 ( the corresponding outcome is uncertain between @xmath69 and @xmath70 ) . let @xmath189 be the last agent and consider her payoff , suppose @xmath191 and @xmath196 . hence , @xmath162 and @xmath164 are satisfied . according to proposition 1 , in state @xmath55 , the outcome implemented in nash equilibrium is @xmath56 , and @xmath197 is not nash implementable although the scr @xmath13 satisfies monotonicity and no - veto . * remark 1 : * some reader may argue that the agreement _ complexmessage _ is a wrapper to the maskin mechanism that changes the game substantially and henceforth it has no implication on the original maskin mechanism . however , this viewpoint is not true . actually , _ complexmessage _ is _ unobservable _ to the designer because it is hidden behind channels and the designer can not prevent the agents from constructing such agreement . from the designer s perspective , no matter whether the agents construct the agreement _ complexmessage _ on the other side of channels or not , the maskin mechanism remain unchanged and the designer acts in the same way : _ i.e. _ , claims the outcome function @xmath40 , receives messages @xmath198 from channels and announces the final outcome @xmath173 . * remark 2 : * although the time and space complexity of _ messagecomputing _ are exponential with the number of agents , _ i.e. _ , @xmath199 , it works well when the number of agents is not very large . for example , the runtime of _ messagecomputing _ is about 0.5s for 15 agents , and about 12s for 20 agents ( matlab 7.1 , cpu : intel ( r ) 2ghz , ram : 3 gb ) . * remark 3 : * the problem of nash implementation requires complete information among all agents . in the last paragraph of page 392 @xcite , serrano wrote : `` _ _ we assume that there is complete information among the agents ... this assumption is especially justified when the implementation problem concerns a small number of agents that hold good information about one another _ _ '' . hence , the fact that _ messagecomputing _ is suitable for small - scale cases ( _ e.g. _ , less than 20 agents ) is acceptable for nash implementation . in this paper , we propose a self - enforcing agreement to help agents avoid a pareto - inefficient social choice rule if agents report messages to the designer through channels and condition @xmath121 is satisfied . put differently , traditional sufficient conditions for nash implementation may fail on internet . with the rapid development of network economics , it will be more and more common that agents communicate with the designer through internet . in the future , there are many works to do to study the self - enforcing agreement further . the author is very grateful to ms . fang chen , hanyue wu ( _ apple _ ) , hanxing wu ( _ lily _ ) and hanchen wu ( _ cindy _ ) for their great support . 99 e. maskin , nash equilibrium and welfare optimality , _ rev . * 66 * ( 1999 ) 23 - 38 . r. serrano , the theory of implementation of social choice rules , _ siam review _ * 46 * ( 2004 ) 377 - 414 . telser , a theory of self - enforcing agreements . _ journal of business _ * 53 * ( 1980 ) 27 - 44 . flitney and l.c.l . hollenberg , nash equilibria in quantum games with generalized two - parameter strategies , _ phys . a _ * 363 * ( 2007 ) 381 - 388 . * lemma 1 * : suppose condition @xmath121 is satisfied and the algorithm _ messagecomputing _ works . consider the following strategy profile : + 1 ) each agent @xmath140 submits @xmath176 , @xmath177 ; + 2 ) each agent @xmath178 submits @xmath200 ; + then this strategy profile is a nash equilibrium of @xmath29 in state @xmath130 , where @xmath130 is specified in condition @xmath122 . _ part 1_. let the last @xmath201 agents be @xmath131 . consider the following strategy profile : each agent @xmath202 submits @xmath203 , @xmath204 ; each agent @xmath205 submits @xmath179 , then we will prove the optimal value of @xmath206 for the @xmath82-th agent is @xmath207 . since condition @xmath122 is satisfied , then @xmath208 . let @xmath209 @xmath210 here we only explicitly show the up - left and bottom - right entries because only these two entries are useful in the following discussions . the other entries in diagonal are simply represented as symbol @xmath62 . note that @xmath211 thus , @xmath212 suppose the @xmath82-th agent chooses arbitrary parameters @xmath213 in his strategy @xmath214 , let @xmath215 then , @xmath216}\cos(\theta/2 ) & * & & & & & \\ ie^{i\frac{(l-1)\pi}{l}}\sin(\theta/2 ) & * & & & & & \\ & & * & * & & & \\ & & * & * & & & \\ & & & & \cdots & & \\ & & & & & * & ie^{-i\frac{(l-1)\pi}{l}}\sin(\theta/2 ) \\ & & & & & * & e^{-i[\frac{(l-1)\pi}{l}+\phi]}\cos(\theta/2 ) \end{bmatrix}_{2^{n}\times 2^{n}}.\end{aligned}\ ] ] recall that @xmath217 thus , @xmath218\overrightarrow{\psi}_{1}=\frac{1}{\sqrt{2}}\begin{bmatrix } e^{i[\frac{(l-1)\pi}{l}+\phi]}\cos(\theta/2)\\ ie^{i\frac{(l-1)\pi}{l}}\sin(\theta/2)\\ 0\\ \cdots\\ 0\\ -e^{-i\frac{(l-1)\pi}{l}}\sin(\theta/2)\\ ie^{-i[\frac{(l-1)\pi}{l}+\phi]}\cos(\theta/2 ) \end{bmatrix}_{2^{n}\times 1},\end{aligned}\ ] ] @xmath219 the probability distribution @xmath118 is computed from @xmath220 : @xmath221 obviously , @xmath222 consider the payoff to the @xmath82-th agent , @xmath223 since @xmath162 is satisfied , _ i.e. _ , @xmath224 , then the @xmath82-th agent chooses @xmath225 to minimize @xmath226 . as a result , @xmath227 . since @xmath164 is satisfied , _ i.e. _ , @xmath166 , then the @xmath82-th agent prefers @xmath228 , which leads to @xmath229 and let @xmath230 be its maximum @xmath153 . therefore , the optimal value of @xmath206 for the @xmath82-th agent is @xmath207 . _ part 2_. let the first @xmath201 agents be @xmath131 . consider the following strategy profile : each agent @xmath231 submits @xmath179 ; each agent @xmath232 submits @xmath203 , @xmath233 , then we will prove the optimal values of @xmath206 and @xmath234 for the @xmath82-th agent are @xmath235 and @xmath236 . as shown before , @xmath237 @xmath238 note that @xmath239 thus , @xmath240 suppose the @xmath82-th agent chooses arbitrary parameters @xmath213 in his strategy @xmath214 , let @xmath215 then , @xmath241 @xmath242\overrightarrow{\psi}_{1}=\frac{1}{\sqrt{2}}\begin{bmatrix } -e^{i\phi}\cos(\theta/2)\\ -i\sin(\theta/2)\\ 0\\ \cdots\\ 0\\ \sin(\theta/2)\\ -ie^{-i\phi}\cos(\theta/2 ) \end{bmatrix}_{2^{n}\times 1},\end{aligned}\ ] ] @xmath243 the probability distribution @xmath118 is computed from @xmath220 : @xmath244,&&p_{c\cdots cd}=\sin^{2}(\theta/2)\\ & p_{d\cdots dc}=0 , & & p_{d\cdots dd}=\cos^{2}(\theta/2)\sin^{2}\phi\end{aligned}\ ] ] obviously , @xmath222 consider the payoff to the @xmath82-th agent , @xmath245+\$_{c\cdots cd}\sin^{2}(\theta/2 ) + \$_{d\cdots dd}\cos^{2}(\theta/2)\sin^{2}\phi\end{aligned}\ ] ] since @xmath162 is satisfied , _ i.e. _ , @xmath224 , then the @xmath82-th agent chooses @xmath246 . as a result , @xmath247 since the @xmath82-th agent belongs to @xmath248 , by condition @xmath138 , @xmath234 can be chosen as @xmath249 . according to step 5 of _ messagecomputing _ , @xmath250 . thus , @xmath251 . in this case , @xmath252 can be chosen as @xmath253 . by symmetry , in state @xmath130 , consider the following strategy : each agent @xmath140 submits @xmath176 , @xmath177 ; each agent @xmath178 submits @xmath179 . then this strategy profile is a nash equilibrium of @xmath29 in state @xmath130 , and the final outcome implemented in nash equilibrium is @xmath254 .	the maskin s theorem is a fundamental work in the theory of mechanism design . in this paper , we propose that if agents report messages to the designer through channels ( _ e.g. _ , internet ) , agents can construct a self - enforcing agreement such that any pareto - inefficient social choice rule satisfying monotonicity and no - veto will not be nash implementable when an additional condition is satisfied . the key points are : 1 ) the agreement is unobservable to the designer , and the designer can not prevent the agents from constructing such agreement ; 2 ) the agents act non - cooperatively , and the maskin mechanism remain unchanged from the designer s perspective . mechanism design ; nash implementation ; social choice .
mapping the qcd phase diagram as a function of temperature @xmath0 and baryochemical potential @xmath1 is one of the fundamental goals of heavy - ion collision experiments . qcd critical point is a distinct singular feature of the phase diagram . it is a ubiquitous property of qcd models based on the chiral symmetry breaking dynamics ( see , e.g. , ref.@xcite for a review and further references ) . locating the point using first - principle lattice calculations is a formidable challenge ( see , e.g. , ref.@xcite for a recent review and references ) . if the critical point is situated in the region accessible to heavy - ion collision experiments it can be discovered experimentally . the search for the critical point is planned at the relativistic heavy ion collider ( rhic ) at bnl , the super proton synchrotron ( sps ) at cern , the future facility for antiproton and ion research ( fair ) at gsi , and nuclotron - based ion collider facility ( nica ) in dubna ( see , e.g. , ref.@xcite ) . the characteristic feature of a critical point is the divergence of the correlation length @xmath2 and of the magnitude of the fluctuations . the simplest measures of fluctuations in heavy - ion collisions are the variances of the event - by - event observables such as multiplicities or mean transverse momenta of particles . the singular , critical contribution to these variances diverges as ( approximately ) @xmath3 , and would manifest in a non - monotonic dependence of such measures as the critical point is passed by during the beam energy scan @xcite . in realistic heavy ion collision the divergence of @xmath2 is cut - off by the effects of critical slowing down @xcite , and the estimates of the maximum correlation length are in the range of at most @xmath4 fm , compared to the natural @xmath5 fm away from the critical point . however , higher , non - gaussian , moments of the fluctuations depend much more sensitively on @xmath2 , according to ref.@xcite . for example , the 4-th moment grows as @xmath6 near the critical point , making it an attractive experimental tool . in this paper we follow up on the results of ref.@xcite to point out that the sign of the 4-th moment could be negative as the critical point is approached from the crossover side of the qcd phase transition . the sign of various moments have been discussed in the literature in related contexts : see , e.g. , discussion of the sign of the 3-rd moment in ref.@xcite or the 6-th and 8-th moments in ref.@xcite and also numerical lattice calculations in ref.@xcite where the possible sign change of kurtosis is noted . in this paper we shall address specifically the sign of the 4-th moment ( or kurtosis ) and do it in a more universal and quantitative way than has been done previously , by using the known parametric form of the universal equation of state near the critical point . we emphasize universality of the behavior of the kurtosis and draw experimental consequences from these results . let us begin , as in ref.@xcite , by describing fluctuations of the order parameter field @xmath7 near a critical point using the probability distribution @xmath8 \sim \exp\left\ { - \omega[\sigma]/t\right\},\ ] ] where @xmath9 is the effective action ( free energy ) functional for the field @xmath10 , which can be expanded in powers of @xmath10 as well as in the gradients ( we chose @xmath11 at the minimum ) : @xmath12 \,.\ ] ] calculating 2-point correlator @xmath13 we find that the correlation length @xmath14 . for the moments of the zero momentum mode @xmath15 in a system of volume @xmath16 we find at tree level @xmath17\ , { \xi}^8\ , . \end{split}\ ] ] where @xmath18 denotes the connected 4-th central moment ( the 4-th cumulant ) . the critical point is characterized by @xmath19 . the central observation in ref.@xcite was that the higher moments ( cumulants ) @xmath20 and @xmath21 diverge with @xmath22 much faster than the quadratic moment @xmath23 . here we shall point out that the _ sign _ of the 4-th moment @xmath21 is negative in a certain sector near the critical point . more precisely , the 4-th cumulant is negative when the critical point is approached from the crossover side . let us demonstrate this in several complementary ways . a simple way to see why the kurtosis is negative is by following the evolution of the probability distribution of @xmath24 as we approach the critical point along the crossover line . in ising scaling coordinates : along @xmath25 , @xmath26 ray . away from the critical point , more precisely for @xmath27 , the central limit theorem dictates that the probability distribution of @xmath24 is gaussian , with a vanishingly small kurtosis . as we approach the critical point the distribution develops non - gaussian shape . this intermediate shape is a deformation of the gaussian towards a two - peak distribution , corresponding to the phase coexistence on the opposite , first - order transition side ( @xmath28 ) of the critical point . such a shape is clearly less `` peaked '' than the gaussian , and thus corresponds to negative kurtosis . more quantitatively , the kurtosis vanishes as @xmath29 at ( almost ) any point away from the critical point , i.e. , @xmath30 the exception is the coexistence line ( @xmath25 , @xmath28 ray ) . the distribution there has two peaks of equal height and its kurtosis is @xmath31 . , at @xmath32 , the value of @xmath33 is independent of @xmath16 ( the correlation length @xmath34 is as large as it can be at given volume ) . this value of @xmath33 is a universal number ( it depends only on the boundary conditions for a given universality class ) and is well - known . it is usually expressed as the value of the binder cumulant @xmath35 , which means @xmath36 . ] it is important to note that this is only true strictly _ on _ the coexistence line @xmath25 , for the moments measured around the symmetric point of the probability distribution of @xmath24 , which is actually a dip , not a peak , for @xmath28 . at any point close to the coexistence line , i.e. , at @xmath37 , @xmath28 , the kurtosis around the dominant peak is positive . in the scaling regime ( close to , but not at the critical point ) where @xmath2 is much greater than the microscopic scale , @xmath38 , but still much less than the linear size of the system : @xmath39 , the coefficient of @xmath40 in eq . ( [ eq : k - def ] ) can be expressed in terms of the couplings @xmath41 using eqs . ( [ eq : sigma - moments ] ) : @xmath42 these couplings , and in fact the shape of the effective potential , is also universal . in particular , @xmath43 scales with @xmath2 as @xmath44 , where the universal value of @xmath45 is known approximately to be 4.0 on the crossover line ( see , e.g. , ref.@xcite for a review ) . , which is only of order few percent . ] since on the crossover line @xmath46 and @xmath47 , it is clear from eq . ( [ eq : sigma - moments ] ) that @xmath48 . away from the crossover line ( @xmath25 , @xmath26 ray ) the distribution is skewed : @xmath49 . this makes the kurtosis positive , according to eq . ( [ eq : k - lambda ] ) , except for a certain sector around the crossover line . to extend this analysis away from the crossover line , i.e. , to take into account @xmath49 in eq . ( [ eq : k - lambda ] ) , we need to know the equation of state , in particular , @xmath21 as a function of both ising variables : reduced temperature @xmath50 and magnetic field @xmath51 . in the scaling regime near @xmath52 this equation of state is also universal . for the ising model it is known to order @xmath53 in the epsilon expansion as well as numerically . before we discuss this universal form , let us keep in mind that the mapping of qcd phase diagram in the @xmath0 , @xmath1 plane into @xmath50 , @xmath51 plane is not universal . however , this mapping is analytic , i.e. , both functions @xmath54 and @xmath55 are analytic at the critical point , which is mapped into the origin , @xmath56 . the standard parametrization , ref.@xcite , of the equation of state in the scaling domain near the critical point is in terms of two new scaling variables @xmath57 and @xmath58 ( it has been applied in the context of qcd before , ref.@xcite ) . denoting the `` magnetization '' by @xmath59 we define @xmath57 and @xmath58 as @xmath60 then the equation of state can be expressed in terms of the single function @xmath61 as @xmath62 unlike the explicit form of the singular equation of state @xmath63 , the function @xmath61 is analytic . it has two zeros . one , at @xmath64 , corresponds to the crossover line ( @xmath26 , @xmath25 ray ) , another , at some @xmath65 , corresponds to the coexistence ( first - order transition ) line ( @xmath28 , @xmath25 ray ) . the function @xmath61 must also be odd since @xmath66 . the simplest function obeying all these requirements is a cubic polynomial @xmath67 where the value @xmath68 is a good approximation to the universal value for the ising model ( and correct up to @xmath69 ) . the choice ( [ eq : h - cubic ] ) is known as the linear parametric model , ref.@xcite . it describes the equation of state with precision quite sufficient for our purposes . the linear parametric model is also known to be exact up to @xmath70 . using this parametric equation of state , we can calculate the moments by taking derivatives at fixed @xmath50 , up to an overall normalization , unimportant in the present context ( it can be fixed by eq . ( [ eq : sigma - moments ] ) ) . in particular , @xmath71 for our purposes , it would be sufficient to use the approximate rational values of critical exponents @xmath72 and @xmath73 , which are within few percent of their exact values in three dimensions . the result of eq . ( [ eq : k4-mh ] ) can then be simplified to @xmath74 we represent @xmath75 graphically as a density plot in fig . [ fig : k4-th ] . we see that the 4-th cumulant ( and kurtosis ) is negative in the sector bounded by two curved rays @xmath76 ( corresponding to @xmath77 ) . and @xmath51 . ] also in fig . [ fig : k4-th ] we show the dependence of @xmath21 along a line which could be thought of as representing a possible mapping of the freezeout trajectory ( fig . [ fig : qcd - pd ] ) onto the @xmath78 plane . although the absolute value of the peak in @xmath21 depends on the proximity of the freezeout curve to the critical point , the ratio of the maximum to minimum along such an @xmath79 curve is a universal number , approximately equal to @xmath80 from eq . ( [ eq : kappa4-theta - r ] ) . the negative minimum is small relative to the positive peak , but given the large size of the latter , ref.@xcite , the negative contribution to kurtosis may be significant . in addition , the mapping of the freezeout curve certainly need not be @xmath79 , and the relative size of the positive and negative peaks depends sensitively on that . the trend described above appears to show in the recent lattice data , ref.@xcite , obtained using pade resummation of the truncated taylor expansion in @xmath1 . as the chemical potential is increased along the freezeout curve , the 4-th moment of the baryon number fluctuations begins to decrease , possibly turning negative , as the critical point is approached ( see fig.2 in ref.@xcite ) . another observation , which we shall return to at the end of the next section , is that @xmath81 grows as we approach the crossover line , corresponding to @xmath25 , @xmath26 on the diagram in fig . [ fig : k4-dp ] . on the qcd phase diagram the freezeout point will move in this direction if one reduces the size of the colliding nuclei or selects more peripheral collisions ( the freezeout occurs earlier , i.e. , at higher @xmath0 , in a smaller system ) . in this section we wish to connect the results for the fluctuations of the order parameter field @xmath10 to the fluctuations of the observable quantities . as an example we consider the fluctuations of the multiplicity of given charged particles , such as pions or protons . for completeness we shall briefly rederive the results of ref.@xcite using a simple model of fluctuations . the model captures the most singular term in the contribution of the critical point to the fluctuation observables . consider a given species of particle interacting with fluctuating critical mode field @xmath10 . the infinitesimal change of the field @xmath82 leads to a change of the effective mass of the particle by the amount @xmath83 . this could be considered a definition of the coupling @xmath84 . for example , the coupling of protons in the sigma model is @xmath85 . the fluctuations @xmath86 of the momentum space distribution function @xmath87 consist of the pure statistical fluctuations @xmath88 around the equilibrium distribution @xmath89 for a particle of a given mass , which itself fluctuates . this gives @xmath90 using this equation we can calculate the most singular contribution from the critical fluctuations to the moments or correlators of @xmath86 . the fluctuation of the multiplicity @xmath91 is given by @xmath92 where @xmath93 is the degeneracy factor ( e.g. , number of spin or charge states of the particle ) . neglecting , for clarity and simplicity , the effects of quantum statistics , i.e. , assuming @xmath94 , we can use poisson statistics for @xmath95 . using additivity of the cumulants ( their defining property ) , and assuming @xmath95 and @xmath82 are uncorrelated , the contribution of the critical fluctuations can be expressed in terms of the corresponding moments of the critical field @xmath10 fluctuations . for example , the contribution to the 4-th moment can be expressed as ( cf . refs.@xcite ) @xmath96 where @xmath97 is the relativistic gamma - factor of a particle with momentum @xmath98 and mass @xmath99 . the first term on the r.h.s . of eq . ( [ eq : n4 ] ) is the poisson contribution . we neglected @xmath100 in the quantum statistics factor @xmath101 for simplicity , and we denoted by `` @xmath102 '' other contributions , less singular at the critical point . the model is admittedly crude , but it illustrates the mechanism and correctly captures the most singular contribution near the critical point . in the region near the critical point where @xmath103 is negative , the 4-th cumulant of the fluctuations will be smaller than its poisson value , @xmath104 . the measure defined in ref.@xcite as @xmath105 will be less than 1 . by how much will depend sensitively on the correlation length ( as @xmath6 ) , i.e. , on how close the freezeout occurs to the critical point , as well as on other factors ( for protons , most significantly , on the value of @xmath1 . ) we shall not attempt to estimate this effect quantitatively in this paper . the analysis of ref.@xcite suggests , however , that this effect for protons can be significant compared to the poisson value already for @xmath106 fm . usual caveats apply : other ( non - trivial ) contributions to moments which do not behave singularly at the critical point can turn out to be relatively large . it is beyond the scope of the paper to estimate these effects . the size of these background contributions could , in principle , be determined experimentally by performing measurements away from the critical point . we conclude by asking an obvious question : has the effect of the negative kurtosis been observed ? data from star indicate that at @xmath107 gev the ratio @xmath108 might be substantially smaller than its poisson value 1 , see fig . 6 in ref.@xcite , while it is very close to 1 at higher @xmath109 ( smaller @xmath1 ) . unfortunately , the statistics accumulated in the short run at @xmath107 gev is clearly not sufficient to make a reliable conclusion . it would be interesting to see if this effect persists with more statistics at this energy . if confirmed , this result could indicate that the critical point is close , at somewhat larger values of @xmath1 ( smaller @xmath109 ) . in this case , as we already discussed at the end of the previous section , the universality would also predict that the negative kurtosis effect should increase in more peripheral collisions at the same @xmath109 . at smaller values of @xmath109 the effect should change sign , increasing kurtosis above its poisson value . the author thanks o. evdokimov for discussions . this work is supported by the doe grant no.de-fg0201er41195 . m. a. stephanov , k. rajagopal and e. v. shuryak , phys . lett . * 81 * , 4816 ( 1998 ) [ arxiv : hep - ph/9806219 ] . m. a. stephanov , k. rajagopal and e. v. shuryak , phys . d * 60 * , 114028 ( 1999 ) [ arxiv : hep - ph/9903292 ] . b. berdnikov and k. rajagopal , phys . d * 61 * , 105017 ( 2000 ) [ arxiv : hep - ph/9912274 ] . m. a. stephanov , phys . lett . * 102 * , 032301 ( 2009 ) [ arxiv:0809.3450 [ hep - ph ] ] . m. asakawa , s. ejiri and m. kitazawa , phys . rev . lett . * 103 * , 262301 ( 2009 ) [ arxiv:0904.2089 [ nucl - th ] ] . b. friman , f. karsch , k. redlich and v. skokov , arxiv:1103.3511 [ hep - ph ] . r. v. gavai and s. gupta , phys . b * 696 * , 459 ( 2011 ) [ arxiv:1001.3796 [ hep - lat ] ] . m. m. tsypin , phys . lett . * 73 * , 2015 ( 1994 ) .	we point out that the quartic cumulant ( and kurtosis ) of the order parameter fluctuations is universally _ negative _ when the critical point is approached on the crossover side of the phase separation line . as a consequence , the kurtosis of a fluctuating observable , such as , e.g. , proton multiplicity , may become smaller than the value given by independent poisson statistics . we discuss implications for the beam energy scan program at rhic .
in this section , we review the definition of random orthonormal basis and relate it to properties of the moment map for the diagonal action of the maximal torus @xmath74 on co - adjoint orbits of @xmath148 . suppose that we have a sequence of hilbert spaces @xmath0 @xmath149 of dimensions @xmath150 . we define the large hilbert space @xmath151 and orthogonal projections @xmath152 we then consider the orthonormal bases of @xmath153 which arise from sequences of orthonormal bases of @xmath0 . let @xmath13 be a zeroth order pseudo - differential operator . by a toeplitz operator we mean the compression @xmath63 of @xmath154 to @xmath0 . given one onb of @xmath0 , @xmath51 can be identified with a hermitian @xmath155 matrix . we fix orthonormal bases @xmath156 of @xmath0 and introduce the random variables : @xmath157 where @xmath158 . we also define @xmath159 evidently , @xmath160 ( where the @xmath161 term is independent of @xmath162 thus , @xcite the ergodic property of an onb @xmath163 @xmath164 is equivalent to : @xmath165 as mentioned in the introduction , it follows by a standard diagonal argument that almost all the individual elements @xmath40 tend to @xmath6 for all @xmath154 . we do not discuss this step since it is nothing new . in the case where the components of @xmath68 are distinct , the covex polytope @xmath82 is the permutahedron determined by @xmath166 , that is , the simple convex polytope defined as the convex hull of the points @xmath167 where @xmath168 runs over the symmetric group on @xmath102 letters ( i.e. the weyl group of @xmath2 ) . the center of mass is the unique point @xmath169 so that @xmath170 where @xmath171 is the vector from @xmath91 to @xmath172 . the center of mass is evidently invariant under @xmath173 , hence has the form @xmath174 for some @xmath175 and clearly @xmath176 in effect , we want to asymptotically calculate the moments of inertia of the sequence of permutahedra associated to a toeplitz operator . the elementary symmetric polynomials of @xmath109 variables are defined by @xmath177 if one replaces @xmath178 by @xmath179 one obtains the complete symmetric polynomials @xmath180 . the schur polynomials are symmetric polynomials defined by @xmath181 where @xmath182 is a dual partition to @xmath166 . we can compute the moments using the fourier transform of the orbital measure on the orbit of @xmath20 . an explicit formuae for @xmath183 is given in the first line of the proof of theorem 5.1 of @xcite : [ ov ] for @xmath184 , @xmath185 here , @xmath186 is the number of rows of the partition @xmath182 . the degree of @xmath187 is @xmath188 , the number of boxes . since we would like to shift the center of mass of @xmath18 to the origin , we mainly consider @xmath189 the fourier transform of the traceless orbit ( see ) . we now prove : [ 2 m ] @xcite let @xmath190 , and let @xmath191 denote the trace zero diagonal matrix with entries . thus , @xmath192 . then @xmath193 where as above , @xmath194 is the normalized haar probability measure on @xmath66 . this lemma was proved in @xcite using the so - called itzykson - zuber - harish - chandra formua for the fourier transform of the orbit , and again using gaussian integrals . the proof we give here generalizes better to higher moments . we also sketch a proof using the weingarten formulae . we use lemma [ ov ] to obtain @xmath195 we sum over the young diagrams with exactly two boxes and @xmath196 rows . there are just two of them : one row of two boxes or two rows of one box each corresponding respectively to the schur functions @xmath197 , @xmath198 . note that @xmath199 is the kth elementary symmetric function and @xmath200 is the complete kth degree symmetric function . we then translate @xmath95 to @xmath201 so that @xmath202 , i.e. we replace @xmath203 by @xmath204 . since the degree @xmath205 , then we can only use @xmath206 and @xmath207 but @xmath208 for each monomial @xmath209 we have @xmath210 thus , @xmath211 and @xmath212 since the schur polynomials are homogeneous of degree 2 , we can remove the @xmath213 under the schur polynomials to get an overall factor of @xmath214 , which is cancelled by the @xmath215 sign from @xmath23 . thus , @xmath216 since @xmath217 we find that @xmath218 here we use that @xmath219 the formula agrees with the one stated in the lemma [ 2 m ] . as a second proof , we use the weingarten formula for integrals of polynomials over @xmath220 @xcite . we denote the eigenvalues of @xmath221 by @xmath201 . then , @xmath222 the weingarten formulae for these special polynomials state that asymptotically @xmath223 is a complex gaussian random variable of mean zero and variance one . thus , to leading order , @xmath224 and @xmath225 since @xmath226 we get @xmath227 we now prove the 4th moment identity in proposition [ 1 ] , which is the main new step in this article . to calculate the variance of @xmath228 we use the expression in lemma [ varform ] in terms of @xmath229 and then use the formula of lemma [ ov ] . a schur polynomial @xmath230 of degree n in @xmath231 variables is parameterized by a partition of of the degree @xmath232 into @xmath231 parts . when @xmath233 and @xmath234 there are 5 partitions : @xmath235 we note that @xmath236 for all @xmath237 , so @xmath238 . also , @xmath239 is a constant vector . so @xmath240 and @xmath241 here , @xmath242 denotes the hessian . we also use that @xmath243 when @xmath244 . also , @xmath245 then , @xmath246 further , @xmath247 , so that @xmath248 we recall newton s identities , @xmath249 we note that @xmath250 for all @xmath237 , so at @xmath251 , @xmath252 by routine calculations and lemma [ ov ] we have , @xmath253 by , we then have @xmath254 recalling and that @xmath255 , we get @xmath256 further recalling that @xmath257 we finally get @xmath258 since the polynomials are homogeneous of degree @xmath259 , the factor of @xmath213 inside the polynomials may be removed , and we get @xmath260 as @xmath118 the leading asymptotics of the outer terms cancel and the middle term is asymptotic to @xmath261 . we note that @xmath262 is bounded . if the the empirical measure of eigenvalues tends to a limit measure , then @xmath262 tends to its second moment . together with lemma [ orbit - int ] , this completes the proof of proposition [ 1 ] . corollary [ varcor ] follows by subtracting the square of the expectation . by the assumption of definition @xmath263 by lemma [ varform]-corollary [ varcor ] , the variances of the independent random variables @xmath264 are bounded . hence , as explained in the introduction ( see also @xcite ) , ( [ ep * ] ) follows from lemma [ varform]-corollary [ varcor ] and the kolmogorov strong law of large numbers , which gives @xmath265 by ( [ y ] ) , @xmath266 hence also @xmath267 if the dimensions @xmath102 grow fast enough so that @xmath268 is summable , then we obtain a stronger form from the fact that @xmath269 is finite hence the general term must tend to zero almost everywhere . it follows again that @xmath270 almost everywhere . the only condition on the eigenspaces for theorem [ maintheo ] is that holds , and we now recall the known results on this problem . given @xmath50 , we denote the eigenspaces on a flat torus , enumerated in order of the eigenvalue by @xmath0 and by @xmath32 the orthogonal projection to @xmath0 . we use the basis @xmath274 with @xmath275 . then @xmath276 hence @xmath277 in dimensions @xmath144 , @xmath145 . it is proved that lattice points of fixed norm on a sphere of radius @xmath278 become uniformly distributed as @xmath279 @xcite . it follows that @xmath280 as in the last step of the proof of theorem [ maintheo ] , @xmath281 is summable when @xmath144 . the liouville limit formula is true in dimension 4 when the number of lattice points grows linearly in @xmath282 . the condition on n is given in @xcite . in dimension 3 , the equidistribution result is proved in @xcite with similar conditions on the sequence of integers @xmath282 . dimension @xmath283 is more complicated . in dimension 2 , the eigenvalues of integers @xmath282 for which there exist lattice points @xmath284 on the circle @xmath285 . it is necessary that all prime factors of @xmath282 are congruent to 1 modulo 4 . in @xcite it is shown that for almost all such n , the lattice points on the circle become uniformly distributed as @xmath279 . in the case of a generic lattice @xmath286 , the multiplicity of eigenvalues of @xmath23 on @xmath287 is two . the analogue of the eigenspaces above are spectral subspaces for @xmath25 of shriking width @xmath288 . thus , one considers the exponentials @xmath289 for @xmath290 with @xmath291 $ ] . it follows from the lattice point results of @xcite that in dimensions @xmath9 , the number of eigenvalues of an irrational flat torus in @xmath292 $ ] is of order @xmath293 . the question whether the trace asymptotics hold for the span of the corresponding eigenfunctions does not appear to have been studied . theorem [ maintheo ] is not restricted to eigenspaces of the laplacian and is equally valid for spaces of quasi - modes . we refer to @xcite for background on quasi - modes . following @xcite , we define a @xmath294 quasimode of infinite order for @xmath295 with index set @xmath296 to be a family @xmath297 of approximate eigenfunctions satisfying @xmath298 it follows by the spectral theorem that for any @xmath299 , there exists at least one eigenvalue of @xmath295 in the interval @xmath300,\ ] ] and @xmath301 here , @xmath302 denotes the spectral projection for @xmath295 corresponding to the interval @xmath303 . we denote the quasi - classical eigenvalue spectrum of @xmath304by @xmath305 since quasi - eigenvalues @xmath306 are only defined up to errors of order @xmath307 , there is a notion of ` multiple quasi - eigenvalue ' defined as follows : we say @xmath308 if @xmath309 and define the multiplicity of @xmath310 by @xmath311 we then introduce slightly larger intervals @xmath312 ( if need be ) so that @xmath313 here , @xmath314 consists of equivalence classes of indices ( corresponding to equivalence classes of quasimodes ) . we denote by @xmath315 the span of the quasimodes @xmath316 . then @xmath317 theorem [ maintheo ] applies to quasi - mode spaces @xmath315 as long as their dimensions tend to infinity and as long as there exists a unique limit state for @xmath318 . one might expect true modes ( eigenfunctions ) with eigenvalues in the intervals @xmath319 to be close to linear combinations of the quasi - modes with quasi - eigenvalues in that interval . the question raised by theorem [ maintheo ] is whether they behave like random linear combinations or not . if they do , theorem [ maintheo ] gives their quantum limits . in particular , this bears on the question whether @xmath23-eigenfunctions of compact riemannian manifolds @xmath12 with kam geodesic flow might be quantum ergodic . it seems unlikely that they are , but we are not aware of a proof that they are not . for such kam @xmath12 , a large family of quasi - modes is constructed in @xcite which localize on the invariant tori of the kam cantor set of tori . without reviewing the results in detail , the ` large ' family has positive spectral density , i.e. the number of quasi - eigenvalues @xmath320 grows like a positive constant times @xmath321 where @xmath322 . to our knowledge , the multiplicities and trace asymptotics for kam quasi - modes have not been studied at this time . as in the discussion of flat tori , one would need to determine the equidistribution law of the tori in the invariant cantor set corresponding to eigenvalues ( or pseudo - eigenvalues ) of @xmath25 in very short intervals @xmath323 $ ] . the orthonormal basis of eigenfunctions is not simple to relate to the near orthonormal basis of quasi - modes in this case , but we might expect that a positive density of the eigenfunctions are mainly given as linear combinations of kam quasi - modes with quasi - eigenvalues very close to the true eigenvalues . whether or not they are quantum ergodic would reflect the extent to which they are sufficiently random combinations of quasi - modes and the extent to which the collection of quasi - modes in @xmath324 is liouville distributed . g. olshanski and a. vershik , ergodic unitarily invariant measures on the space of infinite hermitian matrices . contemporary mathematical physics , 137 - 175 , amer . ser . 2 , 175 , amer soc . , providence , ri , 1996 .	we consider a sequence @xmath0 of finite dimensional hilbert spaces of dimensions @xmath1 . motivating examples are eigenspaces , or spaces of quasi - modes , for a laplace or schrdinger operator on a compact riemannian manifold . the set of hermitian orthonormal bases of @xmath0 may be identified with @xmath2 , and a random orthonormal basis of @xmath3 is a choice of a random sequence @xmath4 from the product of normalized haar measures . we prove that if @xmath1 and if @xmath5 tends to a unique limit state @xmath6 , then almost surely an orthonormal basis is quantum ergodic with limit state @xmath6 . this generalizes an earlier result of the author in the case where @xmath0 is the space of spherical harmonics on @xmath7 . in particular , it holds on the flat torus @xmath8 if @xmath9 and shows that a highly localized orthonormal basis can be synthesized from quantum ergodic ones and vice - versa in relatively small dimensions . the purpose of this article is to prove a general result on the quantum ergodicity of random orthonormal bases @xmath10 of finite dimensional hilbert spaces @xmath11 of dimensions @xmath1 of a compact riemannian manifold @xmath12 . the proof is based on a moment polytope " interpretation of quantum ergodicity from @xcite : the quantum variances of a hermitian observable @xmath13 are identified with moments of inertia of the convex polytopes @xmath14 defined as the convex hull of the vectors @xmath15 of eigenvalues ( in all possible orders ) of @xmath16 where @xmath17 is the orthogonal projection . equivalently , @xmath18 is the image of the coadjoint orbit @xmath19 of the diagonal matrix @xmath20 under the moment map for the hamiltonian action of the maximal torus @xmath21 of diagonal matrices acting by conjugation on @xmath19 . in particular , the main estimates of quantum ergodicity can be formulated in terms of estimates of the first four moments of inertia of @xmath18 . the main result , theorem [ maintheo ] , states that random orthonormal bases are almost surely quantum ergodic as long as @xmath1 and @xmath22 for all @xmath13 , where @xmath6 is the liouville state . more generally , if these traces have any unique limit state , then almost surely it is the quantum limit of a random orthonormal basis . the proof is essentially implicit in @xcite , but we bring it out explicitly here and also give detailed calculations of the moments of intertia , which seem of independent interest . quantum ergodicity of random orthonormal bases is a rigorous result on the ` random wave model ' in quantum chaos , according to which eigenfunctions of quantum chaotic systems should behave like random waves . it also has implications for the approximation of modes by quasi - modes . since eigenfunctions of the laplacian @xmath23 of a compact riemannian manifold @xmath12 form an orthonormal basis , it is natural to compare the orthonormal basis of eigenfunctions to a ` random orthonormal basis ' . in @xcite , the result of this article was proved for the special case where @xmath0 is the space of degree n spherical harmonics on the standard @xmath7 . in @xcite the quantum ergodic property was generalized to any compact riemannian manifold , with @xmath0 the span of the eigenfunctions in a spectral interval @xmath24 $ ] for @xmath25 . related results have recently been proved in @xcite . the dimension of such @xmath0 grows at the rate @xmath26 where @xmath27 and thus a random element of @xmath0 is a superposition of @xmath26 states . the results of this article show that the same quantum ergodicity property holds for sequences of eigenspaces ( or linear combinations ) whose dimensions @xmath28 tend to infinity at any rate . for instance , the results show that random orthonormal bases of eigenfunctions on a flat torus of dimension @xmath29 are quantum ergodic ( for the precise statement , see [ torus ] , and for further discussion , see [ disc ] . ) to explain the moment map interpretation and the variance formula , recall that quantum ergodicity is concerned with quantum variances , i.e. with the dispersion from the mean of the diagonal part of a hermitian matrix @xmath30 on a large dimensional vector space @xmath0 . the matrix @xmath30 is the restriction @xmath31 to @xmath0 of a pseudo - differential operator @xmath13 ; here @xmath32 is the orthogonal projection to @xmath0 and @xmath33 is the space of pseudo - differential operators of order zero . the same methods and results apply to other context such as semi - classical pseudo - differential operators or to toeplitz operators on holomorphic sections of powers of a positive line bundle @xcite . given an onb @xmath10 of @xmath0 we define the quantum variances of the onb ( indexed by @xmath13 by @xmath34 here , @xmath35 where @xmath36 is normalized liouville measure ( of mass one ) . a sequence @xmath37 of onb s of @xmath0 is a quantum ergodic onb of @xmath38 if @xmath39 by a standard diagonal argument , this implies that almost all the individual elements @xmath40 tend to @xmath6 . since this aspect of quantum ergodicity is the same as in @xcite ( e.g. ) we do not discuss it here . to define random orthonormal bases , we introduce the probability space @xmath41 , where @xmath42 is the infinite product of the sets @xmath43 of orthonormal bases of the spaces @xmath0 , and @xmath44 , where @xmath45 is haar probability measure on @xmath43 . a point of @xmath42 is thus a sequence @xmath46 of orthonormal basis . given one orthonormal basis @xmath47 of @xmath0 any other is related to it by a unique unitary matrix . so the probability space is equivalent to the product @xmath48 where @xmath49 is the unit mass haar measure on @xmath2 . here we are working with hermitian orthonormal bases and hermitian pseudo - differential operators . we could also work with real self - adjoint operators and real orthornormal bases , which are then related by the orthogonal group . the results in that setting are essentially the same but the proofs are somewhat more complicated ; for expository simplicity we stick to the unitary hermitian framework . let @xmath50 and denote the eigenvalues of @xmath51 by @xmath52 . the empirical measure of eigenvalues of @xmath51 is defined by @xmath53 its moments are given by @xmath54 to obtain quantum ergodicity , we we put the following constraint on the sequence @xmath55 : we say that @xmath0 has local weyl asymptotics if , for all @xmath13 , @xmath56 in fact , the results generalize to the case where @xmath6 is replaced by any other limit state , i.e. @xmath57 where @xmath58 is another invariant probability measure for the geodesic flow . our main result is : [ maintheo ] let @xmath0 be a sequence of subspaces of @xmath38 of dimensions @xmath59 . assume that @xmath60 for all @xmath13 . then with probability one in @xmath61 , a random orthonormal basis of @xmath3 is quantum ergodic . a natural question ( which we do not study here ) is whether a random orthonormal basis is que , i.e. whether @xmath62 as a tail event the probability of a random orthonormal basis being que is either 0 or 1 . we now explain how to formula theorem [ maintheo ] in terms of moment maps and polytopes . quantum ergodicity of a random orthonormal bases concerns the dispersion from the mean of the diagonal part of @xmath63 . the diagonal part depends on the choice of an orthonormal basis of @xmath0 . once an orthonormal basis is fixed , @xmath64 can be identified with an element @xmath30 of the lie algebra @xmath65 of @xmath66 , and a unitary change of the orthonormal basis results in the conjugation @xmath67 of @xmath30 . if the vector of eigenvalues of @xmath30 is denoted @xmath68 , then the conjugates sweep out the orbit @xmath69 . let @xmath70 denote the cartan subalgebra of diagonal elements in @xmath65 , and let @xmath71 denote the euclidean inner product on @xmath70 . also let @xmath72 denote the orthogonal projection ( extracting the diagonal ) . extracting the diagonal from each element of the orbit is precisely the moment map @xmath73 of the conjugation action of @xmath74 . finally , let @xmath75 for hermitian matrices @xmath76 . we also introduce notation for the diagonal of @xmath77 : @xmath78 thus , @xmath79 with @xmath80 traceless , corresponds to the decomposition @xmath81 . as this description indicates , quantum ergodicity of random orthonormal bases is mainly a result about the asymptotic geometry of the polytopes @xmath82 corresponding to a sequence @xmath63 of toeplitz operators . the pushforward of the @xmath2-invariant normalized measure on @xmath19 to @xmath18 is the so - called duistermaat - heckman measure @xmath83 , a piecewise polynomial measure on @xmath84 . to prove almost sure quantum ergodicity , we prove that for all such sequences @xmath63 and their spectra @xmath85 , the second and fourth moments of inertia of @xmath18 with respect to @xmath86 are bounded . we use the property in definition [ trace ] to replace @xmath6 by the centers of mass , i.e. the scalar matrix with the same trace as @xmath63 . the kolmogorov strong law of large numbers then gives the quantum ergodicity property . in @xcite , we study higher moments and their implication for the limit shape of @xmath18 along a sequence @xmath87 with a limit empirical measure . we asymptotically evalute the moments using the fourier transform @xmath88 of the @xmath89-function on @xmath19 . here , we assume @xmath90 . we may identify @xmath91 with a diagonal matrix , and then @xmath92 , and we get the standard fourier transform . we obviously have : [ varform ] let @xmath23 be the euclidean laplacian of @xmath93 acting in the @xmath91 variable . then , @xmath94 we translate @xmath95 by its center of mass to make the center of mass of @xmath18 equal to @xmath96 , i.e. @xmath97 . using a formula for @xmath98 in terms of schur polynomials , we prove [ 4 ] let @xmath99 be the power functions . assume that @xmath100 . then , @xmath101 the proof of theorem [ maintheo ] follows directly from lemma [ 4 ] and the kolmogorov slln ( strong law of large numbers ) . when @xmath102 grows fast enough it also follows directly from the borel - cantelli lemma . we first introduce notation for the basic random variables : [ yndef ] @xmath103 y^a_n({\bf \psi } ) : = \\|{j_{d_n}}(u_n^ * d({\vec \lambda } ) u_n ) - \overline{d({\vec \lambda } ) } \\|^2 = \\|{j_{d_n}}(u_n^ * d_0({\vec \lambda } ) u_n ) \\|^2 \end{array } \right.\ ] ] then lemma [ varform]-lemma [ 4 ] determine the asymptotics of their mean and variance @xmath104 [ varcor ] @xmath105 the lemma first implies that @xmath106 is bounded for all @xmath13 . hence , @xmath107 as long as @xmath1 . thus , the mean of the quantum variances tends to zero . as in @xcite we then apply the kolmogorov slln ( or the martingale convergence theorem ) . the @xmath108 is a sequence of independent random variables as @xmath109 varies and lemma [ 4 ] shows that they have bounded variance . hence the slln implies that the partial sums , @xmath110 have the property , @xmath111 and this is equivalent to quantum ergodicity of random orthonormal bases . as mentioned above , if @xmath102 grows at a faster rate one can obtain stronger results from the borel - cantelli lemma : e.g. if @xmath112 , one obtains almost sure convergence @xmath113 ( a.s . ) . since the argument above only requires that @xmath114 and @xmath115 is bounded , it does not require any assumption that @xmath116 tends to a limit . our calculations therefore go beyond what is necessary for almost sure quantum ergodicity , and pertain to the asymptotic geometry of the polytopes @xmath18 . there is a natural condition on the this sequence of polytopes : [ limit ] we say that the sequence @xmath55 has szeg asymptotics if , for all @xmath13 , there exists a unique weak * limit , @xmath117 as @xmath118 . here , @xmath119 is the set of probability measures on @xmath120 . under this stronger assumption , lemma [ 4 ] gives moment asymptotics : [ 1 ] let @xmath121 be a sequence of vectors with the property that the empirical measures tend to a weak limit @xmath122 . then @xmath123 this proposition is closely related to the weingarten theorem " that the matrix elements @xmath124 are asymptotically complex normal random variables , where @xmath125 are the matrix elements of @xmath126 @xcite . perhaps this explains why the fourth moment is a constant multiple of the square of the second moment . it would be interesting to see if the pattern continues ; we plan to study @xmath18 further in @xcite . the motivation for proving quantum ergodicity of random orthonormal bases for @xmath0 of any dimensions tending to infinity was prompted by the general question : how many diffuse states ( modes or quasi - modes ) does it take to synthesize localized modes or quasi - modes ? vice - versa , how many localized states does it take to synthesize diffuse states ? we would like to synthesize entire orthonormal bases rather than individual states and measure the dimensions of the space of states in terms of the planck constant @xmath127 . let us consider some examples . in the case of the standard @xmath7 , the eigenspaces @xmath0 of @xmath23 are the spaces of spherical harmonics of degree @xmath109 . they have the well - known highly localized basis @xmath128 of joint eigenfunctions of @xmath23 and of rotations around the @xmath129-axis . by localized we mean that a sequence @xmath130 with @xmath131 microlocally concentrates on the invariant tori in @xmath132 where @xmath133 . here , @xmath134 where @xmath135 generates the @xmath129-axis rotations . on the other hand , it is proved in @xcite that independent random " orthonormal bases of @xmath0 are quantum ergodic , i.e. are highly diffuse in @xmath132 . since @xmath136 , it is perhaps not surprising that the same eigenspace can have both highly localized and highly diffuse orthonormal bases when its dimenson is so large . the question is , how large must it be for such incoherently related bases to exist ? a setting where the eigenvalues have high multiplicity but of a lower order of magnitude than on @xmath7 is that of flat rational tori @xmath137 such as @xmath138 . of course it has an orthonormal basis of localized eigenfunctions , @xmath139 . the key feature of such rational tori is the high multiplicity of eigenvalues of the laplacian @xmath23 of the flat metric . it is well - known and easy to see that the multiplicity is the number of lattice points of the dual lattice @xmath140 lying on the surface of a euclidean sphere . we denote the distinct multiple @xmath23-eigenvalues by @xmath141 , the corresponding eigenspace by @xmath0 and the multiplicity of @xmath142 by @xmath143 . in dimensions @xmath144 , @xmath145 , one degree lower than the maximum possible multiplicity of a @xmath23-eigenvalue on any compact riemannian manifold , achieved on the standard @xmath146 . further , @xmath22 . hence , the results of this article show that despite the relatively slow growth of @xmath102 on a flat rational torus , orthonormal bases of @xmath0 in dimensions @xmath29 are almost surely quantum ergodic . the statement for dimensions @xmath147 is more complicated ( see [ torus ] ) . an interesting setting where the behavior of eigenfunctions is largely unknown is that of kam systems . for these , one may construct a ` nearly ' complete and orthonormal basis for @xmath38 by highly localized quasi - modes associated to the cantor set of invariant tori . it seems unlikely that the actual eigenfunctions are quantum ergodic ; but the results of this article show that if they resemble random combinations of the quasi - mode , then it is possible that they are . further discussion is in [ qm ] .
single - molecule magnets ( smms ) have been extensively studied for the past decade because of both scientific and practical reasons : macroscopic quantum phenomena@xcite and possible utilization as magnetic storage devices.@xcite a prototype of the smms is [ mn@xmath0o@xmath0(ch@xmath4coo)@xmath5(h@xmath6o)@xmath3 ] @xmath82(ch@xmath4cooh)@xmath84(h@xmath6o ) ( hereafter mn@xmath0),@xcite which is a three - dimensional array of identical @xmath1 molecules . very recently , derivatized mn@xmath0 type molecules were successfully deposited on a gold film,@xcite which further enhances the prospects for storing magnetic information in a single molecule . a single molecule of mn@xmath0 has four ferromagnetically coupled mn@xmath9 ions ( @xmath10 ) in the cubane and eight ferromagnetically coupled mn@xmath11 ions ( @xmath12 ) in the crown as schematically shown in fig . [ fig : mn12sch ] . in the ground state , the magnetic moments of the eight mn@xmath11 ions are antiparallel to those of the four mn@xmath9 ions which leads to a total spin of @xmath1 . so far many interesting features of the smm mn@xmath0 such as magnetization steps in the hysteresis loops,@xcite electron paramagnetic resonance ( epr ) transitions,@xcite and low - energy excitations in inelastic neutron scattering,@xcite have been well understood by considering each molecule as an object with an effective spin of @xmath1 . however , magnetic susceptibility measurements@xcite have demonstrated that some experimental data could not be in accord with the @xmath1 single - spin picture and suggested that the first excited - state manifold may be located at 35 k@xmath1340 k above the @xmath1 ground - state manifold within which the energy barrier to magnetization reversal is 65 k in zero field . since the first excited - state manifold overlaps with the ground - state manifold above 35 k@xmath1340 k , an internal many - spin structure of a single molecule should be included to explain the high - energy experimental data . the dimension of the hilbert space of this smm is so large that a simplified eight - spin model with strong dzyaloshinsky - morya interaction was first proposed to include many - spin effects by dimerizing two strongly bonded mn spins.@xcite although this eight - spin model explained some experimental data , it was limited to features below 50 k. at higher temperatures the dimerization scheme breaks down . later all twelve mn spins were included in the heisenberg hamiltonian and excited - state manifolds were clarified by diagonalization of the hamiltonian with constraining the fixed energy gap between the first excited - state and the ground - state manifold of 35 k.@xcite nonetheless , there is still a big controversy over the energy gap between the first excited - state and the ground - state manifold , and there has been less exploration of high - energy features caused by many - spin effects . here `` high - energy '' means low - lying excited manifolds above the @xmath1 ground - state manifold . recent high - energy inelastic neutron scattering measurements exhibited a broad anomalous peak at 10 k , which may have a magnetic origin ( not due to phonons ) and could not be rationalized by the @xmath1 single - spin picture because the first allowed peak by the picture is about 14 k.@xcite more recent high - field epr measurements revealed that several transitions could not be justified by the @xmath1 single - spin manifold but that they were well explained if we assumed the @xmath2 first excited - state manifold to be situated at 10 - 16 k above the ground - state manifold.@xcite additionally , there have been many attempts to determine the signs and magnitudes of the exchange couplings between mn ions from different approaches@xcite but no consensus has arisen . to elucidate the energy gap and examine properties within the low - energy excited manifolds , a single molecule of the mn@xmath0 is considered and the exchange interactions between mn spins are calculated using density - functional theory ( dft ) , which are presented in secs . ii and iii . our calculated values are compared with those from other groups.@xcite through diagonalization of the isotropic heisenberg exchange hamiltonian with our calculated exchange constants using the lanczos method,@xcite excited - state manifolds are manifested and energy gaps between different manifolds are calculated in sec . the magnetic anisotropy barriers for low - energy spin excitations are calculated using dft and the anisotropy barriers for low - lying excited manifolds are discussed in sec . v. our conclusions are presented in sec . our dft calculations@xcite are performed with spin - polarized all - electron gaussian - orbital - based naval research laboratory molecular orbital library ( nrlmol ) @xcite which is ideal for studying a single molecule to a small number of unit cells . here we use the perdew - burke - ernzerhof ( pbe ) generalized - gradient approximation ( gga ) in the exchange - correlation potential.@xcite since the smm mn@xmath0 has fourfold symmetry , for fully symmetrized calculations the total number of inequivalent atoms to consider is reduced to @xmath14 . to save geometry - optimization time without losing interesting physical properties , the following simplified form of the smm mn@xmath0 is used : [ mn@xmath0o@xmath0(hcoo)@xmath5(h@xmath6o)@xmath3 ] ( acetic acids and water molecules of crystallization are not included , and 16 acetates , ch@xmath4coo , are replaced by 16 formates , hcoo ) . the zero - field total anisotropy barrier for the @xmath1 ground - state manifold does not change much with this simplification . the total magnetic moment for the ground state was confirmed to be @xmath15 , which is in good agreement with experiment and it is stable . details of the optimization schemes and electronic properties of the optimized geometry for the ground state were discussed elsewhere.@xcite hereafter , unless specified , our calculations have been carried out for the above simplified form . the fourfold symmetry and the geometry of a single mn@xmath0 molecule indicate that there are three symmetrically inequivalent mn sites and four different exchange interactions between mn spins as shown in fig . [ fig : mn12sch ] . to calculate the exchange constants using dft , it is assumed that the magnetic moments of all mn ions are aligned along a particular direction ( collinear ) . then eleven distinctive spin configurations are constructed by reversing magnetic moments of a few mn ions simultaneously from the @xmath1 ground - state . ( see table [ table:1 ] ) for example , a @xmath16 state ( @xmath17 is an eigenvalue of the total spin operator @xmath18 projected along the particular direction ) can be built by flipping both one mn spin in the cubane and another mn spin in the crown within the @xmath1 ground state : @xmath19 . there are several ways to construct @xmath16 states with collinear spins that are labeled as 9-b , 9-c , and 9-d in table [ table:1 ] . all of the examined collinear spin configurations are _ not _ eigenstates of @xmath20 . mean - field calculations are not directly applicable to optimizing excited states such as @xmath21 in mn@xmath0 , because those eigenstates are represented by linear combinations of many single slater determinants . the geometries for the distinctive spin configurations are taken to be the same as that for the ground state except for the proper spin arrangements . it is confirmed that a slightly different initial geometry for the collinear spin configurations does not siginificantly alter our calculated values of the exchange constants . since fourfold symmetry is broken for the spin configurations considered , unsymmetrized calculations are required with a total number of 100 atoms . the energies of the spin configurations are self - consistently minimized with a small basis set and fine mesh.@xcite our calculations show that the minimized energies range from 0.04 ev to 0.2 ev above the ground - state ferrimagnetic structure . from ref . it is known that for the @xmath1 ground state the lowest electronic excitation is a majority spin excitation between the @xmath22 levels in mn(3d ) states , which is about 0.44 ev ( majority lumo - homo gap ) , and that the energy gap between minority lowest unoccupied molecular orbital ( lumo ) and majority highest occupied molecular orbital ( homo ) is @xmath23 ev . comparison of the minimized energies to the lumo - homo gaps indicates that these different spin configurations are lower - energy spin excitations than moving an electron from the majority homo to the unoccupied majority ( or minority ) orbital . thus they will be called low - energy spin excitations . there are twelve equations to solve for five unknowns ( the background energy , @xmath24 , four exchange constants @xmath25 , @xmath26 , @xmath27 , and @xmath28 ) so a least - square - fit ( lsf ) method@xcite is used . our calculated values of @xmath29 s are shown in table [ table:2 ] in comparison with results of other groups . when the minimized energies of the low - energy spin excitations are recalculated using the calculated values of @xmath29 s , it is found that they are in good agreement with the dft - calculated energies as indicated in table [ table:1 ] . prior to comparison to the results of other groups , let us briefly review the essence of the different approaches from the different groups . in ref . the value of @xmath25 was first determined from the experimental data , and then @xmath26 and @xmath27 were varied with fixed value of @xmath30 to reproduce the @xmath1 ground state . in ref . the values of @xmath29 s were slightly varied from the previously reported values@xcite using diagonalization of heisenberg exchange hamiltonian to provide the @xmath1 ground state and the energy gap of 35 k between the ground - state and the @xmath2 first excited - state manifold . in ref . the exchange parameters were obtained to reproduce their megagauss experimental data with a constraint of the energy gap of 35 k. in ref . the lda@xmath31u method was used to include electron correlations between mn ions , and the values of the intra - atomic hund s exchange parameter @xmath29 and the average coulomb parameter @xmath32 were determined to obtain the correct magnetic moment for the @xmath1 ground state . then the exchange constants were calculated from the values of @xmath29 and @xmath32 . our calculations show that the coupling between one mn spin in the cubane and another mn spin in the crown closest to the mn spin in the cubane , @xmath25 , is confirmed to be the strongest antiferromagnetic coupling , although its absolute magnitude does not agree among different groups . our calculated value of @xmath25 agrees somewhat with those from refs . . the coupling between a mn spin in the cubane and the other type of mn spin in the crown , @xmath26 , is confirmed to be weaker antiferromagnetic than @xmath25 . but the rate of @xmath26 to @xmath25 does not agree among different groups . our value of @xmath26 is between the values obtained from refs . . the coupling between mn spins in the cubane , @xmath27 , turns out to be the weakest and is ferromagnetic . the coupling between mn spins in the crown , @xmath28 , is weakly antiferromagnetic . overall , the dft calculated values are closest to the results obtained from ref . , and they are 20%-40% larger than those from the lda@xmath31u method except for the weakest @xmath27 . however , they are significantly smaller than the other two reported results from refs . . our calculated values of @xmath29 may be overestimated compared to experimental data for the following reasons . it has been found from other smms such as mn@xmath3[ref . ] and v@xmath33[ref . ] that the dft - calculated exchange interactions were overestimated by a factor of 2 and 3 compared to the experimental data respectively.@xcite this overestimated exchange interactions are due to the fact that the pbe generalized - gradient approximation does not fully treat self - interaction corrections of the localized @xmath34 states in the coulomb potential . proper treatment of the electron correlations in mn atoms may improve our dft calculations of the exchange interactions . this is supported by the fact that the lda@xmath31u method produces 20%-40% smaller exchange interactions than our dft calculations and that the exchange interactions decrease with increasing the value of @xmath32 in the lda@xmath31u method.@xcite to calculate the energy gap between the ground - state and the first excited manifold , we use the following isotropic ( i.e. no anisotropy ) heisenberg exchange hamiltonian : @xmath35 where the sum runs over all pairs connected by the exchange interactions and the values of the coupling constants @xmath36 labeled in fig.[fig : mn12sch ] are in table [ table:2 ] . the exchange hamiltonian is a @xmath37 matrix , which is too large to diagonalize using currently available computers . thus , using the fact that @xmath17 is a good quantum number , we classify the total number of the @xmath38 states into 45 different constant @xmath17 states such as @xmath39 . since we are interested in low - lying excited manifolds , we examine @xmath40 , @xmath41 , @xmath42 , @xmath43 , and @xmath44 states only . if the @xmath1 manifold is the ground state , then the lowest - energy eigenstate of the exchange hamiltonian with a basis set of all @xmath42 states corresponds to @xmath45 . additionally the lowest - energy eigenstate of the hamiltonian with a basis set of all @xmath43 states corresponds to @xmath46 and it should have the same energy as @xmath45 . the eigenstates with the same @xmath18 must be degenerate if they belong to the same manifold . the first - excited eigenstate of the hamiltonian with the basis set of all @xmath43 states corresponds to @xmath47 if the @xmath2 manifold is the first - excited manifold above the @xmath1 ground state . there are a total number of 269148 @xmath48 states , 484144 @xmath41 states , 817176 @xmath42 states , and 1299632 @xmath43 states , 1954108 @xmath44 states . we diagonalize the full hamiltonian matrices for the @xmath40 , @xmath41 , @xmath42 , @xmath43 , and @xmath44 states using the lanczos method.@xcite a few calculated low - lying energy eigenvalues are shown in table [ table:3 ] . the lowest energy for the @xmath42 states is identical to those for the @xmath43 and @xmath44 states , but it is much lower than the lowest energies for the @xmath41 and @xmath40 states . the lowest energies for the @xmath49 states will increase with increasing @xmath50 . this indicates that the ground - state has @xmath1 . next higher energies are from the @xmath43 and @xmath44 states . the three excited energy eigenvalues for the @xmath43 states coincide with those for the @xmath44 states which are much lower than the excited energies for the @xmath42 , @xmath41 , and @xmath40 states . this suggests that the three excited states for the @xmath43 ( @xmath44 ) states belong to @xmath2 manifolds . so the first excited - state manifold has @xmath2 located at 40.5 k above the @xmath1 ground state , and there are two more @xmath2 excited manifolds located at 7 k and 40 k above the @xmath2 first excited manifold ( fig . [ fig : elevel ] ) . our calculated values of the energy gaps are upper limits because dft calculations may overestimate the exchange interactions due to the reasons discussed in sec . if the exchange constants are reduced by a half , then the energy gaps will be also reduced by a half . single - ion anisotropy parameters which we did not include in eq . ( [ eq : ham_ms ] ) does not significantly affect the energy gap . the magnetic anisotropy barriers for the eleven low - energy spin excitations as well as the ground state are calculated with the assumption that the barriers are caused by the spin - orbit interaction only . other effects such as noncollinearity of the magnetic moments of mn ions and spin - orbit - vibron coupling@xcite on the anisotropy barrier will not be included in our study . however correlation effects due to the addition of multideterminantal wavefunctions are addressed in part by the further ongoing discussion . the spin - orbit interaction @xmath51 can be decomposed into one - electron operators @xmath52 related to the electric fields caused by nuclei and two - electron operators @xmath53 related to the electric fields due to the rest of electrons as follows : @xmath54 where the summation for @xmath55 ( @xmath56 ) runs over all electrons located at @xmath57 ( all nuclei located at @xmath58 with nuclear number @xmath59 ) , @xmath60 is the speed of light , @xmath61 is the spin operator of the @xmath55th electron , and @xmath62 is the momentum of the @xmath55th electron . within the self - consistent field ( scf ) approximation , we have calculated the wave function @xmath63 satisfying @xmath64 , where @xmath65 is the many - electron hamiltonian without spin - orbit coupling , @xmath66 is a spatial function and @xmath67 is a spinor . therefore , the energy shift , @xmath68 , due to @xmath51 using second - order perturbation theory is written by@xcite @xmath69 where @xmath70 is the energy of the occupied state with spinor @xmath71 , and @xmath72 is the energy of the unoccupied state with spinor @xmath73 . the summation runs over all occupied and unoccupied states of all atoms within a certain energy window and over up and down spinor states . the contributions for the cases that both of @xmath55 and @xmath74 are occupied ( or unoccupied ) to the energy shift cancel each other out . for uniaxial systems such as mn@xmath0 the second - order energy shift is simplified as follows : @xmath75 where @xmath76 is the uniaxial anisotropy parameter which can be calculated from eq . ( [ eq : delta2 ] ) . for the @xmath1 ground state , as shown in table [ table:1 ] the uniaxial parameter is about 0.54 k , which is in good agreement with experiment.@xcite an eigenstate @xmath77 of the many - electron hamiltonian that belongs to an excited - state manifold can be written in terms of a linear combination of many slater determinants : @xmath78 , where @xmath79 is a single slater determinant ( orthonormal basis function ) , and some of @xmath79 correspond to our examined low - energy spin excitations . therefore , the expectation value of the spin - orbit coupling @xmath51 with respect to @xmath77 , is decomposed into diagonal and off - diagonal elements such as @xmath80 where the summations run over spatial coordinates and spin variables of all electrons in mn@xmath0 . @xmath51 is made of one - electron and two - electron operators as shown in eq . ( [ eq : vls-2 ] ) . to facilitate calculations of the diagonal and off - diagonal elements , the off - diagonal elements of the one - electron operator , @xmath81 , are explicitly written in terms of single - electron spin - orbitals . @xcite @xmath82 where @xmath83 is also applied to the single - electron spin - orbitals and the normalization factor @xmath84 cancels out @xmath85 permutations of the @xmath86 single - electron spin - orbitals in @xmath87 . as can be seen from eq . ( [ eq : vls-7 ] ) , because all spin - orbitals @xmath88,@xmath89 @xmath90 , @xmath91,@xmath89 , @xmath92 , are orthonormal to each other , the off - diagonal elements of the one - electron operator @xmath52 are nonvanishing when only one single - electron spin - orbital is different between @xmath79 and @xmath93 . the off - diagonal elements of the two - electron operator @xmath53 are nonvanishing when no more than two single - electron spin - orbitals are different between @xmath79 and @xmath93 . different single slater determinants ( or different low - energy spin excitations ) are not connected by the one- or two - electron operators , so @xmath94 for @xmath95 . hence , the off - diagonal contributions to the energy shift @xmath68 is zero . as shown in table [ table:1 ] , the calculated magnetic anisotropy barriers for the low - energy spin excitations are almost the same as that for the ground state @xmath1 manifold . this indicates that the diagonal elements of the spin - orbit interaction [ eq . ( [ eq : vls-6 ] ) ] do not vary much with different spin excitations : @xmath96 for @xmath97 . then @xmath98 where @xmath99 by definition . so the magnetic anisotropy barriers for low - lying excited - state manifolds must be approximately the same as that for the ground - state manifold . note that we have calculated second - order anisotropy barriers only so that the barriers shown in table [ table:1 ] are all between 54 k and 55 k. the results of this discussion suggest that the dft - based determination of mae is valid under the following conditions : ( i ) single - electron and two - electron excitations cost an order of ev , ( ii ) considered collinear spin excitations cost an order of magnitude less energy than the single - electron and two - electron excitations . we have calculated the intramolecular exchange interactions among mn spins in the single - molecule magnet mn@xmath0 ( [ mn@xmath0o@xmath0(hcoo)@xmath5(h@xmath6o)@xmath3 ] ) considering localized low - energy spin excitations using density - functional theory . our calculated values of the exchange constants agree with the results from refs . . with our calculated exchange constants , we have diagonalized the isotropic heisenberg hamiltonian using the lanczos method . we have confirmed that the ground state is @xmath1 and that there is a @xmath2 first - excited manifold located at 40.5 k above the @xmath1 ground - state manifold . this energy gap ( 40.5 k ) is an upper limit because it is likely reduced by including more exact treatments of the coulomb potential or electron correlations . this indicates that the first excited manifold may be situated much lower than the high energy levels within the @xmath1 ground - state manifold . we have also calculated the magnetic anisotropy barriers for the low - energy spin excitations using dft and the second - order perturbation theory with the assumption that the spin - orbit coupling is the most important interaction . our dft calculations have showed that the anisotropy barriers for the low - energy spin excitations are approximately the same as that for the @xmath1 ground state . from this result and our understanding of the spin - orbit interaction , we conclude that the anisotropy barriers for the low - lying excited manifolds are approximately same as that for the @xmath1 ground state . kp was funded by w. m. keck foundation , mrp and csh were supported in part by onr and the dod hpc chssi program . a. cornia , a. c. fabretti , m. pacchioni , l. zobbi , d. bonacchi , a. caneschi , d. gatteschi , r. biagi , u. del pennino , v. de renzi , l. gurevich , and h. s. j. van der zant , angew . int.ed . * 42 * , 1645 ( 2003 ) . s. hill , j. a. a. j. perenboom , n. s. dalal , t. hathaway , t. stalcup , and j. s. brooks , phys . lett . * 80 * , 2453 ( 1998 ) ; j. a. a. j. perenboom , j. s. brooks , s. hill , t. hathaway , and n. s. dalal , phys . b * 58 * , 330 ( 1998 ) d. n. hendrickson , g. christou , e. a. schmitt , e. libby , j. s. bashkin , s. wang , h. -l . tsai , j. b. vincent , p. d. w. boyd , j. c. huffman , k. folting , q. li , and w. e. streib , j. am . chem . soc . * 114 * , 2455 ( 1992 ) . .here @xmath100 denotes @xmath101 , where @xmath17 is an eigenvalue of the total spin operator projected along the easy axis and `` c '' is attached to distinguish between different types of @xmath101 states . from the second column , shown are flipped spins to create low - energy spin excitations relative to the @xmath1 ground - state labeled in fig . , ising energy expressions , dft - calculated energies relative to the ground state , @xmath102 ( in units of ev ) , the energy differences between dft results and least - square - fit ( lsf ) , e@xmath103@xmath13e@xmath104 , and the magnetic anisotropy barriers ( mae ) ( in units of k ) . [ cols="^,^,^,^,^,^",options="header " , ] . the empty circles denote mn@xmath9 ions , the filled circles denote mn@xmath11 ions , and each mn ion is numerically labeled . the bond length between spin 1 and 9 is shorter than that between spin 1 and 6 . the thick solid lines are for @xmath25 , the thick dashed lines are for @xmath26 , the thin solid lines in the inner cubane are for @xmath27 , and the thin solid lines in the outer crown are for @xmath28.,scaledwidth=40.0% ]	most experimental data on the single - molecule magnet mn@xmath0 acetate have been successfully explained by the assumption that the mn@xmath0 acetate has an effective ground - state spin of @xmath1 . however , the effect of the low - lying excited manifolds caused by interactions between mn spins has not been well understood . to investigate the features of the low - lying excited manifolds , the intramolecular exchange interactions are calculated using density - functional theory ( dft ) . with the calculated exchange parameters , the energy gap between the @xmath1 ground - state and the first excited - state manifold is calculated by diagonalization of the heisenberg hamiltonian . the upper limit on the energy gap is about 40.5 k which is likely to be overestimated due to incomplete treatment of the coulomb potential within dft . it is found that there are several @xmath2 low - energy excited - state manifolds above the @xmath1 ground - state manifold . the magnetic anisotropy barriers for the low - lying spin excitations are calculated using dft . based on the calculations , it is found that the anisotropy barriers for the low - lying excited manifolds are approximately the same as that for the ground - state manifold , which is applicable to other single - molecule magnets such as mn@xmath3 .
the wigner @xcite function @xmath0 is a representation of the quantum density operator @xmath1 as a real function on phase space , for a non - relativistic dynamical system with classical - quantum correspondence and @xmath2 degrees of freedom . here , all the underlying geometry is euclidean and we can include the wigner function within the general framework of weyl representation of operators : @xmath3 , so @xmath4 , where @xmath5 denotes a point in phase space . in this context , the quantum evolution of the density operator @xmath6 effected by an autonomous external hamiltonian @xmath7 can be written in terms of its weyl @xcite representation as : @xmath8 where @xmath9 , @xmath10 is the weyl symbol of @xmath7 and @xmath11 is the classical poisson bracket . even though the wigner function @xmath0 is generally nonpositive and thus can not be taken as a classical probability density ( in contrast to its lagrangian averages @xmath12 , etc ... ) , equation ( [ ham ] ) is the source of a somewhat widespread belief that the propagation of @xmath13 coincides in the semiclassical limit ( @xmath14 ) with the classical ( @xmath15 ) propagation of a liouville probability density . in fact , for a quadratic hamiltonian @xmath16 the semiclassical ( exact ) propagation of @xmath13 is indeed classical . that is , if @xmath17 is the classical hamiltonian flow generated by @xmath18 , then @xmath19 for a general hamiltonian @xmath7 ( whose weyl symbol @xmath10 does not necessarily coincide with the classical hamiltonian @xmath20 ) the classical propagation ( [ folk ] ) induced by ( [ ham ] ) remains a good approximation ( semiclassically correct ) if @xmath0 is a fairly smooth function ( as with the weyl representation of an observable ) , in which case the corrections to ( [ ham ] ) can be semiclassically ignored . such a smoothness condition can be realized for highly mixed statistical states , in which case @xmath0 will look very much like a classical probability density in phase space and not surprisingly its propagation will be nearly classical . on the other hand , pure states are not in general represented by smooth wigner functions . indeed , the semiclassical limit @xmath21 for the wigner function of a pure state @xmath22 in one degree of freedom @xcite is : @xmath23 where @xmath24 is the symplectic area between an arc of a ( bohr - sommerfeld quantized ) curve @xmath25 and its corresponding chord centered at @xmath26 , and @xmath27 is a smooth amplitude function . such an expression for @xmath21 is highly oscillatory , specially at this semiclassical limit , and its successive derivatives knock off all the favorable powers of @xmath28 in the `` corrections '' to ( [ ham ] ) . the important exception is when @xmath26 lies very close to @xmath25 in which case @xmath24 is very close to zero and constant , and @xmath29 is locally smooth . however , the regions inside a closed leaf @xmath25 , where @xmath21 is highly oscillatory , are of utmost importance for they neatly exhibit the nonpositive and thus nonclassical aspect of @xmath0 which accounts for quantum interference and coherence phenomena . thus , ( [ folk ] ) provides an inadequate description of the propagation of wigner functions at the semiclassical level . most of these problems with the propagation of semiclassical wigner functions were already clearly discussed by heller @xcite in 1976 , even before the semiclassical approximations for @xmath0 were properly developed @xcite@xcite . the asymptotic expansion ( [ ham ] ) in @xmath28 was then partially resummed to obtain an improved propagation , though in a not very general context . from another approach , starting with the integral expression of the evolution equation , marinov @xcite derived in 1991 a path integral representation for the propagator of wigner functions , obtaining its semiclassical limit , which leads to ( [ folk ] ) . in this paper we develop the geometrical explanation of why the semiclassical limit of the propagation of wigner functions can not be derived from the semiclassical limit of their propagator . thus , in @xmath30 we present the integral equation for the wigner evolution , defining its kernel , the wigner propagator , whose stationary phase evaluation necessarily leads to classical propagation . then , in @xmath31 we recollect the constructions of semiclassical wigner functions which will permit , in @xmath32 , a correct stationary phase evaluation of the full integral equation for their evolution . this leads , in @xmath33 , to a simple geometrical prescription for the semiclassical limit of the propagation of a wigner function in terms of the classical flow of the tips of the chords centered on the arguments of the initial semiclassical wigner function . we conclude in @xmath34 with a discussion on the geometrical meaning of the difference between classical liouville and semiclassical wigner propagation . the starting point of our analysis is the expression for the product of operators in the center ( or weyl ) representation . thus , if @xmath35 and @xmath36 are the center representations of @xmath37 and @xmath38 , then the center representation for @xmath39 is given @xcite@xcite by the integral moyal product : @xmath40(x ) = \int dx ' dx '' a(x ' ) b(x '' ) exp(i\delta(x , x',x'')/\hbar ) \ , \ ] ] where @xmath41 is the symplectic area of the triangle with these midpoints . iterating ( [ prod ] ) we obtain the integral equation for the evolution of wigner functions as the weyl transform of the quantum evolution equation @xmath42 : @xmath43 where @xmath44 is the weyl propagator , i.e. the weyl transform of the unitary evolution operator @xmath45 . the @xmath46-function prescribes the 4 points in @xmath47 as vertices of a parallelogram . the phase is twice the area of this parallelogram ( times @xmath48 ) . but this is also the area of any element of a continuous family of quadrilaterals circumscribed to the given parallelogram . in other words , we can identify the symplectic area of the parallelogram with vertices at @xmath49 as half of the symplectic area of any quadrilateral whose sides are centered on these points . this brings the product rule for three operators in line with that for two operators and , indeed , the product of any number of operators will depend on the corresponding circumscribed polygon @xcite . thus , denoting by @xmath50 the area of a quadrilateral as function of its midpoints , we have that @xmath51 and this function is well defined only on the subset @xmath52 , isomorphic to @xmath53 , determined by the @xmath46-function . fixing one ( say , the first , @xmath26 ) of these points , we obtain a subset @xmath54 , isomorphic to @xmath55 . denoting its induced measure by @xmath56 , we can rewrite ( [ tp1 ] ) as @xmath57 similarly , we can reparametrize the parallelogram by identifying : @xmath58 , @xmath59 , @xmath60 , in which case we can use ( [ tp1 ] ) to get an expression for the wigner propagator or kernel @xmath61 , via its defining formula @xmath62 , in the following form : @xmath63 marinov @xcite has derived an explicit path integral representation for the wigner propagator @xmath61 . this can also be achieved by introducing in ( [ l ] ) the path integral representation @xcite for the weyl propagator @xmath44 . for this latter , there exists a well - established semiclassical limit @xcite@xcite : @xmath64 ( x ) \right\ } \ , \ ] ] where @xmath65 is the symplectic area between the classical trajectory @xmath66 ( determined by the weyl hamiltonian @xmath10 ) and the chord centered on @xmath26 , with @xmath67 being the energy of this trajectory , and where , again , @xmath68 is a slow - varying real amplitude function . for sufficiently small times we can guarantee the existence of a single trajectory . eventually there may be bifurcations , in which case ( [ scw ] ) must be replaced by a sum of similar terms with appropriate morse indices @xcite . the naive semiclassical limit for the propagation of wigner functions can be seen as a consequence of trying to get a semiclassical limit for the wigner propagator ( [ l ] ) itself by stationary phase , either via its path integral representation @xcite or directly by using ( [ scw ] ) . in any case we are lead to a semiclassical wigner propagator which is indeed classical , i.e. of the singular form : @xmath69 where @xmath70 is the inverse of the classical hamiltonian flow of @xmath10 . obviously , ( [ naive ] ) implies ( [ folk ] ) . therefore , in order to obtain the correct semiclassical limit for the propagation of wigner functions , one must apply stationary phase arguments to the full integral equation ( [ tp1]-[tp2 ] ) for the wigner evolution , using the correct semiclassical expressions for the wigner functions themselves . for one degree of freedom , the simple form of the semiclassical wigner function ( [ wig ] ) depends on there existing only a single chord centered on each point @xmath26 , besides the fulfilling of the semiclassical condition itself , i.e. the area @xmath24 being large in comparison to planck s constant @xmath28 . these conditions hold for points not too close to a convex leaf ( curve ) @xmath25 , outside the cusped triangular curve well in the interior of @xmath25 known as the wigner caustic @xcite . inside the caustic there are three chords centered on each point and the semiclassical wigner function becomes a superposition of contributions of the same form as ( [ wig ] ) , one for each chord and its associated area . along the wigner caustic itself , two chords coalesce and the amplitude in ( [ wig ] ) , defined as : @xmath71 blows up . here , @xmath72 is the classical frequency of motion along the curve @xmath25 , whereas @xmath73 are the phase space velocities at the tips @xmath74 of the corresponding chord centered on @xmath26 . the wigner caustic is hence the locus of the centers of the chords between points on @xmath25 with parallel or anti - parallel tangents . the caustic condition thus leads to the vanishing of the skew - product @xmath75 and since this also happens when @xmath26 converges onto @xmath25 , this latter can be considered as a separate branch of the wigner caustic , i.e. the breakdown of ( [ wig ] ) also takes place on @xmath25 itself . berry @xcite derived a uniform approximation based on the airy function that is also oscillatory and asymptotically equivalent to ( [ wig ] ) inside @xmath25 , rises to a smooth maximum near to this curve and decays exponentially outside . the same picture holds for integrable systems with @xmath76 degrees of freedom @xcite . the novelty is that the caustic still has codimension @xmath77 , so the invariant @xmath76-torus arises as a higher singularity or catastrophe of the wigner caustic . another important feature is that multiple chords may be centered on points that lie arbitrarily close to the invariant torus . still , sufficiently far within the energy - shell we retrieve the asymptotic form @xcite@xcite @xmath78 where again the actions @xmath79 are bounded by the @xmath80-th chord centered at @xmath26 and any arc on the quantized torus ( again a lagrangian leaf ) between the chord tips . the amplitudes are now given by @xmath81^{-1/2 } \ , \ ] ] and @xmath82 is the signature of the matrix in ( [ mamp ] ) . here , the @xmath83 action functions @xmath84 are defined in terms of the @xmath76 action functions @xmath85 as @xmath86 . note that if @xmath87 ( [ mamp ] ) is identical with ( [ amp ] ) . no immediate generalization of the semiclassical expression for pure - state wigner function is available for chaotic or general nonintegrable systems . however , the superposition of pure wigner functions in a classically narrow energy range @xmath88 determines the spectral wigner function @xmath89 which has the now familiar semiclassical limit @xcite@xcite @xmath90 \ , \ ] ] where again @xmath79 is the symplectic area between a chord centered on @xmath26 and a classical path between the chord tips . in this case , this is an arc of a classical trajectory along the energy shell , traversed in the time @xmath91 . for @xmath26 close to the energy shell , the chords are all small and there is one arc traversed in a small time and a succession of longer arcs winding around the energy shell . moving the center @xmath26 leads to the crossing of many wigner caustics where pairs of chords coalesce and disappear or vice versa . along these caustics , one can again implement uniform approximations which coincide , off - caustics , with the sum ( [ spcw ] ) in which the amplitude for each chord is @xmath92 \| \right\}^{-1/2 } \ , \ ] ] where @xmath93 is the matrix for the linearized classical map near the @xmath80-th trajectory arc on a special @xmath94-dimensional section that is centro - symmetric with respect to @xmath26 . thus we have a persistent overall picture : superposition of rapidly oscillating functions with wave vectors @xmath95 $ ] , where @xmath96 is the standard symplectic matrix in @xmath97 , and amplitudes depending only on the relation between the velocity vectors at the tips of each respective chord . having recollected the general form of the semiclassical expressions for the various kinds of wigner functions , we are now in a position to correctly approximate ( [ tp1]-[tp2 ] ) by stationary phase . the crucial point here is how to _ geometrically _ interpret formulas ( [ tp1]-[tp2 ] ) in the semiclassical approximation . as mentioned earlier , the phase of the exponential in the integrand , which is twice the area of the parallelogram ( times @xmath48 ) , corresponds to the area of _ any _ quadrilateral circumscribed to this parallelogram . furthermore , recall from the previous section that the argument @xmath26 of the semiclassical wigner function ( [ wig],[mwig],[spcw ] ) corresponds to the center of the chord whose tips lie on the energy shell or the lagrangian leaf @xmath25 corresponding to the quantum state @xmath22 , so that the phase of its oscillatory factor corresponds to the symplectic area between @xmath25 and the chord centered on @xmath26 ( minus @xmath98 ) . on the other hand , the semiclassical weyl propagator ( [ scw ] ) also has a phase corresponding to the symplectic area between the classical trajectory @xmath66 and the chord centered on @xmath26 ( minus @xmath99 ) . therefore , all the semiclassical terms of the integrand in ( [ tp1]-[tp2 ] ) have oscillatory factors whose phases depend on the chords centered on their arguments . this suggests interpreting the phase @xmath50 of the integrator in ( [ tp1]-[tp2 ] ) as the area of the circumscribed quadrilateral that _ fits _ all the pertinent chords discussed above . to be totally consistent , such fitting must take into account all the correct orientations , so that one should first dismember the wigner function as @xmath100 , where @xmath101 , and propagate each member separately , then add them up for the total evolution of @xmath21 . the situation for a single degree of freedom is illustrated in figure 1 . stationary phase condition for the propagation of wigner functions . 1.a : single degree of freedom . the various chords of @xmath102 , @xmath103 , @xmath66 and @xmath104 centered on @xmath105 , @xmath26 , @xmath106 and @xmath107 , respectively , match perfectly with the quadrilateral whose sides are centered on these points to yield , via eq.@xmath108 , the area between @xmath25 and the chord centered on @xmath26 , the phase of the propagated wigner function at @xmath26 . 1.b : the same situation but with all the orientations reversed . the really important issue is that these perfect matchings correspond precisely to the stationary phase condition , as we now show . let us concentrate on the configuration in fig.1.a , corresponding to the semiclassical evolution of @xmath109 , determined by the leaf @xmath110 , via formulas ( [ tp1]-[tp2 ] ) . the same analysis holds for the other case ( fig.1.b ) as well as more dimensions . inserting expressions ( [ wig ] ) and ( [ scw ] ) for @xmath111 and @xmath112 in ( [ tp2 ] ) , we obtain : @xmath113 @xmath114(x ' ) + s_{\psi ' ' } ( x '' ) -\pi\hbar /4 + [ s_{\gamma_t } - e_{\gamma_t}t](x ' '' ) + \delta_4 ( x , x',x'',x ' '' ) \ . \vspace{0.2cm}\ ] ] the stationary phase condition : @xmath115 implies that @xmath116}{\partial x ' } = - \frac{\partial\delta_4 } { \partial x ' } \ \ , \ \ \frac{\partial s_{\psi ' ' } } { \partial x '' } = - \frac{\partial\delta_4 } { \partial x '' } \ \ , \ \ \frac{\partial [ s_{\gamma_t } - e_{\gamma_t}t]}{\partial x ' '' } = - \frac{\partial\delta_4 } { \partial x ' '' } \ \ .\ ] ] these equations mean @xcite that the side of the quadrilateral @xmath50 whose midpoint is @xmath105 coincides with the chord of @xmath117 centered on @xmath105 , with the same orientation as @xmath110 . this is the chord from @xmath118 to @xmath119 in fig(1.a ) . similarly for the other cases , the side of @xmath50 centered on @xmath107 is the chord connecting @xmath120 to @xmath118 and the side of @xmath50 centered on @xmath106 is the chord connecting @xmath119 to @xmath121 . therefore , the stationary phase condition coincides with the perfectly matching scenario . furthermore , note that @xmath122 is the area of the `` curvilinear '' quadrilateral which is formed by the paths @xmath110 , @xmath66 , @xmath103 , @xmath104 , where @xmath25 is the immage of @xmath102 under the hamiltonian flow for a time @xmath123 . it follows by direct geometrical inspection on the perfectly matching configuration that @xmath124 t \ = \ s_{\psi}(x ) \ \big\}_{match}\ ] ] and thus @xmath125 . and of course , @xmath126 . therefore , we retrieve a new wigner function @xmath127 whose phase is of the same general form given in ( [ wig],[mwig],[spcw ] ) for @xmath128 , in terms of the new action @xmath24 . of course , this is just as we should expect within the general rules of semiclassical self - consistency in which all integrations are carried out within the stationary phase approximation . it should now be feasible to proceed with the full stationary phase evaluation of ( [ w+ ] ) for each of the different types of wigner functions ( [ wig],[mwig],[spcw ] ) previously described , using the semiclassical expression @xcite@xcite for the amplitude @xmath129 of the weyl propagator . such a thorough verification of semiclassical self - consistency has already been carried out @xcite for the pure state condition @xmath130 . it relies on such complicated geometrical constructions that we consider more appropriate , at this point , to just assume semiclassical self - consistency for the propagated amplitudes , as well . accordingly , each semiclassically propagated wigner function @xmath128 retains its respective form ( [ wig],[mwig],[spcw ] ) , with its corresponding amplitude function ( [ amp],[mamp],[spamp ] ) . the beautiful consequence of this natural result is that it is sufficient to propagate the two tips of the chord of the original wigner function @xmath111 , centered on @xmath131 , in order to obtain the value of the new wigner function @xmath128 at @xmath132 . then , using ( [ proph ] ) we obtain the new phase @xmath133 from @xmath134 . similarly for the amplitudes , the actions @xmath135 and @xmath136 determine the matrices @xmath137 and @xmath138 for the linearized classical maps from @xmath139 to @xmath140 and from @xmath141 to @xmath142 , respectively , by @xcite @xmath143[1 + j(\d^2\mathcal{s}_{\gamma_{\pm t}}(y_{\pm})/\d y_{\pm}^2)]^{-1 } \ , \ ] ] where @xmath144 , @xmath145 . hence , the corresponding phase - space velocity vectors for the tips of the propagated chord are obtained from the original ones by @xmath146 which gives the new amplitudes @xmath147 from @xmath148 via their defining equations ( [ amp],[mamp],[spamp ] ) . when the central action @xmath149 or @xmath150 goes through a caustic , one may rely on the chord actions @xmath151 or @xmath152 instead , for @xmath153 , where the chord actions are the legendre transform of the central actions : @xmath154 , to get @xcite @xmath155[1 - j(\d^2\tilde{\mathcal{s}}_{\gamma_{\pm t}}(\xi_{\pm})/\d \xi_{\pm}^2)]^{-1 } \ .\ ] ] note , however , that in this case one must be very careful in correctly counting the maslov indices . therefore , our analysis clearly leads to interpreting the correct point - to - point propagation of a semiclassical wigner function as given by the simple geometrical picture : @xmath156 where @xmath157 is the midpoint of @xmath158 , while @xmath159 stands for the tips of the original chord centered on @xmath160 . this prescription based on the tips - of - the - chord flow provides a precise semiclassical evaluation of @xmath161 from @xmath162 . specifically , ( [ proph ] ) is used to propagate the phase of @xmath21 along the path @xmath163 . for short times , the smooth amplitude may be taken as approximately constant along this path , as illustrated below . for longer times the amplitude must also be properly propagated along this path via ( [ m]-[m2 ] ) . in this way , ( [ corr ] ) is straightforwardly used to obtain a quantitatively precise semiclassical propagation of @xmath21 `` mostly everywhere '' . but , wherever @xmath128 goes through an inner wigner caustic , its analysis must be carefully completed by uniform approximations . even then ( [ corr ] ) provides most valuable information because ( [ m]-[m2 ] ) can be used to determine these inner wigner caustic points and ( [ proph ] ) determine the contributing new phases which , from the knowledge of the corresponding caustics and uniform approximations for @xmath111 and after a careful qualitative analysis of the flow @xmath164 , can lead to the correct ( airy , pearcey ... @xcite ) functions representing @xmath128 in these regions . the on - shell wigner caustics are quite simpler to deal with because there @xmath165 and ( [ corr ] ) coincides with ( [ folk ] ) . thus , we can say that ( [ corr ] ) is quantitatively precise `` mostly everywhere '' and qualitatively precise everywhere , at the semiclassical level . as a simple partial illustration of the method , consider the hamiltonian @xmath166 . operator ordering is not relevant , modulo a constant , and @xmath167 . the trajectories are circles around the origin , but @xmath168 so the classical flow of @xmath169 is @xmath170 . if @xmath160 is the center of the chord whose tips , lying on @xmath171 , are @xmath172 , @xmath173 , then the new center @xmath174 of @xmath175 is given by @xmath176 , @xmath177 , with @xmath178 . thus , not only @xmath179 , but also @xmath180 . despite the oddity of this flow , @xmath161 is mostly everywhere obtained from @xmath162 in a straightforward manner from the knowledge of @xmath181 . hence , modulo nontrivial maslov changes , the phase difference @xmath182 independs on @xmath171 or @xmath111 and is ( via ( [ proph ] ) ) given by : @xmath183 on the other hand , the new amplitude depends on the velocities of the tips of the chord determined by @xmath171 and thus depends on @xmath171 and @xmath111 itself in an intrinsic way . however , for short enough times such that @xmath157 is not too different from @xmath184 , we can approximate the slow - varying amplitude by @xmath185 , `` using '' ( [ folk ] ) in a more justified way . together with ( [ difase ] ) this gives a first approximation for the semiclassical propagation of _ any _ wigner function , mostly everywhere and for short times , under the @xmath186 hamiltonian . for longer times , we obtain @xmath187 from @xmath188 via ( [ amp ] ) and ( [ popvel ] ) , with @xmath189 determined by ( [ m ] ) from @xmath190/4 $ ] with @xmath191 , and by ( [ m2 ] ) from the corresponding @xmath192 . as for formula ( [ folk ] ) , again the precise error will depend on @xmath171 and @xmath160 , but if @xmath160 is far from the leaf @xmath171 , we can approximate @xmath193 . thus , for any pair @xmath194 such that @xmath195 and @xmath196 differ significantly , ( [ folk ] ) can be considerably wrong even for a short time propagation . we have shown that the propagation of a semiclassical wigner function is correctly determined by the classical flow of the _ tips _ of the chord whose center is the argument of the initial wigner function , not by the classical flow of the argument itself . this reveals the irrelevance of the semiclassical limit for the wigner propagator . accordingly , in evaluating the integral equation for the evolution of a wigner function , we found that satisfying the stationary phase condition was tantamount to matching four areas around an appropriate quadrilateral . however , by reducing the propagation to a single trajectory , traversed in positive and negative time , the side of the quadrilateral facing the chord of the wigner function shrinks to a point and so it could only be matched by a zero length chord . thus we found that the relevant trajectories for a precise semiclassical propagation are explicitly determined by the specific semiclassical wigner function being propagated . indeed , one way to correctly evolve a semiclassical wigner function @xmath111 determined by a leaf @xmath171 in phase space is to classically evolve @xmath171 via the hamiltonian flow : @xmath164 and then evaluate the new semiclassical wigner function @xmath128 at each point from the knowledge of @xmath197 via ( [ wig],[amp]-[spamp ] ) , as shown in @xcite . thus , to obtain the value of the new semiclassical wigner function @xmath128 at @xmath26 one needs to know , at the very least , the corresponding chord of @xmath197 centered on @xmath26 . in fact , our analysis shows that the knowledge of the flow of each pair of points in @xmath171 is sufficient to determine the evolved semiclassical wigner function @xmath128 at all points . though these two prescriptions for the evolution of @xmath21 are equivalent , in practice the semiclassical point - to - point propagation of @xmath21 is often simpler to determine , both qualitatively and quantitatively , by ( [ proph]-[m2 ] ) via tips - of - the - chord flow ( [ corr ] ) . on the other hand , a mere knowledge of the hamiltonian flow of an argument of @xmath111 is not enough to reconstruct the new chord and hence the new value of @xmath128 . the exceptions are either linear flow or when the two tips of the corresponding chord coalesce into its center , in other words , when the argument of @xmath111 lies on the leaf @xmath171 . for these points the total chord flow coincides with the center flow , thus ( [ folk ] ) is verified in the regions of high amplitude near the leaf @xmath171 , but not otherwise for nonlinear flows . it is easy to see why : the phase of @xmath162 corresponds to the symplectic area between @xmath171 and the chord centered on @xmath160 . under a nonlinear hamiltonian flow , the symplectic area of this closed circuit is preserved , however the immage of the chord is no longer a straight segment and thus the symplectic area between @xmath197 and a new chord centered on @xmath184 will generally differ from the corresponding area evaluated at @xmath198 . this difference can be large , even for short times , if the tips of the initial chord are sufficiently far apart , i.e. when the argument of the initial wigner function is sufficiently far from the initial classical leaf . therefore , generally ( [ folk ] ) is simply wrong , semiclassically . in sharp contrast , the prescription based on the tips - of - the - chord flow ( [ corr ] ) provides a precise semiclassical evaluation of @xmath161 from @xmath162 . only when @xmath199 do ( [ folk ] ) and ( [ corr ] ) coincide . berry , m.v . : _ phil . _ * 287 * , 237 ( 1977 ) . berry , m.v . : in _ chaos and quantum physics _ ( session lii , les houches ) , + m.j.giannoni , a.voros and j.zinn-justin eds . ( north holland , amsterdam,1991 ) . berry , m.v . and balazs , n.l . : _ j. phys . _ a * 12 * , 625 ( 1979 ) . heller , e.j . : _ j. chem . phys . _ * 65 * , 1289 ( 1976 ) . heller , e.j . : _ j. chem . phys . _ * 67 * , 3339 ( 1977 ) . marinov , m.s . rep . _ * 60 * , 1 ( 1980 ) marinov , m.s . : _ phys . lett . _ a * 153 * , 5 ( 1991 ) . moyal , j.e . : _ proc . _ * 45 * , 99 ( 1949 ) ozorio de almeida , a.m. : _ physica _ * 110a * , 501 ( 1982 ) . ozorio de almeida , a.m. : _ phys . * 295 * , 265 ( 1998 ) . ozorio de almeida , a.m. and hannay , j.h . _ ny * 138 * , 115 ( 1982 ) . weyl , h. : _ group theory and quantum mechanics_. dover ( 1931 ) . wigner , e.p . _ * 40 * , 749 ( 1932 ) .	we establish the difference between the propagation of semiclassical wigner functions and classical liouville propagation . first we re - discuss the semiclassical limit for the propagator of wigner functions , which on its own leads to their classical propagation . then , via stationary phase evaluation of the full integral evolution equation , using the semiclassical expressions of wigner functions , we provide the correct geometrical prescription for their semiclassical propagation . this is determined by the classical trajectories of the tips of the chords defined by the initial semiclassical wigner function and centered on their arguments , in contrast to the liouville propagation which is determined by the classical trajectories of the arguments themselves .
strongly lensed galaxies are very important in the study of galaxy formation and evolution because they permit derivation of important physical parameters such as the total mass of the lensing object , without any assumption on the dynamics . cases in which the einstein ring ( er ) is almost complete and the central lensing galaxy isolated are rare ; these permit constraining with great accuracy the enclosed mass within the projected einstein radius @xmath6 @xcite . @xcite predicted several @xmath7 optical er to be detectable over the whole sky , down to a magnitude limit of @xmath8 and a lower limit for the enclosed mass of @xmath9 . this notwithstanding , despite extensive surveys ( see for example @xcite ) only a few tens of complete or nearly complete optical ers have been identified so far , and among these objects , only a few show a close similarity , in morphology and elongation of the ring , to the one we discuss in the present work . the first er to be discovered is the radio source mg1131 + 0456 @xcite . @xcite report the discovery of a partial er ( @xmath10 ) with @xmath11 ; the background oii emitting galaxy at @xmath12 is lensed by an elliptical massive galaxy at @xmath13 . this is the first known case in the literature of a er discovered at optical wavelengths . discovered an almost complete er ( @xmath14 ) with @xmath15 produced by a massive and isolated elliptical galaxy at @xmath16 . the source galaxy is a starburst at @xmath17 . then , a similar er to the one we report in this letter , in morphology , but not in the physics of the source galaxy , a bx galaxy , is the so called `` cosmic horseshoe '' @xcite ; the ring extension is similar to the one we report here , ( @xmath0 ) , but the einstein radius is double , @xmath18 ; the lensing galaxy has a huge mass of @xmath19 . other partial er discovered recently are : the `` cosmic eye '' @xcite , the `` 8 oclock arc '' @xcite and the `` clone '' @xcite . here we report the discovery of iac j010127 - 334319 , an optical , almost complete er , that we refer to as the `` canarias einstein ring '' , noticed as a peculiar object in decam images . no previous reference to the object has been found in the literature . subsequently we observed it with osiris@gtc for a spectroscopic confirmation of its nature . in this letter we provide the first physical parameters of this system . in the following discussion we assume a flat cosmology with @xmath20 , @xmath21 and @xmath22 . l r lens & + right ascension(j2000 ) : & @xmath23 + declination(j2000 ) : & @xmath24 + redshift : & @xmath25 + surface brightness lens ( @xmath26,@xmath27 ) @xmath28}$ ] : & @xmath29 + apparent magnitude ( @xmath26,@xmath27 ) & @xmath30 + absolute magnitude ( @xmath26,@xmath27 ) & @xmath31 + + ring & + redshift : & @xmath32 + einstein radius : & @xmath33 + enclosed mass @xmath34}$ ] : & @xmath35 + surface brightness a ( @xmath26,@xmath27 ) @xmath28}$ ] : & @xmath36 + surface brightness b ( @xmath26,@xmath27 ) @xmath28}$ ] : & @xmath37 + surface brightness c ( @xmath26,@xmath27 ) @xmath28}$ ] : & @xmath38 + apparent magnitude ( @xmath26,@xmath27 ) & @xmath39 + + [ tab : valori ] the serendipitous discovery of iac j010127 - 334319 was made while performing photometry on stacked images , in @xmath26 and @xmath27 filters , taken with decam @xcite at the blanco 4 m telescope at the cerro tololo inter - american observatory ( ctio ) , reduced with the noao community pipeline @xcite and obtained from the noao science archive @xcite . the total exposure time is @xmath40 s in the @xmath26 filter and @xmath41 s in the @xmath27 filter . figure [ fig : color ] shows the resulting color composite which translates to @xmath42 . three peaks , a , b and c , are clearly visible ( bottom right panel of figure [ fig : color ] ) ; they are located respectively at @xmath43 , @xmath44 and @xmath45 from north counterclockwise . in table [ tab : valori ] all the derived parameters for the object are listed . from the decam photometry we estimated an apparent magnitude for the lens in both @xmath26 and @xmath27 bands of @xmath46 , @xmath47 . the color @xmath48 indicates that this galaxy is probably a luminous red galaxy @xcite . all the details about the photometric calibration will be given in a forthcoming paper ( bettinelli et al . in prep . ) . in order to confirm the lensing nature of this system we performed a spectroscopic follow - up at the @xmath2 m gran telescopio canarias ( gtc ) on roque de los muchachos observatory ( la palma , spain ) using the optical system for imaging and low - intermediate - resolution integrated spectroscopy ( osiris ) spectrograph . osiris has a mosaic of two e2v ccd42 - 82 ( @xmath49 ) . all the obtained spectra were registered on the second detector , which is the default for long - slit spectroscopy . we used a binning of @xmath50 providing a pixel size of @xmath51 , and the grism r300b , which provides a spectral coverage of @xmath52 and a nominal dispersion of @xmath53 @xmath54 . the slit width was @xmath55 . long - slit spectral observations were performed on 2015 december 2 in good seeing conditions of @xmath56 . the slit was placed along the n - s direction , in order to minimize the effects of atmospheric differential refraction at culmination . the total exposure time was 3600 s divided into 6 exposures of 600 s each . in each of the 6 exposures the two components , ring and lensing galaxy , have been detected and in particular their spectra were not overlapping . the position of the slit was such that the spectra obtained for the ring refers to peak b. for the pre - reduction we have used the osiris offline pipeline software ( oops ) ; sky subtraction and flux calibration were performed using iraf . we performed wavelength calibration using standard hgar+ne+xe arc lamps ; the resulting error on wavelength determination has been measured to be consistent with the above spectral resolution . we corrected the extracted spectra for instrumental response using observations of the spectrophotometric standard star gd140 , a white dwarf , obtained the same night . the fluxes of this standard star are available in @xcite . in order to derive the redshifts for the two components we noted the strong emission line in the source spectrum and the 4000 balmer discontinuity in the lens spectrum . this led us to choose template models for a starburst galaxy and a early - type galaxy respectively , as specified below . following line identification , we determined redshifts . using the template spectra by @xcite results that the spectrum of the lens galaxy fits well the spectrum of a s0 galaxy ( see figure [ fig : spec ] ) , a typical early - type galaxy characterized by a large increase in flux from the uv part of the spectrum to the optical . the @xmath57 balmer discontinuity at @xmath58 is noticeable . the redshift of the lens galaxy is @xmath59 and it has been determined from the measurements of : @xmath60 , ca k @xmath61 , ca h @xmath62 , @xmath63 , g - band @xmath64 , mg - b2 @xmath65 , mg - b1 @xmath66 ( marked red features from left to right in figure [ fig : spec ] middle panel ) . for the source galaxy we used the template spectra by @xcite and we found that the spectrum best fitting our observed spectrum corresponds to a starburst galaxy in the case of _ clumpy scattering slab _ , where it is assumed that clumped dust is located close to the source of radiation . in such circumstances , @xcite show that scattering into the line of sight dominates over absorption by the dust , providing a significant positive contribution to the emerging radiation . this template spectrum fits well the strong @xmath67 emission line . we also identified the following lines : @xmath68 , @xmath69 , 11 @xmath70 , @xmath71,10 @xmath72 ( marked blue features from left to right in figure [ fig : spec ] upper panel ) . according to these features , we derived for the source galaxy a redshift of @xmath73 . we note that the selected slit position enable us to extract only the portion of the spectrum corresponding to peak b ( see figure [ fig : color ] ) ; this notwithstanding , the emission coming from the opposite side of the ring can be easily noted in our spectra . the strong circular symmetry of our object ( see figure [ fig : color ] ) suggests that it can be approximated to the case of a circularly symmetric lens , with source and lens in the line of sight . under these assumptions , for an arbitrary mass profile @xmath74 , ( i.e. without assuming any particular model for the potential ) , we can apply the following relation @xcite and solve it for the mass . @xmath75 here @xmath76 is the einstein radius in radians ; @xmath74 is the mass enclosed within the einstein radius ; @xmath77 , @xmath78 , @xmath79 are the angular diameter distances respectively of source - lens , lens - observer and source - observer . these last quantities are related to the relative comoving distances and , in general , this relation depends on the assumed curvature of the universe @xcite . in our case @xmath80 has been assumed and the resulting angular diameter distances are @xmath81 , @xmath82 and @xmath83 . we calculated a total mass @xmath5 where the error on the mass ( @xmath84 ) is overwhelmingly due to the measurement error in the determination of the einstein radius , that we have estimated to be @xmath85 which corresponds to @xmath86 . the error on the redshift derives from the error estimated on the wavelength calibration which is @xmath87 . this value is consistent with the spectral resolution ( @xmath53 @xmath54 ) of the grating r300b that we used . under the assumption of a singular isothermal sphere ( sis ) it is possible to give an estimate of the magnification of the ring : @xmath88 , where @xmath89 is the source size . from @xcite the average size of a starburst galaxy in our redshift range is @xmath90 kpc which corresponds to @xmath91 . the derived magnification is @xmath92 . we determined also the mass - to - light ratio of the lens in the @xmath93 band ; a k - correction of 2.1 has been derived for the lens using the ned calculator @xcite . the resulting ratio is @xmath94 . the former mass estimate can be improved by applying to the system a singular isothermal ellipsoid ( sie ) model using the gravlens / lensmodel software @xcite . this software allows to fit a sie model using only the image positions and fluxes . the obtained best - fit model is plotted in figure [ fig : mass ] . the solutions have been derived for pointlike sources ; the ring shape is due to the fact that the source should actually be more extended with respect to the caustics than is shown here . the best fit ellipticity is @xmath95 calculated as @xmath96 where @xmath97 is the axis ratio ; the associated position angle is -57 deg , angle measured from north to east . the best fit @xmath98 value is 6.12 , calculated setting to 0 the weights relative to image fluxes . the derived einstein radius is @xmath99 , which translates in an enclosed mass of @xmath100 , hence in excellent agreement with our previous estimate . we report the discovery of an almost complete ( @xmath0 ) circular optical einstein ring in the constellation of sculptor . the gravitational lens is a massive luminous red galaxy at @xmath101 . the source galaxy is a starburst at redshift @xmath102 ; its spectrum is dominated by a strong emission line . using these redshift determinations and the einstein radius @xmath103 we calculated the total enclosed mass that produced the lensing effect : @xmath5 . all the parameters we determined for iac j010127 - 334319 are listed in table [ tab : valori ] . the authors thank the anonymous referee for the constructive comments that significantly improved the manuscript . the authors are grateful to all the gtc staff and in particular to dr . a. cabrera - lavers for his support in refining the spectroscopical observations during the phase-2 . the authors also thank dr . j. falcn - barroso for the helpful discussion . this letter is based on observations made with the gtc telescope , in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias , under director s discretionary time . this project used data obtained with the dark energy camera ( decam ) , which was constructed by the dark energy survey ( des ) collaboration . funding for the des projects has been provided by the doe and nsf ( usa ) , mise ( spain ) , stfc ( uk ) , hefce ( uk ) , ncsa ( uiuc ) , kicp ( u. chicago ) , ccapp ( ohio state ) , mifpa ( texas a&m ) , cnpq , faperj , finep ( brazil ) , mineco ( spain ) , dfg ( germany ) and the collaborating institutions in the dark energy survey , which are argonne lab , uc santa cruz , university of cambridge , ciemat - madrid , university of chicago , university college london , des - brazil consortium , university of edinburgh , eth zrich , fermilab , university of illinois , ice ( ieec - csic ) , ifae barcelona , lawrence berkeley lab , lmu mnchen and the associated excellence cluster universe , university of michigan , noao , university of nottingham , ohio state university , university of pennsylvania , university of portsmouth , slac national lab , stanford university , university of sussex , and texas a&m university . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .	we report the discovery of an optical einstein ring in the sculptor constellation , iac j010127 - 334319 , in the vicinity of the sculptor dwarf spheroidal galaxy . it is an almost complete ring ( @xmath0 ) with a diameter of @xmath1 . the discovery was made serendipitously from inspecting dark energy camera ( decam ) archive imaging data . confirmation of the object nature has been obtained by deriving spectroscopic redshifts for both components , lens and source , from observations at the @xmath2 m gran telescopio canarias ( gtc ) with the spectrograph osiris . the lens , a massive early - type galaxy , has a redshift of @xmath3 while the source is a starburst galaxy with redshift of @xmath4 . the total enclosed mass that produces the lensing effect has been estimated to be @xmath5 . [ firstpage ] galaxies : evolution galaxies : distances and redshifts galaxies : elliptical and lenticular , cd galaxies : starburst gravitational lensing : strong
it is believed that the perturbative qcd is applicable in the deconfined phase at large enough temperatures @xmath2 , where the effective coupling constant @xmath3 is small [ 1 ] , while at small @xmath2 ( in the confined phase ) the nonperturbative effects instead are most important . however even at large t the physics is not that simple : some effects , like screening ( electric gluon mass ) , need a resummation of the perturbative series [ 2 ] , while the effects connected with the magnetic gluon mass demonstrate the infrared divergence of the series [ 3 ] . there is a substantial amount of data from lattice calculations in the deconfined phase , which can be explained only by nonperturbative effects ( see discussion below in section 6 ) . therefore one can assume that nonperturbative effects are of vital importance both for small and large @xmath2 and should be taken into account before any dynamical perturbative scheme is applied . we suggest to use for this purpose the background field formalism [ 4 ] , in connection with the cluster expansion for background fields [ 5 ] and the feynman - schwinger representation ( fsr ) [ 6 ] for the green s function . this method has been used successfully for qcd at @xmath4 and it contains confinement as an essential ingredient , appearing through the area law of wilson loops in fsr . it is a purpose of this paper to apply the method to the case of @xmath0 by proper modifications of fsr and the background field method ( bfm ) . for this purpose we write in section 2 the familiar bfm expressions for the free energy with the only important new element : we do not consider the background field as classical ( as it is usually done ) but rather make an average over an ensemble of background fields , so that the latter enter final expressions for the free energy in the form of gauge - invariant correlators . as a consequence the simple loop diagram the leading term @xmath5 of the usual perturbation theory becomes a set of diagrams representing multiple loops interacting via nonperturbative correlators . this is done in section 3 for gluons and in section 4 for quarks . the important new element there is a new form of fsr for a gluon or a quark green s function appropriate for @xmath0 : it contains a path integral winding on a torus which compactifies the fourth euclidean direction . this enables us to compute gauge - invariant green s function for hadrons and response functions for @xmath0 . higher order amplitudes are discussed in section 5 . here the behaviour of the qcd coupling constant @xmath6 in strong background field at all distances is exploited [ 7 ] . in section 6 we identify the leading contributions to the free energy in the limit when @xmath7 and @xmath6 is small . it is gratifying that the phase transition of the first order occurs already in this oversimplifying case and the resulting transition temperature @xmath8 does not depend on @xmath1 when @xmath1 is large . clearly the mechanism of the phase transition is the evaporation of a part of the gluonic condensate into a gas of almost free gluons [ 8 ] . the predicted value of @xmath8 agrees well with monte carlo calculations both in the case of gluodynamics [ 9 ] and in the case of qcd [ 10 ] for different number of flavours . in conclusion possible prospectives of the proposed method are outlined . we start with standard formulas of the background field formalism [ 4 ] generalized to the case of nonzero temperature . we assume that the gluonic field @xmath9 can be split into the background field @xmath10 and the quantum field @xmath11 @xmath12 both satisfying periodic boundary conditions @xmath13 where @xmath14 is an integer and @xmath15 . the partition function can be written as @xmath16 @xmath17 where @xmath18 denotes all set of fields @xmath19 is a normalization constant , and the sign @xmath20 means some averaging over ( nonperturbative ) background fields @xmath10 , the exact form of this averaging is not needed for our purposes . furthermore , we have @xmath21 where @xmath22 here @xmath23 are ghost fields , @xmath24- gauge fixing constant , @xmath25 contains 3gluon and 4gluon vertices , and we keep the most general background field @xmath10 , not satisfying classical equations , hence the the appearence of @xmath26 . the inverse gluon propagator in the background gauge is @xmath27 where @xmath28 we cosider first the case of pure gluodinamics , @xmath29 . integration over ghost and gluon degrees of freedom in ( 3 ) yields @xmath30_{a=\frac{\delta}{\delta j } } \times \\ \times \ { 1 + \sum^{\infty}_{l=1 } \frac{s_{int}^l}{l ! } ( a= \frac{\delta}{\delta j } ) \ } exp ( -\frac{1}{2 } j w^{-1}j)_{j_{\mu}= \;\;\;\;\;d_{\mu}(b)f_{\mu\nu}(b)}\end{aligned}\ ] ] one can consider strong background fields , so that @xmath31 is large ( as compared to @xmath32 ) , while @xmath33 in that strong background is small at all distances [ 7 ] . in this case eq . ( 7 ) is a perturbative sum in powers of @xmath34 , arising from expansion in @xmath35 . in what follows we shall discuss the feynman graphs for the free energy @xmath36 connected to @xmath37 via @xmath38 as will be seen , the lowest order graphs already contain a nontrivial dynamical mechanism for the deconfinement transition , and those will be considered in the next section . to the lowest order in @xmath39 ( and keeping all dependence on @xmath31 explicitly ) one has @xmath40 where using ( 7 ) @xmath41 can be written as @xmath42 in ( 10 ) @xmath43 implies summing over all variables ( lorentz and color indices and coordinates ) , @xmath44 is a regularizing factor , one can use also the pauli - villars form for @xmath45 . graphically the first term on the r.h.s . of ( 10 ) is shown in fig.1 and is a gluon loop in the background field , the second term is a ghost loop , shown in fig.2 . let us turn now to the averaging procedure in ( 9 ) . with the notation @xmath46 , one can exploit in ( 9 ) the cluster expansion [ 5 ] @xmath47 + 0(\varphi^3 ) \}.\end{aligned}\ ] ] to get a closer look at @xmath48 we first should discuss termal propagators of gluon and ghost in the background field . we start with the thermal ghost propagator and write the fsr for it [ 6 ] @xmath49 here @xmath50 is the parallel transporter in the adjoint representation along the trajectory of the ghost : @xmath51 also @xmath52 and @xmath53 is a path integration with boundary conditions imbedded ( this is marked by the subscript @xmath54 ) and with all possible windings in the euclidean temporal direction . one can write it explicitly as @xmath55 here @xmath56 . one can check that in the free case , @xmath57 , eq.(12 ) reduces to well - known form of the free propagator @xmath58 with @xmath59 using the poisson summation formula @xmath60 one finally gets the jackiw - templeton s form [ 1,2 ] @xmath61 note that as expected the propagator ( 12),(17 ) corresponds to a sum of ghost paths with all ppossible windings around the torus . the momentum integration in ( 14 ) asserts that the sum of all infinitesimal `` walks '' @xmath62 should be equal to the distance @xmath63 modulo @xmath14 windings in the conpactified fourth coordinate . for the gluon propagator in the backgrounvd gauge we obtain similarly to ( 12 ) @xmath64 where @xmath65 and the operators @xmath66 are used to order insertions of @xmath67 on the trajectory of the gluon [ 11,12 ] . now we come back to the first term in ( 11 ) , @xmath48 , which can be represented with the help of ( 12 ) and ( 18 ) as @xmath68\ ] ] where the sign @xmath69 implies summation over lorentz and color indices . in the appendix 1 we show that eq.(20 ) yields for @xmath70 the usual result of the free gluon gas : @xmath71 as a next example we consider the case of nonzero component @xmath72 , while @xmath73 . we also neglect for simplicity the difference between @xmath74 and @xmath50 , i.e. we put @xmath75 . actually the interactions of @xmath67 represent the interaction of the gluon spin with the background field [ 11 , 12 ] and gives rise to short - range correlations ( in the confining phase it also yields spin - orbit interaction , the nonperturbative part of which , the thomas precession , is long range , @xmath76 ) with the notation @xmath77 one obtains from ( 20 ) @xmath78 in the case @xmath79 we come back to the free case , eq.(21 ) , since @xmath80 the expression ( 24 ) for a constant field @xmath72 yields the well - known result @xmath81 where @xmath82 means trace over color indices . one can notice in ( 26 ) , ( 24 ) that in general the presence of field @xmath72(or @xmath83 ) descreases the modulus of @xmath84 , i.e. is not advantageous from the point of view of the minimum of @xmath84 . thus @xmath85 in the deconfined regime represents a loop contribution of the gluon gas in the background field . now we turn to the confining regime and again calculate @xmath48 . it is convenient to choose another point @xmath86 on the loop in addition to the initial point @xmath87 , as shown in fig.3 and to write @xmath48 as @xmath88 where we have introduced the area law for the closed contour average @xmath89 it is important at this point to stress that in the confining regime the total contour @xmath90 shown in fig.4 , made by trajectories @xmath91 and @xmath92 in ( 27 ) , should be a closed contour , or in other words , in the representation @xmath93 one should choose @xmath94 in this case both trajectories @xmath91 and @xmath95 are winding in the same way on the torus and the minimal surface @xmath96 lies on the surface of the torus , winding around it @xmath14 times . in the case @xmath97 there appear a piece of length @xmath98 where a gluon is winding and contributing an infinite amount to the free energy ( since we assume that the free energy of an isolated quark or gluon is ininite ) . thus for @xmath99 one can integrate in ( 27 ) , ( 29 ) over @xmath100 as follows ( note that @xmath101 changes in the limits @xmath102 $ ] ) @xmath103= ( 2\pi)^4 \delta^4(p+p')\ ] ] we obtain for @xmath48 @xmath104 where we have introduced @xmath105 one can notice that temperature enters only as a factor @xmath106 in ( 32 ) and it is cancelled in the free energy @xmath107 one can also rewrite @xmath108 through the two - gluon green s function + @xmath109 , where gluons start at the point @xmath110 and meet at the point @xmath87 : @xmath111 in the free case it is equal to @xmath112 . note however that this result coincides with the loop diagram for @xmath4 , which is entirely absorbed by the subtraction constant in calculating ( 21 ) . in case of confinement one obtains instead @xmath113 where @xmath114 are wavefunction at origin and glueball mass respectively . in any case our result for the gluon loop with confinement , eq.(34 ) , does not depend on temperature and should be included in the value of the free energy for zero temperature , which as we shall argue below is given by the scale anomaly [ 14 ] , i.e. by the gluonic condensate . note that physically this correspondence is quite understandable , since the gluon loop contributimon ( 34 ) is expressed through the two - gluon bound state at the total momentum zero , i.e. through the condensate of bound gluon - gluon pairs ( glueballs ) , and the latter is most generally given by the gluonic condensate @xmath115 via the scale anomaly expression . integrating over quark fields in ( 3 ) is done trivially and leads to the following additional factor in ( 7 ) @xmath116 where we have used the symmetry property of eigenvalues of @xmath117 in the lowest approximation we omit @xmath11 in ( 37 ) and write the free energy contribution @xmath41 similary to the gluon contribution ( 9 - 10 ) as @xmath118 where the sign @xmath43 has the same meaning as in ( 10 ) and @xmath119 our aim now is to exploit the fsr to represent ( 38 ) in a form of the path integral , as it was done for gluons in ( 14 ) . the equivalent form for quarks must implement the antisymmetric boundary conditions pertinent to fermions . it is easy to understand that the correct form for quark is @xmath120 where @xmath121 @xmath122 one can easily check that in the case @xmath123 one is recovering the well - known expression for the free quark gas @xmath124 where @xmath125 is the number of flavors . the derivation of ( 42 ) starting from the path - integral form(40 ) is done similarly to the gluon case given in the appendix 1 . the loop @xmath126 in ( 40 ) corresponds to @xmath14 windings in the fourth direction . above the deconfinement transition temperature @xmath8 one can visualize in ( 40 ) the appearance of the factor @xmath127 ( note the absence of the dash sign in ( 43 ) as compared to ( 22 ) implying that in ( 43 ) @xmath128 is taken in the fundamental representation ) . for the constant field @xmath128 and @xmath129 one obtains @xmath130 this result coincides with the obtained in the literature [ 13 ] . we unify under this title two different types of contributions : i ) higher order cumulants from the cluster expansion ( 11 ) for the determinantal term ( 9 ) ; ii ) terms of the higher order in @xmath6 , appearing in the perturbative expansion ( 7 ) . we discuss first the terms i ) . a generic case is given by the quadratic cumulant @xmath131 which can be written as ( we neglect the gluon spin terms ) @xmath132 @xmath133 where @xmath50 is defined in ( 13 ) . the cumulant in the integral ( 45 ) corresponds to the interaction between two gluon loops , each loop being a gauge - invariant object . hence the interaction of the loops , given by the cumulant , is suppressed as @xmath134 both in the confining and nonconfining regimes(as compared to the contribution @xmath48 in ( 11 ) ) . in the confining regime two gluons form a bound state a glueball . one can choose the initial glueball state connecting the loops @xmath135 and @xmath136 by the stright line at some fixed @xmath137 , and then ( 45 ) can be rewritten as @xmath138 where @xmath139 and @xmath140 are central and relative coordinates of two gluons defined in accordance with ( 33 ) . the resulting integrals in ( 46 ) are similar to those in ( 32 ) with one essential difference ; in ( 32 ) the integration over all @xmath141 as in ( 31 ) is performed leading to the total momentum of two gluons @xmath142 . in contrast to that in ( 46 ) the initial and final central point @xmath139 is the same and therefore all values of total momentum are possible as in the one gluon case , see ( 15 ) and appendix 1 . as a result one obtains the temperature dependence and the final result ( see appendix b for details ) is the free energy of the glueball state [ 15 ] @xmath143 where @xmath144 is the glueball mass in the glueball state @xmath145 . note that ( 47 ) is @xmath146 since it is a color singlet contribution , in contrast to the gluon contribution ( 21 ) . in the same way one can consider the next irreducible average , @xmath147 . the contribution is again a color singlet , corresponds to the three - gluon glueball in the confining regime and is of the order @xmath146 . the same features are common to higher order cumulants . in a similar manner one can study the quark contributions to @xmath148 . in this case one has to replace in ( 45 ) the adjoint parallel transporters @xmath149 by the fundamental ones and use the fermion boundary conditions in @xmath53 as in ( 41 ) . in the confining regime one obtains the meson contribution to the free energy which has the same form as in ( 47 ) . it is of the order of @xmath150 . the largest contribution at small temperatures , @xmath151 comes from the pion gas , where neglecting the pion mass we have @xmath152 one should note , that in the set i ) introduced above the gluons interact whith each other via nonperturbative field @xmath10 , which can be exemplified by the correlators @xmath153 etc . in contrast to that in the set ii ) all perturbative exchanges are taken into account . these are proportional to @xmath34 and in the usual treatment @xmath154 is small at large temperature while it may diverge as @xmath155 [ 1 ] @xmath156 in the strong background field however one has an additional scale parameter @xmath157 which defines dynamics both at small and large distances and temperatures . one can show that when @xmath158 this scale defines the value of the coupling constant @xmath159 and @xmath159 remains small even when @xmath2 tends to zero [ 7 ] . therefore one may hope that the perturbative terms ( set ii ) ) which yield relatively small contribution ( @xmath160 at @xmath161 , remain of the same order at all temperatures . in this section we estimate the deconfinement temperature keeping only leading terms in both phases confining and deconfining . results given below have been published in a short form in [ 8 ] . we start with the zero temperature and note that the @xmath163 energy density @xmath164 is connected via the scale anomaly [ 14 ] to the gluonic condensate [ 16 ] ( we neglect the quark masses since only light quarks are important at low temperatures ) @xmath165 one can associate with @xmath164 the zero temperature limit of the free ( helmholtz ) energy @xmath166 and write @xmath167 at nonzero temperature as @xmath168 where @xmath169 is to be computed using the formulas ( 8) and subsequent ones . since @xmath169 is regularized by subtracting the zero - temperature limit @xmath170 and all divergencies are present in the latter term , eq.(51 ) is a definition of the regularized limit @xmath166 where only @xmath163 contribution in the gluonic condensate is present . a check of correctness of our normalization of @xmath166 is given by the computation of the leading contribution in section 3 . there the perturbative contribution ( 21 ) goes to zero as @xmath171 in accordance with our prescription that perturbative contributions are normalized to vanish at @xmath171 . if however one takes into account the confining background , one gets for @xmath171 a nonzero contribution ( 34 ) , which is exactly due to a pair of gluons connected by the adjoint string ( 28 ) and with zero total momentum ( see ( 31 ) ) . this is exactly the contribution which one expects from a gluon pair forming a condensate . the low temperature phase consists of the gluonic condensate ( 50 ) , glueballs ( 47 ) , mesons ( we keep only the pion ( 48 ) ) and their interacting conglomerates : @xmath172 here the first term is the leading in @xmath1 ; it grows as @xmath173 [ 17 ] , while the ideal gas of glueballs and mesons contributes @xmath146 . note that the interaction between white objects is suppressed at @xmath174 [ 18 ] and is presented in ( 52 ) by the last term . the high temperature phase is belived to be the gas of quarks and gluons , interacting perturbatively [ 1,2 ] . we shall now argue that i ) there is at @xmath175 another component of the phase , namely a part of the gluonic condensate , which has not evaporated during the phase transition ; ii ) there is a @xmath163 interaction between quarks and gluons which is important in some situations . to start we remind that the @xmath163 interaction is governed by the correlators [ 6 ] @xmath176>\ ] ] which at @xmath4 contain two independent lorentz structures with scalar coefficients @xmath177 and @xmath178 [ 5,6 ] @xmath179 d_1(u)\ ] ] at @xmath0 the lorentz invariance is broken and one obtains two independent correlators for electric and magnetic fields respectively [ 19 ] @xmath180 d^{(e)}_1 ( u)\ ] ] @xmath181 d^{b}_1 ( u)\ ] ] for the gluon condensate one must put @xmath182 in ( 54 - 55 ) with the result @xmath183 out of four functions in ( 56 ) only one , @xmath184 , is connected to the confinement , since the string tension @xmath185 is expressed through it [ 5,6 ] @xmath186 where dots imply contributions of higher - order cumulants . note that the integral ( 58 ) is confined to the region @xmath187 , since @xmath184 falls off exponentially outside . the value of @xmath188 the gluonic correlation length is the order of @xmath189 [ 20 ] . therefore for the temperatures @xmath190 the periodicity in the fourth direction does not change @xmath185 significantly , and the deconfinement occurs when @xmath184 disappears at some @xmath191 together with @xmath185 . there is no reason why @xmath192 disappear at @xmath191 or at @xmath193 . moreover , there is evidence [ 21 ] that @xmath194 , which defines the area law of the spacial wilson loops , should be nonzero at @xmath175 . additional evidence comes from the hadronic screening lengths [ 22 ] , also implying the confining dynamics in the spacial directions ( we refer the reader to [ 9,19 ] for more discussion of this point ) . thus we can assume that only @xmath184 disappears at @xmath191 . how gluonic condensate changes with temperature ? the answer comes from lattice calculations [ 23 ] and shows that @xmath195 is rather stable and only a part of it disappears near @xmath191 . this is understandable from the theoretical point of view , since the natural scale of @xmath196 is given by the lowest glueball mass approximately equal to the mass of the dilaton [ 24 ] which is of the order of @xmath197 , and therefore @xmath196 should not strongly change in the range @xmath198 . thus we come to the following model for high temperature phase : @xmath199 where @xmath200 since @xmath201 enters the @xmath163 tensor force of quarkonia [ 11 ] one can make some estimate of @xmath202 the lattice calculations [ 20 ] at @xmath4 yield @xmath203 . therefore it seems reasonable to neglect @xmath204 and @xmath205 ( note that @xmath206 at @xmath4 ) and having in mind that @xmath207 at @xmath4 , one has an estimate @xmath208 as another option with @xmath209 , one can choose @xmath210 . we have neglected in ( 59 ) i ) higher order terms in @xmath34 , ii ) higher order cumulants @xmath211 with @xmath212 and iii ) the influence of the term @xmath213 ( 24 ) and ( 43 ) . as we discussed above , we believe that i ) @xmath214 is not large , @xmath215 at all temperatures ii ) higher cumulants are suppessed as @xmath216 , iii ) from the structure of ( 24 ) and ( 43 ) one can notice that @xmath217 tends to decrease @xmath218 and since in the equilibrium the phase chooses the minimal @xmath167 ( maximal @xmath219 ) it will choose by that @xmath220 . note also , that studies of @xmath167 for constant @xmath128 [ 25 ] seem to support our conclusion ( in lowest orders graphs ) . of course our argument here is very qualitative and can be used only for a rough estimate of @xmath8 . a more detailed study of the role of @xmath213 is now in progress [ 26 ] . we now can plot both curves @xmath221 ( 59 ) , @xmath222 ( 52 ) or rather the pressure @xmath223 as a function of temperature and use again the principle of minimal @xmath167 ( maximal @xmath224 ) . one can see in fig.4 that there exists indeed some critical temperature @xmath191 and the system prefers to be in the low phase at @xmath225 and in the high phase at @xmath175 . the transition is of the first order , and @xmath8 can be computed equalizing ( 59 ) and ( 52 ) in the leading order @xmath226 one obtains : @xmath227 we note that @xmath8 is @xmath146 which is reasonable since strong interactions ( except for baryons ) have finite limit both in masses and the range of interaction when @xmath228 . with the standard value [ 16 ] . @xmath229 we have for @xmath230 @xmath231 however number of flavours @xmath232 gluonic gluonic condensate should be 2 - 3 times larger [ 16 ] , which yields @xmath233 and agrees well with the lattice calculations for gluodynamics [ 9 ] . in the next order @xmath234 we take into account the quark contribution and still disregard all meson and glueball contributions ( as well as nonideal gas corrections for gluons and quarks ) . we obtain @xmath235 for the choice ( 61 ) and @xmath236 we have @xmath237 while for @xmath238 we have @xmath239 , which agrees well with recent lattice calculations [ 9,10 ] @xmath240(lattice)=@xmath241 and @xmath242 respectively . we have started in the paper the background field formalism for @xmath0 with the purpose to take into account the np contributions in a systematic way . the perturbation theory was formulated where expansion is of two different kinds : the cluster expansion for nonperturbative interaction and the usual expansion in powers of the coupling constant for perturbative quarks and gluons in the strong np background . we have shown in the previous section that already a rough approximtion , where only terms i ) and ii ) are kept , yields a reasonable picture of the deconfining phase transition where a ) the order of the transition b ) the numerical value of @xmath8 c ) existence of non - perturbative effects at @xmath175 , are correctly predicted . we also demonstrated that the terms i ) and ii ) are leading in the large @xmath1 limit and the resulting value of @xmath8 does nt depend on @xmath1 in this limit . we plan to expand the formalism to include higher order terms and check the stability of our predictions . first of all , the role of gluonic ( adjoint ) polyakov line @xmath213 is being clarified [ 26 ] . this role is especially important for the question of the latent heat , which comes out too large with @xmath243 as compared to lattice calculations . in the vaccum background field of qcd the confinement at @xmath175 is still present in the special planes [ 19 ] which cures the infrared divergencies and can solve in principle the problem of magnetic mass . this will be discussed in detail in a subsequent paper . note added : after the original version of this manuscript has been published as a preprint [ 27 ] , the author became aware of a similar estimate of @xmath8 in [ 28 ] , where it was assumed that no nonperturbative configurations are present for @xmath175 in the high temperature phase . this corresponds to @xmath244 and results in @xmath245% higher values of @xmath8 as compared to ours . the author is grateful for discussions to the members of the theoretical seminar of itep . a part of this work has been done while the author was a guest of the max - planck institut fr kernphysik in heidelberg . it is a pleasure for him to thank professor h.a.weidenmueller and all the staff of the insitute for a kind hospitality . fruitful discussions with h.g.dosch , h.leutwyler , p.minkowski and h.d.pirner are gratefully acknowleged . the pressure @xmath256 ( corresponds to @xmath257 eq.(52 ) ) as a function of temperature - dash- dotted line . the pressure @xmath258 corresponding to @xmath221 , eq.(59 ) ) with the quark contribution -solid line , without it - dashed line . transition temperatures @xmath259 and @xmath8 obtained with and without quark contribution respectively . 99 j.kapusta , finite temperature field theory , cambridge univ . press , 1989 . + l.mclerran , rev . * 58 * ( 1986 ) 1021 . r.jackiw and s. templeton , phys . rev.d23 ( 1981 ) 22 + e.braaten and r.pisarski , phys * 64 * ( 1990 ) 1338 , nucl * b339 * ( 1990 ) 310 , ibid * b 357 * ( 1990 ) 569 . a.d.linde , phys . lett . * 96b * ( 1980 ) 289 . b.s.de witt , phys . rev . * 162 * ( 1967 ) 1195 , 1239 . + j.honerkamp , nucl . phys . * b36 * ( 1971 ) 130 , * b48 * ( 1972 ) 269 + g.thooft , nucl . phys.62 ( 1973 ) 444 + l. abbot , nucl . phys . * b185 * ( 1981 ) 189 . n.g.van kampen , phys.rep . * c24 * , ( 1976 ) 171 + h.g.dosch , phys . lett.b190 ( 1987 ) 177 , h.g.dosch and yu.a.simonov , phys b205 * ( 1988 ) 339 . yu.a.simonov , nucl . b307 * , ( 1988 ) 512 for earlier refs . see in yu.a.simonov , yad.fiz.54 ( 1991 ) 224 . yu.a.simonov , hp - thep-93 - 16 jept lett . * 57 * ( 1993 ) 525 yu.a.simonov , pisma jetf , * 55 * , ( 1992 ) 605 s.gottlieb et al . phys.rev.lett . * 59 * , ( 1987 ) 1513 + f.karsch , hlrz 92 - 62 s.gottlieb , nucl . suppl ) 20 , 247 ( 1991 ) yu.a.simonov , nucl b324 * , ( 1989 ) 67 yu.a.simonov , z. phys . * 53c * , ( 1992 ) 419 d.j.gross , r.d.pisarski and l.g.yaffe , rev . mod . phys.53 , ( 1981 ) 43 r.crewther , phys.rev.lett . * 28 * ( 1972)1421 ; m.chanowitz , j.ellis , phys . lett . * b40 * ( 1972)397 ; j.collins , l.duncan and s.joglecar , phys . rev * d16 * ( 1977 ) 438 r.d.pisarski , phys . rev.29 , ( 1984 ) 1222 m.a.shifman , a.i.vainshtein and v.i.zakharov , nucl . phys.b147 ( 1979)385,448 w.bardeen and v.i.zakharov , phys . lett.b91 ( 1980)111 e.witten , nucl . phys . * b156 * ( 1979 ) 269 . yu.a.simonov , pisma jetf , * 54 * , ( 1991 ) 256 m.campostrini , a.di.giacomo and g.mussardo , z.phys.c25 ( 1984)179 + a.di giacomo and h.panagopoulos , phys.lett . * b 285 * ( 1992 ) 133 j.polonyi and e.manousakis , phys rev . lett . * 58(1987 ) 847 c.de tar , j.kogut , phys . rev . 36 * , 2828(1987 ) + s.gottlieb et al . phys.rev.lett . * 59 * , ( 1990 ) 1881 + t.de grand et al.,phys.rev.lett . * 68 * , ( 1992 ) 2125 . m.campostrini , a.di.giacomo et al . phys.lett . * b197 * ( 1987)403 n.a.agasyan , pisma jetf * 57 * ( 1993 ) 200 n.weiss , phys . rev . * d24 * ( 1981 ) 475 ; v.m.belyaev , ic/89/321 . h.g.dosch , h.j.pirner and yu.a.simonov ( in preparation ) . yu.a.simonov , preprint itep393 . p.minkowski , chech . j. phys . * 40 * ( 1990 ) 1003	the background field formalism is used to implement nonperturbative qcd contributions into diagrammatic technic at @xmath0 . the leading terms both in the confining and nonconfining phase are identified at large @xmath1 and the transition temperature is calculated in this limit , which appears to be in good agreement with the monte carlo calculations .
the astrophysical data from distant ia supernovae observations @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , cosmic microwave background anisotropy @xcite , @xcite , @xcite , @xcite , and large scale galaxy surveys @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , all indicate that the current universe is not only expanding , it is accelerating due to some kind of negative - pressure form of matter known as dark energy ( @xcite , @xcite,@xcite,@xcite , @xcite , @xcite ) . the combined analysis of cosmological observations also suggests that the universe is spatially flat , and consists of about @xmath0 of dark matter ( the known baryonic and nonbaryonic dark matter ) , distributed in clustered structures ( galaxies , clusters of galaxies , etc . ) and @xmath1 of homogeneously distributed ( unclustered ) dark energy with negative pressure . despite the high percentage of the dark energy component , its nature as well as its cosmological origin remain unknown at present and a wide variety of models have been proposed to explain the nature of the dark energy and the accelerated expansion ( see @xcite-@xcite for review ) . among the different models of dark energy , the holographic dark energy approach is quite interesting as it incorporates some concepts of the quantum gravity known as the holographic principle ( @xcite),which first appeared in the context of black holes @xcite and later extended by susskind @xcite to string theory . according to the holographic principle , the entropy of a system scales not with its volume , but with its surface area . in the cosmological context , the holographic principle will set an upper bound on the entropy of the universe @xcite . in the work @xcite , it was suggested that in quantum field theory a short distance cut - off @xmath2 is related to a long distance cut - off ( infra - red cut - off @xmath3 ) due to the limit set by black hole formation , namely , if is the quantum zero - point energy density caused by a short distance ( uv ) cut - off , the total energy in a region of size @xmath3 should not exceed the mass of a black hole of the same size , thus @xmath4 . applied to the dark energy issue , if we take the whole universe into account , then the vacuum energy related to this holographic principle is viewed as dark energy , usually called holographic dark energy @xcite @xcite , @xcite . the largest @xmath3 allowed is the one saturating this inequality so that the holographic dark energy density is defined by the equality @xmath5 , where @xmath6 is a numerical constant and @xmath7 . + as is well known , the hubble horizon @xmath8 as the infrared cut - off , gives an equation of state parameter ( eos ) equal to zero , behaving as pressureless matter which can not give accelerated expansion , and the particle horizon gives an eos parameter larger than @xmath9 , which is not enough to satisfy the current observational data . the infrared cut - off given by the future event horizon yields the desired result of accelerated expansion with an eos parameter less than @xmath9 , despite the fact that it is based on non local quantities and has problems with the causality @xcite . an holographic dark energy model which is based on local and non local quantities have been considered in @xcite , @xcite . based mainly on dimensional arguments , in @xcite , @xcite an infrared cut - off for the holographic density was proposed , which describes the dark energy in good agreement with the astrophysical data , and may explain the cosmic coincidence . this model exhibits quintom nature ( for some values of the parameters ) without the need to introduce any exotic matter , and has also proven to be useful in the reconstruction of different scalar field models of dark energy which reproduce the late time cosmological dynamics in a way consistent with the observations @xcite . + despite the available proposals , the theoretical root of the holographic dark energy is still unknown mostly because our ignorance about the microscopic nature of quantum gravity and the origin of dark energy . this translates into a lack of details about the holographic cut - off , and therefore in the present paper we assume that this cut - off should be encoded in a function that depends on natural cosmological quantities like the hubble parameter , its time derivative and also should contain a constant term , which represents the first approximation to vacuum energy density ( which could be interpreted , in the context of the holographic principle as giving quantum gravity nature to the cosmological constant ) . the constant term would act as the first natural approximation to the infrared cut - off for the holographic density , since this term may be identified with a constant vacuum energy density , that adjusts very well to the current observational data . a discussion about the cosmological constant and the holographic principle is given in @xcite . we propose an holographic density given by a general function of the form @xmath10 , where @xmath11 should be expanded in powers of his arguments , based on their smallness ( compared to the squared planck mass @xmath12 ) . the coefficients of the different terms of this expansion must be found by fitting the model with current observations . to obtain the cosmological constraints on the model , we will use the snia constitution sample ( @xcite ) , complemented with the cmb ( cosmic micro wave background ) anisotropy and bao ( baryon acoustic oscillation ) data . in section 2 we review the main aspects of the model , in section 3 we include the matter content and fit the parameters using the joint constitution , cmb and bao data analysis . in section 4 we give a discussion . the main inequality discussed above @xmath13 , leads to the restriction involving the uv and ir cut - offs @xmath14 . saturating this inequality we come to the concept of holographic dark energy @xmath15 . to define the infrared cut - off and hence the holographic dark energy density , we propose for @xmath16 a general function which should depend on the main cosmological parameters @xmath17 and @xmath18 , and is measured in units of @xmath19 , defining the holographic density as follows ( from now on we will set @xmath20 ) @xmath21 setting @xmath22 , we can rewrite ( [ eq5 ] ) as follows @xmath23 introducing the scaled hubble expansion rate @xmath24 , where @xmath25 is the present value of the hubble parameter ( for @xmath26 ) , the above friedman equation becomes @xmath27 where @xmath28 solving the equation ( [ eq7 ] ) , we obtain @xmath29 where @xmath30 , @xmath31 is an integration constant and the last equality is written in the redshift variable ( by using @xmath32 ) . note that even without the matter component , this solution is interesting enough as it contains the cosmological constant , and the second term may give matter - like behavior if the constants @xmath33 and @xmath34 satisfy the restriction @xmath35 , giving rise to the @xmath2cdm model . evaluating the eq . ( [ eq8 ] ) at @xmath26 ( the flatness condition ) , it follows [ eq9 ] _ thus @xmath36 represents the constant component of vacuum energy density and @xmath31 stands for the new form of `` matter '' , both quantities coming from the holographic principle . additional constraints may be used to find appropriate values for the parameters . these constraints are related with the present observational value of the dark energy eos , which between the margins of errors can be set to @xmath37 , and the transition deceleration - acceleration , which occurs in the region @xmath38 ( @xmath39 is the transition redshift ) . the eos in terms of the redshift is given by [ eq10 ] ( z)=-1 + replacing @xmath40 from ( [ eq8 ] ) it follows that [ eq11 ] ( z)=-1- from eq . ( [ eq8 ] ) the deceleration parameter @xmath41 , is given by [ eq12 ] q(z)=-1- note that the hubble parameter ( [ eq8 ] ) and the eqs . ( [ eq11 ] ) and ( [ eq12 ] ) depend on @xmath33 and @xmath34 only through the combination @xmath42 . therefore , for a given @xmath43 and @xmath44 , we can find the combination @xmath42 and @xmath36 . note also that @xmath45 gives @xmath46 , as expected from a universe dominated by the cosmological constant . it is worth examining the special case of @xmath47 . integrating the eq . ( [ eq7 ] ) for @xmath47 , and using ( [ eq10 ] ) to the define the corresponding eos , gives [ eq12a ] ( z)=-1 + , where @xmath48 is the integration constant . at @xmath49 gives @xmath50 . thus depending on the relative sign of @xmath34 and @xmath48 , we can have quintessence or phantom behavior . an interesting limit of eq . ( [ eq7 ] ) can be obtained when @xmath51 , which is equivalent to one of the slow - roll conditions used in inflationary models . in this limit @xmath52 , which gives rise to exponential expansion ( @xmath53 ) . in order to construct a more realistic scenario , we include the contribution of matter , and by using the observational data we can define the parameters of the model . adding the contribution of matter , the friedmann equation becomes [ eq13 ] h^2=_m0 e^-3x++h^2 + , with the solution given by ( in terms of the scaled hubble parameter ) [ eq14 ] ^2=_m0 e^-3x+_+e^-2x(-1)/+ e^-3x where @xmath54 , @xmath31 is the integration constant and the last three terms represent the contribution of the holographic dark energy . the third term provides the possibility of crossing the phantom barrier , giving rise to quintom behavior . note that the last term in ( [ eq14 ] ) makes contribution to the matter sector , which indicates that the present holographic density also contributes to the presureless matter component of the universe , which is about 27% of the total . in what follows we will consider the sum of the terms proportional to @xmath55 as the matter component ( in fact is an effective matter component ) and the eq . ( [ eq14 ] ) will be written as [ eq15 ] ^2=_m0 e^-3x+_+e^-2x(-1)/ , where [ eq16 ] _ m0=_m0 and @xmath56 should be adjusted with the data . according to current observations , the main contribution to dark energy comes from the constant term in eq . ( [ eq15 ] ) , and the third term may be responsible for possible quintom behavior . the coefficients in eq . ( [ eq15 ] ) are subject to the flatness condition ( at @xmath26 ) [ eq17 ] _m0+_+=1 it is important to note here that the net contribution to the present density parameter of the dark energy comes from the sum of the last two terms in eq . ( [ eq17 ] ) , which we can resume as [ eq18 ] _ de=_+so that the meaning of each separate term in ( [ eq18 ] ) is not important , but what matters is their sum which should contribute about 73% to the total balance in ( [ eq17 ] ) . note also that as in the previous case of de dominance , @xmath33 and @xmath34 appear in the combination @xmath42 in ( [ eq15 ] ) , and hence we can define the parameter @xmath57 . this effectively reduce the number of the free parameters of the model to three , as can be seen from ( [ eq15 ] ) by using the above replacement for @xmath33 and @xmath34 and the flatness condition ( [ eq17 ] ) . at the same time this simplifies the procedure of fitting the parameters with the current observational data . in terms of the redshift @xmath58 the hubble parameter takes the form [ eq19 ] ^2=_m0(1+z)^3+_+(1+z)^2 , and the respective total eos becomes [ eq20 ] _ t(z)=-1 + where the flatness condition ( [ eq17 ] ) was used . before beginning with the fixing of the parameters , we illustrate the behavior of the model by choosing some representative values of the parameters , guided by the redshift transition in the deceleration acceleration parameter and by the present observational value of the dark energy eos parameter . from ( [ eq19 ] ) the deceleration parameter is given by [ eq21 ] q(z)=-1 + the condition @xmath44 gives @xmath31 in terms of @xmath59 and @xmath60 : [ eq22 ] = let s turn to the equation of state of the dark energy sector . + from ( [ eq19 ] ) it follows the expression for the holographic dark energy density as [ eq24 ] _ de=3h_0 ^ 2(_+(1+z)^2 ) using the continuity equation @xmath61 in terms of the redshift @xmath58 we find the de pressure density , and the following expression for the eos of the de [ eq25 ] _ de==-1+=-1 + at @xmath49 ( [ eq25 ] ) becomes [ eq26 ] _ de_0=-1 + where the flatness condition ( [ eq17 ] ) was used . assuming the current observational restrictions on the de eos @xcite , we can write @xmath62 , where @xmath63 represents the deviation from @xmath64 which may be in the range @xmath65 . replacing in ( [ eq26 ] ) we obtain the relation [ eq27 ] = = for a given @xmath63 and @xmath39 , and using the relations ( [ eq17 ] ) , ( [ eq22 ] ) and ( [ eq27 ] ) , the number of free parameters appearing in the solution ( [ eq19 ] ) reduces to one . note that for @xmath66 it follows that @xmath67 or @xmath68 , both of them conducing to the cosmological constant . and negative values of @xmath63 ( indicating quintom behavior of the model ) , are possible if @xmath69 ( or even @xmath70 ) . + replacing @xmath31 from ( [ eq27 ] ) in ( [ eq22 ] ) we obtain an equation to determine @xmath59 for a given @xmath39 , @xmath63 and @xmath60 : [ eq28 ] 3(1-_m0)=thus , solving the eq . ( [ eq28 ] ) for @xmath71 , @xmath72 and @xmath73 gives @xmath74 . quintom behavior can be obtained for @xmath75 ( i.e. @xmath76 ) , @xmath72 and @xmath77 , giving @xmath78 . from ( [ eq27 ] ) @xmath31 takes the values @xmath79 and @xmath80 respectively , and the dark energy density parameter can be found from : @xmath81 , completing in this manner the determination of the original model parameters as displayed in ( [ eq19 ] ) . to obtain @xmath59 we have assumed concrete values for @xmath60 , but we will use the observational data to fix the three parameters of the model , without assuming any prior values . + * cosmological constraints from snia , cmb and bao . * + we turn now to use the data sets to constrain the parameters @xmath59 , @xmath82 and @xmath60 . we performed a @xmath83 analysis to obtain the constraints on @xmath59 @xmath82 and @xmath60 using the constitution ( 397 ) snia data set @xcite , the shift parameter of the cosmic microwave background ( cmb ) form wmap5 observations @xcite and the baryon acoustic oscillations from the sloan digital sky survey ( sdss ) @xcite . the details of the @xmath83 technique and definitions of the main observational quantities used to constraint the cosmological parameters , can be found in @xcite ( see @xcite for holographic dark energy models ) . we will just quote here that the luminosity distance times @xmath25 for our model is given by [ eq30 ] d_l(z)=(1+z)_0^z where @xmath84 is given by ( [ eq19 ] ) , after replacing @xmath31 in terms of @xmath82 through the relation ( [ eq26 ] ) . after minimizing the @xmath83 function with respect to the model parameters @xmath59 , @xmath82 and @xmath60 , the following best fit values were found : @xmath85 , @xmath86 , @xmath87 . this values define the best fit value for @xmath88 : ( the hubble constant @xmath25 in units of 100 km / s / mpc ) @xmath89 the behavior of the hubble parameter against the experimental bars ( see @xcite and references therein ) is shown in fig . 1 . 1 _ the behavior of @xmath17 for the best fit values @xmath85 , @xmath86 and @xmath87 , with @xmath90 . the error bars correspond to the sample of passively evolving galaxies _ @xcite . in this work we propose a new model of holographic density which incorporates a constant term as the first approximation to the infrared ( ir ) cut - off , and has notorious advantages with respect to the previous models , as it contains the @xmath2cdm model and additional de term which could play an important role in the late time dynamics of the universe . the cosmological constant in the context of the holographic principle have been discussed in @xcite ( where using thermodynamical arguments it was showed that the most probable value of the cosmological constant is zero , but a very small cosmological constant is still allowed ) , and the presence of a constant term in the holographic density brings a quantum gravity nature to the cosmological constant . moreover , the current astrophysical data indicates that the constant cosmological constant adjusts very well to the observations , and this fact supports the presence of constant term in the holographic density . adding this constant term to the ir cut - off may improve even models like the hubble scale which was unable to produce accelerated expansion @xcite , @xcite . in this case if we add a constant term to the squared hubble scale , we obtain the @xmath2cdm model . in order to reduce the number of free parameters , we grouped the terms proportional to @xmath55 in an effective matter density parameter , and the dependence on the initial model parameters @xmath33 and @xmath34 reduces to dependence on the combination @xmath57 . by using the astrophysical data we can fix @xmath60 , and this means that one of the parameters @xmath33 or @xmath34 still free . if for instance we let @xmath34 free , then the relevant cosmological parameters which dictate the dynamical evolution of the universe do not depend on @xmath34 , and there is apparently not dynamical criteria to fix @xmath34 . to fix @xmath34 we can use geometrical criteria in order to convert the last two terms in ( [ eq5 ] ) in a term proportional to the ricci scalar . in this case the constant @xmath34 is automatically defined . from the relation @xmath91 , if we set @xmath92 then the last two terms in ( [ eq5 ] ) become @xmath93 . + we have also shown that the holographic density contributes to the matter density defining in this way the `` effective '' matter density containing the usual baryonic and dark matter ( see ( [ eq14]-[eq16 ] ) ) . we can interpret this result as meaning that the present holographic density may account for the dark matter content , and in this sense previously considered models of holographic de based on local quantities @xmath94 and @xmath18 can give rise only to constant de eos parameter . the model parameters ( @xmath95 ) were constrained using the observational data including the constitution sample of snia , the cmb shift parameter given by wmap5 , and bao measurement from sdss . we found the best fit values @xmath96 , @xmath86 and @xmath87 , with @xmath97 . the hubble parameter versus the redshift is plotted in fig . @xmath98 . a detailed study of the available data to constraint the parameters will be done elsewhere . + resuming , the present model could be relevant because of its simplicity ; it introduces a `` constant '' cosmological term in the infrared cut - off of the holographic density , which puts the cosmological constant in the context of the holographic principle , attributing quantum gravity nature to the cosmological constant ; it improves previously considered holographic models of de based on local quantities ; it may reproduce the standard @xmath2cdm model even without explicitly giving matter content ( see ( [ eq8 ] ) at @xmath35 ) . we expect that future high precision experiments allow accurately determine the parameters of the model and define its quintessence or quintom nature . this work was partially supported by fundacion seneca ( spain ) in the frame of pctrm 2007 - 2010 . riess , et al . j. * 116 * , 1009 ( 1998 ) , [ astro - ph/9805201 ] a.g . riess , et al . j. * 117,707 ( 1999 ) s.perlmutter et al , nature 391 * , 51 ( 1998 ) s. perlmutter et al . [ supernova cosmology project collaboration ] , astrophys . j. * 517 * , 565 ( 1999 ) [ astro - ph/9812133 ] p. astier et al . , astron . astrophys . * 447 * , 31 ( 2006 ) [ astro - ph/0510447 ] . m. kowalski , et . al . , astrophys . journal , * 686 * , p.749 ( 2008 ) , arxiv:0804.4142 r. m. amanullah et . arxiv:1004.1711 [ astro-ph.co ] m. hicken et al . , astrophys . j. * 700 * , 1097,2009 ; arxiv:0901.4804v3 d. n. spergel et al . [ wmap collaboration ] , astrophys . 148 , 175 ( 2003 ) [ astro - ph/0302209 ] d. n. spergel et al . , astrophys . * 170 * , 377 ( 2007 ) ; astro - ph/0603449 . e. komatsu et al . [ wmap collaboration ] , astrophys . * 180 * , 330 ( 2009 ) , arxiv:0803.0547 d. larson et al . , astrophys . j. supp . * 192 * , 16 ( 2011 ) ; arxiv:1001.4635v2 [ astro-ph.co ] m. tegmark et al . [ sdss collaboration ] , phys . rev . d 69 , 103501 ( 2004 ) [ astro - ph/0310723 ] m. tegmark et al . [ sdss collaboration ] , astrophys . j. * 606 * , 702 ( 2004 ) , astro - ph/0310725 k. abazajian et al . [ sdss collaboration ] , astron . j. * 128 * , 502 ( 2004 ) [ astro - ph/0403325 ] k. abazajian et al . [ sdss collaboration ] , astron . j. * 129 * , 1755 ( 2005 ) [ astro - ph/0410239 ] . u. seljak et al . [ sdss collaboration ] , phys . d * 71 * , 103515 ( 2005 ) , astro - ph/0407372 . d. j. eisenstein et al . [ sdss collaboration ] , astrophys . j. 633 , 560 ( 2005 ) [ astro - ph/0501171 ] . m. tegmark et al . [ sdss collaboration ] , phys . d * 74 * , 123507 ( 2006 ) , astro - ph/0608632 . peebles and b. ratra , rev . * 75 * ( 2003 ) 559 [ astro - ph/0207347 ] edmund j. copeland , m. sami and shinji tsujikawa , int . d * 15 * 1753 - 1936 ( 2006 ) , arxiv : hep - th/0603057 j.a . frieman , m.s . turner and d. huterer , ann . . astrophys . * 46 * , 385 ( 2008 ) , arxiv:0803.0982[astro - ph ] varun sahni and alexei starobinsky , int.j.mod.phys.d*9 * , 373 - 444 , 2000 , astro - ph/9904398 ; v. sahni , lect . notes phys . * 653 * , 141 - 180 ( 2004 ) , arxiv : astro - ph/0403324v3 t. padmanabhan , phys . rept * 380 * , 235 ( 2003 ) , [ hep - th/0212290 ] . j. d. bekenstein , phys . d * 7 * , 2333 ( 1973 ) . g. t hooft , [ gr - qc/9310026 ] . r. bousso , _ jhep _ * 9907 * , 004 ( 1999 ) , [ hep - th/9905177 ] . a. cohen , d. kaplan and a. nelson , phys . 4971 ( 1999 ) , [ hep - th/9803132 ] . l. susskind , j. math . ( n. y ) * 36 * , 6377 ( 1994 ) . w. fischler and l. susskind , [ hep - th/9806039 ] . s. d. h. hsu , phys . b * 594 * , 13 ( 2004 ) , [ hep - th/0403052 ] . m. li , phys . lett . b * 603 * , 1 ( 2004 ) , [ hep - th/0403127 ] . e. elizalde , s. nojiri , s.d . odintsov and p. wang , phys . d * 71 * , 103504 ( 2005 ) , hep - th/0502082 s. nojiri and s. d. odintsov , gen . grav . * 38 * , 1285 ( 2006 ) , hep - th/0506212 . l. n. granda and a. oliveros , phys . b * 669 * , 275 ( 2008 ) , gr - qc/0810.3149 . l. n. granda and a. oliveros , phys . b * 671 * , 199 ( 2009 ) , gr - qc/0810.3663 . l. n. granda , int.j.mod.phys.d , v.18 , no.11 , 1749 ( 2009 ) ; arxiv:0811.4103 [ gr - qc ] l. n. granda and a. oliveros , arxiv:0901.0561 [ hep - th ] p. horava and d. minic , phys . * 85 * , 1610 ( 2000 ) , hep - th/0001145 j. r. bond , g. efstathiou and m. tegmark , mon . not . 291 * , l33 ( 1997 ) , astro - ph/9702100 y. wang and p. mukherjee , astrophys . j. * 650 * , 1 ( 2006 ) , astro - ph/0604051 s. nesseris and l. perivolaropoulos , phys.rev.d*72 * , 123519 ( 2005 ) ; arxiv : astro - ph/0511040 s. nesseris and l. perivolaropoulos , jcap 0702 , ( 2007 ) 025 ; arxiv : astro - ph/0612653 m. li , x. li , s. wang and x. zhang , jcap 0906 , ( 2009 ) 036 ; arxiv:0904.0928 d. stern et al . , jcap 02 ( 2010 ) 008	we propose an infrared cut - off for the holographic density which incorporates among others a constant term , that produces the effect of the cosmological constant , improving the results of previously considered holographic models based on local quantities . the inclusion of constant term is interpreted as a natural first approximation for the infrared cutoff which is associated with the vacuum energy , and the additional terms guarantee an appropriate evolutionary scenario that fits the astrophysical observations . cosmological constraints on the model have been studied using the observational data . _ pacs : 98.80.-k , 95.36.+x _ +
the interactions between the constituents of matter are successfully described by the four forces : the weak , electromagnetic and strong forces and the gravitational force . the weak and the strong interactions occur at small atomic to subatomic distances , the electromagnetic interaction is observed at subatomic to macroscopic distances while effects of gravitation only play a role at macroscopic distances . the strong interaction , the main focus of this review , is responsible for the existence of all composite elementary particles ( hadrons ) by providing the binding force between the constituents and also for most of the short lived hadron decays . furthermore , the binding of protons and neutrons in nuclei may be explained in analogy to chemical binding of molecules based on the strong interaction of the proton and neutron constituents . the constituents of hadrons are known as partons or quarks . the known spectrum of hadrons can be explained as composites of five ( out of a total of six ) quark flavours , where each flavour occurs in three variants distinguished by the so - called colour . in table [ tab_quarks ] the basic properties of the six known quarks are shown . . basic properties of the six quarks in the standard model [ cols="^,^,^,^,^",options="header " , ] jets originating from quarks or gluons should have different properties due to the different colour charge carried by quarks or gluons , see section [ sec_qg ] . in an analysis by delphi @xcite the scaling violation of the fragmentation function ( ff ) in gluon and quark jets at different energies is compared . from a sample of @xmath9 hadronic @xmath1 decays planar 3-jet events with well reconstructed jets are selected using the durham or cambridge algorithms , but without imposing a fixed value of @xmath10 in order to reduce biases @xcite ( see section [ sec_excljets ] ) . in events with the two angles between the most energetic jet and the other jets in the range between 100@xmath11 and 170@xmath11 a b - tagging procedure is applied to the jets . gluon jets are identified indirectly as the jets without a successful b - tag . jets originating from light ( udsc ) quarks are taken from events which failed the b - tagging . a correction procedure based on efficiencies and purities determined by monte carlo simulation yields results for pure gluon or quark jets . the jet energy scale is calculated according to @xmath12 , where @xmath13 is the angle w.r.t . to the closest jet . this definition takes colour coherence effects into account @xcite . the ff in a jet @xmath14 is given by the distribution of @xmath15 for the hadrons assigned to the jet . the evolution of the ffs of gluon and quark jets with jet energy scale @xmath16 is studied in intervals of @xmath17 . figure [ fig_4jqgff ] ( right ) presents the quantity @xmath18 , i.e. the slopes of the scaling violation for a given interval of @xmath17 , for quark and gluon jets . the steeper slopes corresponding to stronger scaling violations of gluon compared to quark jets are clearly visible . a fit of the scaling violations based on the lo qcd prediction ( dglap equation ) allows to extract the ratio @xmath19 resulting in @xmath20 . in this analysis @xcite the decomposition of the @xmath21+nlla qcd predictions for event shape observables into terms proportional to the colour factors is used . since the sensitivity of event shape distributions measured at lep 1 alone is not sufficient @xcite data from @xmath22 to 189 gev are used . in this way the colour structure of the running of the strong coupling contributes as well . hadronisation corrections are implemented using power corrections , see section [ sec_powcor ] . the advantage of using power corrections instead of monte carlo model based hadronisation corrections is that the colour structure of the power corrections is known and can be varied in the fit . in the analysis simultaneous fits of @xmath5 , @xmath23 and @xmath24 to data for the event shape observables @xmath25 at @xmath22 to 189 gev and @xmath26at @xmath27 to 189 gev are performed . the data for @xmath25 and @xmath26are analysed separately and the results are combined . the results are @xmath28 , @xmath29 and @xmath30 and are shown on figure [ fig_cacf ] below . the errors are dominated by uncertainties from the hadronisation correction and from experimental effects . the measurements of the colour factors @xmath23 and @xmath24 or of @xmath31 discussed above or in section [ sec_qg ] can be combined into average values of @xmath23 and @xmath24 taking into account correlations between @xmath23 and @xmath24 as well as between different experiments . the variables @xmath32 and @xmath33 are used to define the @xmath34 function @xmath35 where @xmath36 and @xmath37 are the averages , the indices @xmath38 and @xmath39 count experiments , @xmath40 , @xmath41 and @xmath42 are elements of the inverses of the corresponding covariance matrices for the @xmath43 and @xmath44 within experiment @xmath38 and for the @xmath43 and @xmath44 between experiments . the averages @xmath36 and @xmath37 are converted to the average values @xmath45 and @xmath46 after the fit is performed . the input data for @xmath32 and @xmath47 are directly taken from @xcite while the results from @xcite have to be converted , @xmath48 , @xmath49 and @xmath50 . ] . the covariance matrices between experiments are constructed as follows : the aleph and opal 4-jet analyses have experimental and hadronisation errors partially and theory errors fully correlated , the 4-jet analyses and the event shape analysis have hadronisation errors partially and theory errors fully correlated , and the 4-jet and event shape analyses and the delphi ff and the opal @xmath51 analyses have their theory errors partially correlated . the averaging fit is done using only the first term of equation ( [ equ_covxy ] ) to determine the averages @xmath45 and @xmath46avoiding possible biases from our assumptions on correlations of systematic uncertainties between different sets of measurements . the fit is repeated using only statistical correlations in the first term to determine the statistical errors and using the full covariance matrix to determine the total errors . the final results are @xmath52 with correlation coefficient @xmath53 . the relative total uncertainties are about 8% for both @xmath45 and @xmath46 . the fit using the full covariance matrix yields @xmath54 and @xmath55 in good agreement with our main results . figure [ fig_cacf ] presents the results of the individual analysis in a @xmath24 vs. @xmath23 plane together with the combined result and the expectations of qcd based on the su(3 ) gauge symmetry and various other gauge symmetries . the correlation coefficients for @xcite were calculated from the references with the results @xmath56 ( aleph ) and @xmath57 ( opal ) . the error ellipses refer to 90% cl . the combined result is in good agreement with the individual analyses and with standard su(3 ) qcd while the total uncertainties are substantially reduced . the other possibilities for gauge symmetries shown on the figure are clearly ruled out . and @xmath24 discussed in this report . the ellipses show the correlated measurements using 4-jet events @xcite or event shape distributions @xcite while the lines represent the results of determinations of @xmath19 from delphi @xcite ( dashed ) and opal @xcite ( solid ) . the upper solid and dashed lines overlap . the grey filled ellipsis displays the combined result for @xmath45 and @xmath46 ( see text ) . the solid triangle and squares show the expectations for various assumptions for the gauge symmetry of qcd as indicated on the figure . , scaledwidth=75.0% ] we have shown in this report that hadron production in @xmath0annihilation is a useful and fruitful environment to test the theory of strong interactions , qcd . the absence of interference between initial and final states and the large range of cms energies probed by the experiments make many stringent tests of the theory possible . the early results from hadron production in @xmath0 annihilation at low energies were crucial to establish qcd as the theory of strong interactions and thus as an integral part of the standard model of high energy physics . studies of differences between quark and gluon jets reveal many properties of the gauge bosons of qcd , the gluons , which are correctly predicted by the theory . the measurements of jet production rates and event shape observables using theoretically and experimentally well behaved observables allow direct tests of advanced perturbative qcd predictions and precise determinations of the value of the strong coupling constant @xmath58 . a fundamental prediction by qcd is asymptotic freedom of the coupling at high energies and this has been verified directly using data over a large range of cms energies . more indirect tests of asymptotic freedom stem from successful comparison of precision determinations of @xmath58 at different energy scales . the most reliable and precise determinations of @xmath58 in @xmath0annihilation to hadrons employ inclusive observables such as the hadronic branching ratios of the gauge bosons of the electroweak theory , @xmath59 , @xmath1 and @xmath60 . the inclusive observables are mostly unaffected by hadronisation corrections and they are accessible to nnlo perturbative qcd calculations with 2-loop radiative corrections . the final result for @xmath5 from inclusive observables calculated in nnlo perturbative qcd is @xmath61 this result is in good agreement with recent world averages of measurements of the strong coupling constant @xmath62 @xcite or @xmath63 @xcite which are based on all available analyses . the best measurements of @xmath5 including ours shown above yield errors of about 2% . the other determinations of @xmath58 discussed in this report are found to be in good agreement with our main result and we note that the results from analyses of the 4-jet rate using the durham or cambridge algorithms are of similar precision . further improvements in the near future may be expected from nnlo calculations for 3-jet observables which are expected to reduce the otherwise dominating theoretical uncertainties and might allow total uncertainties of 1% @xcite . studies of the gauge structure of qcd using angular correlations in 4-jet final states , 3-jet observables at different energy scales and quark and gluon jet differences were discussed . the results for the colour factors @xmath23 and @xmath24 were observed to be in good agreement and combined values were calculated . the results are @xmath64 with correlation @xmath53 . these results are in good agreement with expectations from qcd with the su(3 ) gauge symmetry @xmath65 and @xmath66 . in summary , @xmath0 annihilation to hadrons has been and still is indispensable to stringently probe the theory of strong interactions . qcd has been established as a part of the standard model of high energy physics and has been tested to a precision of 2% with some well suited observables . this impressive result has been achieved by the lep experiments , sld , the re - analysis of jade data and the parallel theoretical work stimulated by the availability of precise data . future high energy physics programmes such as the lhc or the ilc will benefit greatly from this achievement , because precise and reliable qcd predictions will be essential for disentangling possible new physics signals from standard processes . 100 kluth s 2003 in _ 2003 qcd and high energy hadronic interactions _ ed aug e and vn j t t pages 2530 ( the gioi publishers , vietnam ) contributed to xxxviiith recontre de moriond , qcd and high energy hadronic interactions , les arcs , france , march 2229 , 2003	the current status of tests of the theory of strong interactions , quantum chromo dynamics ( qcd ) , with data from hadron production in @xmath0 annihilation experiments is reviewed . the lep experiments aleph , delphi , l3 and opal have published many analyses with data recorded on the @xmath1 resonance at @xmath2 gev and above up to @xmath3 gev . there are also results from sld at @xmath2 gev and from reanalysis of data recorded by the jade experiment at @xmath4 gev . the results of studies of jet and event shape observables , of particle production and of quark gluon jet differences are compared with predictions by perturbative qcd calculations . determinations of the strong coupling constant @xmath5from jet and event shape observables , scaling violation and fragmentation functions , inclusive observables from @xmath1 decays , hadronic @xmath6 decays and hadron production in low energy @xmath0annihilation are discussed . updates of the measurements are performed where new data or improved calculations have become available . the best value of @xmath5 is obtained from an average of measurements using inclusive observables calculated in nnlo qcd : @xmath7 where the error is dominated by theoretical systematic uncertainties . the other measurements of @xmath5 are in good agreement with this value . finally , investigations of the gauge structure of qcd are summarised and the best values for the colour factors are determined : @xmath8 with errors dominated by systematic uncertainties and in good agreement with the expectation from qcd with the su(3 ) gauge symmetry .
we acknowledge support from the australian research council . 99 j. lebowitz , _ ann . * 19 * , 389 - 418 ( 1968 ) . d. ruelle , _ statistical mechanics . rigorous results _ ( world scientific , london 1999 ) . j. frenkel , _ kinetic theory of liquids _ ( oxford university press , london , 1946 ) ; f. sausset , g. biroli and j. kurchan , _ phys . _ * 140 * , 718 - 727 ( 2010 ) . s. r. williams and d. j. evans , _ j. chem . phys . _ * 131 * , 024115(1 - 9 ) ( 2009 ) ; _ ibid _ , _ j. chem . phys . _ * 132 * , 184105(1 - 14 ) ( 2010 ) ; s. r. williams , _ j. chem . phys . _ * 135 * , 131102(1 - 4 ) ( 2011 ) . d. r. squire , a. c. holt and w. g. hoover , _ physica _ * 42 * , 388 - 397 ( 1969 ) . i. fuereder and p. ilg , _ j. chem . phys . _ * 142 * , 144505 ( 2015 ) . a. lemaitre and c. maloney , _ j. stat . phys . _ * 123 * , 415- 453 ( 2006 ) . h. yoshino , _ phys . _ * 136 * , 214108(1 - 36 ) ( 2012 ) . r. zwanzig and r. d. mountain , _ j. chem . phys . _ * 43 * , 4464 - 4471 ( 1965 ) . d. n. perera , and p. harrowell , _ phys . . e _ * 59 * , 5721 - 5743 ( 1999 ) . s. plimpton , _ j. comp . phys . _ * 117 * , 1 - 19 ( 1995 ) ; http://lammps.sandia.gov . s. abraham and p. harrowell , _ phys . _ * 137 * , 014506(1 - 8 ) ( 2012 ) . barrat , j .- roux , j .- hansen and m. l. klein , _ europhys . lett . _ * 7 * , 707 - 712(1988 ) . j. p. wittmer , h. xu , p. polinska , f. weysser and j. baschnagel , _ j. chem . phys . _ * 138 * , 12a533(1 - 18 ) ( 2013 ) ; j. p. wittmer , h. xu , o. benzerara and j. baschnagel , _ mol . phys . _ * 113 * , 2881 - 2893 ( 2015 ) . a. zaccone and e. m. terentjev , _ phys . lett . _ * 110 * , 178002(1 - 5 ) ( 2013 ) . a. zaccone , j. r. blundell and e. m. terentjev , _ phys . b _ * 84 * , 174119 ( 2011 ) . h. yoshino and f. zamponi , _ phys . e _ * 90 * , 022302(1 - 14 ) ( 2014 ) . y. p. mitrofanov , v. a. khonik , a. v. granato , d. m. joncich and s. v. khonic , _ j. app . phys . _ * 109 * , 073518(1 - 4 ) ( 2011 ) . j. c. dyre , _ rev . * 78 * , 953- 972 ( 2006 ) . f. a. lindemann , _ phys . z. _ * 11 * , 609 ( 1910 ) . s. saw and p. harrowell , cond - mat arxiv:1602.02855 .	motivated by the formal argument that a non - zero shear modulus is the result of averaging over a constrained configurations space , we demonstrate that the shear modulus calculated over a range of temperatures and averaging times can be expressed ( relative to its infinite frequency value ) as a single function of the mean squared displacement . this result is shown to hold for both a glass - liquid and a crystal - liquid system . glasses are rigid and liquids are not . the difficulty with any distinction of two phases based solely on rigidity is that the property is not an equilibrium one . over 45 years ago , lebowitz@xcite and ruelle@xcite pointed out that , in the thermodynamic limit , the free energy of a phase can not depend on the shape of the sample and so the equilibrium value of the shear modulus must vanish for all phases @xmath0 crystals as well as glasses @xmath0 in the limit of large n. rationalising the obvious point that rigid materials do in fact exist , a number or researchers@xcite have concluded that a non - zero shear modulus is a property of a metastable state and hence rigidity is observable only for observation times shorter than the lifetime of that state . since the observation of a non - zero shear modulus depends crucially on this lifetime , it would seem that any theoretical treatment of the mechanical properties of a material will depend on solving the onerous problem of slow relaxation in a condensed phase . williams and evans@xcite , acknowledging this difficulty , suggested that the shear modulus be formally calculated as an equilibrium average over a constrained space of configurations . this perspective suggests the attractive possibility that the magnitude of the shear modulus might be expressed as an explicit function of the magnitude of the configurational constraint applied , a relation that includes a threshold degree of constraint , below which rigidity vanishes . in this paper we establish just such a relationship between the shear modulus and the configurational constraint , measured here by the mean squared displacement , for both a glass - liquid and crystal - liquid system . the squire - holt - hoover expression@xcite for the ( constrained ) equilibrium shear modulus @xmath1 of a solid is @xmath2\end{aligned}\ ] ] where @xmath3 is the shear stress , @xmath4 , @xmath5 is volume and @xmath6 is the infinite frequency ( or born ) shear modulus given by@xcite @xmath7 -\frac{d^2\phi}{dr_{ij}^2 } \frac{x_{ij}^2 y_{ij}^2}{r_{ij}^2 } \big ) > \end{aligned}\ ] ] where @xmath8 and @xmath9 is a spherically symmetric inter - particle potential . note that the shear modulus @xmath1 is reduced , relative to the high frequency value , by an amount associated with variance of the shear stress fluctuations . in the context of elastic theory , these fluctuations correspond to non - affine contributions to the modulus@xcite . what is measured in a typical experiment is the stress relaxation function @xmath10 , where @xmath11 is a applied strain and @xmath12 is the resulting time dependent shear stress . the relation between @xmath13 and @xmath1 is given by the following expression@xcite , @xmath14\end{aligned}\ ] ] where the shear stress autocorrelation function @xmath15 equals @xmath16 when t=0 and @xmath17 in the limit @xmath18 . it follows from eq . 3 that @xmath1 represents a lower bound to the observed modulus @xmath13 with @xmath19 . this long time limit refers only to the _ explicit _ time dependence arising from the shear stress autocorrelation function . it does not include any _ implicit _ time dependence associated with the observation time used to construct the averages in @xmath1 ( see eq . 1 ) . so , the averages @xmath20 in eq . 1 - 3 are understood to be taken over some constrained configuration space . in the absence of a constraint , @xmath21 and @xmath22@xcite so that @xmath23 . the model liquid used in this study is a 2d system of soft disks with a pair interaction potential , @xmath24 , between species @xmath25 and @xmath26 . in the case of the binary equimolar mixture we use @xmath27 , @xmath28 and @xmath29 and all particle with unit mass , a model that has been extensively studied@xcite in the context of the glass transition . the temperature is reported in units of @xmath30 and time in units of @xmath31 . simulations were carried out under constant nvt conditions using lammps@xcite with a nose - hoover thermostats at reduced densities @xmath32 ( binary mixture ) and @xmath33 ( single component ) with a potential cut - off distance of @xmath34 . the system consisted of a total of n = 1024 particles in the case of the binary mixture and @xmath35 for the single component system . previously@xcite , we established that these values of n were sufficient for accurate calculation of the stress fluctuations . at low temperatures , the trajectories are non - ergodic for all accessible values of the averaging time @xmath36 . in order to fairly sample the configuration space at these low temperatures we have averaged trajectories over statistically distinct initial configurations . for the binary mixtures , @xmath37 uncorrelated configurations were generated by cooling a liquid , equilibrated at @xmath38 , to @xmath39 at a cooling rate of @xmath40 and then minimizing the potential energy of the resulting @xmath41 liquid by conjugate gradient minimization . the minima , referred to as inherent structures ( is ) , were statistically independent as established by the average shear stress ( at @xmath42 ) of the inherent structures equalling zero . to calculate @xmath1 and @xmath6 at a given temperature @xmath43 we randomly assigned momenta from the boltzmann distribution consistent with a temperature @xmath43 to the particles in each of the n is configurations and then determined the moduli for each individual is using eqs . 1and 2 , respectively , by averaging over a trajectory run for a time interval t. to obtain our final values of @xmath1 and @xmath6 , we averaged the moduli for the individual is configurations over all @xmath37 is configurations . in the case of the single component system , this protocol was modified as follows . for the crystal phase we only used a single inherent structure , that of the perfect crystal . for the liquid phase data we simply carried out averages over md trajectories of the equilibrated liquid . and ( b ) @xmath6 vs @xmath43 for the binary mixture over different averaging times . note the significant effect of the averaging time in the case of @xmath1 in contrast to @xmath6 , were the influence of averaging time has saturated within a short time @xmath44 . insert in fig . 1b : the virial pressure @xmath45 as a function of @xmath43 . , title="fig : " ] and ( b ) @xmath6 vs @xmath43 for the binary mixture over different averaging times . note the significant effect of the averaging time in the case of @xmath1 in contrast to @xmath6 , were the influence of averaging time has saturated within a short time @xmath44 . insert in fig . 1b : the virial pressure @xmath45 as a function of @xmath43 . , title="fig : " ] in fig . 1 we plot the values of @xmath1 and @xmath6 as a function of @xmath43 for a binary mixture of soft disks in @xmath46 at a fixed density . we find that @xmath6 increases linearly with @xmath43 and shows no significant variation with the averaging time t. ( note that the infinite frequency modulus referred to experimentally is not @xmath6 but the value of @xmath13 in the plateau region . ) the equilibrium modulus @xmath1 , in contrast , exhibits a strongly nonlinear decrease with increasing temperature , to finally vanish at a sufficiently high temperature . the family of curves presented in fig . 1 is evidence of the significance of the time t used to calculate the statistics of the stress fluctuations . a number of papers@xcite have discussed the loss of rigidity of a glass as characterised by the disappearance of @xmath1 on heating in the context of a possible thermodynamic instability , analogous to the softening in a superheated crystal@xcite , or as an un - jamming transition associated with the thermal expansion of the amorphous solid@xcite . ( we remind the reader that our calculations have been carried out at fixed density so that this latter proposal is not directly relevant here . ) while the role of the observation time is discussed@xcite , it is the temperature that is treated as the essential control parameter for the transition . following on from our opening discussion , we shall explore the idea that this decrease of the equilibrium shear modulus , either through increasing t or the observation time , is most fundamentally expressed as a result of the changing degree of configurational constraint associated with the averaging . to begin we note that temperature dependence of @xmath6 is not associated with configurational constraint since , by construction , the infinite frequency modulus depends only on the sampling of the local curvature of the potential energy surface . the increase in @xmath6 with temperature at constant volume demonstrated in fig . 1b is directly associated with the increase of the virial pressure with t under the constant volume constraint ( see insert fig . 1b ) . to eliminate this additional temperature dependence we shall therefore consider the reduced modulus @xmath47 . next , we need a measure of the configurational constraint . the simplest such measure is the particle mean squared displacement , @xmath48 where the time @xmath36 here is the same as the observation time used to calculate the stress averages and , in the case of a binary mixture , the average is over both species . vs @xmath49 for the binary mixture . in each case , the mean squared displacement is calculated over the same time interval as that use to evaluate @xmath1 . ] in fig . 2 we plot @xmath47 vs @xmath49 where we have used the data from fig . 1 for a range of temperatures and observation times . we find that all of the data from fig . 1 collapses onto a single curve . this result provides strong support the twin propositions of this paper , i.e. that the ( reduced ) shear modulus is simply a consequence of configurational constraint and that the mean squared displacement provides a useful measure of this constraint . and @xmath49 ( both averaged over a time @xmath50 ) for the single component 2d soft disk system . due to the constant density constraint , there is a range of temperatures corresponding to a two phase coexistence , indicated by the two vertical dashed lines . the freezing transition is marked by step - like change in the modulus and the mean squared displacement . ] a glass forming liquid is convenient for our purposes because it can access the entire range of @xmath1 without encountering a thermodynamic singularity . our argument relating shear modulus and configurational constraint , however , should apply equally to crystallizing liquids . to demonstrate this point , we consider a single component soft disk liquid in 2d which crystallizes readily into a triangular lattice . in fig . 3 we plot the values of @xmath1 and @xmath49 for the system as a function of @xmath43 , using an observation time @xmath51 . the presence of the @xmath52 order freezing at @xmath53 is clearly evident in both quantities . in fig . 4 we plot @xmath47 vs @xmath49 for a range of temperatures ( using crystal and liquid configurations for @xmath43 below and above @xmath54 , respectively ) and a range of observation times . again , we find the data collapsed onto a common curve , this in spite of the discontinuity of the modulus and @xmath49 with respect to temperature . it is worth noting the striking difference in the low @xmath43 limit of @xmath47 for the crystal ( fig . 4 ) and the glass ( fig . 2 ) . the reason for the considerable softening of the glass relative to @xmath6 even at @xmath42 is due to ( i ) the higher density of crystal and ( ii ) the large non - affine motions@xcite in the glass relative to those in the crystal . the presence of non - affine motions in the amorphous phase and their effective absence in the crystal is due to the absence of inversion symmetry in the local structure of the amorphous phase and its presence in the crystal@xcite . the dependence of the reduced shear modulus @xmath47 on @xmath49 is found , empirically , to be well described by the following relation , @xmath55 \big)\end{aligned}\ ] ] the success of this function is shown in fig . 5 for the glass forming mixture with the fitted values @xmath56 and @xmath57 . . 5 also provides an excellent fit to @xmath47 vs @xmath49 for both the crystal and liquid phases of the single component system ( see fig . 4 ) , but with different parameters , @xmath58 and @xmath59 . viewed as an emergent property of restricted particle fluctuations , the derivation of the dependence of @xmath47 on the degree of configurational constraint must represent a problem of fundamental importance . on @xmath49 for the single component crystal and liquid . as for fig . 2 , each point corresponds to a choice of @xmath43 and the averaging time . liquid state data i.e. @xmath60 are presented by filled symbols and crystal data by open symbols . the curve corresponds of a fit of eq . 5 to the single component data with @xmath58 and @xmath59 . ] we have argued here that the value of @xmath1 ( relative to @xmath6 ) is a consequence of constraint . this is the opposite to the account provided within harmonic models of solids in which the elastic constants ( or the bond force constants ) are prescribed in the model and the mean squared displacement are determined as a consequence . this latter treatment , however , is only possible because of the implicit configurational constraints ( i.e. assumed elasticity , unbreakable harmonic bonds , etc . ) on which such models rely . for the harmonic solid , @xmath61 ( at fixed density ) . as shown in fig . 5 ( insert ) , this relation holds only for @xmath62 , a result that underscores the inclusion of anharmonic effects in the empirical relations demonstrated in fig . 4 and 5 . yoshino and zamponi@xcite have recently derived a power law relationship between the shear modulus and the mean squared displacement in a granular model that applies within a metabasin , a restricted range of configuration space corresponding , roughly , to @xmath63 . vs @xmath49 for the binary mixture . ( the correspondence between symbols and temperatures are the same as in fig . 2 ) . the expression in eq . 5 ( solid curve ) provides a good description over the entire range of @xmath49 with @xmath64 and @xmath57 . insert : the quantity @xmath65 vs @xmath49 . the harmonic approximation , indicated by a constant value ( dashed line ) , breaks down for @xmath66 . ] in conclusion , we have verified that our two propositions : ( 1 ) the degree of configurational constraint determines the magnitude of the shear modulus ( relative to the low temperature limit ) , and ( 2 ) the mean squared displacement provides a useful measure of this constraint , do indeed represent a consistent physical picture for both a glass forming liquid and one that undergoes freezing . this result represents a fundamental unification of the physical basis of rigidity . the presence of a non - zero shear modulus is not , we argue , the consequence of a low temperature , a high frequency measurement or even the presence of long range order . rather , each of these factors is important only in as far as they contribute to an implicit constraint on the volume of configuration space that can be explored by stress fluctuations . it is this constraint , however it is achieved , that determines the value of the equilibrium shear modulus . this is a powerful result with a number of interesting consequences . first , accounts of the temperature dependence of the shear modulus of metallic glasses@xcite have relied on the language of anharmonic effects borrowed from crystal physics . in the picture we present here , the decrease in the glass modulus on heating is associated the increase in @xmath49 by harmonic or anharmonic motions ( along with any decrease in @xmath6 associated with thermal expansion when a constant pressure is employed as in ref . @xcite ) . second , we have argued that the shear modulus should be regarded as a mechanical manifestation of restricted motion . couple this idea with a description of the role that elastic behaviour plays in determining the rate of particle motion ( e.g. the shoving model of dyre@xcite ) and there is possibility of a self consistent theory in which the modulus is , itself , a consequence of the very particle mobilities that it acts to constrain . thirdly , these results suggest a reassessment of the empirical lindemann criterion@xcite , i.e. the observation that crystal order is lost once the mean squared displacement exceeds some threshold value . our results here suggest that it is rigidity , not structure _ per se _ , that vanishes as the mean square displacement increases . finally , since our account of rigidity places no special condition on how the configuration space is accessed it is possible that non - thermal contributions to particle mobility such as the non - affine motion due to applied strain should result in an analogous reduction in @xmath47@xcite . each of these lines of inquiry is currently under investigation . in this paper we have established that the collection of factors @xmath0 time , temperature and order @xmath0 associated with the observation of rigidity in a dense phase can be replaced by a single tangible length that characterizes the degree of configurational constraint . while we have established that @xmath49 provides a workable measure of this constraint length , further work is required to establish whether there is a better measure of this constraint and whether we can derive from first principles the mathematical relationship between this measure and the shear modulus .
in stellar evolution computations , and in particular in the case of stars more massive than the sun , it is generally sufficient to use a simple equation of state . the plasma of the stellar interior is treated as a mixture of perfect gases of all species ( atoms , ions , nuclei and electrons ) , and the saha equation is solved to yield the degrees of ionization or molecular formation . in the case of low mass stars however , non ideal effects , such as coulomb interactions become important . it is then necessary to use a more adequate equation of state than the one employed in the geneva code for more massive stars . this simple equation of state essentially contains a mixture of ideal gases , ionization of the chemicals is dealt with by the saha equation , excited states and molecules are neglected , complete pressure ionization is artificially imposed above certain temperatures and pressures , and no coulomb - pressure correction is included ( see schaller et al . 1992 ) . for the present grids of models of 0.4 to 1.0 m@xmath0 stars these assumptions are obviously inadequate . for such stars , the most useful equations of state , as far as their smooth realization and versatility are concerned , are ( i ) the so - called mihalas - hummer - dppen ( mhd ) equation of state ( hummer and mihalas 1988 ; mihalas et al . 1988 ; dppen et al . 1988 ) , and ( ii ) the opal equation of state , the major alternative approach developed at livermore ( rogers 1986 , and references therein ; rogers et al . 1996 ) . a brief description of these two equations of state is given in the next section . here , we chose the mhd equation of state . first , we were able to compute very smooth tables specifically for our cases of chemical composition , instead of relying on pre - computed , relatively coarse tables that would require interpolation in the chemical composition . second , our choice was forced by the fact that the currently available opal equation of state tables do not allow to go below stars less massive than @xmath10.8 m@xmath0 . third , we validated our choice by a comparative calculation with opal at its low - mass end . we found results that are virtually indistinguishable from mhd . fourth , we examined in a parallel theoretical study ( trampedach & dppen 1998 ) the arguments about the validity of the mhd equation of state down to the limit of our calculation of 0.4 m@xmath0 ( see below ) . therefore we do not have to include a harder excluded - volume term such as the one included in the saumon - chabrier ( sc ) equation of state ( saumon & chabrier 1991 , 1992 ) . although the mhd equation of state was originally designed to provide the level populations for opacity calculations of stellar _ envelopes _ , the associated _ thermodynamic quantities _ of mhd can none the less be reliably used also for stellar cores . this is due to the fact that in the deeper interior the plasma becomes virtually fully ionized . therefore , in practice , it does not matter that the condition to apply the detailed hummer - mihalas ( 1988 ) occupation formalism for bound species is not fulfilled , because essentially there are no bound species . other than that , the mhd equation of state includes the usual coulomb pressure and electron degeneracy , and can therefore be used for low - mass stars and , in principle , even for envelopes of white dwarfs ( w. stolzmann , _ private communication _ ) . the present paper , with its mhd - opal comparison ( see 3 ) corroborates this assertion . this broad applicability of the mhd equation of state for entire stars was specifically demonstrated by its successes in solar modeling and helioseismology ( christensen - dalsgaard et al . 1988 , charbonnel & lebreton 1993 , richard et al . 1996 , christensen - dalsgaard et al . a solar model that is based on the mhd equation of state from the surface to the center is in all respects very similar to one based on the opal equation of state , the major alternative approach developed at livermore ( rogers 1986 , and references therein ; rogers et al . this similarity even pertains to the theoretical oscillation frequencies that are used in comparisons with the observed helioseismic data . although the difference between the mhd and opal equations of state is of helioseismological relevance , it has no importance for the lower - mass stellar modeling of the present analysis . this is explicitly validated in the present paper . for the much finer helioseismological analyses , it turned out that in some respect the opal model seems to be closer than the mhd model to the one inferred from helioseismological observations ( christensen - dalsgaard et al . 1996 , basu & christensen - dalsgaard 1997 ) . however , we stress that for the present stellar modelling these subtle differences are no compelling reason to abandon the convenience of our ability to compute mhd equation of state tables ourselves , and to go below the range of the available opal tables ( @xmath10.8 m@xmath0 ) . not only helioseismology , but also fine features in the hertzsprung - russell diagram of low- and very low - mass stars impose strong constraints on stellar models ( lebreton & dppen 1988 , dantona & mazzitelli 1994 , 1996 , baraffe et al . 1995 , saumon et al . they have all confirmed the validity of the principal equation - of - state ingredients employed in mhd ( coulomb pressure , partial degeneracy of electrons , pressure ionization ) . finally , we have checked that even at the low - mass end of our calculations the physical mechanism for pressure ionization in the mhd equation of state is still achieved by the primary pressure ionization effect of mhd ( the reduction of bound - state occupation probabilities due to the electrical microfield ; see hummer & mihalas 1988 ) . such a verification was necessary to be sure that our results are not contaminated by the secondary , artificial pressure - ionization device included in mhd for the very low - temperature high - density regime ( the so - called @xmath2 term of mihalas et al . a parallel calculation has confirmed that in our models a contamination by this @xmath2 term can be ruled out ( trampedach & dppen 1998 ) . in the present paper , we expand the current mass range of the geneva evolution models from 0.8 down to 0.4 m@xmath0 , by using a specifically calculated set of tables of the mhd equation of state . this work aims to complete the base of extensive grids of stellar models computed by the geneva group with up - to - date input physics [ z=0.020 and 0.001 , schaller et al . ( 1992 ) , bernasconi ( 1996 ) , and charbonnel et al . ( 1996 ) ; z=0.008 , schaerer et al . ( 1992 ) ; z=0.004 , charbonnel et al . ( 1993 ) ; z=0.040 , schaerer et al . ( 1993 ) ; z=0.10 , mowlavi et al . ( 1998 ) ; enhanced mass loss rate evolutionary tracks , meynet et al . ( 1994 ) ] . in sect . 2 , we present the characteristics of our equation of state and recall the physical ingredients used in our computations . in sect . 3 , we summarize the main characteristics of the present models and discuss the influence of the equation of state on the properties of low mass stars . finally , we compare our solar - metallicity models with recent models computed by other groups and with observations in 4 . the basic physical ingredients used for the complete set of grids of the geneva group are extensively described in previous papers ( see schaller et al . 1992 , hereafter paper i ) . with the exception of the equation of state , described in detail in the following subsection , we therefore just mention the main points . we have already justified our choice of the mhd equation of state in the introduction . as mentioned there , mhd is one of the two recent equations of state that have been especially successful in modeling the sun under the strong constraint of helioseismological data . historically , the mhd equation of state was developed as part of the international `` opacity project '' [ op , see seaton ( 1987 , 1992 ) ] . it was realized in the so - called _ chemical picture _ , where plasma interactions are treated with modifications of atomic states , _ i.e. _ the quantum mechanical problem is solved before statistical mechanics is applied . it is based on the so - called free - energy minimization method . this method uses approximate statistical mechanical models ( for example the nonrelativistic electron gas , debye - hckel theory for ionic species , hard - core atoms to simulate pressure ionization via configurational terms , quantum mechanical models of atoms in perturbed fields , etc ) . from these models a macroscopic free energy is constructed as a function of temperature @xmath3 , volume @xmath4 , and the concentrations @xmath5 of the @xmath6 components of the plasma . the free energy is minimized subject to the stoichiometric constraint . the solution of this minimum problem then gives both the equilibrium concentrations and , if inserted in the free energy and its derivatives , the equation of state and the thermodynamic quantities . the other of these two equations of state is the one underlying the opal opacity project ( see 2.2 ) . the opal equation of state is realized in the so - called _ physical picture_. it starts out from the grand canonical ensemble of a system of the basic constituents ( electrons and nuclei ) , interacting through the coulomb potential . configurations corresponding to bound combinations of electrons and nuclei , such as ions , atoms , and molecules , arise in this ensemble naturally as terms in cluster expansions . any effects of the plasma environment on the internal states are obtained directly from the statistical - mechanical analysis , rather than by assertion as in the chemical picture . more specifically , in the chemical picture , perturbed atoms must be introduced on a more - or - less _ ad - hoc _ basis to avoid the familiar divergence of internal partition functions ( see _ e.g. _ ebeling et al . in other words , the approximation of unperturbed atoms precludes the application of standard statistical mechanics , _ i.e. _ the attribution of a boltzmann - factor to each atomic state . the conventional remedy of the chemical picture against this is a modification of the atomic states , _ e.g. _ by cutting off the highly excited states in function of density and temperature of the plasma . such cut - offs , however , have in general dire consequences due to the discrete nature of the atomic spectrum , _ i.e. _ jumps in the number of excited states ( and thus in the partition functions and in the free energy ) despite smoothly varying external parameters ( temperature and density ) . however , the occupation probability formalism employed by the mhd equation of state avoids these jumps and delivers very smooth thermodynamic quantities . specifically , the essence of the mhd equation of state is the hummer - mihalas ( 1988 ) occupation probability formalism , which describes the reduced availability of bound systems immersed in a plasma . perturbations by charged and neutral particles are taken into account . the neutral contribution is evaluated in a first - order approximation , which is good for stars in which most of the ionization in the interior is achieved by temperature [ the aforementioned study ( trampedach & dppen 1998 ) has verified the validity of this assumption down to the lowest mass of our calculation ] . for colder objects ( brown dwarfs , giant planets ) , higher - order excluded - volume effects become very important ( saumon & chabrier 1991 , 1992 ; saumon et al . 1995 ) . in the common domain of application of the saumon et al . ( 1995 ) and mhd equations of state , chabrier & baraffe ( 1997 ) showed that both developments yield very similar results , which strongly validates the use of the mhd equation of state for our mass range of 0.4 to 1.0 m@xmath0 . despite undeniable advantages of the physical picture , the chemical picture approach leads to smoother thermodynamic quantities , because they can be written as analytical ( albeit complicated ) expressions of temperature , density and particle abundances . in contrast , the physical picture is normally realized with the unwieldy chemical potential as independent variable , from which density and number abundance follow as dependent quantities . the physical - picture approach involves therefore a numerical inversion before the thermodynamic quantities can be expressed in their `` natural '' variables temperature , density and particle numbers . this increases computing time greatly , and that is the reason why so far only a limited number of opal tables have been produced , only suitable for stars more massive than @xmath10.8 m@xmath0 . therefore we chose mhd for its smoothness , availability , and the possibility to customize it directly for our calculation , despite the in principle sounder conceptual foundation of opal . * the initial helium content is determined by y=0.24+(@xmath24y/@xmath24z)z , where 0.24 corresponds to the current value of the cosmological helium ( audouze 1987 ) . we use the value of 3 for the average relative ratio of helium to metal enrichment ( @xmath24y/@xmath24z ) during galactic evolution . this leads to ( y , z ) = ( 0.300,0.020 ) and ( 0.243,0.001 ) . in addition , computations were also performed with ( y , z ) = ( 0.280,0.020 ) . * the relative ratios for the heavy elements correspond to the mixture by grevesse & noels ( 1993 ) used in the opacity computations by iglesias & rogers ( 1996 ) . * choosing initial abundance values for d and @xmath16he is more complex . pre - solar abundances for both elements have been reviewed in geiss ( 1993 ) , however galactic chemical models face serious problems to describe their evolution ( see e.g. tosi 1996 ) , and no reliable prescription exists to extrapolate their values in time . deuterium is only destroyed by stellar processing since the big bang nucleosynthesis , so that its abundance decreases with time ( i.e. with increasing metallicity ) . on the other hand , the actual contribution of stars of different masses is still subject to a large debate ( hogan 1995 , charbonnel 1995 , charbonnel & dias 1998 ) , and observations of ( @xmath16he / h ) in the proto - solar nebulae ( geiss 1993 ) and in galactic hii regions present a large dispersion . we adopt the same ( d / h ) and ( @xmath16he / h ) initial values for both metallicities , namely 2.4@xmath25 and 2.0 @xmath25 respectively . at the low mass range considered in this paper , the observed quasi - static contraction begins rather close to the theoretical deuterium main sequence , once the stars emerge from their parental dense gas and dust . for the present purposes then , we take as starting models polytropic configurations on the hayashi boundary , neglecting the corrections brought to isochrones and upper tracks by the modern accretion paradigm of star formation ( palla & stahler 1993 , bernasconi & maeder 1996 ) . for the mass range considered here , these corrections are likely not to exceed 3@xmath26 of the kelvin - helmholtz timescale for the pre - main sequence contraction times ( bernasconi 1996 ) . we note , however , that the predicted upper locus for the optical appearance of t tauri stars in the hr diagram can be as much as half less luminous than the deuterium ignition luminosity on the convective tracks ( @xmath27 0.3 ) . the hr diagrams for pre - main sequence evolution and for the following phases are given in fig . 1 and 2 respectively for both metallicities . for each stellar mass , table 1 displays the lifetimes in the contraction phase and in the deuterium- and hydrogen - burning phases . note that we did not complete the main sequence evolution computations for the less massive stars which have a h - burning phase longer than the age of the universe ; for these stars , our last computed model corresponds to an age of 20 gyr . * in the stellar mass range we consider , the ignition of deuterium burning ( indicated in fig.1 ) takes place in a fully convective interior . during pre - main sequence evolution , a radiative core develops . however a proper radiative branch is absent , since these stars maintain a convective envelope all along contraction until the zams has been reached , and further on . * from table 1 one can see that the contraction time lasts less than 4 thousandths of the hydrogen - burning stage . * due to opacity effects , the entire evolution ( pre - main sequence , main sequence and red giant branch ) occurs at higher luminosity and effective temperature for a given stellar mass when the initial metallicity is smaller , or when the hydrogen content is lower for the same initial metallicity . * as a consequence , the contraction phase and the deuterium- and hydrogen - burning phases for a given stellar mass are shorter at lower metallicity ( see table 1 ) , and at lower hydrogen content for the same value of z. @xmath28 \multicolumn{2}{c}{\mbox{z } } & \multicolumn{2}{c}{\mbox{initial } } & \multicolumn{2}{c}{\mbox{contraction } } & \multicolumn{2}{c}{\mbox{d - burning } } & \multicolumn{2}{c}{\mbox{h - burning } } & \multicolumn{2}{c}{\mbox { { t$_c$ / t$_h$ } } } \\ \multicolumn{2}{c}{\mbox{y } } & \multicolumn{2}{c}{\mbox{mass } } & \multicolumn{2}{c}{\mbox{phase } } & \multicolumn{2}{c}{\mbox{phase } } & \multicolumn{2}{c}{\mbox{phase } } & \multicolumn{2}{c } { } \\[1 mm ] \hline \\[0.5 mm ] 0&001 & 0&4 & 160&10 & 0&417 & \tiret & \tiret \\ 0&243 & 0&5 & 81&11 & 0&321 & \tiret & \tiret \\ & & 0&6 & 51&89 & 0&277 & \tiret & \tiret \\ & & 0&7 & 36&44 & 0&240 & \tiret & \tiret \\ & & 0&8 & 31&20 & 0&214 & 14338&03 & 0&0022 \\ & & 0&9 & 22&82 & 0&196 & 9051&01 & 0&0025 \\ & & 1&0 & 19&34 & 0&187 & 6847&20 & 0&0028 \\[1 mm ] \hline 0&020 & 0&4 & 153&84 & 0&605 & \tiret & \tiret \\ 0&300 & 0&5 & 117&34 & 0&476 & \tiret & \tiret \\ & & 0&6 & 88&26 & 0&389 & \tiret & \tiret \\ & & 0&7 & 68&80 & 0&330 & \tiret & \tiret \\ & & 0&8 & 58&15 & 0&302 & 22713&11 & 0&0026 \\ & & 0&9 & 44&30 & 0&276 & 14070&96 & 0&0031 \\ & & 1&0 & 32&80 & 0&252 & 9059&52 & 0&0036 \\[1 mm ] \hline 0&020 & 0&4 & 161&79 & 0&640 & \tiret & \tiret \\ 0&280 & 0&5 & 124&50 & 0&484 & \tiret & \tiret \\ & & 0&6 & 93&75 & 0&407 & \tiret & \tiret \\ & & 0&7 & 72&86 & 0&344 & \tiret & \tiret \\ & & 0&8 & 62&34 & 0&309 & 26372&72 & 0&0024 \\ & & 0&9 & 48&26 & 0&286 & 16421&94 & 0&00002 \\ & & 1&0 & 37&98 & 0&263 & 10662&62 & 0&00002 \\[1 mm ] \hline \end{array}}\ ] ] ccccc + & & & & + + 0.020 & 1.0 & 9059.5 & 8968.1 & 9970.1 + 0.020 & 0.8 & 22713.3 & 22361.4 & 25151.3 + 0.001 & 0.8 & 14338.0 & 13986.0 & + when we first compare the results obtained with the mhd and with the simple geneva equations of state ( see fig . 3 and 4 , and table 2 ) , we obtain essentially the same results than lebreton & dppen ( 1988 ) . firstly , the fact that mhd contains @xmath29 molecules , and the simple geneva code does not , is reflected in a shift essentially _ along _ the zams . on the other hand , the coulomb pressure correction , also contained in mhd , causes a slight shift of the zams , clearly visible for higher masses , where there are no hydrogen molecules in the photosphere . this coulomb effect has been well discussed in the case of helioseismology ( _ e.g. _ christensen - dalsgaard et al . conformal to the effect of the mhd equation of state to push the apparent position on the zams upward , it also decreases the lifetime on the zams ( see table 2 ) . for comparison , we have computed with the opal equation of state two 0.8m@xmath0 models ( the lowest mass that can be computed with the current opal tables ) , for both metallicities . as can be seen in fig.3 , the corresponding tracks are very close to those obtained with the mhd equation of state , the use of the opal equation of state leading to slightly higher effective temperature on the zams . as far as their internal structure is concerned , the models computed with mhd equation of state have slightly deeper convection zones . the main sequence lifetime obtained with the mhd equation of state is slightly higher than the one obtained with the opal equation of state ( table 2 ) . the comparison shows that down to 0.8m@xmath0 all is fine with the mhd pressure ionization . as mentioned in the introduction , trampedach & dppen ( 1998 ) predict a correct functioning of pressure ionization in mhd even for much smaller masses . with the present comparison , we have validated their prediction at least to 0.8m@xmath0 . we now compare our models with recent models computed by other groups and with various observations . the strongest approximation in our models lies in the treatment of the atmosphere and of the surface boundary conditions , which are specified by the eddington approximation . in figs . 5 and 6 we compare our results with the models of dantona & mazzitelli ( 1994 ) and tout et al . ( 1996 ) which also rely on a simple treatment of the boundary conditions . for the mass range considered in our work we obtain predictions very similar to dantona & mazzitelli . sophisticated model atmospheres for the computation of very low mass stars have been developed recently ( baraffe et al . 1995 , 1998 ; brett 1995 , allard et al . their use becomes crucial for very low mass stars ( @xmath30 m@xmath0 ) down to the brown dwarf limit . we refer to chabrier & baraffe ( 1997 ) and baraffe et al . ( 1998 ) for a discussion of the physical basis of the differences . as can be seen in fig.5 , the eddington approximation we use results in higher t@xmath31 ( from 2 to 5 @xmath26 depending on the metallicity ) compared to the chabrier & baraffe ( 1997 ) models at our low mass end ; our models also have slightly smaller radius ( see fig.6 ) . in this mass range the predictions agree well with the observations of popper ( 1980 ) and leggett et al . ( 1996 ) and do not allow to disregard one model with respect to the other . in fig.7 we show the predicted mass - luminosity relations for the v and k band for different metallicities and ages . the magnitudes were derived from the standard stellar library for evolutionary synthesis of lejeune et al.(1998 ) , which provides empirically calibrated colours for solar metallicity and a semi - empirical correction for non - solar metallicities . the comparison with the observations of andersen ( 1991 ) and henry & mccarthy ( 1993 ) shows a good agreement with our predictions for both the v and k band ( fig.7 ) . again a similar agreement is obtained with the models of brocato et al . ( 1998 ) and baraffe et al . ( 1998 ) . from the comparisons in figs.5 to 8 , we conclude ( in agreement with alexander et al . 1997 ) that for the mass range considered in this work an approximate treatment of the stellar atmosphere leads to a satisphactory agreement between theoretical predictions and observations . independently ab initio stellar interior and atmosphere models allowing the detailed predictions of all observational properties are of fundamental importance for our understanding of very low mass stars which are out of the scope of the present grids . the results of this work and of our previous grids ( papers i to vi ) are published by astronomy and astrophysics at the centre de donnes spatiales ( cds at strasbourg ) where the corresponding tables are available in electronic form : an ensemble of models were selected to describe each evolutionary track . for each model the tables display the age , actual mass , log @xmath32 , log @xmath33 , the surface abundances in mass fraction of h , @xmath34he , @xmath35c , @xmath36c , @xmath37n , @xmath38o , @xmath39o , @xmath40o , @xmath41ne , @xmath42ne , the core mass fraction qcc , log@xmath43 ( where @xmath44 is the mass loss rate on the red giant branch ) , log @xmath45 ( where @xmath45 is the central density ) , log t@xmath46 ( where t@xmath46 is the central temperature ) , and the central abundances in mass fraction of the above elements . a detailed description of the models selection and of the tables contents is given in paper i. we now also provide photometric data in verious systems for all tracks and isochrones ( see schaerer & lejeune 1998 ) . alexander d.r . , fergusson j.w . , 1994 , apj 437 , 879 alexander d.r . , brocato e. , cassisi s. , castellani v. , ciacio f. , deglinnocenti s. , 1997 , a&a 317 , 90 allard f. , hauschildt p.h . , alexander d.r . , starrfield s. , 1997 , ara&a 35 , 137 andersen j. , 1991 , a&a rev 3 , 91 audouze j. , 1987 , in observational cosmology , iau symp . 124 , eds . et al . , reidel publ . p.89 baraffe i. , chabrier g. , allard f. , hauschildt p.h . , 1995 , apj 446 , l35 baraffe i. , chabrier g. , allard f. , hauschildt p.h . , 1998 , a&a , submitted ( astro - ph/9805009 ) basu s. , christensen - dalsgaard j. , 1997 , a&a , 322 , l5 bernasconi p.a . 1996 , a&as , 120 , 57 bernasconi p.a . , maeder a. , 1996 , a&a , 307 , 829 brett j.m . , 1995 , a&a 295 , 736 brocato e. , cassisi s. , castellani v. , 1998 , mnras 295 , 711 caughlan g.r . , fowler w.a . , 1988 , atomic data nuc . data tables 40,283 chabrier g. , baraffe i. , 1997 , a&a 327 , 1039 charbonnel c. , 1995 , apj 453 , l41 charbonnel c. , dias do nascimento j. , 1998 , a&a , 336 , 915 charbonnel c. , lebreton y. , 1993 , a&a 280 , 666 charbonnel c. , meynet g. , maeder a. , schaller g. , schaerer d. , 1993 , a&a 102 , 339 charbonnel c. , meynet g. , maeder a. , schaerer d. , 1996 , a&as 115 , 339 christensen - dalsgaard , j. , dppen , w. , lebreton , y. , 1988 , nature , 336 , 634 christensen - dalsgaard j. , dppen , w. , and the gong team , 1996 , science , 272 , 1286 dantona f. , mazzitelli i. , 1994 , apjs 90 , 467 dantona f. , mazzitelli i. , 1996 , apj 456 , 329 dppen , w. , mihalas , d. , hummer , d.g . & mihalas , b.w . : 1988 , astrophys . j. 332 , 261 ebeling , w. , kraeft , w.d . & kremp , d. , 1976 , _ theory of bound states and ionization equilibrium in plasmas and solids _ , akademie verlag , berlin , ddr geiss j. , 1993 , origin and evolution of the elements , eds . n.prantzos , e.vangioni-flam & m.cass , cambridge university press , 89 graboske h.c . , de witt h.e . , grossman a.s . , cooper m.s . , 1973 , apj 181 , 457 grevesse n. , noels a. , 1993 , in origin and evolution of the elements " , eds prantzos n. , vangioni - flam e. , cass m. , cup henry t.j . , mccarthy d.w . , 1993 aj 106 , 773 hogan c. , 1995 , apj 441 , l17 hummer , d.g . & mihalas , d. : 1988 , apj , 331 , 794 iglesias c.a . , rogers f.j . , 1996 , apj 464 , 943 lebreton y. , dppen w. , 1988 , seismology of the sun and the sun - like stars , eds.v.domingo , e.j.rolfe , esa sp-286 , 661 leggett s.k . , allard f. , berriman g. , dahn c.c . , hauschildt p.h . , 1996 , apjs 104 , 117 lejeune t. , cuisinier f. , buser r. , 1997 , a&as , in press ( astro - ph/9710350 ) maeder a. , meynet g. , 1989 , a&a 210 , 155 meynet g. , maeder a. , schaller g. , schaerer d. , charbonnel c. , 1994 , a&as 103 , 97 mihalas d. , hummer d.g . , dppen w. , 1988 , apj 331 , 815 mowlavi n. , schaerer d. , meynet g. , bernasconi p.a . , maeder a. , charbonnel c. , 1998 , a&as 128 , 1 palla f. , stahler s.w . 1993 , apj , 418 , 414 popper d.m . , ann.rev.astr.ap . 18 , 115 reimers d. , 1975 , mem . soc . lige 8 , 369 richard o. , vauclair s. , charbonnel c. , dziembowski w.a . , 1996 , a&a 312 , 1000 rogers , f.j . : 1986 , apj . 310 , 723 rogers , f.j . , swenson , f.j . and iglesias , c.a . , 1996 , apj , 456 , 902 saumon d. , chabrier g. , 1991 , _ phys . rev _ , a44 , 5122 saumon d. , chabrier g. , 1992 , _ phys . rev _ , a46 , saumon d. , chabrier g. , vanhorn h.m . , 1995 , apjs 99 , 713 schaerer d. , lejeune t. , 1998 , in preparation schaerer d. , meynet g. , maeder a. , schaller g. , 1992 , a&as 98 , 523 schaerer d. , charbonnel c. , meynet g. , maeder a. , schaller g. , 1993 , a&as 102 , 339 schaller g. , schaerer d. , meynet g. , maeder a. , 1992 , a&as 96 , 269 seaton , m. : 1987 , j. phys . b : atom . molec . 20 , 6363 seaton , m.j . : 1992 , in _ astrophysical opacities _ , eds . c. mendoza & c. zeippen ( _ revista mexicana de astronoma y astrofsica _ ) 180 tosi m. , 1996 , from stars to galaxies : the impact of stellar physics on galaxy evolution , asp conf . series , vol . 78 , eds . c. leitherer , u. fritze - von alvensleben , j. huchra , p. 299 tout c.a . , pols o. , eggleton p.p . , han z. , 1996 , mnras 281 , 257 trampedach r. , dppen w. , 1998 , a&a , submitted	we present stellar evolutionary models covering the mass range from 0.4 to 1 m@xmath0 calculated for metallicities z=0.020 and 0.001 with the mhd equation of state ( hummer & mihalas , 1988 ; mihalas et al . 1988 ; dppen et al . 1988 ) . a parallel calculation using the opal ( rogers et al . 1996 ) equation of state has been made to demonstrate the adequacy of the mhd equation of state in the range of 1.0 to 0.8 m@xmath0 ( the lower end of the opal tables ) . below , down to 0.4 m@xmath0 , we have justified the use of the mhd equation of state by theoretical arguments and the findings of chabrier & baraffe ( 1997 ) . we use the radiative opacities by iglesias & rogers ( 1996 ) , completed with the atomic and molecular opacities by alexander & fergusson ( 1994 ) . we follow the evolution from the hayashi fully convective configuration up to the red giant tip for the most massive stars , and up to an age of 20 gyr for the less massive ones . we compare our solar - metallicity models with recent models computed by other groups and with observations . the present stellar models complete the set of grids computed with the same up - to - date input physics by the geneva group [ z=0.020 and 0.001 , schaller et al . ( 1992 ) , bernasconi ( 1996 ) , and charbonnel et al . ( 1996 ) ; z=0.008 , schaerer et al . ( 1992 ) ; z=0.004 , charbonnel et al . ( 1993 ) ; z=0.040 , schaerer et al . ( 1993 ) ; z=0.10 , mowlavi et al . ( 1998 ) ; enhanced mass loss rate evolutionary tracks , meynet et al . ( 1994 ) ] . # 1#2#3=#3 cm
in combinatorics on words , the most famous class of infinite words is certainly that of sturmian words . they code digital straight lines in the discrete plane having irrational slope , and are characterized by the property of having exactly @xmath0 factors of length @xmath1 , for every @xmath2 . it is well known ( see @xcite , proposition 2.1.17 ) that a finite word @xmath6 is a factor of some sturmian word if and only if @xmath6 is a binary _ balanced _ word , that is , for every letter @xmath7 and every pair of factors of @xmath6 of the same length , @xmath4 and @xmath8 , one has that @xmath4 and @xmath8 contain the same number of @xmath7 s up to one , i.e. , @xmath9 finite sturmian words ( finite factors of sturmian words ) have _ at most _ @xmath0 factors of length @xmath1 , for every @xmath2 . however , this property does not characterize them , as shown by the word @xmath10 , which is not sturmian since the factors @xmath11 and @xmath12 do not verify . the finite words defined by the property of having at most @xmath0 factors of length @xmath1 , for every @xmath2 , are called _ trapezoidal words_. the name comes from the fact that the graph of the function counting the distinct factors of each length ( usually called the factor complexity ) defines , for trapezoidal words , an isosceles trapezoid . trapezoidal words were defined by de luca @xcite by means of combinatorial parameters on the repeated factors of finite words . for a finite word @xmath6 over a finite alphabet @xmath13 , let @xmath14 be the minimal length of an unrepeated suffix of @xmath6 and let @xmath15 be the minimal length for which @xmath6 does not contain right special factors , i.e. , factors having occurrences followed by distinct letters . de luca proved that for any word @xmath6 of length @xmath16 one always has @xmath17 and studied the case in which the equality holds . this is in fact the case in which @xmath6 is trapezoidal , since one can prove ( see proposition [ prop : trap ] later ) that a word @xmath6 is trapezoidal if and only if @xmath18 then , de luca proved that finite sturmian words verify and are therefore trapezoidal . the non - sturmian trapezoidal words were later characterized by dalessandro @xcite . since a non - sturmian word is not balanced , it must contain a pair of factors having the same length and not verifying . such a pair is called a _ pathological pair_. let @xmath6 be a non - sturmian word and @xmath19 its pathological pair of minimal length . dalessandro proved that @xmath6 is trapezoidal if and only if @xmath20 for some @xmath21 and @xmath22 , where @xmath23 is the reversal of the fractional root @xmath24 of @xmath25 and @xmath26 is the fractional root of @xmath27 . this characterization is quite involved and will be discussed in detail later . the characterization by dalessandro allows us to give an enumerative formula for trapezoidal words . indeed , it is known that the number of sturmian words of length @xmath1 is equal to @xmath28 ( cf . @xcite ) , where @xmath29 is the euler totient function , i.e. , @xmath30 is the number of positive integers smaller than or equal to @xmath1 and coprime with @xmath1 . in section [ sec : formula ] we prove that the number of non - sturmian trapezoidal words of length @xmath1 is @xmath31 and thus the number of trapezoidal words of length @xmath1 is @xmath32 ( theorem [ theor : enum ] ) . we then distinguish trapezoidal words into two distinct classes , accordingly with the definition below . [ def : closed ] let @xmath6 be a finite word over an alphabet @xmath13 . we say that @xmath6 is _ closed _ if it is empty or it has a factor ( different from the word itself ) occurring exactly twice in @xmath6 , as a prefix and as a suffix , that is , with no internal occurrences . a word which is not closed is called _ open_. for example , the words @xmath33 and @xmath34 are closed , whereas the word @xmath35 is open . the notion of closed word is closely related to the concept of _ complete return _ to a factor @xmath4 in a word @xmath6 , as considered in @xcite . a complete return to @xmath4 in @xmath6 is any factor of @xmath6 having exactly two occurrences of @xmath4 , one as a prefix and one as a suffix . hence @xmath6 is closed if and only if it is a complete return to one of its factors ; such a factor is clearly both the longest repeated prefix and the longest repeated suffix of @xmath6 . the notion of closed word is also equivalent to that of _ periodic - like _ word @xcite . a word @xmath6 is periodic - like if its longest repeated prefix does not appear in @xmath6 followed by different letters . we derive some properties of open and closed trapezoidal words in section [ sec : openclosed ] . we prove that the set of closed trapezoidal words is contained in that of sturmian words ( proposition [ prop : clostur ] ) . we then characterize the special factors of closed trapezoidal words ( lemma [ lem : clo1 ] ) by showing that every closed trapezoidal word @xmath6 contains a bispecial factor @xmath4 such that @xmath4 is a central word and the left ( resp . right ) special factors of @xmath6 are the prefixes ( resp . suffixes ) of @xmath4 . we show that trapezoidal palindromes ( hence , sturmian palindromes ) are closed words ( theorem [ theor : palclo ] ) , and that a closed trapezoidal word is a sturmian palindrome if and only if its longest repeated prefix is a palindrome ( proposition [ prop : palh ] ) . in section [ sec : sesqui ] we introduce _ semicentral words _ , that are trapezoidal words in which the longest repeated prefix , the longest repeated suffix , the longest left special factor and the longest right special factor all coincide . we prove in theorem [ theor : sesqui ] that a trapezoidal word @xmath6 is semicentral if and only if it is of the form @xmath36 , for a central word @xmath4 and two different letters @xmath5 . we then study , in section [ sec : fibo ] , the prefixes of the fibonacci infinite word @xmath25 , and show that the sequence of the numbers of consecutive prefixes of @xmath25 that are closed ( resp . open ) is the sequence of fibonacci numbers ( theorem [ theor : fibo ] ) . the paper ends with conclusions and open problems in section [ sec : conc ] . an _ alphabet _ , denoted by @xmath13 , is a finite set of symbols , called _ letters_. a _ word _ over @xmath13 is a finite sequence of letters from @xmath13 . the subset of the alphabet @xmath13 constituted by the letters appearing in @xmath6 is denoted by @xmath37 . we let @xmath38 denote the @xmath39-th letter of @xmath6 . given a word @xmath40 , with @xmath41 for @xmath42 , the nonnegative integer @xmath1 is the _ length _ of @xmath6 , denoted by @xmath16 . the empty word has length zero and is denoted by @xmath43 . the set of all words over @xmath13 is denoted by @xmath44 . the set of all words over @xmath13 having length @xmath1 is denoted by @xmath45 . for a letter @xmath46 , @xmath47 denotes the number of @xmath7 s appearing in @xmath6 . the word @xmath48 is called the _ reversal _ of @xmath6 . palindrome _ is a word @xmath6 such that @xmath49 . in particular , we assume @xmath50 and so the empty word is a palindrome . a word @xmath51 is a _ factor _ of a word @xmath6 if @xmath52 for some @xmath53 . in the special case @xmath54 ( resp . @xmath55 ) , we call @xmath51 a prefix ( resp . a suffix ) of @xmath6 . an occurrence of a factor of a word @xmath6 is _ internal _ if it is not a prefix nor a suffix of @xmath6 . the set of prefixes , suffixes and factors of the word @xmath6 are denoted respectively by @xmath56 , @xmath57 and @xmath58 . a _ border _ of the word @xmath6 is any word in @xmath59 different from @xmath6 . the _ factor complexity _ of a word @xmath6 is the function defined by @xmath60 , for every @xmath61 . a factor @xmath4 of @xmath6 is _ left special _ ( resp . _ right special _ ) in @xmath6 if there exist @xmath62 , @xmath63 , such that @xmath64 ( resp . @xmath65 ) . a factor is _ bispecial _ in @xmath6 if it is both left and right special . a _ period _ for the word @xmath6 is a positive integer @xmath66 , with @xmath67 , such that @xmath68 for every @xmath69 . since @xmath16 is always a period for @xmath6 , we have that every non - empty word has at least one period . we can unambiguously define _ the _ period of the word @xmath6 as the smallest of its periods . the period of @xmath6 is denoted @xmath70 . the _ fractional root _ @xmath71 of a word @xmath6 is its prefix whose length is equal to the period of @xmath6 . a word @xmath6 is _ a power _ if there exists a non - empty word @xmath4 and an integer @xmath72 such that @xmath73 . a word which is not a power is called _ primitive_. for example , . the left special factors of @xmath6 are @xmath75 , @xmath7 , @xmath76 , @xmath77 and @xmath78 . the right special factors of @xmath6 are @xmath75 , @xmath7 , @xmath76 and @xmath77 . therefore , the bispecial factors of @xmath6 are @xmath75 , @xmath7 , @xmath76 and @xmath77 . the period of @xmath6 is @xmath79 and the fractional root of @xmath6 is @xmath80 . the following parameters were introduced by aldo de luca @xcite : given a word @xmath81 , @xmath82 ( resp . @xmath83 ) denotes the minimal length of a prefix ( resp . a suffix ) of @xmath6 which occurs only once in @xmath6 . given a word @xmath81 , @xmath84 ( resp . @xmath85 ) denote the minimal length for which there are no left special factors ( resp . no right special factors ) of that length in @xmath6 . [ ex:1 ] let @xmath86 . the longest left special factor of @xmath6 is @xmath87 , and it is also the longest repeated suffix of @xmath6 ; the longest right special factor of @xmath6 is @xmath88 , and it is also the longest repeated prefix of @xmath6 . thus , we have @xmath89 , @xmath90 , @xmath91 and @xmath92 . notice that the values @xmath93 and @xmath15 are positive integers , unless for the words that are powers of a single letter , in which case @xmath94 and @xmath95 . moreover , one has @xmath96 and @xmath97 ( cf . @xcite , corollary 4.1 ) . the following proposition is from de luca ( @xcite , proposition 4.2 ) . [ prop : del ] let w be a word of length @xmath16 such that @xmath98 and set @xmath99 and @xmath100 . the factor complexity @xmath101 is strictly increasing in the interval @xmath102 $ ] , is nondecreasing in the interval @xmath103 $ ] and strictly decreasing in the interval @xmath104 $ ] . moreover , for @xmath39 in the interval @xmath105 $ ] , one has @xmath106 . if @xmath107 , then @xmath101 is constant in the interval @xmath103 $ ] . proposition [ prop : del ] allows one to give the following definition . a non - empty word @xmath6 is _ trapezoidal _ if * @xmath108 for @xmath109 , * @xmath110 for @xmath111 , * @xmath112 for @xmath113 . trapezoidal words were considered for the first time by de luca @xcite . the name _ trapezoidal _ was given by dalessandro @xcite . the choice of the name is motivated by the fact that for these words the graph of the complexity function defines an isosceles trapezoid ( possibly degenerated in a triangle ) . . one has @xmath114 and @xmath115.,height=226 ] it follows from the definition that a trapezoidal word is necessarily a binary word , since @xmath116 , the number of distinct letters appearing in @xmath6 , must be @xmath117 or @xmath118 . the word @xmath86 considered in example [ ex:1 ] is trapezoidal . see [ fig : graph ] . in the following proposition we gather some characterizations of trapezoidal words . [ prop : trap ] let @xmath6 be a word . the following conditions are equivalent : 1 . @xmath6 is trapezoidal ; 2 . @xmath119 ; 3 . @xmath120 ; 4 . @xmath6 is binary and has at most one left special factor of length @xmath1 for every @xmath61 ; 5 . @xmath6 is binary and has at most one right special factor of length @xmath1 for every @xmath61 ; 6 . @xmath6 has at most @xmath0 distinct factors of length @xmath1 for every @xmath61 ; 7 . @xmath121 for every @xmath61 . the equivalence of @xmath122 , @xmath123 , @xmath124 , @xmath125 and @xmath126 is in @xcite and @xcite . the equivalence of @xmath126 and @xmath127 follows from elementary considerations on the factor complexity of binary words . indeed , it is easy to see the number of distinct factors of length @xmath0 of a binary word @xmath6 is at most equal to the number of distinct factors of length @xmath1 plus the number of right special factors of length @xmath1 . the equivalence of @xmath128 and @xmath122 comes directly from the definitions and from proposition [ prop : del ] . from now on , we fix the alphabet @xmath129 . recall that a word @xmath130 is _ sturmian _ if and only if it is balanced , i.e. , verifies . the following proposition is from de luca ( @xcite , proposition 7.1 ) . [ prop : sturmtrap ] every sturmian word is trapezoidal . the inclusion in proposition [ prop : sturmtrap ] is strict , since there exist trapezoidal words that are not sturmian , e.g. the word @xmath86 considered in example [ ex:1 ] . recall that a word @xmath6 is _ rich _ @xcite ( or _ full _ @xcite ) if it has @xmath131 distinct palindromic factors , that is the maximum number of distinct palindromic factors a word can contain . the following proposition is from de luca , glen and zamboni ( @xcite , proposition 2 ) . [ prop : traprich ] every trapezoidal word is rich . again , the inclusion in proposition [ prop : traprich ] is strict , since there exist binary rich words that are not trapezoidal , e.g. the word @xmath132 . we now discuss the case of non - sturmian trapezoidal words . we first recall the definition and some properties of _ central words _ @xcite . a word @xmath130 is a central word if and only if it has two coprime periods @xmath66 and @xmath133 and length equal to @xmath134 . [ prop : ipcent ] a word @xmath130 is a central word if and only if it satisfies one of the following equations , for some @xmath135 and @xmath61 : 1 . @xmath136 2 . @xmath137 3 . @xmath138 moreover , if case @xmath139 holds one has that @xmath4 and @xmath8 are central words . it is known that if @xmath140 is a central word , then the longest between @xmath4 and @xmath8 has not internal occurrences in @xmath6 . thus , @xmath6 is a closed word . note that this is also a consequence of our theorem [ theor : palclo ] . [ prop : newcent ] a word @xmath130 is a central word if and only if @xmath6 is a palindrome and @xmath141 and @xmath142 are both sturmian words . [ prop : centrbis ] a word @xmath130 is a central word if and only if there exists a sturmian word @xmath143 such that @xmath6 is a bispecial factor of @xmath143 . recall that a word @xmath6 is not balanced if and only if it contains a _ pathological pair _ @xmath19 , that is , @xmath25 and @xmath27 are factors of @xmath6 of the same length but they do not verify . moreover , if @xmath25 and @xmath27 are chosen of minimal length , there exists a palindrome @xmath4 such that @xmath144 and @xmath145 , for two different letters @xmath7 and @xmath77 ( see ( * ? ? ? * proposition 2.1.3 ) or ( * ? ? ? * lemma 3.06 ) ) . we can also state that @xmath25 and @xmath27 are sturmian words , since otherwise they would contain a pathological pair shorter than @xmath146 and hence @xmath6 would contain such a pathological pair , against the minimality of @xmath25 and @xmath27 . so the word @xmath4 is a palindrome such that @xmath147 and @xmath148 are sturmian words , that is , by proposition [ prop : newcent ] , @xmath4 is a central word . moreover , we have the following let @xmath6 be a binary non - sturmian word and @xmath19 a pathological pair of @xmath6 of minimal length . then there exists a unique central word @xmath4 such that @xmath144 and @xmath145 . suppose by contradiction that there exists a word @xmath149 , with @xmath150 , such that @xmath6 contains as factors @xmath147 , @xmath148 , @xmath151 and @xmath152 . let @xmath153 be the longest common prefix between @xmath4 and @xmath154 . since @xmath155 there exist distinct letters @xmath156 such that @xmath157 is a prefix of @xmath4 and @xmath158 is a prefix of @xmath154 . then @xmath6 would contain as factors @xmath159 , @xmath160 , @xmath161 and @xmath162 . therefore , @xmath6 would contain a pathological pair @xmath163 shorter than @xmath19 , a contradiction . the following lemma is attributed to aldo de luca in @xcite . we provide a proof for the sake of completeness . [ lem : separation ] let @xmath6 be a non - sturmian word and @xmath19 the pathological pair of @xmath6 of minimal length . then @xmath25 and @xmath27 do not overlap in @xmath6 . by contradiction , suppose that @xmath6 contains a factor @xmath164 such that @xmath165 and @xmath166 , with @xmath167 . since @xmath168 , then @xmath169 would be a pathological pair of @xmath6 shorter than @xmath19 . we then have : [ prop : centralroot ] let @xmath6 be a binary non - sturmian word . then there exists a unique central word of minimal length @xmath4 such that @xmath170 , for @xmath62 distinct letters and @xmath171 . we call the word @xmath4 the _ central root _ of @xmath6 . the following is the characterization of trapezoidal non - sturmian words given by dalessandro @xcite . [ theor : dal ] let @xmath6 be a binary non - sturmian word . then w is trapezoidal if and only if @xmath172 where @xmath19 is the pathological pair of @xmath6 of minimal length . in particular , if @xmath6 is trapezoidal , then @xmath173 and the longest right special factor of @xmath6 is the prefix of @xmath6 of length @xmath174 , that is , the longest proper prefix of @xmath66 . hence , trapezoidal words are either sturmian or of the form described in theorem [ theor : dal ] . let @xmath86 be the non - sturmian trapezoidal word considered in example [ ex:1 ] . we have @xmath175 and @xmath176 , so that @xmath177 and @xmath178 . the word @xmath6 can be written as @xmath20 , with @xmath179 and @xmath180 . from now on , according to theorem [ theor : dal ] , @xmath19 is the pair of pathological factor of @xmath6 of minimal length . we now show that the words @xmath66 and @xmath133 of the factorization are sturmian words . we premise the following known result : let @xmath6 be a finite word and @xmath71 its fractional root . then @xmath6 is sturmian if and only if so is every power of @xmath71 . [ lem : pk ] let @xmath6 be a trapezoidal non - sturmian word . then the words @xmath66 and @xmath133 of the factorization are sturmian words . since @xmath25 and @xmath27 are sturmian , by the preceding proposition all powers of @xmath24 and @xmath26 are sturmian as well . observe that @xmath181 can be restated as @xmath182 . as prefixes and reversals of sturmian words are sturmian , the assertion follows . the following result , due to de luca , glen and zamboni @xcite , states that trapezoidal palindromes are all sturmian . we give a new proof based on theorem [ theor : dal ] . [ theor : trappal ] the following conditions are equivalent : 1 . @xmath6 is a trapezoidal palindrome ; 2 . @xmath6 is a sturmian palindrome . let @xmath6 be a trapezoidal palindrome . if @xmath6 is non - sturmian we can write , by theorem [ theor : dal ] , @xmath183 , with @xmath184 such that @xmath185 and @xmath186 . without loss of generality , we can assume @xmath187 ; then @xmath188 contains @xmath25 and @xmath27 as factors , a contradiction since , by lemma [ lem : pk ] , @xmath66 is sturmian . so @xmath6 can not be a non - sturmian word . hence , we proved that trapezoidal palindromes are sturmian . since by proposition [ prop : sturmtrap ] , sturmian words are trapezoidal , the assertion follows . mignosi @xcite proved the following formula ( see also @xcite ) for the number @xmath189 of sturmian words of length @xmath1 : @xmath190 where @xmath29 is the euler totient function . therefore , by proposition [ prop : sturmtrap ] , in order to count the number of trapezoidal words of length @xmath1 , it is sufficient to count the number of non - sturmian trapezoidal words of that length . recall from theorem [ theor : dal ] that every non - sturmian trapezoidal word @xmath6 can be uniquely written as @xmath20 , with @xmath191 and @xmath22 , with @xmath144 , @xmath145 , where @xmath4 is the central root of @xmath6 and @xmath19 is the pathological pair of @xmath6 of minimal length . our goal is to count , for a given central word @xmath4 , how many non - sturmian trapezoidal words exist with central root @xmath4 and length @xmath1 . let us premise the following technical lemma , which is a straightforward consequence of theorem [ theor : dal ] . [ lem : tech ] let @xmath4 be a central word , and @xmath156 different letters . the non - sturmian trapezoidal words having central root @xmath4 are exactly the words of the form @xmath192 , for any @xmath193 with @xmath194 , and @xmath195 with @xmath196 . [ theor : enum ] for all @xmath61 , the number of trapezoidal words of length @xmath1 is @xmath197 we first notice that for any word @xmath51 and any @xmath198 , there exists a unique word of length @xmath199 in @xmath200 ( resp . in @xmath201 ) . from this remark and lemma [ lem : tech ] , we deduce that for any @xmath202 and for any central word @xmath4 of length @xmath203 , there are exactly @xmath204 non - sturmian trapezoidal words of length @xmath1 having central root @xmath4 . since there are @xmath205 central words of length @xmath39 ( see @xcite ) , we have that the number of non - sturmian trapezoidal words of length @xmath1 is @xmath206 since the number of trapezoidal words of length @xmath1 is @xmath32 , by and the statement follows . in table [ table : trap ] we give the number of trapezoidal words of each length up to 20 . c * 30@l @xmath1 & 1 & 2 & 3 & 4 & 5 & 6 & 7 & 8 & 9 & 10 & 11 & 12 & 13 & 14 & 15 & 16 & 17 & 18 & 19 & 20 + ' '' '' @xmath207 & 2 & 4 & 8 & 16 & 28 & 46 & 70 & 102 & 140 & 190 & 250 & 318 & 398 & 496 & 602 & 724 & 862 & 1018 & 1192 & 1382 + ' '' '' in this section we derive some properties of open and closed trapezoidal words . recall that a word @xmath6 is closed if it is empty or it has a factor ( different from @xmath6 ) occurring exactly twice in @xmath6 , as a prefix and as a suffix ; otherwise it is open . let @xmath6 be a non - empty word over @xmath13 . the following characterizations of closed words follow easily from the definition : 1 . @xmath6 has a factor occurring exactly twice in @xmath6 , as a prefix and as a suffix of @xmath6 ; 2 . the longest repeated prefix of @xmath6 does not have internal occurrences in @xmath6 , that is , occurs in @xmath6 only as a prefix and as a suffix ; 3 . the longest repeated suffix of @xmath6 does not have internal occurrences in @xmath6 , that is , occurs in @xmath6 only as a suffix and as a prefix ; 4 . the longest repeated prefix of @xmath6 is not right special in @xmath6 ; 5 . the longest repeated suffix of @xmath6 is not left special in @xmath6 ; 6 . @xmath6 has a border that does not have internal occurrences in @xmath6 ; 7 . the longest border of @xmath6 does not have internal occurrences in @xmath6 ; 8 . @xmath6 is the complete return to its longest repeated prefix ; 9 . @xmath6 is the complete return to its longest border ; 10 . @xmath208 , with @xmath209 non - empty , and @xmath210 . [ prop : rev ] let @xmath6 be a word . then @xmath6 is open ( resp . closed ) if and only if @xmath211 is open ( resp . closed ) . straightforward . the following proposition gives a characterization of open trapezoidal words . [ prop : openspe ] let @xmath6 be a trapezoidal word . then the following conditions are equivalent : 1 . @xmath6 is open ; 2 . the longest repeated prefix of @xmath6 is also the longest right special factor of @xmath6 ; 3 . the longest repeated suffix of @xmath6 is also the longest left special factor of @xmath6 . let @xmath213 be the longest repeated prefix of @xmath6 and @xmath214 the letter such that @xmath215 is a prefix of @xmath6 . since @xmath6 is not closed , @xmath213 has a second non - suffix occurrence in @xmath6 followed by a letter @xmath216 . since @xmath213 is the longest repeated prefix of @xmath6 , we have @xmath217 . therefore , @xmath213 is right special in @xmath6 . suppose that @xmath6 has a right special factor @xmath218 longer than @xmath213 . since @xmath6 is trapezoidal , @xmath6 has at most one right special factor for each length ( proposition [ prop : trap ] ) . since the suffixes of a right special factor are right special factors , we have that @xmath213 must be a proper suffix of @xmath218 . since @xmath218 is right special in @xmath6 , it has at least two occurrences in @xmath6 followed by different letters . this implies a non - prefix occurrence of @xmath215 in @xmath6 , against the definition of @xmath213 . let @xmath213 be the longest repeated prefix of @xmath6 . since @xmath213 is a right special factor of @xmath6 there exist different letters @xmath7 and @xmath77 such that @xmath220 and @xmath221 are factors of @xmath6 . this implies that @xmath213 can not appear in @xmath6 only as a prefix and as a suffix , that is , @xmath6 can not be a closed word . @xmath222 can be proven symmetrically . [ prop : hrkl ] let @xmath6 be a trapezoidal word . if @xmath6 is open , then @xmath223 and @xmath224 . if @xmath6 is closed , then @xmath225 and @xmath226 . the claim for open trapezoidal words follows directly from proposition [ prop : openspe ] . suppose that @xmath6 is a closed trapezoidal word . then @xmath225 ( since @xmath6 is closed ) and therefore @xmath226 ( since @xmath6 is trapezoidal ) . the latter proposition , however , does not characterize open and closed trapezoidal words . actually , one can have @xmath227 , as is the case in the word @xmath228 , which is closed , and also in the word @xmath229 , which is open . open trapezoidal words can be sturmian ( e.g. @xmath230 ) or not ( e.g. @xmath231 ) . closed trapezoidal words , instead , are always sturmian , as shown in the following proposition , that can also be found in a paper of the first two authors together with aldo de luca ( cf . @xcite , proposition 3.6 ) . here we report a proof following a totally different approach , based on theorem [ theor : dal ] . [ prop : clostur ] let @xmath6 be a trapezoidal word . if @xmath6 is closed , then @xmath6 is sturmian . suppose that @xmath6 is not sturmian . then , by theorem [ theor : dal ] , @xmath20 , @xmath191 , @xmath22 , with @xmath19 being the pathological pair of @xmath6 of minimal length , @xmath173 ( and hence @xmath232 since @xmath6 is trapezoidal ) and the longest right special factor of @xmath6 is the prefix of @xmath6 of length @xmath174 . let @xmath233 be the suffix of @xmath133 of length @xmath234 , that is , the longest repeated suffix of @xmath6 . since @xmath6 is closed , the only other occurrence of @xmath233 in @xmath6 is as a prefix of @xmath6 . if @xmath235 , then @xmath233 is a prefix of the longest right special factor of @xmath6 . this is a contradiction with the fact that @xmath233 does not have internal occurrences in @xmath6 . if @xmath107 , then @xmath66 is a prefix of @xmath236 and hence @xmath66 is a factor of @xmath133 this implies that @xmath25 is a factor of @xmath133 , and therefore @xmath133 contains both @xmath25 and @xmath27 as factors . hence @xmath133 would be non - sturmian , contradicting lemma [ lem : pk ] . as a corollary of proposition [ prop : clostur ] , we have that every trapezoidal word is open or sturmian . we now focus on closed trapezoidal words and their special factors . [ lem : clo1 ] let @xmath6 be a closed trapezoidal word and @xmath4 the longest left special factor of @xmath6 . then @xmath4 is also the longest right special factor of @xmath6 ( and thus @xmath4 is a bispecial factor of @xmath6 ) . moreover , @xmath4 is a central word . let @xmath4 be the longest left special factor of @xmath6 , so that there exist different letters @xmath62 such that @xmath237 and @xmath238 are factors of @xmath6 . we claim that @xmath237 and @xmath238 both occur in @xmath6 followed by some letter . indeed , suppose the contrary . then one between @xmath237 and @xmath238 , say @xmath237 , appears in @xmath6 only as a suffix . let @xmath233 be the longest repeated suffix of @xmath6 . since @xmath4 is a repeated suffix of @xmath6 , we have @xmath239 . if @xmath240 , then @xmath241 and since @xmath6 is closed , @xmath4 appears in @xmath6 only as a prefix and as a suffix , against the hypothesis that @xmath238 is a factor of @xmath6 . so @xmath242 and therefore @xmath237 must be a suffix of @xmath233 . this implies an internal occurrence of @xmath237 in @xmath6 , a contradiction . so , there exist letters @xmath5 such that @xmath243 and @xmath244 are factors of @xmath6 . now , we must have @xmath245 , since otherwise @xmath246 would be a left special factor of @xmath6 longer than @xmath4 . thus @xmath4 is right special in @xmath6 . since @xmath6 is closed , we have , by proposition [ prop : hrkl ] , @xmath226 , and therefore @xmath4 is the longest right special factor of @xmath6 . by proposition [ prop : clostur ] , @xmath6 is sturmian . since @xmath4 is a bispecial factor of a sturmian word , we have by proposition [ prop : centrbis ] that @xmath4 is a central word . let @xmath247 . the longest repeated prefix of @xmath6 is @xmath229 , which is also the longest repeated suffix and does not have internal occurrences , hence @xmath6 is a closed trapezoidal word . the longest left special factor of @xmath6 is @xmath87 , which is also the longest right special factor of @xmath6 and it is a central word . by theorem [ theor : trappal ] , trapezoidal palindromes coincide with sturmian palindromes , so the next results will be stated in terms of sturmian palindromes but one can replace the word `` sturmian '' with the word `` trapezoidal '' . some results on sturmian palindromes can be found in @xcite . in particular , we recall here the following one . [ theor : delucas ] a palindrome @xmath81 is sturmian if and only if its minimal period @xmath248 satisfies @xmath249 . the next theorem shows that sturmian palindromes , and so , by theorem [ theor : trappal ] , trapezoidal palindromes , are all closed words . [ theor : palclo ] let @xmath6 be a sturmian palindrome . then @xmath6 is closed . by contradiction , suppose that @xmath6 is open . then @xmath213 , the longest repeated prefix of @xmath6 , is also the longest right special factor of @xmath6 ( proposition [ prop : openspe ] ) . since @xmath6 is a palindrome , we have that the longest repeated suffix of @xmath6 is @xmath250 , the reversal of @xmath213 . in particular , then , @xmath251 . by proposition [ prop : hrkl ] , @xmath223 and @xmath224 . thus we have @xmath252 , since @xmath6 is trapezoidal ( see proposition [ prop : trap ] ) . it follows that @xmath253 , for a letter @xmath254 . by theorem [ theor : delucas ] , the period of @xmath6 is @xmath255 , so that @xmath256 . therefore , we have @xmath257 . since @xmath213 is right special in @xmath6 , there exists a letter @xmath217 such that @xmath258 is a factor of @xmath6 . now , any occurrence of @xmath258 in @xmath6 can not be preceded by the letter @xmath214 , since @xmath259 is the longest repeated suffix of @xmath6 . thus @xmath6 contains the factor @xmath260 . hence , @xmath6 contains both @xmath261 and @xmath260 as factors , and this contradicts the fact that @xmath6 is sturmian . from theorem [ theor : palclo ] and lemma [ lem : clo1 ] , we derive the following characterization of the special factors of sturmian palindromes . [ cor : sturpal ] let @xmath6 be a sturmian palindrome . then the longest left special factor of @xmath6 is also the longest right special factor of @xmath6 and it is a central word . let us recall the following characterization of rich words from @xcite : [ prop : compret ] a word @xmath6 is rich if and only if every factor of @xmath6 which is a complete return to a palindrome is palindromic itself . we have the following result : [ prop : palh ] let @xmath6 be a closed trapezoidal word and @xmath213 its longest repeated prefix . then @xmath6 is a sturmian palindrome if and only if @xmath213 is a palindrome . the word @xmath6 is a complete return to @xmath213 , and it is rich by proposition [ prop : traprich ] . since trapezoidal palindromes coincide with sturmian palindromes by theorem [ theor : trappal ] , the preceding proposition implies the assertion . the general structure of a closed trapezoidal word is depicted in [ fig : clotrap ] . , is also the longest repeated suffix and does not have internal occurrences . the longest left special factor , @xmath4 , is also the the longest right special factor and it is a central word ( lemma [ lem : clo1 ] ) . the word is a palindrome if and only if @xmath213 is a palindrome ( proposition [ prop : palh]).,height=37 ] [ rem : richpal ] using proposition [ prop : compret ] , one can prove that theorem [ theor : palclo ] can be generalized to rich palindromes . indeed , let @xmath6 be a rich palindrome , and @xmath4 the longest border of @xmath6 . in order to prove that @xmath6 is closed , it is sufficient to prove that @xmath4 does not have internal occurrences in @xmath6 . suppose the contrary . then @xmath6 would contain a proper prefix of the form @xmath262 . by proposition [ prop : compret ] , @xmath262 is a palindrome . hence , @xmath262 would be a suffix of @xmath6 , against the maximality of @xmath4 . notice that the converse is not true . the word @xmath263 is a closed palindrome of length @xmath264 but contains only @xmath264 palindromic factors , so it is not rich . in what follows , we let by @xmath265 respectively denote the longest repeated prefix , the longest repeated suffix , the longest right special factor and the longest left special factor of the word @xmath6 . an intriguing class of trapezoidal words is that of words in which all these factors coincide . let @xmath6 be a trapezoidal word . we say that @xmath6 is _ semicentral _ if @xmath266 . notice that , by proposition [ prop : openspe ] , semicentral words can also be defined as those open trapezoidal words in which the longest repeated prefix is also the longest repeated suffix . the following theorem provides a characterization of semicentral words . [ theor : sesqui ] a trapezoidal word @xmath6 is semicentral if and only if there exists a central word @xmath4 such that @xmath36 with @xmath267 , @xmath245 . by , if @xmath6 is semicentral we can write @xmath36 , with @xmath268 and @xmath5 ( possibly equal ) letters in @xmath13 . since @xmath4 is a bispecial factor of @xmath6 , it must have a third occurrence in @xmath6 ; as @xmath269 , such an occurrence can not be preceded by @xmath216 or followed by @xmath214 . we have then @xmath270 where @xmath271 , @xmath272 and @xmath273 . if @xmath274 , then @xmath275 , so that @xmath245 and @xmath276 . thus @xmath277 for some @xmath61 , so that it is a central word . similarly , if @xmath278 it follows that @xmath245 and @xmath279 is central . let then @xmath280 . from it follows that there exists a prefix @xmath281 of @xmath4 such that @xmath282 , and a suffix @xmath283 of @xmath4 such that @xmath284 . hence , as @xmath285 , we obtain @xmath286 again , this implies @xmath245 , otherwise the number of occurrences of @xmath214 in @xmath287 and @xmath288 would be different . thus , @xmath289 is central by proposition [ prop : ipcent ] . conversely , let @xmath4 be a central word and @xmath36 , with @xmath290 . we claim that @xmath291 occurs in @xmath6 . this is clear if @xmath4 is a power of @xmath214 or @xmath216 ; by proposition [ prop : ipcent ] , the other possibility is @xmath289 for some words @xmath292 , so that @xmath293 gives the desired occurrence of @xmath291 . hence @xmath4 is a right special factor and a repeated suffix of @xmath6 , so that @xmath294 and @xmath295 . since @xmath296 , it follows from that @xmath6 is trapezoidal and @xmath297 . analogously , since @xmath4 is a left special factor and a repeated prefix of @xmath6 , one proves that @xmath298 . by proposition [ prop : ipcent ] , every central word @xmath4 different from a power of a single letter can be written as @xmath289 for central words @xmath299 and @xmath300 and distinct letters @xmath5 . theorem [ theor : sesqui ] provides a similar characterization for semicentral words ( and this motivates the choice of the name ) . it is also worth noticing that semicentral words are non - strictly bispecial sturmian words ( see @xcite ) , that is , for any semicentral word @xmath6 one has that @xmath301 and @xmath302 are all sturmian words , but nevertheless either @xmath303 or @xmath304 is not sturmian . the semicentral words of length @xmath305 are : @xmath306 , @xmath307 , @xmath308 , @xmath309 , @xmath310 , @xmath311 , @xmath312 and @xmath313 . as a consequence of theorem [ theor : sesqui ] , we have that the number @xmath314 of semicentral words of length @xmath1 is : @xmath315 in this section , we give an interesting example of the open / closed dichotomy in trapezoidal words , by analyzing prefixes of the fibonacci infinite word under this new light . the fibonacci infinite word @xmath25 is the word @xmath316 obtained as the limit of the substitution @xmath317 , @xmath318 . note that every prefix of the fibonacci word is sturmian , and therefore trapezoidal . we investigate which prefixes of @xmath25 are open and which are closed . if one writes up the list of the prefixes of @xmath25 for each length ( see table [ tab : fibo ] ) , can notice that the sequences of `` open '' and `` closed '' alternate following the fibonacci sequence @xmath319 ( defined by @xmath320 and @xmath321 for every @xmath322 ) . more precisely , the numbers of consecutive prefixes that are closed ( resp . open ) , that is , the lengths of the runs of `` c '' and `` o '' in table [ tab : fibo ] , are the fibonacci numbers . in theorem [ theor : fibo ] we show this fact . r9@c11c @xmath1 & 1 & 2 & 3 & 4 & 5 & 6 & 7 & 8 & 9 & 10 & 11 & 12 & 13 & 14 & 15 & 16 & 17 & 18 & 19 & 20 + + @xmath323 & c & o & c & o & c & c & o & o & c & c & c & o & o & o & c & c & c & c & c & o + [ tab : fibo ] [ theor : fibo ] let @xmath6 be a prefix of the fibonacci word @xmath25 . then @xmath6 is open if and only if there exists @xmath39 such that @xmath324 . first , let us recall some well known properties of the fibonacci word . let @xmath325 be the set of palindromic prefixes of @xmath25 . it is well known that the elements of @xmath326 are the prefixes @xmath327 of @xmath25 such that @xmath328 , @xmath329 . the words in @xmath326 are all central words . also , for every @xmath330 we have @xmath331 where @xmath332 , @xmath333 for even @xmath39 , and @xmath334 , @xmath335 for odd @xmath39 . our assertion is trivially verified if @xmath336 . hence we only need to show that for all @xmath330 , a prefix @xmath6 of @xmath25 with @xmath337 is 1 . open if @xmath338 , and 2 . closed otherwise , i.e. , when @xmath339 . let us fix @xmath330 and set @xmath340 , @xmath341 as in . we know that the prefix of @xmath25 of length @xmath342 is @xmath343 . by theorem [ theor : palclo ] , it is a closed word and its longest repeated prefix is @xmath327 . moreover , @xmath327 is also the longest repeated prefix of the semicentral word @xmath344 , that is the prefix of @xmath25 of length @xmath345 . therefore , @xmath327 is the longest repeated prefix for all @xmath6 s in between , i.e. , for each @xmath346 with @xmath338 . as @xmath327 has an internal occurrence in all such prefixes , they are all open , proving point 1 . it remains to show that any prefix of @xmath347 longer than @xmath344 is closed . indeed , such words have a border which is strictly longer than @xmath327 , and which can not have any internal occurrences for otherwise @xmath348 would have a second occurrence in @xmath344 . .the structure of the prefixes of @xmath25 with respect to the palindromic prefixes @xmath327 . [ cols="<,<,^,<,^",options="header " , ] the longest word in a run of closed prefixes of @xmath25 is a central word . the longest word in a run of open prefixes of @xmath25 is a semicentral word . in this paper we have investigated trapezoidal words , which are a natural generalization of finite sturmian words . we exhibited a formula for counting trapezoidal words and studied open and closed trapezoidal words separately . trapezoidal words form a subclass of the class of rich words , that are words containing the maximum number of palindromic factors . a challenging problem is that of finding an enumerative formula for rich words . this problem is still open , even in the binary case . we think that separating rich words in open and closed could give further insights on these words . more generally , we think that the open / closed dichotomy can be useful in the study of other classes of words . for instance , it would be interesting to generalize the results obtained in section [ sec : fibo ] for the fibonacci word to the class of standard words . m. bucci , a. de luca , a. de luca , rich and periodic - like words , in : v. diekert , d. nowotka ( eds . ) , dlt 2009 , 13th international conference on developments in language theory , vol . 5583 of lecture notes in comput . sci . , springer , 2009 , pp . 145155 . a. carpi , a. de luca , central sturmian words : recent developments , in : c. de felice , a. restivo ( eds . ) , dlt 2005 , 9th international conference on developments in language theory , vol . 3572 of lecture notes in comput . sci . , springer , 2005 , pp . 3656 . g. fici . a characterization of bispecial sturmian words , in : b. rovan , v. sassone , and p. widmayer ( eds . ) , mfcs 2012 , 37th international symposium on mathematical foundations of computer science , vol . 7464 of lecture notes in comput . sci . , springer , 2012 , pp . 383394 . g. fici , a classification of trapezoidal words , in : p. ambro , . holub , z. maskov ( eds . ) , words 2011 , 8th international conference on words , no . 63 in electronic proceedings in theoretical computer science , 2011 , pp . 129137 .	trapezoidal words are words having at most @xmath0 distinct factors of length @xmath1 for every @xmath2 . they therefore encompass finite sturmian words . we give combinatorial characterizations of trapezoidal words and exhibit a formula for their enumeration . we then separate trapezoidal words into two disjoint classes : open and closed . a trapezoidal word is closed if it has a factor that occurs only as a prefix and as a suffix ; otherwise it is open . we investigate open and closed trapezoidal words , in relation with their special factors . we prove that sturmian palindromes are closed trapezoidal words and that a closed trapezoidal word is a sturmian palindrome if and only if its longest repeated prefix is a palindrome . we also define a new class of words , _ semicentral words _ , and show that they are characterized by the property that they can be written as @xmath3 , for a central word @xmath4 and two different letters @xmath5 . finally , we investigate the prefixes of the fibonacci word with respect to the property of being open or closed trapezoidal words , and show that the sequence of open and closed prefixes of the fibonacci word follows the fibonacci sequence . sturmian words , trapezoidal words , rich words , closed words , special factors , palindromes , enumerative formula .
indirect evidence for the existence of massive black holes ( mbhs ) in active galactic nuclei ( agns ) has been growing over the years ( e.g. rees 1984 ) . however , the most conclusive evidence for the existence of massive black holes has been recently obtained in the milky way ( e.g. genzel et al . 1997 ) , and in ngc 4258 ( miyoshi et al . 1995 ) , a weakly - active galaxy . this new evidence is based on high spatial resolution observations of stellar and gas kinematics . similar estimates could not be employed in quasars and bright seyfert galaxies as the stellar kinematics close to the black hole is hopelessly lost behind the glare of the active nucleus . a rough estimate of the black hole mass in agns can be obtained based on the size and the typical velocities in the broad line region ( blr , e.g. dibai 1981 ; wandel & yahil 1985 ; joly et al . 1985 ; padovani & rafanelli 1988 ; koratkar & gaskell 1991 ) . however , this method is susceptible to various systematic errors , and there is currently no independent way to estimate its accuracy . compact massive dark objects , most likely mbhs , were inferred in the cores of many nearby normal galaxies based on stellar kinematics and the observed light distribution ( see review by kormendy & richstone 1995 ) . in a recent comprehensive study of the stellar dynamics of a large sample of nearby galaxies magorrian et al . ( 1998 ) found that a mbh may exist in the cores of nearly all bulges . they also confirmed the strong correlation between the black hole mass and the bulge mass , consistent with @xmath1 . if quasars reside in normal galaxies , then their black hole mass and bulge mass should also follow this correlation . it is not possible to explore this correlation directly in quasars since stellar velocity distributions in the host bulges have not been measured yet . however , since there is a strong correlation between @xmath2 and @xmath3 in galaxies ( faber et al . 1997 ) , one can instead test if quasars follow the @xmath4 versus @xmath3 correlation found for normal galaxies . rather accurate determinations of quasar host galaxy luminosities were recently obtained by bahcall et al . ( 1997 ) for a representative sample of 20 bright low redshift quasars using the _ hubble space telescope ( hst)_. in this _ letter _ i show that the quasar host galaxy luminosity appears to be significantly correlated with the h@xmath5-deduced black hole mass , @xmath6 , and that this correlation is very similar to the @xmath3 versus @xmath4 relation determined by magorrian et al . for nearby normal galaxies . this similarity provides a `` calibration '' for the @xmath4 estimates in agns . the @xmath6 estimation method , its application to the bahcall et al . sample , and the correlation analysis results are given in 2 , and the implications are discussed in 3 , together with some predictions of host luminosities which can be tested with _ hst _ in the near future . the black hole mass can be estimated using the the velocity dispersion of the h@xmath5 emitting clouds in the blr and the size of this region , together with the assumption that the clouds motion are virialized , i.e. , @xmath8 where @xmath9 is the size of the -emitting region in the blr and @xmath10 is the observed velocity dispersion . kaspi et al . ( 1996 ) find @xmath11 pc , where @xmath12 is the @xmath13 luminosity in units of @xmath14 erg s@xmath15 , assuming @xmath16 km s@xmath15 mpc@xmath15 . this relation is equivalent to @xmath17 where @xmath18 is the bolometric luminosity in units of @xmath19 erg s@xmath15 ( using @xmath20 , e.g. fig.7 in laor & draine 1993 ) , and @xmath21 km s@xmath15 mpc@xmath15 which is used throughout the paper . the @xmath22 scaling is also indicated by the weak luminosity dependence of agn emission line spectra , and is also expected based on dust sublimation which occurs at @xmath23 ( laor & draine 1993 ; netzer & laor 1993 ) . thus , just based on @xmath10 , taken here as the observed h@xmath5 fwhm , and @xmath18 , one obtains the following estimate for the black hole mass from eqs.1 & 2 , @xmath24 where @xmath25 , and + @xmath26 fwhm/3000 km @xmath15 . quasar hosts have been studied extensively from the ground ( e.g. mcleod & rieke 1994a , 1994b ; dunlop et al . however , separating out the quasar host galaxy clearly requires a high angular resolution , and thus measurements with the _ hst _ can provide the most accurate determination of quasar host properties . i use the sample of 20 luminous low redshift ( @xmath27 ) quasars studied by bahcall et al . with the hst wide field / planetary camera-2 ( wfpc2 ) . this sample is likely to represent the properties of nearby bright quasars . in addition , the large sample size , the uniform and detailed reduction , and the detection of all quasar hosts , make this sample the best one available for exploring the @xmath4 versus @xmath7 relation in quasars . some of the host galaxies morphologies were identified by bahcall et al . as elliptical , and for these galaxies @xmath28 ( taken from the best fit 2-d model in their table 5 ) . other hosts were identified as spirals , or interacting , and the value of @xmath3 for these objects , required for a direct comparison with the magorrian et al . results , is not available . an estimate of @xmath3 for the objects which are best fit by an exponential disk is obtained by subtracting the @xmath29 kpc annular magnitude ( their table 8) from the total magnitude ( their table 5 ) , yielding @xmath30(inner host ) . the mean @xmath31(host@xmath32inner host ) is 0.5 mag , which is smaller than @xmath33(total@xmath32bulge)@xmath34 mag for early type spiral galaxies ( simien & de vaucouleurs 1986 ) , suggesting that @xmath30(inner host ) overestimates @xmath3 . although , @xmath30(host ) may be underestimated , since the fit does not included a bulge component . table 1 lists the bahcall et al . quasars used for the correlation analysis together with their @xmath35 , @xmath30(bulge ) for objects with a de vaucouleurs fit , or @xmath30(inner host ) for objects with an exponential disk fit , h@xmath5 fwhm , bolometric luminosity , @xmath4 ( ) as deduced from equation 3 , and the host morphology from bahcall et al . the h@xmath5 fwhm is obtained from boroson & green ( 1992 ) which provide high quality and uniformly reduced spectra of all 87 @xmath36 pg quasars ( schmidt & green 1983 ) , of which 14 overlap with the bahcall et al . sample . continuum fluxes are available for all of these 14 pg quasars in neugebauer et al . ( 1987 ) , which provides accurate and uniformly reduced continuum spectrophotometry for most pg quasars . the luminosity at 3000 is converted to @xmath37 using @xmath38(3000 ) ( see fig.7 in laor & draine ) . there is no uniform data set with the continuum flux and h@xmath5 fwhm for 5 additional quasars from the bahcall et al . sample . different papers quote parameters that can differ by @xmath39% for a given object . these objects were therefore not included in the analysis as they may be subject to significant systematic deviations . the upper panel in figure 1 shows @xmath30(bulge / inner host ) [ hereafter @xmath30(b / ih ) ] versus @xmath4 for the 19 bahcall et al . only the 14 quasars marked with filled squares were used in the analysis . the spearman rank order correlation coefficient is @xmath40 which has a probability of 0.005 to occur for unrelated parameters . a simple least squares fit to the data gives @xmath41 three quasars which are best fit by an exponential disk , pks 1302 - 102 , pg 1307 + 085 , and pg 1444 + 407 ( bahcall et al . table 5 ) , may have an elliptical morphology ( table 1 ) . a least squares fit for the 14 quasars using the three quasars de vaucouleurs fit @xmath42 yields the coefficients @xmath43 and @xmath44 . the middle panel in fig.1 shows the @xmath30(bulge ) vs. @xmath4 relation for nearby normal galaxies from magorrian et al . the dashed line represents the relation @xmath45 as deduced from the @xmath46 and @xmath47 relations found by magorrian et al . , and the standard relation @xmath48 . the quasar correlation is flatter than the magorrian et al . correlation ( @xmath49 vs. @xmath50 ) . this may be partly due to the fact that all the quasar hosts at @xmath51 are disk galaxies , where @xmath30(inner host ) may overestimate @xmath30(bulge ) ( 2.2 ) . the lower panel in figure 1 compares directly the distributions of the 32 magorrian et al . galaxies and of the 19 quasars in the @xmath4 versus @xmath30(b / ih ) plane . the two distributions overlap surprisingly well . the overlap of the distributions of quasars and of normal galaxies in the @xmath4 versus @xmath30(b / ih ) plane is the main result of this paper . this overlap is remarkable as bright quasars and nearby galaxies are apparently different types of objects , and since the @xmath4 estimates in quasars and in nearby galaxies are based on very different methods ( the blr versus stellar dynamics ) . the overlap is also surprising given the crudeness of the @xmath4 estimates for both populations . as stressed by magorrian et al . , their data are fit with a simplified , axisymmetric , stellar dynamics model , and a more general model may yield @xmath4 which could be off by an order of magnitude , or may even not require a `` massive dark object '' at all . the @xmath6 estimate is even cruder . large systematic errors could be induced if the blr velocity field or the optical - uv continuum are anisotropic , if the scaling of @xmath9 with @xmath52 does not hold in bright quasars , or if the h@xmath5 dynamics is affected by non gravitational forces ( e.g. radiation pressure , magnetic fields ) . the overlap suggests a number of interesting implications . first , concerning the blr : 1 . the h@xmath5 dynamics are most likely dominated by gravity ; 2 . the h@xmath5 velocity field and the observed optical - uv emission are not likely to be strongly anisotropic , and 3 . the @xmath9 versus @xmath52 relation most likely holds in quasars . second , concerning the @xmath6 estimate ; the overlap allows a `` calibration '' of this mass estimate and suggests it is probably accurate to within @xmath0 on log scale . third , concerning quasar hosts ; 1 . the scatter in the @xmath6 versus @xmath30(b / ih ) correlation suggests that @xmath30(b / ih ) can be estimated to within @xmath0 mag based on the quasar luminosity and line width . the overlap of the two distributions suggests that quasar hosts are similar to normal , nearby galaxies , and thus that @xmath30(b / ih ) is generally not strongly affected by processes such as a nuclear starburst , or distortions due to a tidal interaction . this correlation may also be useful for surveys of the large scale structure of the universe ( 2df , sdss ) . quasars can be used as bright markers of galaxies out to high @xmath35 whose bulge luminosity and mass can be deduced from the quasar emission spectra , allowing studies of clustering as a function of mass . the correlation found here , @xmath53 ( or @xmath54 ) translates using eq.7 to @xmath55 ( or @xmath56 ) , which is steeper than the magorrian et al . relation @xmath57 . the slope of the quasar relation has a relatively large uncertainty due to the small range in @xmath4 available ( @xmath58 ) , but it is interesting to note that at the high mass end ( @xmath59 ) the magorrian et al . galaxies appear to follow the quasar relation quite well ( see fig.1 ) . at the low black hole mass end one has the two best @xmath4 estimates available , in the galaxy and in ngc 4258 , where @xmath60 ( miyoshi et al . 1995 ; genzel et al . 1997 ) and @xmath61 ( bahcall & soneira 1980 ; rc2 catalogue + simien & de vaucouleurs 1986 ) . these galaxies follow the quasar relation significantly better than the magorrian relation ( see fig.1 ) . thus , the data in the range @xmath62 appears to agree better with the steeper quasar relation . the quasar relation is also interestingly close to the haehnelt , natarajan & rees ( 1998 ) prediction of @xmath63 . there are some objects , such as m 32 , which appear to agree better with the nearby galaxies relation ( fig.1 ) . a number of seyfert 1 galaxies with @xmath64 , as deduced by reverberation mappings ( peterson et al . 1998 ) , are 1 - 2 mag brighter than expected based on the quasar relation ( ho 1998 ) . subtraction of the agn light from the host light would bring them closer to the quasar relation . the distribution of quasars in the absolute quasar @xmath65 band magnitude @xmath66 versus the absolute host @xmath67 band magnitude @xmath68 plane appears to be bounded such that @xmath69 ( e.g. mcleod & rieke 1995 , their figure 6 ) . mcleod ( 1998 ) suggested that the reason for this bound is that objects where @xmath70 `` have a maximum allowed black hole mass for their galaxy mass and that the black hole is accreting at the eddington rate . '' this idea is broadly consistent with the correlation found here . for example , a quasar with @xmath71 mag typically has @xmath72 mag ( using @xmath73 mag for our 14 pg quasar hosts , with @xmath68 from mcleod & rieke 1994b ) . equation 4 then gives @xmath74 ( or @xmath75 ) . a magnitude of @xmath76 mag translates to @xmath77)=45.55 , and @xmath78 , which corresponds to 0.44 or 0.75 of @xmath79 . the above estimates are rather rough since the @xmath68 versus @xmath80 correlation has a significant scatter . how can the analysis presented here be improved ? the crude `` inner host '' estimate for the bulge luminosity , used here for disk galaxies , can be improved by fitting a disk+bulge model to the _ hst _ images . this may be feasible for early type spiral hosts where the typical effective radius of the bulge is @xmath81 kpc ( simien & de vaucouleurs ) , or @xmath82 for the bahcall et al . . however , it will not be feasible for late type spiral hosts , where @xmath84 kpc , and a sample of lower @xmath35 agns will be required . one also needs to measure the h@xmath5 fwhm and luminosity simultaneously to guard against variability , and to obtain a more accurate estimate of the ionizing luminosity to use in the @xmath9 relation ( eq.2 ) . obscuration effects are well established in agns , and these may increase the scatter , if not accounted for . in particular , objects with a very narrow h@xmath5 line may have their blr partly obscured , or may be strongly dominated by emission from the narrow line region . using the variable h@xmath5 component profile can overcome such biases . future observations with _ hst _ can address the following questions : 1 . does the black hole mass vs. host bulge luminosity correlation extends to quasars with higher and lower black hole masses ? 2 . can the scatter in the correlation be reduced with a more careful analysis , or is it intrinsically large , as suggested for galaxies ? and 3 . does @xmath85 , as suggested here , or is @xmath57 as suggested by magorrian et al . ? the pg quasars sample may be particularly useful for such future explorations with _ hst _ since a high s / n homogeneous spectroscopic data base is already available from neugebauer et al . and boroson & green . using this data set and eqs.3 & 4 one can predict that of the 87 @xmath86 pg quasars , some of the lowest luminousity hosts should be found in pg 1244 + 026 , pg 1404 + 226 , and pg 1448 + 273 ( predicted @xmath80=@xmath87 to @xmath88 ) , while some of the highest luminousity hosts should be found in pg 1704 + 608 , pg 1425 + 267 , and pg 2308 + 098 ( @xmath89 to @xmath90 ) . one can also predict that the hosts of pg 2304 + 042 and pg 2209 + 184 should be 2 - 3 magnitudes brighter than the hosts of pg 1244 + 026 and pg 1448 + 273 respectively , although the former and later quasars have , respectively , similar luminosities . bahcall , j. n. , & soneira , r. m. 1980 , , 44 , 73 bahcall , j. n. , kirhakos , s. , saxe , d. h. , & schneider , d. p. 1997 , , 479 , 642 boroson , t. a. & green , r. f. 1992 , , 80 , 109 dibai , e. a. 1981 , soviet astron . , 24 , 389 dunlop , j. s. , taylor , g. l. , hughes , d. h. , & robson , e. i. 1993 , , 264 , 455 faber , s. m. , et al . 1997 , , 114 , 1771 genzel , r. , eckart , a. , ott , t. , eisenhauer , f. , 1997 , mnras , 291 , 219 haehnelt , m.g . , natarajan , p. , & rees , m. j. 1998 , , in press ( astro - ph/9712259 ) ho , l. c. 1998 , in observational evidence for black holes in the universe , ed . s. k. chakrabarti ( dordrecht : kluwer ) , in press joly , m. , collin - souffrin , s. , masnou , j. l. , & nottale , l. 1985 , a&a , 152 , 282 kaspi , s .. smith , p. s. , maoz , d. , netzer , h. , & jannuzi , b. t. 1996 , , 471 , l75 koratkar , a. p. , & gaskell , c. m. 1991 , , 370 , 61 kormendy , j. , & richstone , d. o. 1995 , ara&a , 33 , 581 laor , a. , & draine , b. t. 1993 , , 402 , 441 magorrian , j. , et al . , 1998 , , 115 , 2285 mcleod , k. k. , & rieke , g. h. 1994a , , 420 , 58 mcleod , k. k. , & rieke , g. h. 1994b , , 431 , 137 mcleod , k. k. , & rieke , g. h. 1995 , , 441 , 96 mcleod , k. k. , 1998 , in quasar hosts , eds . d. clements and i. perez - fournon ( berlin : springer - verlag ) , in press miyoshi , m. , et al . 1995 , nature , 373 , 127 netzer , h. , & laor , a. 1993 , , 404 , l51 neugebauer , g. , et al . 1987 , , 63 , 615 padovani , p. , & rafanelli , a&a , 205 , 53 peterson , b. m. , et al . 1998 , , 501 , 82 rees , m. j. 1984 , ara&a , 22 , 471 simien , f. , & de vaucouleurs , g. 1986 , , 302 , 564 schmidt , m. , & green , r. f. 1983 , , 269 , 352 wandel , a. , & yahil , a. 1985 , , 295 , l1 lccrccl object & @xmath35 & @xmath91 & h@xmath92 & @xmath93 & @xmath94 & @xmath95 + pg 0052 + 251 & 0.155 & @xmath3220.89 & 5.20 & 46.02 & 8.74 & sb + phl 909@xmath96 & 0.171 & @xmath3221.48 & 11.0 & 46.37 & 9.57 & e4 + nab 0205 + 02@xmath96 & 0.155 & @xmath3219.33 & 1.05 & 46.41 & 7.55 & s0 ? + ph 0923 + 201 & 0.190 & @xmath3221.48 & 7.61 & 46.06 & 9.09 & e1 + pg 0953 + 414 & 0.239 & @xmath3220.29 & 3.13 & 46.39 & 8.49 & ? + pks 1004 + 130 & 0.240 & @xmath3222.48 & 6.30 & 46.39 & 9.10 & e2 + pg 1012 + 008 & 0.185 & @xmath3219.91 & 2.64 & 45.85 & 8.07 & int . + he 1029@xmath32140@xmath96 & 0.086 & @xmath3220.98 & 7.50 & 46.49 & 9.30 & e1 + pg 1116 + 215 & 0.177 & @xmath3221.88 & 2.92 & 46.38 & 8.42 & e2 + pg 1202 + 281 & 0.165 & @xmath3220.98 & 5.05 & 45.51 & 8.46 & e1 + 3c 273 & 0.158 & @xmath3222.58 & 3.52 & 46.96 & 8.87 & e4 + pks 1302@xmath32102 & 0.286 & @xmath3221.15 & 3.40 & 46.77 & 8.75 & e4 ? + pg 1307 + 085 & 0.155 & @xmath3220.51 & 2.36 & 45.99 & 8.04 & e1 ? + pg 1309 + 355 & 0.184 & @xmath3221.23 & 2.94 & 45.83 & 8.15 & sab + pg 1402 + 261 & 0.164 & @xmath3219.95 & 1.91 & 45.96 & 7.84 & sbb + pg 1444 + 407 & 0.267 & @xmath3220.49 & 2.48 & 46.13 & 8.16 & e1 ? + 3c 323.1 & 0.266 & @xmath3221.48 & 7.03 & 46.34 & 9.17 & e3 ? + pks 2135@xmath32147@xmath96 & 0.200 & @xmath3221.58 & 5.50 & 46.61 & 9.09 & e1 + pks 2349@xmath32014@xmath96 & 0.173 & @xmath3222.58 & 5.50 & 46.57 & 9.07 & int . + @xmath97 host absolute magnitude from bahcall et al . ( 1997 , table 5 ) , calculated for @xmath21 km s@xmath15 mpc@xmath15 . + @xmath98 h@xmath5 fwhm in units of 1000 km s@xmath15 from boroson & green ( 1992 ) . + @xmath99 log bolometric luminosity in erg s@xmath15 , based on @xmath100 at rest frame 3000 from neugebauer et al . + @xmath101 log of black hole mass in units of @xmath102 ( see eq.3 ) . + @xmath103 host morphology from bahcall et al . + @xmath96 object not included in the correlation analysis ( see text ) .	the mass of massive black holes in quasar cores can be deduced using the typical velocities of -emitting clouds in the broad line region ( blr ) and the size of this region . however , this estimate depends on various assumptions and is susceptible to large systematic errors . the -deduced black hole mass in a sample of 14 bright quasars is found here to correlate with the quasar host galaxy luminosity , as determined with the _ hubble space telescope ( hst)_. this correlation is similar to the black hole mass vs. bulge luminosity correlation found by magorrian et al . in a sample of 32 nearby normal galaxies . the similarity of the two correlations is remarkable since the two samples involve apparently different types of objects and since the black hole mass estimates in quasars and in nearby galaxies are based on very different methods . this similarity provides a `` calibration '' of the -deduced black hole mass estimate , suggesting it is accurate to @xmath0 on log scale . the similarity of the two correlations also suggests that quasars reside in otherwise normal galaxies , and that the luminosity of quasar hosts can be estimated to @xmath0 mag based on the quasar continuum luminosity and the line width . future imaging observations of additional broad - line active galaxies with the _ hst _ are required in order to explore the extent , slope , and scatter of the black hole mass vs. host bulge luminosity correlation in active galaxies .
nuclear matter produces a restoration of chiral symmetry which is remarkably large . the quark condensate , which is the order parameter of the spontaneous symmetry breaking , decreases in magnitude by about 1/3 at the normal density @xmath1@xcite . the bulk of this restoration comes from the additive effect of independent nucleons . an effort has been made to study the restoration effect due to the interaction between the nucleons @xcite . in particular , chanfray and ericson @xcite made the observation that the quark condensate in a dense medium ( of density @xmath2 ) is governed by the nuclear sigma commutator with the exact relation : @xmath3 where @xmath4 is the quark condensate density in the medium and @xmath5 its vacuum value . the pion mass and decay constant are @xmath6 and @xmath7 ( = 94 mev ) . we have denoted by @xmath8 the nuclear sigma commutator per unit volume with the definition for a piece of nuclear matter of density @xmath2 : @xmath9\big ] \vert \psi_a \rangle \label{dcom}\ ] ] where @xmath10 is the nuclear ground state with total nucleon number @xmath11 , @xmath12 the hamiltonian and @xmath13 the axial charge . in this way the quantity @xmath14 , which is a function of the density , is the effective sigma commutator per nucleon in the nuclear medium . thus the study of the in - medium quark condensate amounts to that of the sigma commutator . in the low density limit , @xmath15 reduces to @xmath16 the nucleon sigma commutator and the relation ( [ lin ] ) leads to the linear density dependence given in refs . @xcite . notice that since @xmath15 is of order @xmath17 the relative modification of the condensate survives in the chiral limit @xmath18 . using the partially conserved axial current hypothesis ( pcac ) , chanfray and ericson have related @xmath15 to the soft pion amplitude on the nucleus . they have thus shown that a correction to the additive assumption , _ i.e. _ @xmath19 , arises from the contribution of the pions exchanged between the nucleons which also participate in the restoration of chiral symmetry . this simple result does not depend on the original assumption that pcac holds @xcite . pushing the consequences of pcac , m. ericson has suggested the existence of a reaction of the nuclear medium against the restoration of chiral symmetry , arising from the distortion of the soft pion wave @xcite . this possibility was examined by birse and mc govern @xcite in the linear sigma model where pcac holds . the distortion effect is indeed present but these authors have shown that it is overcompensated by exchange effects , linked to the @xmath20 meson , such that there is an overall acceleration of the restoration instead of a hindrance . on the other hand t. ericson has shown that the soft pion amplitude can be appreciably affected by the correlations @xcite , eventually cancelling , as expected , the distortion effect in the limit of nucleons totally confined inside the correlation hole . the aim of the present work is twofold . we first want to complete the studies of refs . @xcite by introducing the correlations in the linear sigma model in a consistent way , which was not done previously . we will show that the effect of the correlations should not be discussed alone , but in connection with those of the @xmath20 meson exchange introduced in ref . @xcite . using the information provided by the scattering of physical pions on nuclei we will discuss the limitations of the linear sigma model and show that it can not be trusted for a quantitative discussion . the second and main aim of this work is then to perform an exploration of the nuclear sigma commutator in the non - linear sigma model , which does not assume the existence of a scalar meson . in the non - linear version the pcac relation does not automatically hold . this hypothesis is abandoned as a systematic assumption and only the chiral properties of the interaction are emphasized . however for certain representations it is possible to make a transformation on the field in such a way that pcac is valid . in this case the nuclear sigma commutator coincides with the soft pion amplitude and it is possible to elucidate the interplay between the distortion and the exchange terms . we will show that physical quantities , such as the condensate and the in - medium pion mass are independent of the representation , as expected from a general theorem @xcite . we will also introduce the correlations in a consistent way in this model . we limit our discussion to the s - wave isospin symmetric pion - nucleon amplitude . we work in the linear sigma model , where the sigma commutator is proportional to the soft pion amplitude . we want to evaluate the scattering amplitude for soft pion on a nucleus , taken for simplicity as a piece of nuclear matter of density @xmath2 . we denote @xmath21 the amplitude per unit volume so that @xmath22 . we work up to second order in the density . for simplicity we limit our calculation to the tree level . the graphs contributing to second order in the density to the soft pion scattering amplitude are represented in fig . [ figa]a - d . the coherent rescattering term is represented in fig . the graphs linked to @xmath20 exchange introduced in ref . @xcite are shown in the graphs of fig . [ figa]b - d . the short range correlations add the contributions of graphs [ figb]a - d , where the wiggled line may be interpreted as the iterated exchange of a heavy meson , such as the @xmath23 meson , responsible for the short range repulsion . we take for the correlation function the convenient schematic form @xmath24 ( with @xmath25 ) introduced by weise @xcite , which has the simple fourier transform @xmath26 . denoting @xmath27 the sum of the sigma exchange contributions in the correlated medium we obtain in the hartree approximation which we follow throughout this paper for the clarity of the argument : @xmath28 + o(m_\pi^4)\ ; , \label{sigex}\ ] ] where @xmath29 is the meson - nucleon coupling constant . on the other hand , the sum of the coherent and incoherent rescattering is ( fig . [ figa]a and [ figb]a ) : @xmath30 + o(m_\pi^4)\ ; . \label{sigres}\ ] ] both terms @xmath27 and @xmath31 are strongly affected by the correlations , as apparent from the presence of a term proportional to @xmath32 which does not vanish in the chiral limit @xmath33 . however , when the sum of the second order contributions is performed , the two terms in @xmath32 exactly cancel each other . what this cancellation reveals is that the role of the incoherent rescattering ( graph [ figb]a ) should not be discussed alone , otherwise chiral symmetry would be violated since the amplitude @xmath21 would not vanish in the chiral limit . instead incoherent rescattering should be associated with the contribution of the correlations in the @xmath20 meson exchange graphs [ figb]b - d so as to make the soft amplitude go as @xmath34 . this cancellation has not been discussed in the previous works of refs . @xcite but it is necessary to reach a consistent description . adding the contribution of first order in the density to obtain the total amplitude we arrive at the following expression for the density evolution of the quark condensate : @xmath35 \left ( 1 - \frac{m^2_\pi } { m^2_\sigma } \right)\ > . \label{consig}\ ] ] here we have introduced the sigma commutator of the nucleon , @xmath36 which in the linear sigma model writes : @xmath37 . notice that @xmath15 could also be obtained directly as the expectation value of the symmetry breaking piece of the lagrangian , i.e. of @xmath38 , in a correlated medium . the toy model of ericson @xcite corresponds to the limit of a large correlation hole compared to the extension of the condensate . in the linear sigma model this corresponds to the condition @xmath39 . one sees on eq . ( [ consig ] ) that the term in @xmath40 then goes to zero and @xmath41 . the meson exchange terms disappear as expected , but also the incoherent rescattering cancels the distortion , as found in ref . @xcite . in this case the quark condensate density in the nuclear medium evolves linearly with density , as given by the eq . ( [ lin ] ) if we replace @xmath15 by @xmath36 . this is an extreme situation and the question is how close it is to a realistic one . in fact the size of the condensate , or equivalently the scalar radius of the nucleon , has been extracted from experimental data @xcite . its value , @xmath42 fm is sensibly larger than the typical correlation radius @xmath43 fm . there is indeed a large overlap of the nucleon condensates which is neither reproduced in the linear sigma model nor in the toy model of ref . @xcite . we will discuss below in more detail the limitations of the linear sigma model for a proper description of the problems of the quark condensate and physical pion scattering . we first give in table [ tab1 ] the value of the quark condensate in this model at nuclear matter density @xmath44 @xmath45 . for the numerical evaluation we take the parameters of the linear sigma model in such a way as to reproduce the experimental value of the sigma commutator : @xmath46 45 mev @xcite . this fixes the @xmath20 mass at @xmath47 = 740 mev . on the other hand we take for the cut - off parameter @xmath25 = 782 mev . the acceleration of chiral symmetry restoration by @xmath20 exchange terms is appreciable , 0.54 instead of 0.65 in the linear approximation eq . ( [ lin ] ) . since the linear sigma model is a consistent one which fulfils the requirements of chiral symmetry we find of interest to give also in table [ tab1 ] the effective nucleon mass @xmath48 as evaluated from the graphs of fig . contrary to a common belief it does not follow exacly the same evolution as the condensate even to first order in the density . correlations do affect the first order in the mass but not in the condensate . the linear sigma model has serious shortcomings when it comes to a quantitative description of physical quantities . we mention some of the problems . _ i ) _ the physics we are discussing is intimately linked to the behaviour of the isospin symmetric @xmath0n s - wave amplitude . the sigma model grossly fails to reproduce the corresponding scattering length . the measured value is slightly repulsive @xmath49@xcite arising essentially from the pseudovector ( pv ) nucleon pole terms . the model gives instead a much larger repulsion , @xmath50 . _ ii ) _ in this model , the range of the condensate is determined by the @xmath20 mass , which the value of the sigma commutator fixes at 740 mev . the r.m.s . radius of the condensate follows , @xmath51 fm . as already mentioned the value deduced from experiments is much larger @xmath52 fm . with a small extension of the condensate , one is closer to the limit where the condensate remains inside the correlation hole . we have seen that in this limit the incoherent rescattering cancels the coherent one , which is not the case in a realistic situation . the linear sigma model thus overestimates the incoherent rescattering effect , and more generally overstresses the role of the correlations . _ iii ) _ the last point concerns the @xmath20 meson exchange effects . is their magnitude realistic ? we can confront them with the experimental value of the s - wave optical potential . since the @xmath20 exchange graphs of fig . [ figa]c - d are sea - gull terms , they are independent of the energy , thus keeping the same value for physical pions . this implies the existence of a many - body correction due to @xmath20 exchange in the s - wave optical potential . the latter is commonly parametrized as @xcite : @xmath53 since the quantity @xmath54 incorporates many - body effects of various sources , it is a function of the density . the value of @xmath54 deduced from the @xmath0-mesic data is @xmath55 @xcite . the typical density explored by these data is somewhat lower than the nuclear matter one @xcite , _ i.e. _ @xmath56 . at this density , the linear sigma model predicts an exchange contribution ( eq . ( [ sigex ] ) with the last term in @xmath57 removed ) @xmath58 . this would mean that most of the observed potential would have this origin . let us discuss this point in more detail . in the low density limit the parameter @xmath54 should reduce to the free nucleon value @xmath59 . at finite density , there is one correction of well known origin . it is the rescattering with charge exchange , linked to the pauli correlations @xcite . at @xmath60 this adds @xmath61 . we have also to consider the effect of correlations , @xmath62 linked to the isospin symmetric amplitude , i.e. the incoherent rescattering term previously discussed and contained in eq . ( [ sigres ] ) . this is not a sea - gull term . therefore it depends on the energy , but its dependence is mild and we neglect it . removing the term in @xmath63 and the coherent rescattering in eq . ( [ sigres ] ) , we find : @xmath64 a small contribution . finally we have to consider also the exchange correction due to pion excess . it represents the s - wave scattering on exchanged pions discussed by chanfray and ericson in the case of soft pions @xcite . this is also a sea - gull term which keeps the same value for physical ones . with an excess number of pions per nucleon of 0.1 the corresponding contribution is @xmath65 . adding all the known contributions @xmath66 we arrive already at @xmath67 , exactly the experimental value . unless there is a large attractive correction ( but we do not see from where ) there is no room for the large @xmath20 meson exchange correction implied by the linear sigma model , which is therefore unrealistic for @xmath0-nuclear scattering problem as well . thus we can safely conclude that the meson exchange terms in the scattering amplitude of soft , or physical pions , linked to the exchange of a scalar - isoscalar object , can not be as large as given by the linear sigma model . this was our motivation to investigate the non - linear sigma model . though non - linear sigma models have an already long story , their systematic application to nuclear forces and pion nuclear interactions has been recently popularized by weinberg @xcite . we follow the guideline provided by an effective lagrangian introduced in this spirit by lynn in his study of nuclear matter @xcite . we start from the linear sigma model with the lagrangian density @xmath68 where the chiral symmetric part writes in standard notations : @xmath69 where the limit @xmath70 is to be taken to get the non - linear version . in order to recover the non - linear model of lynn , we modify the chiral symmetry breaking term by adding a term bilinear in the nucleon field with the same chiral transformation property as the usual term linear in the sigma field : @xmath71 with , as before , @xmath16 = 45 mev the nucleon sigma commutator . we use the standard procedure to go to the non - linear version , writing @xmath72 and @xmath73 where @xmath74 is an arbitrary polynomial in @xmath75 , which contains only odd powers of @xmath76 . we also transform the fermion field according to @xmath77 where @xmath78 . a convenient choice for the function @xmath74 , introduced by weinberg is @xmath79 @xcite . the corresponding chiral symmetry breaking lagrangian is : @xmath80 in this representation the divergence of the axial current does not satisfy pcac since one has : @xmath81 the commutator of the axial charge operator with its time derivative , _ i.e. _ the @xmath20-term which is a measure of explicit symmetry breaking reads : @xmath82\big ] \nonumber \\ & = & \frac{1 } { 2}\;m_\pi^2 f_\pi^2\ ; \int d{\hbox{\boldmath $ r$\unboldmath}}\ ; \left(\frac{{\hbox{\boldmath $ \phi$\unboldmath}}^2(r)/f_\pi^2 } { ( 1+{\hbox{\boldmath $ \phi$\unboldmath}}^2(r)/4f_\pi^2 ) } - 2\right ) \left(1 - \frac{\sigma_n \bar n n(r ) } { f_\pi^2 m_\pi^2}\right)\ > . \label{sigop}\end{aligned}\ ] ] the variation of the condensate can be evaluated from the expectation value of this operator , @xmath83 , per unit volume of nuclear matter ( cf . eq . [ lin ] ) divided by the squared pion mass . as already mentioned in the introduction , this does not mean that the medium dependence has to disappear in the chiral limit since @xmath84 is finite . the situation is very similar to that encountered in the gell - mann - oakes - renner relation where the pion decay constant @xmath7 which is non zero in the symmetry limit is related to the vacuum expectation value of the symmetry breaking hamiltonian . to lowest order in the density the value of @xmath15 is that of the free nucleon , @xmath36 . we use perturbation theory to evaluate the second order terms in the density . we retain only in eq . ( [ sigop ] ) the lowest order terms in the pion field . the expectation value of the part @xmath85 of @xmath86 has been discussed by chanfray and ericson @xcite , restricting the interaction hamiltonian to the pseudovector @xmath0nn ( and also @xmath0n@xmath87 ) coupling . they showed that this contribution to the sigma commutator is nothing else than the sigma commutator of the pion excess . we extend here their calculation to incorporate the s - wave contact term @xmath88 of the interaction hamiltonian . there are interference terms between the s - wave and p - wave operators , as represented in the graph of fig . [ figd ] . we ignore them in the present work . our result for uniform nuclear matter , in the hartree approximation , is expressed as the following integral : @xmath89 we have introduced the dispersive form in the right hand side of the second equality in order to facilitate the comparison with the feynman - hellmann theorem . the physical interpretation of this term is the same as the one given in ref . it is the contribution of the excess pions linked to the s - wave contact term , as represented in the graph [ fige ] . another contribution arises from the term @xmath90 of @xmath91 . it leads to a logarithmically divergent integral : @xmath92 the sum of these two contributions could also be obtained by the feynman - hellmann theorem from the expectation value of the part of the two pion exchange potential linked to the s - wave contact term : @xmath93 we have to derive this expression with respect to the squared pion mass . the derivation of the integrand leads to the pion excess contribution discussed previously eq . ( [ ds1 ] ) . it adds a very small relative correction to @xmath36 : @xmath94 , which is not surprising since the s - wave term ( [ v2pi ] ) is only a weak component of the two pion exchange potential . a second piece arises from the derivation of the quantity @xmath95 , @xmath16 being proportional to @xmath96 ; it leads to the previous diverging integral eq . ( [ ds2 ] ) . in order to evaluate it we introduce a cut - off mass @xmath97 = 1 gev . the corresponding value of @xmath98 is again small ( a relative correction of @xmath99 to @xmath100 ) . summarizing the results obtained with the weinberg type lagrangian , the evolution of the quark condensate with the density is mostly given , up to the normal density , by the linear term in the density : @xmath101 the higher order corrections arise essentially from the modification of the pion number in the nucleus , which is expected to produce at normal density a mild acceleration of the restoration process ( at density @xmath1 , @xmath15 differs from @xmath16 by @xmath102 10% ) . there is no reaction against the restoration , contrary to what can be expected from pcac . in the next paragraph we go to a representation which possesses pcac , so as to elucidate the role of the distortion . the pcac representation corresponds to @xmath103 , with a subsequent transformation of the pion field : @xmath104 the chiral symmetry breaking lagrangian , expressed in the new variable @xmath105 reads : @xmath106^{1/2 } \nonumber \\ & \approx & f_\pi^2m_\pi^2 \left ( 1 - \frac { \sigma_n \bar n n } { f_\pi^2 m_\pi^2}\right ) -\frac{1 } { 2}\;m_\pi^2\;\frac{{\hbox{\boldmath $ \varphi$\unboldmath}}^2 } { ( 1 - \sigma_n\bar n n/ f_\pi^2 m_\pi^2 ) } + o(\varphi^4 ) \ ; . \label{lpcac}\end{aligned}\ ] ] note that in this representation the contact term @xmath107 , and hence also the soft pion amplitude on the nucleon , has the opposite sign as compared to the previous weinberg type representation . however the physical amplitude is the same as will be shown below . the pcac relation which is now valid , allows the calculation of the nuclear sigma commutator as a soft pion amplitude . if we ignore for the moment the one - pion and two - pion exchange contributions , the soft amplitude per unit volume is in the born approximation : @xmath108 as follows readily from the interaction part of the lagrangian eq . ( [ lpcac ] ) . it displays a renormalization by higher order terms in the density , similar , but not equal , to the @xmath20 meson exchange terms of birse and mc govern @xcite . here they appear from contact terms in the lagrangian since the @xmath20 meson mass is sent to infinity in the non - linear version . the distortion of the soft pion wave modifies the born amplitude in such a way that the free nucleon amplitude is recovered : @xmath109 the soft pion amplitude per nucleon and hence the sigma commutator @xmath15 is the same as the free one . here the distortion is exactly cancelled by the contact term , contrary to what happens in the linear sigma model where it is overcompensated . the expression of the quark condensate density is the same as in the weinberg representation , as it should @xcite . the introduction of pion exchange effects does not alter this result at least to second order in the density . indeed the one - pion contribution is obviously the same . as for the two - pion exchange eqs . ( [ ds1 ] ) and ( [ ds2 ] ) , the change of sign of the contact term does not influence the result as it appears quadratically in @xmath110 . it is illustrative to check the independence on the lagrangian also on the pion effective mass in the medium , @xmath111 , which obeys the equation @xmath112 where @xmath113 is the pion self energy , taken at the energy @xmath114 and at zero three momentum . to find this last quantity we need the off - shell @xmath0n amplitude . for simplicity , we neglect the small nucleon pole terms which would not change the validity of the argument . in the representation of weinberg the amplitude keeps the same value , @xmath115 , off- and on - shell . in this case the effective mass has a very simple linear density dependence : @xmath116 in the pcac representation instead , the @xmath0n amplitude has a complex off shell behaviour . indeed it gets also a contribution from the pion kinetic energy term , @xmath117 which expressed in the new field @xmath118 variable writes : @xmath119 where the operators @xmath120 are covariant derivatives of the pion field . the presence of the nucleon fields in this expression leads to new contributions to the @xmath0n amplitude which takes the form : @xmath121 where @xmath122 and @xmath123 are the initial and final pion four momenta ( @xmath124 ) and the bar over @xmath125 indicates that the pv nucleon poles have not been considered . this expression satisfies adler consistency conditions , together with the well known sign change between the soft and cheng - dashen points . notice that the physical amplitude is @xmath126 as in the weinberg representation . the nuclear born amplitude @xmath127 , which is also the pion self - energy @xmath128 , writes : @xmath129 where @xmath130 . in this expression , as in eq . ( [ tbsoft ] ) , the first piece in @xmath17 derives from the interaction part in the pion mass term of eq . ( [ lpcac ] ) . the second piece in @xmath131 arises in the same way from the interaction part of the kinetic energy term ( eq . ( [ pikin ] ) , first term ) . the effective mass then obeys the equation : @xmath132 which leads to the same result as the weinberg lagrangian eq . ( [ mstlin ] ) , with the linear density dependence of @xmath133 . the mass is indeed independent on the representation as already known from previous work @xcite . notice however that the simple linear drop of eq . ( [ mstlin ] ) which could lead to s - wave pion condensation @xcite does not occur in reality , because the non - linear lagrangian as introduced previously is incomplete . it fails to reproduce the ( nearly ) vanishing observed of the non - born @xmath0n amplitude at threshold ( see ref . @xcite for a general low energy amplitude with this property ) . it is then necessary to introduce extra terms , which are chirally invariant , so as to produce the threshold cancellation . following refs . @xcite as already done in ref . @xcite , we add to the previous lagrangian a piece @xmath134 : @xmath135 where @xmath136 is the nucleon four - velocity which reduces to ( 1,0,0,0 ) in the nucleon rest frame . this extra piece does not change the condensate but it modifies the effective mass : @xmath137 the threshold cancellation of the non - born amplitude implies @xmath138 such that @xmath139 . this relation was first derived on a multiple scattering basis @xcite and then in a chiral lagrangian approach @xcite ( see also @xcite ) . we summarize the discussion of this section in table [ tab2 ] where we compare the results obtained with the weinberg and pcac lagrangians for the amplitudes and the observables . we also give there the expression of the inverse pion propagator @xmath140 in the two cases . they are identical up to a factor @xmath141 , with in particular the same dispersion law . we recover here the expression of the propagators already derived along somewhat different lines by thorsson and wirzba @xcite in their comparative study between two lagrangians of weinberg and pcac types . we could have discussed also the behaviour of the nucleon effective mass @xmath48 . retaining only the s - wave pion - nucleon interaction we would get to first order in the density : @xmath142 terms in @xmath143 . it is seen that this is a very small correction which bears no resemblance with the evolution of the condensate ( [ linn ] ) . in fact , in the effective lagrangian approach @xcite,@xcite the nucleon effective mass is mainly governed by a contact term @xmath144 which is a chiral invariant and thus does not contribute to the @xmath20 term ( [ sigop ] ) . we get then : @xmath145 with no relation to the condensate , contrary to what occurred , at least in the absence of correlations , in the linear sigma model ( see table [ tab1 ] ) . the latter situation is however quite peculiar in that the nucleon mass is generated by spontaneous symmetry breaking . it may happen of course that , at the deeper level of qcd , the phenomenological coefficients @xmath146 of the non - linear lagrangians are connected so that the evolution of the nucleon effective mass and that of the quark condensate become similar . we will now discuss the influence of the correlations on the condensate . in the representation of weinberg it is clear that the quark condensate can not be affected by the incoherent rescattering of the soft pions discussed in ref . @xcite since the nuclear sigma commutator bears no relation to the soft pion amplitude . on the other hand it is interesting to understand in the representation in which pcac holds how the effect of incoherent rescattering can be cancelled in the soft pion amplitude which is the sigma commutator . this investigation is also interesting in connection with the scattering of physical pions . indeed the isospin symmetric optical potential is strongly influenced by the correlations , as the corresponding @xmath0n scattering length is very small . the incoherent rescattering effect linked to the charge - exchange amplitude has been discussed a long time ago @xcite . the one linked to the isospin - symmetric amplitude is more debatable as the off - shell behaviour of this amplitude is model dependent . in the pcac representation it is given by expression ( [ toff ] ) . we will first discuss the role of the correlations on the soft pion amplitude and then we extend the result to the physical threshold pions . we limit our consideration to two - body correlations and hence we work to second order in the density . to this order and in the soft limit the relevant pieces of the lagrangian in the pcac representation are : @xmath147 in a correlated medium the amplitude can be evaluated from the graphs of fig . [ figf ] and [ figg ] . those of fig . [ figf ] are the contact terms while those of fig . [ figg ] represent the coherent and incoherent rescattering . the latter corresponds to the lorentz - lorenz effect well known in the p - wave case . in the absence of correlations and in the static approximation only the first two pieces of the lagrangian eq . ( [ lsoft ] ) contribute an amount ( graphs [ figf]a and [ figf]b ) : @xmath148 which is nothing else than the expansion of @xmath149 of formula ( [ tbsoft ] ) . with correlations the graph [ figf]c gives : @xmath150 where the two parts in the integrand arise from the second and third pieces of the lagrangian respectively . the rescattering terms [ figg]a and [ figg]b add : @xmath151 summing all the pieces the total soft pion amplitude thus reduces to : @xmath152 the overall effect of the correlations disappears due to cancellation between the lorentz - lorenz term ( _ i.e. _ the incoherent rescattering of graph [ figg]b ) and the correlation contribution in the contact terms . the soft amplitude and hence the sigma commutator are affected neither by distortion nor by correlations . thus we recover the result of the weinberg type lagrangian . the validity of our statement about correlations is however limited to the simplest case where the chiral lagrangians ( e.g. our eqs.([lsig]-[lwsb ] ) ) do not contain terms of second order or more in nucleon fields bilinears @xmath153 . as we will be comment later on , correlations effects can not be excuded on the sole basis of chiral symmetry for lagrangians extended to higher order . it is natural to extend our study to the case of threshold physical pions , as we did in the linear sigma model . our aim is first to understand if , in the non - linear model , chiral symmetry imposes the existence of two - body terms in the s - wave optical potential . this was the case in the linear sigma model . the second point is to make a quantitative evaluation of the influence of the correlations linked to the isospin symmetric @xmath0n amplitude . concerning the first point , we recall that in the pcac representation and for soft pions contact terms are needed to cancel the distortion . the question is then : are these contact terms present in the s - wave optical potential as it is the case in the linear sigma model ? in the weinberg representation contact terms are totally absent . their cancellation is therefore plausible in the pcac representation . the problem is to understand the mechanism for cancellation . for simplicity we ignore the extra terms in @xmath154 and @xmath155 which do not change the essence of the discussion . in the pcac representation the pion self energy has been given in eq . ( [ tborn ] ) . it displays a priori many - body terms . threshold pions correspond to the situation where @xmath156 which are valid outside the nucleus . for the inside conditions we take the example of a spherical nucleus with a uniform density . in the interior of the nucleus the pion keeps its energy and acquires a momentum @xmath157 solution of the dispersion equation : @xmath158 _ { \omega = m_\pi } = 0 \ ; , \label{disp}\ ] ] which gives @xmath159 such that @xmath160 . in the usual expression of the potential @xmath161 the momentum dependence is not explicitly taken into account but it has to be implicitly incorporated in the parameter @xmath162 as follows : @xmath163 = - { \rho\sigma_n } { f_\pi^2}\ ; . \label{b0eff}\ ] ] this expression displays no two - body terms and is the same as for the weinberg lagrangian . thus in the pcac representation of the non - linear sigma model the momentum dependence of the self energy is such that many - body terms are absent in the s - wave optical potential . in the linear model instead , the @xmath0n amplitude @xmath164 which also satisfies the adler condition , depends only on the four momentum transfer variable @xmath165 so that in the forward direction there is no three momentum dependence . hence the self energy is momentum independent and there is no cancellation of the contact terms . in that respect the non - linear and linear sigma models are quite different . we now turn to the influence of correlations on the s - wave pion - nucleus optical potential . for this purpose it is simpler to use the weinberg lagrangian supplemented with the @xmath134 term introduced previously in eq . ( [ deltal ] ) . indeed in this case there are no two - body contact terms and correlations enter only as a lorentz - lorenz effect . for pions at rest with energy the amplitude in the born approximation ( _ i.e. _ the pion self - energy ) takes the following form in a correlated medium : @xmath166 \ > . \label{tphys}\ ] ] here @xmath167 is the full off - shell @xmath0n amplitude including the nucleon poles in pseudovector coupling ( first term on the right hand side of the following equation ) : @xmath168 the s - wave pion optical potential is obtained from the self - energy eq . ( [ tphys ] ) taken at @xmath169 : @xmath170 . the pole terms can be ignored in the part with the correlation integral which gives a very small contribution . as for the remaining terms the effect of the correlations is totally negligible due to the cancellation which occurs at threshold @xmath171 . the same conclusion would be obtained in a more complicated way with the pcac representation . thus we can conclude that only the incoherent rescattering with charge exchange , which has not been introduced in the present discussion , is relevant for the s - wave optical potential , at variance with the findings of salcedo _ et al . _ @xcite . one conclusion of our study with the non - linear sigma model is the absence of two - body terms in the nuclear sigma commutator @xmath15 ( apart from those due to pion exchange ) and in the s - wave optical potential . the meaning of this statement is that chiral symmetry does not impose them . they can nevertheless be present . indeed we have used the simplest forms of the chiral lagrangians . in the nuclear medium , at the same order in the chiral expansion one can add terms quadratic in the nucleon density or higher , either breaking chiral symmetry such as @xmath172 or conserving it as e.g. @xmath173 . in the spirit of chiral perturbation expansions all the coefficients are phenomenological and should be determined by experiment though their order of magnitude can be guessed by dimensional counting arguments @xcite . a priori they can be of any sign , accelerating or hindering chiral symmetry restoration . we are planning to explore which experiments can be informative on these terms , beyond the obvious one of pion - nucleus scattering . in the same way nucleon - nucleon interaction terms built with squared bilinears @xmath174 ( @xmath175 matrices ) should be present as introduced e.g. in refs @xcite to simulate heavy exchanges . with such lagrangians , the effects of correlations can be buried in the coefficients of some contact terms or / and have to be explicitly computed as loop contributions . to conclude the present study , we have investigated the evolution with density of the quark condensate beyond the approximation of non interacting nucleons . in the non - linear sigma model we have found that the calculable correction to the linear dependence of the quark condensate arises from the one - pion and two - pion exchange potentials . the first one has been already discussed by chanfray and ericson @xcite . as for the latter we have evaluated it in the framework of the non linear model and found it to be small . however we stress that this is only a part of the two - pion exchange potential , the part linked to the contact s - wave @xmath0-nucleon amplitude . the bulk of the potential arises from the p - wave coupling terms which is outside the present approach . the evaluation of its role is in progress . symmetry does not impose other two - body contributions contrary to what occurs in the linear sigma model . they can nevertheless be present but have to be determined empirically . in particular there is no systematic reaction against the restoration of chiral symmetry . since the quark condensate is the order parameter of chiral symmetry , it is intimately linked to the physics of the pion which is the goldstone boson of this symmetry in the broken phase . it is thus natural to apply the concepts developed for the condensate in the nuclear medium to the pion - nucleus interaction . the absence of two - body terms imposed by chiral symmetry in the condensate reflects in their absence in the @xmath0-nucleus optical potential . the possible ones have to be determined empirically . we have also found that correlations act only through the well known incoherent rescattering with charge exchange . thus the non - linear sigma model has been quite helpful to make clear several intriguing points . we have displayed on an example how the independence of the condensate and other physical quantities on the particular form of the lagrangian is achieved . in the version where pcac holds , this property occurs through a subtle cancellation which eliminates both coherent ( distortion ) and incoherent rescattering ( effect of correlations ) in the soft pion amplitude . we have made a quantitative estimate of the ( small ) contribution of the contact two - pion exchange potential @xmath176 to the condensate , which can be extended to the full two - pion exchange . though pcac may be misleading in the sense that it introduces artificial effects eventually cancelling each other , the derivation of the condensate properties from a soft pion amplitude has the advantage of stimulating the application of the same methods to the domain of physical pions . following this line we have been lead to question the use of the linear sigma model for the estimate of heavy meson exchanges . such cross - fertilization between physics related to qcd and pion physics will likely bring new results in both fields . we are grateful to drs . t. ericson , m. lutz and m. rho for stimulating discussions . e.g. drukarev and e.m . levin , nucl . a511 ( 1990 ) 679 ; progr.part.nucl . 27 ( 1991 ) 77 . cohen , r.j . furnstahl and d.k . griegel , phys . c45 ( 1992 ) 1881 . m. lutz , s. klimt and w. weise , nucl . a542 ( 1992 ) 521 . g. chanfray and m. ericson , nucl . a556 ( 1993 ) 427 . m. ericson , phys . lett . b301 ( 1993 ) 11 . g. q. li and c. m. ko , phys . b338 ( 1994 ) 118 . r. brockmann and w. weise , phys . b367 ( 1996 ) 40 . birse and j.a . mc govern , phys . b309 ( 1993 ) 231 . t. ericson , phys . b321 ( 1994 ) 312 . v. thorsson and a. wirzba , nucl . a589 ( 1995 ) 633 . s. coleman , j. wess and b. zumino , phys . rev . 177 ( 1969 ) 2239 . w. weise , nucl . a278 ( 1977 ) 402 . j. gasser , h. leutwyler and m.e . sainio , phys . b253 ( 1991 ) 260 . j. gasser , h. leutwyler and m.e . sainio , phys . b253 ( 1991 ) 252 . t. ericson and w. weise , pions in nuclei ( clarendon press , oxford 1988 ) . m. ericson and t. ericson , annals of physics ( n.y . ) 36 ( 1966 ) 323 . r. seki and k. masutani , phys . c27 ( 1983 ) 2799 . s. weinberg , phys . b251 ( 1990 ) 288 ; nucl . b363 ( 1991 ) 3 . lynn , nucl . b402 ( 1993 ) 281 . s. weinberg , phys . 33 ( 1968 ) 1568 . brown , v. koch and m. rho , nucl . a535 ( 1991 ) 701 . j. delorme , m. ericson and t. ericson , phys . b291 ( 1992 ) 379 . j. gasser , m.e . sainio and a. varc , nucl . b307 ( 1988 ) 779 . v. bernard , n. kaiser and u .- g . meiner , phys . b309 ( 1993 ) 421 . c. adami and g.e . brown , phys . repts . 234 ( 1993 ) 1 . brown and m. rho , phys . ( in press ) . salcedo , k. holinde , e. oset and c. schtz , phys . b353 ( 1995 ) 1 . a. manohar and h. georgi , nucl . b234 ( 1984 ) 189 . & @xmath182 & @xmath183 + & & + @xmath184 & @xmath185 & @xmath186 + & & @xmath187 + & & + & & + @xmath188 & @xmath189 & @xmath190 + & & + & & + @xmath191 & @xmath192 & @xmath193 + & @xmath194 & @xmath195 + & & + & & + @xmath196 & @xmath197\omega^2 $ ] & @xmath198\omega^2 - ( 1 + 2x_3){\hbox{\boldmath $ q$\unboldmath}}^2 $ ] + & @xmath199 & @xmath200 /(1-x)^2 $ ] + & & + & & + @xmath201 & @xmath202 & @xmath203 + & & + & & + @xmath204 & @xmath205 & @xmath206 + & & +	we study the evolution with density of the quark condensate in the nuclear medium with interacting nucleons and including the short range correlations . we work with two chiral models , the linear sigma model and the non - linear one . for the last one we use two versions , one which does not satisfy pcac , and another one which does . we show that the quark condensate , as other observables , is independent on the variant selected . the application to physical pions excludes the linear sigma model as a credible one . in the non - linear models restricted to pure s - wave pion - nucleon scattering , our conclusions are the following : \i ) chiral symmetry does not imply a systematic reaction against symmetry restoration . in representations where pcac holds , this absence results from a subtle cancellation . \ii ) in the density evolution of the quark condensate , the two - pion exchange potential , restricted to s - wave @xmath0n interaction , has a negligible effect , much smaller than that of the one - pion exchange . at variance with the linear sigma model , two - body contact terms simulating sigma meson exchange are not constrained by chiral symmetry . the same conclusion applies in the scattering of physical pions on nuclei , _ i.e. _ to the s - wave optical potential . \iii ) on the other hand for the effects of correlations calculable from chiral symmetry , those linked to the isospin symmetric amplitude are negligible . = 0.1 cm = 40 mm pacs numbers : 12.39.fe 24.85.+p 25.80.dj lycen 9604 + cern - th/96 - 60 + nucl - th/9603005
some high - t@xmath1 superconductors , such as bi@xmath4sr@xmath4ca@xmath4cu@xmath4o@xmath5 , absorb very strongly in the microwave regime even at temperatures @xmath6 far below @xmath7@xcite . at a fixed frequency as a function of @xmath6 , the real conductivity typically has two features : a sharp peak near @xmath7 , and a broad , frequency - dependent background extending to quite low @xmath6 . the peak near t@xmath1 is usually ascribed to critical fluctuations arising from the superconducting transition . the origin of the broad background is less clear . in conventional s - wave superconductors , there is no such background , because the material can not absorb radiation at frequencies below the energy gap for pair excitations , @xmath8 . but in high - t@xmath1 materials , which are thought to have a @xmath9 order parameter@xcite , the gap vanishes in certain nodal directions in @xmath10-space . hence , gapless nodal quasiparticles can be excited at arbitrarily low @xmath6 , and these can absorb at low frequencies . but the observed absorption appears to be stronger than expected from the quasiparticles alone in a two - fluid model@xcite . several authors have suggested quenched inhomogeneities as the origin of this extra absorption . such inhomogeneities could arise from statistical fluctuations in the local densities of holes or impurities , or from the presence of superfluid - suppressing impurities such as zn . et al_@xcite found that such fluctuations in bi@xmath4sr@xmath4ca@xmath4cu@xmath4o@xmath5 displace about 30% of the spectral weight from the condensate to the conductivity at finite frequencies . in addition , several explicit models have been presented which produce such a displacement . for example , van der marel and tsvetkov@xcite have calculated the fluctuation conductivity due to a periodic one - dimensional modulation of the phase stiffness . the present authors@xcite have calculated the fluctuation conductivity of a two - dimensional array of josephson junctions with quenched randomness in the critical currents , using the kubo formalism@xcite . they found an excess contribution to the conductivity below @xmath7 arising from this inhomogeneity , which arose from displacement of some of the superfluid contribution from zero to finite frequencies . more recently , orenstein@xcite considered a two - fluid model of superconductivity with quenched , anticorrelated inhomogeneity in both the normal and superfluid densities ; he found that some of the spectral weight of the conductivity is displaced from zero to finite frequencies , and gave explicit expressions for this extra conductivity . han@xcite has given a field - theoretic treatment of a randomly inhomogeneous two - dimensional d - wave superconductor , using a replica formalism to treat the assumed weak disorder . he obtained an extra lorentzian peak in the real conductivity , centered at zero frequency , arising from this disorder , and a corresponding reduction in the superfluid density . all such models are made more plausible by imaging experiments which have directly observed spatial fluctuations in the local superconducting energy gap in underdoped cuprate superconductors@xcite . in this paper , we study the low - frequency complex conductivity of several models for two - dimensional superconductors with quenched inhomogeneities . we use a straightforward approach to obtain the effective complex conductivity , based on standard treatments of randomly inhomogeneous media in the quasistatic regime@xcite . from this approach , we obtain sum rules for the complex conductivity . we also find , in a number of cases , explicit expressions for the extra conductivity @xmath11 due to the quenched inhomogeneities . where the models are comparable , our explicit form for @xmath11 reduces to that of han , even though obtained using a quite different approach . the remainder of this paper is organized as follows . in section ii , we develop sum rules for the optical conductivity of a two - dimensional superconductor with quenched inhomogeneities , and apply them to several explicit models . in section iii , we present some approximate calculations for the effective conductivity of such superconductors@xcite . a concluding discussion is given in section iv . we consider a two - dimensional superconductor , in which the local conductivity has normal and superfluid contributions which act in parallel , but , in contrast to the usual two - fluid model , these terms are spatially varying . the complex local conductivity @xmath12 is taken as a scalar function of position @xmath13 and frequency @xmath14 of the following form : @xmath15 @xmath16 might be a contribution from the nodal quasiparticles expected in a superconductor with @xmath9 order parameter symmetry , while @xmath0 represents the perfect - conductivity response of the superconductor . our goal is to calculate the position - independent effective complex conductivity of this medium , which we denote @xmath17 we first obtain a sum rule for @xmath18 . we start from the kramers - kronig relations obeyed by any function , such as @xmath19 , which describes a causal response : @xmath20 and @xmath21 here p denotes the principal part , and @xmath22 lim@xmath23 . we have assumed that lim@xmath24 position - independent , and have used the fact that , at high frequencies , @xmath25 , where @xmath26 is the effective superfluid density . at sufficiently large @xmath14 , the right hand side of eq . ( [ eq : kk2 ] ) takes the form @xmath27d\omega^\prime\right ) . \label{eq : kk3}\ ] ] we first write down a perturbation result for @xmath19 , which will be needed below . if the spatial fluctuations in @xmath28 are small in magnitude compared to it spatial average , @xmath29 , then to second order in the @xmath30 , @xmath31 where @xmath32 denotes a space average of the quantity in brackets@xcite . suppose first that @xmath33 is real , nonzero , frequency - independent , and spatially uniform . then @xmath34 , and , at high @xmath14 , however large the fluctuations in @xmath35 , the fluctuations in the _ local _ complex conductivity , @xmath36 , are small compared to the space - averaged conductivity @xmath37 ; we can thus use eq . ( [ eq : weak ] ) . since only @xmath0 is fluctuating , and not @xmath16 , eq . ( [ eq : weak ] ) becomes @xmath38 where @xmath39 . thus , to leading order in @xmath40 , @xmath41 . equating this expression to the right - hand side of eq . ( [ eq : kk3 ] ) , we obtain @xmath42d\omega^\prime = \frac{\pi}{2}\left(\rho_{s , av } - \rho_{s , e}\right ) \label{eq : sige1}\ ] ] one consequence of eq . ( [ eq : sige1 ] ) is that , if @xmath35 is spatially varying , @xmath35 , @xmath43 , the right - hand side of eq . ( [ eq : sige1 ] ) is positive , and _ there will be an additional contribution to @xmath44 , above @xmath16_. the sum rule ( [ eq : sige1 ] ) requires only that the conductivity fluctuations are asymptotically small at large @xmath14 , and not necessarily that @xmath0 have small fluctuations . if , however , the fluctuations in @xmath0 are , in fact , small , then @xmath26 can be calculated approximately using the analog of eq . ( [ eq : weak ] ) , namely , @xmath45 from this approximate result , eq . ( [ eq : sige1 ] ) becomes @xmath46 d\omega^\prime \sim \frac{\pi}{4}\rho_{s , av}\left(\frac{\langle(\delta \rho_s)^2\rangle } { \rho_{s , av}^2}\right ) . \label{eq : sige1p}\ ] ] for fixed mean - square fluctuations @xmath47 , this integral is proportional to the average superfluid density @xmath48 . that is , the extra integrated fluctuation contribution to @xmath18 , above the mean contribution of the normal fluid , is proportional to @xmath48 . a similar result has been reported in experiments carried out over a range of mean superfluid densities@xcite . next , we consider a frequency - dependent @xmath49 , assuming that lim@xmath50 , but allowing for a spatially varying @xmath49 . then eq . ( [ eq : kk2 ] ) becomes @xmath51 where the last result is again valid at sufficiently large @xmath14 . we now assume a drude form for @xmath49 , i. e. , @xmath52 where @xmath53 is a suitable normal fluid density , and @xmath54 is a quasiparticle relaxation time . then at sufficiently high @xmath14 , @xmath55}{\omega},\ ] ] while @xmath56 falls off at least as fast as @xmath57 . similarly , @xmath18 also falls off at least as fast as @xmath57 at large @xmath14 . hence , @xmath58 at high frequencies , where @xmath59 here @xmath60 denotes the effective superfluid density of a hypothetical material whose spatially varying local superfluid density is @xmath61 . equating the coefficients of @xmath62 on right hand sides of eqs . ( [ eq : sige2pp ] ) and ( [ eq:14 ] ) yields @xmath63,\ ] ] which gives the integrated spectral weight of @xmath44 . to calculate this weight , we first introduce @xmath64 , and note that @xmath65 even if @xmath54 is position - dependent . hence , @xmath66 d\omega^\prime = \frac{\pi}{2}\left[(\rho_s+ \rho_n)_e -\rho_{s , e}-\rho_{n , av}\right ] . \label{eq : sigesn}\ ] ] thus @xmath67 again has some extra spectral weight beyond that of @xmath68 . we first estimate expression ( [ eq : sigesn ] ) in the small - fluctuation regime where @xmath69 , assuming that @xmath70 is non - fluctuating . following steps analogous to those leading up to eq . ( [ eq : sige1p ] ) , we obtain @xmath71d\omega^\prime \rightarrow \frac{\pi}{4}\langle(\delta \rho_s)^2\rangle \left(\frac{1}{\rho_{s , av}}-\frac{1}{\rho_{s , av } + \rho_n}\right).\ ] ] in the limit @xmath72 , this expression reduces to the right - hand side of eq . ( [ eq : sige1p ] ) . in this regime , the extra spectral weight is again proportional to @xmath48 . a frequency - independent @xmath16 is a special case of this limiting behavior . in the opposite limit ( @xmath73 ) , the behavior is @xmath74 d\omega^\prime \rightarrow \frac{\pi}{4}\rho_n \frac{\langle(\delta \rho_s)^2\rangle}{\rho_{s , av}^2}. \label{eq : sumrule2}\ ] ] thus , in this limit , for fixed mean - square fluctuations in @xmath0 , the extra spectral weight is _ not _ proportional to @xmath48@xcite . if both @xmath0 and @xmath70 are spatially varying , eq.([eq : sigesn ] ) can still be easily treated when spatial fluctuations in @xmath0 and @xmath70 are small . then , to second order in these fluctuations , @xmath75 ^ 2\rangle } { \rho_{s , av } + \rho_{n , av } } , \label{eq : rhoe1}\ ] ] while @xmath26 is given approximately by eq . ( [ eq : rhose ] ) . substituting eqs . ( [ eq : rhoe1 ] ) and ( [ eq : rhose ] ) into eq . ( [ eq : sigesn ] ) gives @xmath76d\omega^\prime = \frac{\pi}{4}\left[\frac{\langle(\delta\rho_s)^2\rangle}{\rho_{s , av } } - \frac{\langle[\delta(\rho_s+\rho_n)]^2\rangle}{\rho_{s , av } + \rho_{n , av}}\right ] . \label{eq : flucn}\ ] ] to interpret eq . ( [ eq : flucn ] ) , we assume that @xmath0 and @xmath70 are correlated according to @xmath77 . thus @xmath78 and @xmath79 represent perfect correlation and perfect anticorrelation . the predictions of eq . ( [ eq : flucn ] ) are particularly striking in the latter case , as postulated in some models of inhomogeneity@xcite . in this case the _ sum _ @xmath80 is _ not _ fluctuating . then the second term on the right - hand side of eq . ( [ eq : flucn ] ) vanishes , and the extra spectral weight due to the inhomogeneity is again proportional to @xmath81 . in the opposite case @xmath78 , the right - hand side of eq . ( [ eq : flucn ] ) may be negative , i. e. the total spectral weight is _ smaller _ than that given by @xmath82 . the preceding arguments apply equally well to a _ tensor _ conductivity , as expected if the superconducting layer contains charge stripes or other types of static charge density waves . in this case , the local conductivity should be a @xmath83 tensor@xcite of the form @xmath84 , where @xmath85 is a diagonal @xmath83 matrix with diagonal components @xmath86 , @xmath87 , and @xmath88 is a @xmath89 rotation matrix describing the local orientation of the stripes . if this phase has a domain structure , the stripes would have either of two orientations relative to the layer crystalline axes , with equal probability . in a macroscopically isotropic layer , this domain structure leads to an scalar effective conductivity @xmath19 . if , for example , the conductivity along the j@xmath90 principal axis is @xmath91 , then the superfluid inhomogeneity produces an extra spectral weight at finite @xmath14 given by eq . ( [ eq : sigesn ] ) . we now supplement these sum rules with some explicit expressions for @xmath19 , beginning with weak inhomogeneity , @xmath92 . in this case , @xmath93 is given by eq . ( [ eq : weak ] ) . if only @xmath0 is spatially fluctuating , this equation reduces to eq.([eq : hifreq ] ) and @xmath94 thus , the extra absorption is a lorentzian centered at @xmath95 , of half - width @xmath96 , and strength @xmath97 . a result equivalent to eq . ( [ eq : weak1 ] ) has also been obtained by han@xcite using a replica formalism . indeed , even the parameters of the two results are identical , provided we interpret @xmath98 in our model as equal to the quantity @xmath99 in the model of ref . @xcite . next , we consider the case where both @xmath0 and @xmath16 have weak spatial variations . we assume that @xmath16 is given by eq . ( 12 ) , that @xmath70 is spatially varying , but not @xmath100 , and that @xmath70 and @xmath0 are correlated according to the rule @xmath101 . then @xmath93 is given approximately by eq . ( 6 ) , with @xmath102 and @xmath103 some representative plots of @xmath44 resulting from eqs . ( [ eq : weak ] ) , ( [ eq : fluc2a ] ) and ( [ eq : fluc2b ] ) are shown in fig . 1 for @xmath78 ( perfect correlation ) @xmath79 ( perfect anticorrelation ) , and @xmath104 ( no fluctuations in @xmath105 ) . the case @xmath79 appears most plausible physically , since it implies that the sum @xmath106 is non - fluctuating . it also produces the largest increase in @xmath18 at any given @xmath14 . for @xmath107 , eqs.([eq : weak ] ) , ( [ eq : fluc2a ] ) , and ( [ eq : fluc2b ] ) lead to the analytical result @xmath108 here @xmath109 and we have neglected corrections of order @xmath110 in the numerator . ( [ eq:25 ] ) is again identical with han s result@xcite , obtained using a replica formalism , if we make the identification @xmath111 . next , we calculate @xmath19 using a simple model for the conductivity of a cuo@xmath4 layer proposed in ref.@xcite , but generalized to include inhomogeneities . we assume that @xmath28 is given by eq . ( [ eq : model ] ) , but for simplicity that it can have only two values , @xmath112 and @xmath113 , with probabilities @xmath114 and @xmath115 . the corresponding conductivities are @xmath116 ( j = a , b ) , where @xmath117 following ref . @xcite , we take @xmath118 . we assume @xmath119 , for @xmath120 , as expected for nodal quasiparticles in a @xmath9 superconductor , and @xmath121 . we also take @xmath122 . this linear @xmath6-dependence may be expected if @xmath123 is determined primarily by quasiparticle - quasiparticle scattering in two dimensions@xcite , rather than impurity scattering , which might give a @xmath6-independent @xmath123 . for the superfluid density , we write @xmath124 , @xmath125 , where @xmath126 . this form ensures that @xmath127 decreases linearly with @xmath6 at low @xmath6 , as observed experimentally@xcite , and that it vanishes at a critical temperature @xmath7 as @xmath128 . we use the experimental values of the parameters @xmath7 , @xmath129 , @xmath130 , @xmath131 and @xmath132 ( as quoted in ref.@xcite ) for our numerical estimates . finally , we have arbitrarily assumed that the inhomogeneity parameter @xmath133 . we compute @xmath19 using the bruggeman effective - medium approximation ( ema)@xcite : @xmath134 the physically relevant solution is obtained by requiring that @xmath19 be continuous in @xmath14 and that @xmath135 . we also arbitrarily assume that @xmath136 . our form for @xmath137 and @xmath138 guarantees that the ema solution @xmath139 . 2 shows the resulting @xmath140 for several frequencies ranging from 0.2 to 0.8 thz , the range measured in ref.@xcite . also shown are @xmath141 for @xmath142thz and @xmath143thz . finally , we plot @xmath144 for these frequencies , as well as @xmath145 . clearly , @xmath44 is substantially increased beyond the quasiparticle contribution , because of spatial fluctuations in @xmath0 . we next consider @xmath146 near a percolative superconductor to normal ( s - n ) transition . such a transition might occur , for example , in a single cuo@xmath4 layer doped by a non - superconducting element such as zn , as has already been discussed by several authors@xcite . we assume @xmath147 to be given by eq . ( [ eq : model ] ) , with a position - independent drude @xmath16 given by eq . ( [ eq : drude ] ) , and a @xmath35 which has one of two values , @xmath0 or zero , with probabilities @xmath114 and @xmath148 . we calculate @xmath19 using the ema , eq . ( [ eq : ema1 ] ) . the ema predicts an s - n transition at @xmath149 . for all @xmath114 on either side of @xmath150 , @xmath151 has a peak at @xmath95 , whose height diverges and whose half - width decreases to zero , as @xmath152 . this behavior is illustrated in fig . 3 , where we plot @xmath151 for several @xmath153 , as calculated in the ema . the other parameters are indicated in the figure caption . the behavior of @xmath18 is qualitatively similar to that shown in fig . 2 , but there is additional critical behavior near the s - n transition , which is absent from the weakly disordered system . more exact results near @xmath150 can be obtained with the help of a standard scaling hypothesis@xcite . for the present model , it takes the form @xmath154 here @xmath155 and @xmath156 denote the complex conductivities in the s and n regions of the layer ; @xmath157 , @xmath158 and @xmath159 are standard percolation critical exponents , which depend on dimensionality , and possibly on other structural details of the layer ; and @xmath160 are characteristic scaling functions which apply above and below @xmath150 . the expected behavior@xcite is @xmath161 for @xmath162 , and @xmath163 for @xmath162 , while @xmath164 for @xmath165 . from these forms of the scaling functions , we can infer@xcite that @xmath18 has a peak at @xmath95 whose half - width on both sides of the percolation threshold is @xmath166 which vanishes at the percolation threshold , where @xmath167 . the height of this zero - frequency peak diverges according to the law @xmath168 on either side of the percolation threshold . for two - dimensional bond percolation , @xmath169 , while in the ema in any dimension , @xmath170 . thus , we expect that @xmath151 is characterized near @xmath150 by a line of diverging height , and half - width which goes to zero near @xmath150 . precisely this type of behavior is seen in the ema curves of fig . 3 . the results shown in fig . 3 correspond to an in - plane scalar superfluid density @xmath35 with a random spatial variation . if instead @xmath35 were a @xmath83 _ tensor _ with principal axes varying randomly with position ( as might occur for random quenched domains of stripes ) , a similarly enhanced @xmath18 would be expected . if the two principal components of the conductivity tensor are given by eq . ( [ eq : condj ] ) , and the stripes can point with equal probability along either crystallographic direction , then @xmath19 is given in the ema by eq . ( [ eq : ema1 ] ) with @xmath171 . the resulting @xmath44 would behave exactly like that shown in fig . 2 . it is useful to compare our results to those of previous workers , considering first the work of han@xcite . for weak fluctuations in @xmath0 , and a frequency - independent non - fluctuating @xmath16 , our results are identical to his , provided we identify our parameter @xmath98 with his quantity @xmath99 . however , our formal derivation is more elementary than that of ref . @xcite . because of this simplicity , we can easily consider a wide range of quenched inhomogeneities . since the two approaches yield identical results for comparable types of inhomogeneities , we infer that our calculations ( which are based on standard methods of treating random inhomogeneities in classical field equations such as those of electrostatics ) can also be done using field - theoretic techniques . this field - theoretic approach may therefore be useful in treating other classical problems in heterogeneous media . we briefly discuss relation between the present model and that of ref . @xcite . in that earlier paper , the cuo@xmath4 layer was treated as an array of small `` grains '' coupled together by overdamped resistively shunted josephson junctions with langevin thermal noise ; @xmath172 was then computed using the classical kubo formalism . it consists of two parts : a sharp peak near the kosterlitz - thouless transition at @xmath7 , due to breaking of vortex - antivortex pairs , and a broader contribution , for @xmath120 , which exists only if the critical currents have quenched randomness . this second contribution corresponds to that considered here for spatially varying @xmath0 . for @xmath6 well below @xmath7 , there are few vortices in the josephson array ; hence , the josephson links in the model of ref . @xcite behave like inductors , and the array acts like an inhomogeneous @xmath173 network . this network is simply a discretized version of the two - fluid model considered here : the resistances @xmath174 represent the normal quasiparticle channel , and the inductances @xmath175 corresponds to the superfluid . to summarize , we have shown that , in a superconducting layer with a spatially varying @xmath0 , there is extra absorption beyond that from low - lying quasiparticles alone . such quasiparticles are expected in a @xmath9 superconductor . for several models of weak inhomogeneity , we have shown that the extra spectral weight at finite frequencies is proportional to the quantity @xmath3 , as reported for @xmath120 in underdoped samples of bisr@xmath4ca@xmath4cu@xmath4o@xmath176@xcite . furthermore , in this regime , the extra weight generally appears as a lorentzian centered at zero frequency , with half - width proportional to @xmath177 . similar behavior is predicted for tensor inhomogeneities , as is expected in stripe geometries . we also predict that , near an inhomogeneous superconductor - normal transition , the height of the inhomogeneity peak in @xmath178 diverges while its half - width goes to zero . thus , superfluid inhomogeneity may be the principal source of the extra absorption beyond the two - fluid model which is reported in the high - t@xmath1 cuprates . this work has been supported by nsf grant dmr01 - 04987 , and by the u. s./israel binational science foundation . we are most grateful for useful conversations with prof . david bergman .	we discuss the low - frequency absorption arising from quenched inhomogeneity in the superfluid density @xmath0 of a model superconductor . such inhomogeneities may arise in a high - t@xmath1 superconductor from a wide variety of sources , including quenched random disorder and static charge density waves such as stripes . using standard classical methods for treating randomly inhomogeneous media , we show that both mechanisms produce additional absorption at finite frequencies . for a two - fluid model with weak mean - square fluctuations @xmath2 in @xmath0 and a frequency - independent quasiparticle conductivity , the extra absorption has oscillator strength proportional to the quantity @xmath3 , as observed in some experiments . similar behavior is found in a two - fluid model with anticorrelated fluctuations in the superfluid and normal fluid densities . the extra absorption typically occurs as a lorentzian centered at zero frequency . we present simple model calculations for this extra absorption under conditions of both weak and strong fluctuations . the relation between our results and other model calculations is briefly discussed .
the history of star formation in the universe is of intense interest to many in astrophysics , and it is natural to pursue pushing the boundary of observations to as early of times as possible . our understanding of this history is increasing , with a consistent picture now emerging up to redshift @xmath1 , as summarized in fig . the cosmic star formation rate ( sfr ) measurements from the compilation of @xcite are shown , along with new high-@xmath0 measurements based on observations of color - selected lyman break galaxies ( lbg ) @xcite and ly@xmath4 emitters ( lae ) @xcite . much current interest is on this high-@xmath0 frontier , where the primeval stars that may be responsible for reionizing the universe reside . due to the difficulties of making and interpreting these measurements , different results disagree by more than their quoted uncertainties . instead of inferring the formation rate of massive stars from their observed populations , one may directly measure the sfr from their death rate , since their lives are short . while it is not yet possible to detect ordinary core - collapse supernovae at high @xmath0 , long - duration gamma - ray bursts , which have been shown to be associated with a special class of core - collapse supernovae @xcite , have been detected to @xmath5 . the brightness of grbs across a broad range of wavelengths makes them promising probes of the star formation history ( sfh ) ( see , e.g. , the early works of @xcite ) . in the last few years , _ _ swift _ _ @xcite has spearheaded the detection of grbs over an unprecedentedly - wide redshift range , including many bursts at @xmath6 . surprisingly , examination of the _ swift _ data reveals that grb observations _ are not _ tracing the sfh directly , instead implying some kind of additional evolution @xcite . data ( the lae points only sample ly@xmath4 emitters ) . the results of this work , as inferred using bright _ swift _ gamma - ray bursts , are shown with dark diamonds . the solid line is our new high-@xmath0 fit given by eq . [ sfh ] ] grbs can still reveal the overall amount of star formation , provided that we know how the grb rate couples to the sfr . in this study , we use the portion of the sfh that is sufficiently well - determined to probe the range beyond @xmath7 . we do this by relating the many bursts observed in @xmath8 to the corresponding sfr measurements , and by taking into account the possibility of additional evolution of the grb rate relative to the sfr . this calibration eliminates the need for prior knowledge of the absolute conversion factor between the sfr and the grb rate and allows us to properly relate the grb counts at @xmath9 to the sfr in that range . additionally , we make use of the estimated grb luminosities to exclude faint low-@xmath0 grbs that would not be visible in our high-@xmath0 sample , i.e. , to compare `` apples to apples '' . our results show that the sfr must be relatively high in the range @xmath2 when compared to sfr measurements made using more conventional techniques . while the grb statistics at high @xmath0 are relatively low , they are high enough , permitting an approach complementary to other sfr determinations , which themselves must contend with presently unknown extinction corrections , cosmic variance , and selection effects , most importantly that flux - limited surveys necessarily probe the brightest galaxies , which may only contain a small fraction of the star formation activity at early epochs . we discuss the implications these results in the context of reionization and broader applications . the relationship between the comoving grb and star formation rate densities can be parametrized as @xmath10 , where @xmath11 reflects the fraction of stars that produce long - duration grbs and any additional evolutionary effects . importantly , the sfh is well known for @xmath12 , and we use the @xcite fit in this range . the expected ( all - sky ) redshift distribution of grbs can be cast as @xmath13 where @xmath14 summarizes the ability both to detect the initial burst of gamma rays and to obtain a redshift from the optical afterglow , and includes observational factors such as instrumental sensitivities . beaming rendering some fraction of grbs unobservable from earth is accounted for through @xmath15 @xcite , the @xmath16 is due to cosmological time dilation of the observed rate , and @xmath17 is the comoving volume per unit redshift , where @xmath18 is the comoving distance , @xmath19 , @xmath20 , and @xmath21 km / s / mpc . ] . both @xmath22 and @xmath11 are discussed in detail in @xcite , in which it has been shown that the former can be set to a constant ( @xmath23 ) by focusing on a bright subset of the grb sample , allowing the latter to be parametrized as @xmath24 , with @xmath25 ( an enhanced evolution of grbs ) preferred over @xmath26 . here @xmath27 is a ( unknown ) constant that includes the absolute conversion from the sfr to the grb rate in a given grb luminosity range . the evolutionary trend described by @xmath4 may arise from several mechanisms @xcite , such as a grb preference for low - metallicity environments @xcite , and must be accounted for to characterize @xmath28 . grbs are not visible ) , as we determine from the data of butler et al . ( 2007 ) , marked according to the primary method of redshift determination . the ratios of the number counts of grbs in the three shaded boxes are used to estimate the sfr density at high @xmath0 ( 45 , 57 ) by normalizing to the @xmath29 sfr , as discussed in the text . the boxes contain ( 21 , 4 , 4 ) bursts , respectively . [ liso ] ] fig . [ liso ] shows the luminosity - redshift distribution of the 63 long - duration ( @xmath30 s ) grbs from the catalog of @xcite . we compute source - frame grb luminosities , @xmath31 $ ] , based on the values of @xmath32 , the isotropic equivalent ( uncorrected for beaming ) @xmath33 kev energy release in the rest frame , and @xmath34 , the interval containing 90% of the prompt grb emission . using another quantity , such as the peak isotropic equivalent flux ( n. butler , private communication ) yields quantitatively similar results . as only very bright bursts can be seen from all redshifts , we define a cut , based on the luminosities of the high-@xmath0 events , of @xmath35 erg s@xmath36 , for the subsample that we use to estimate the high-@xmath0 sfr , a lower @xmath37 threshold was used . ] . ( a naive normalization using all grbs in @xmath38 without regard to luminosity would yield values @xmath39 times smaller . ) this circumvents the need for an exact definition of the @xmath0-dependent threshold for detecting a grb with a redshift measurement . the complicated detection threshold of _ swift _ ( e.g. , @xcite ) and the human factor involved with optical observations makes a detailed treatment based on constructing the grb luminosity function at low @xmath0 and marginalizing over this distribution to obtain the sfr at high-@xmath0 practically challenging , if not impossible . the net effect would probably only be to make our eventual results somewhat larger . the shaded boxes in fig . [ liso ] show three groups of grbs defined by this @xmath37 cut in @xmath40 @xmath41 , @xmath42 , and @xmath43 ( the high-@xmath0 boxes chosen to have equal counts ) . the grbs in @xmath29 act as a `` control group '' to base the grb to sfr conversion , since this range has both good sfr measurements and grb counting statistics . we calculate the `` expected '' number of grbs in this range as @xmath44 in which @xmath45 , an unknown normalization , depends on the total live - time , @xmath46 , and the angular sky coverage , @xmath47 . using the average sfr density , @xmath48 , a similar relation can be written for @xmath49 and @xmath43 as @xmath50 our interest is in @xmath48 , which we find by dividing out @xmath51 , using eq . [ nz1-z2 ] . taking the observed grb counts , @xmath52 , to be representative of the expected numbers , @xmath53 , we find @xmath54 written in terms of the data for the groups of grbs . the decrease of @xmath55 at @xmath56 increasingly amplifies the significance of distant observed grbs . we show our new determinations of the high-@xmath0 sfr in fig . [ sfh ] , and how they compare with most of the conventional sfr results in this range would necessitate a rescaling ( e.g. , @xcite ) . ] . while evolution with @xmath57 ( 0 ) would decrease ( increase ) these values by a factor @xmath39 , it would also disagree with @xmath58 data . other changes in the details of our analysis , while changing the results somewhat , do not change the important point that the high-@xmath0 sfr must be relatively large to accommodate the several observed high-@xmath0 grbs . these details include the exclusion of a particular grb or changes in redshift assignments or ranges . taking into account the poisson confidence interval for 4 observed events , and slightly enlarging it , we assign a statistical uncertainty of a factor 2 up or down . as the number of grbs increases , this uncertainty will correspondingly decrease , and more attention will need to be placed on systematics.4 grbs from butler et al . ( 2007 ) , none were observed in the @xmath59 1 year since then , indicative of a possible systematic effect . this will not change our results more than the quoted uncertainties . ] . our luminosity cut happens to restrict most of the grb redshifts used to have been obtained by absorption lines , reducing one possible source of error . we provide an update of the sfh fit of @xcite based on our grb results ( solid line in fig . [ sfh ] ) . here we adopt a continuous form of a broken power law , @xmath60^{1/\eta } \label{fit}\end{aligned}\ ] ] where @xmath61 ( @xmath62 , @xmath63 ) is the logarithmic slope of the first ( middle , last ) piece , and the normalization is @xmath64 yr@xmath36 mpc@xmath65 . using @xmath66 smoothes the transitions , while @xmath67 would recover the kinkiness of the original form . the breaks at @xmath68 and @xmath69 correspond to @xmath70 and @xmath71 . our sfr results , particularly the @xmath72 point , are at the high end of sfr results obtained with more conventional methods , which show significant differences among themselves . likely , these methods are measuring different populations of galaxies , with the grbs probing low - luminosity galaxies that are otherwise not accounted for fully . as detailed , the statistics of the grb data are adequate , and it is unlikely that our grb results are overestimated , and if anything , the true sfr is even larger . we assume that a steep drop in the sfr is not being hidden by a steep rise in the fraction of stars that produce grbs , beyond the evolution already taken into account ; that would be even more interesting . the extent of extinction by dust on high-@xmath0 sfr measurements is not yet strongly constrained . there are a number of indications that dust is ubiquitous over the range @xmath73 ( e.g. , @xcite ) , and the dust correction in this range can be up to a factor of @xmath74 . at higher redshift , there are no strong observational constraints . the dust corrections assumed in @xcite are generally small at high @xmath0 ; on the other hand , observations verify the existence of at least one heavily - obscured galaxy at @xmath75 @xcite , and uv - selected samples would be expected to be biased against such objects . we also note that the recent mid - ir detection of the progenitor of sn 2008s implies that some short - lived stars do produce dust @xcite . for all of the lbg - based sfr results in fig . [ sfh ] , we have included the dust corrections indicated by the respective authors . while the true dust corrections may be larger , it seems likely that an additional factor of several is implausible . the lbg surveys are most sensitive to the brightest galaxies , and great efforts are made to define the uv galaxy luminosity function ( lf ) and how well it is sampled at the faint end . for example , the results shown in our fig . [ sfh ] from @xcite are integrated down @xmath76 , for @xmath77 defined at @xmath78 . for the three lowest-@xmath0 points , they also integrate down to @xmath79 , yielding sfr results a factor of a few higher . correcting these measurements to account for the contributions from even fainter galaxies is necessary but difficult due to the poor observational constraints on the faint - end slope of the lf . depending on the value of this slope , and the lower limit of the integration , the necessary correction may be as little as a factor of two , or as much as an order of magnitude ( see @xcite ) . it also seems that ly@xmath4 emitters do not contribute significantly to the total sfr . we developed an empirical method for estimating the high-@xmath0 sfr using grb counts , improving on earlier estimates of the high-@xmath0 sfr using grb data ( e.g. , @xcite ) in several ways , not least of which by using significantly updated sfr @xcite and/or grb @xcite data . taking advantage of the improved knowledge of the sfr at intermediate @xmath0 , we were able to move beyond the assumption of a simple one - to - one correspondence between the grb rate and the sfr , accounting for an increasing evolutionary trend . the higher statistics of the recent _ swift _ grb data allowed the use of luminosity cuts to fairly compare grbs in the full @xmath0 range , eliminating the uncertainty of the unknown grb luminosity function . by comparing the counts of grbs at different @xmath0 ranges , normalized to sfr data at intermediate @xmath0 , we based our results squarely on data , eliminating the need for knowledge of the absolute fraction of stars that produce grbs . our grb - based sfr value in @xmath49 is comparable to more conventional results , which may be taken as a validation of our method , i.e. , possible selection effects have been adequately controlled ( see @xcite for more details ) , while our sfr value in @xmath72 is larger , which has important implications . a significant population of low - luminosity star - forming galaxies picked out by grbs but missed in other surveys could reconcile these differences ( searches utilizing cluster lensing , e.g. , @xcite , may help , if cosmic variance can be understood ) . @xcite argue that the correction for the faint - end of the lf should be large to correct for dwarf galaxies missed in lbg surveys . since grbs are observed to favor subluminous host galaxies @xcite , they may be probing such faint galaxies . the universe was fully reionized by @xmath80 @xcite , and it appears that agn could not have been responsible @xcite . can stars have reionized the universe ? in @xcite , the sfr density required to produce a sufficient ionizing photon flux is parametrized by two factors : the photon escape fraction ( @xmath81 ) and the clumpiness of the igm ( @xmath82 ) . these enter as a ratio , @xmath83 , which is not known precisely . for @xmath84 , the required sfr at @xmath85 is @xmath86 yr@xmath36 mpc@xmath65 , just at the level of our grb - based sfr . this ratio may be higher ( due to either @xmath82 or @xmath81 ) , however , which would render our measured value of @xmath87 too low , reintroducing the concern that star formation may be insufficient to achieve reionization ( e.g. , @xcite , c. a. faucher - giguere et al . 2008 , in preparation ) . is there any way around this ? one can plausibly increase our sfr determination by considering the increasing incompleteness of the grb sample with @xmath0 due to our @xmath37 cut , since for fixed luminosity , higher-@xmath0 bursts are relatively more difficult to detect @xcite and the universe will at some @xmath0 quickly become opaque blueward of ly@xmath4 ( e.g. , @xcite ) . if there is a _ metallicity for forming a grb , then the evolutionary trend may saturate at high enough redshift ; arbitrarily assuming no evolution beyond @xmath6 would give an additional factor of @xmath88 . finally , requiring a _ minimum _ amount of metals in a star for a successful grb would suppress the grb rate at increasingly high-@xmath0 , necessitating a higher underlying sfr . a similar argument could be made concerning whether early , metal - poor stars possessed sufficient angular momentum . overall , more data are needed concerning the varying ionization fraction with redshift to establish if star formation is quantitatively acceptable as an explanation for reionization , instead of simply by default if agn have been eliminated . the possibility of continuing evolution in the grb rate relative to the sfr invites some astrophysical speculations . while the overall black hole production rate in core - collapse supernovae is poorly understood empirically @xcite , bright grbs might be regarded as tracing the large angular momentum end of the black hole birth distribution ( e.g. , for collapsars ; @xcite ) . evolution would mean that the high-@xmath0 universe was more efficient at producing such black holes than usually considered . this may have implications for the nucleosynthetic yields from these explosions ( e.g. , @xcite ) . the effect of supernovae on the gas in high-@xmath0 galaxies is an ongoing field of research ( e.g. , @xcite ) . following from the considerations above , it may be that small galaxies had a disproportionately large rate of grbs relative to normal supernovae . in this case , it would be interesting to examine the fate of such galaxies after including multiple injections of highly - asymmetric , relativistic ejecta , as well as implications for enrichment of the igm ( i.e. , `` grb feedback '' ) . we thank nat butler , kris stanek , todd thompson , and haojing yan for helpful discussions . we acknowledge use of the _ swift _ public archive . hy and jfb were supported by nsf career grant phy-0547102 to jfb , mdk by doe grant de - fg02 - 91er40690 , and amh by the australian research council in the form of a qeii fellowship ( dp0557850 ) . ando , m. , et al . 2006 , , 645 , l9 band , d. l. 2006 , , 644 , 378 berger , e. , et al . 2005 , , 634 , 501 blain , a. w. , & natarajan , p. 2000 , , 312 , l35 bloom , j. s. 2003 , , 125 , 2865 bloom , j. s. , frail , d. a. , & kulkarni , s. r. 2003 , , 594 , 674 bouwens , r. j. , et al . 2007 , , 670 , 928 bouwens , r. j. , et al . 2008 , , submitted ( arxiv:0803.0548 ) bromm , v. , & loeb , a. 2002 , , 575 , 111 butler , n. r. , et al . 2007 , , 671 , 656 cen , r. , & fang , t. 2007 , preprint ( arxiv:0710.4370 ) chary , r .- r . , stern , d. , & eisenhardt , p. 2005 , , 635 , l5 chary , r. , berger , e. , & cowie , l. 2007 , , 671 , 272 ciardi , b. , & loeb , a. 2000 , , 540 , 687 coward , d. m. , et al . 2008 , , 386 , 111 daigne , f. , rossi , e. , & mochkovitch , r. 2006 , , 372 , 1034 fan , x. , et al . 2006 , , 132 , 117 fiore , f. , et al . 2007 , , 470 , 515 firmani , c. , et al . 2004 , , 611 , 1033 fruchter , a. s. , et al . 2006 , , 441 , 463 fynbo , j. p. u. , et al . 2003 , , 406 , l63 gehrels , n. , et al . 2004 , , 611 , 1005 gnedin , n. y. 2008 , , 673 , l1 guetta , d. , & piran , t. 2007 , jcap , 7 , 3 hjorth , j. , et al . 2003 , , 423 , 847 hopkins , a. m. 2006 , astro - ph/0611283 hopkins , a. m. , & beacom , j. f. 2006 , , 651 , 142 hopkins , p. f. , et al . 2008 , , 175 , 356 jakobsson , p. , et al . 2006 , , 447 , 897 kistler , m. d. et al . 2008 , , 673 , l119 kocevski , d. , & butler , n. 2008 , , in press ( arxiv:0707.4478 ) kochanek , c. s. , et al . 2008 , , in press ( arxiv:0802.0456 ) lamb , d. q. , & reichart , d. e. 2000 , , 536 , 1 langer , n. , & norman , c. a. 2006 , , 638 , l63 le , t. , & dermer , c. d. 2007 , , 661 , 394 le floch , e. , et al . 2003 , , 400 , 499 li , l. x. 2007 , , in press ( arxiv:0710.3587 ) macfadyen , a. , & woosley , s. e. 1999 , , 524 , 262 madau , p. , haardt , f. , & rees , m. j. 1999 , , 514 , 648 mannucci , f. , et al . 2007 , , 461 , 423 natarajan , p. , et al . 2005 , , 364 , l8 nomoto , k. , et al . 2006 , nuclear physics a , 777 , 424 ota , k. , et al . 2008 , , 677 , 12 ouchi , m. , et al . 2004 , , 611 , 660 porciani , c. , & madau , p. 2001 , , 548 , 522 prieto , j. l. , et al . 2008 , apjl , in press ( arxiv:0803.0324 ) richard , j. , et al . 2008 , , submitted ( arxiv:0803.4391 ) salpeter , e. e. 1955 , , 121 , 161 salvaterra , r. , & chincarini , g. 2007 , , 656 , l49 salvaterra , r. , et al . 2008 , preprint ( arxiv:0805.4104 ) stanek , k. z. , et al . 2003 , , 591 , l17 stanek , k. z. , et al . 2006 , acta astron . , 56 , 333 totani , t. 1997 , , 486 , l71 verma , a. , et al . 2007 , , 377 , 1024 whalen , d. , et al . 2008 , , submitted ( arxiv:0801.3698 ) wijers , r. a. m. , et al . 1998 , , 294 , l13 wilkins , s. m. , trentham , n. , & hopkins , a. m. 2008 , , 385 , 687 wyithe , j. s. b. , loeb , a. , & carilli , c. 2005 , , 628 , 575 yan , h. , & windhorst , r. a. 2004 , , 600 , l1 yksel , h. , & kistler , m. d. 2007 , , 75 , 083004	while the high-@xmath0 frontier of star formation rate ( sfr ) studies has advanced rapidly , direct measurements beyond @xmath1 remain difficult , as shown by significant disagreements among different results . gamma - ray bursts , owing to their brightness and association with massive stars , offer hope of clarifying this situation , provided that the grb rate can be properly related to the sfr . the _ swift _ grb data reveal an increasing evolution in the grb rate relative to the sfr at intermediate @xmath0 ; taking this into account , we use the highest-@xmath0 grb data to make a new determination of the sfr at @xmath2 . our results exceed the lowest direct sfr measurements , and imply that no steep drop exists in the sfr up to at least @xmath3 . we discuss the implications of our result for cosmic reionization , the efficiency of the universe in producing stellar - mass black holes , and `` grb feedback '' in star - forming hosts .
the hawking radiation @xcite is one of the most robust and important predictions of quantum field theory in curved space - time . here we would like to study some ( local ) properties of a generic friedmann - lemaitre - robertson - walker ( flrw ) space - time with non - flat topology . let us remind some basic facts about the formalism . any spherically symmetric four dimensional metric can be expressed in the form : @xmath0 with @xmath1 a tensor describing a two - dimensional space - time with coordinates @xmath2 , @xmath3 being the `` areal radius '' and @xmath4 encoding the metric of a two - dimensional sphere orthogonal respect to the first one . the dynamical trapping horizon -if exists- is located in the correspondence of @xmath5 thus , one may define the quasi - local misner - sharp gravitational energy as @xmath6\,.\ ] ] for example , the mass of a black hole described by this formalism results to be @xmath7 . the killing vector fields @xmath8 are the generators of the isometries with @xmath9 : in the static case , with the time - like killing vector field @xmath10 , the killing surface gravity @xmath11 is given by @xmath12 in the dynamical case , the real geometric object which generalizes the killing vector field is the kodama vector field @xcite , @xmath13 thus , the hayward surface gravity associated with dynamical horizon is @xcite @xmath14 the hawking radiation is a thermal radiation of the black holes due to quantum effects . in the static case , all derivations of hawking radiation lead to a semi - classical expression for the radiation rate @xmath15 in terms of the exchange @xmath16 of the killing energy @xmath17 and the killing / hawking temperature @xmath18 , @xmath19 in the dynamical case one may suggest the kodama / hayward temperature : @xmath20 an important example that demonstrates the covariance of the formalism is given by the de sitter space - time . the static patch reads @xmath21 where @xmath22 and the horizon is located at @xmath23 with surface gravity @xmath24 . the second patch is given by the expanding coordinates of the flat flrw metric , @xmath25 where @xmath26 and the dynamical ( cosmological ) horizon is @xmath23 with @xmath24 . finally , the global patch in non - flat flrw metric is given by @xmath27 with @xmath28 , and @xmath23 and @xmath24 again . now we will see how it is possible to associate a temperature to the dynamical horizon of flat and non - flat de sitter space - time in ( [ flat])([global ] ) . we recall the quantization of a conformal coupled massless scalar field in the frlw space - time . the metric reads @xmath29 where @xmath30 is the conformal time , @xmath31 is a mass scale and the topology of the spacial section can be flat , spherically or hyperbolic for @xmath32 , respectively . given a massless scalar field , @xmath33 such that the modes are conformal invariant , namely @xmath34 , the associated wightman function @xmath35 results to be @xmath36 the wightman function satisfies the following rule for the conformal transformations of the metric : @xmath37 we may also take @xmath38 due to the homogeneity and isotropy of flrw space - times . let us consider the spherical case ( @xmath39 ) in ( [ metric ] ) , @xmath40 this metric is conformally related to the minkowski space - time , @xmath41 with @xmath42 thus , by starting from the well - known wightman function in minkowski space - time , one can use ( [ conf ] ) and derive for the spherical flrw metric @xmath43 the hyperbolic case @xmath44 is obtained with the substitution @xmath45 , while the flat case @xmath46 corresponds to the limit @xmath47 . let us consider a free massless quantum scalar field @xmath48 in thermal equilibrium at the temperature @xmath49 in flat space - time . we know that finite temperature field theory effects of this kind can be investigated by given that the scalar field defined in the euclidean manifold @xmath50 , where the imaginary time is @xmath51 , compactified in the circle @xmath52 with period @xmath53 . we briefly review the local quantity @xmath54 , which is a divergent quantity due to the product of valued operator distribution in the same point @xmath55 . by making use of the zeta - function regularization procedure , the quantum fluctuations read @xcite : @xmath56 where @xmath57 is the local zeta - function associated with the operator @xmath58 . it is easy to see that the analytic continuation of the local zeta - function is regular at @xmath59 and finally one gets @xmath60 in this way we obtain the temperature of the quantum field in thermal equilibrium from the zeta - function renormalized vacuum expectation value , namely we have a quantum thermometer . now we extend the argument to generic flrw metric . the off - diagonal wightman function ( [ wsph ] ) leads to @xmath61 with @xmath62 where @xmath63 is the conformal factor . thus , in the limit @xmath64 , one has @xmath65 it is possible to show that @xmath66 where @xmath67 the dot being the derivative respect to the proper time , @xmath68 being the usual hubble parameter , and @xmath69 the radial acceleration . therefore , after the regularization for the divergent part at @xmath70 , @xmath71 this result is quite general and it is valid also for spatial curvature @xmath72 . in the case of de sitter space - time with @xmath39 , we may put @xmath73 and the expression for quantum fluctuations reads @xmath74 where @xmath75 is the areal radius of the kodama observer and the acceleration has been computed as @xmath76 for a kodama observer with @xmath77 we recover the gibbons - hawking temperature associated with de sitter space - time , @xmath78 this is an important check of our approach , since it shows the coordinate independence of the result for the important case of de sitter space - time . the minkowski space - time may be written in a frlw form with hyperbolic section @xmath44 ( milne universe ) , @xmath79 making use of the hayward formalism , it is easy to verify that there is no dynamical horizon and the surface gravity is vanishing . in this case we obtain @xmath80 namely only the radial acceleration @xmath69 is present and the temperature is defined as @xmath81 recovering the well known unruh effect . this proceeding is based on the paper @xcite . i thank sergio zerbini for helpful discussions .	in this paper we study a quantum field theoretical approach , where a quantum probe is used to investigate the properties of generic non - flat frlw space time . the fluctuations related to a massless conformal coupled scalar field defined on a space - time with horizon is identified with a probe and the procedure to measure the local temperature is presented .
for almost fifty years , @xmath2-sphere bundles over the @xmath3-sphere have been of interest to both topologists and geometers . in 1956 , milnor @xcite proved that all @xmath0-bundles over @xmath1 with euler class @xmath4 are homeomorphic to @xmath5 . he also showed that some of these bundles are not diffeomorphic to @xmath5 , he thereby exhibited the first examples of exotic spheres . shortly thereafter , in 1962 , eells and kuiper @xcite classified all such bundles with euler class @xmath4 up to diffeomorphism and showed that @xmath6 of the @xmath7 seven dimensional exotic spheres can be described as @xmath0-bundles over @xmath1 . in 1974 gromoll and meyer @xcite constructed a metric with non - negative sectional curvature on one of these sphere bundles , exhibiting the first example of an exotic sphere with non - negative curvature . very recently grove and ziller @xcite showed that the total space of every @xmath0-bundle over @xmath1 admits a metric with non - negative sectional curvature . motivated by these examples they asked for a classification of these manifolds up to homotopy equivalence , homeomorphism and diffeomorphism ( see problem 5.2 in @xcite ) . although a partial classification of this type is contained in @xcite and @xcite , it seems that this has not been carried out in the literature in complete generality . it is the purpose of this note to do this . in the following description we use the notation of tamura @xcite . let @xmath8 denote the fibre bundles over the @xmath3-sphere @xmath9 with total space @xmath10 , fibre the @xmath2-sphere @xmath11 and structure group @xmath12 . equivalence classes of such bundles are in one - to - one correspondence with @xmath13 . more precisely , the generators @xmath14 and @xmath15 of @xmath16 are given by @xmath17 where @xmath18 and @xmath19 denote the quaternions with norm 1 , i.e. we have have identified @xmath11 with the unit quaternions . with this choice of generators each pair of integers @xmath20 gives the element @xmath21 and thus determines the bundle @xmath22 the definition we use is different from the one given by milnor in @xcite . he uses two integers @xmath23 corresponding to a different choice of generators of @xmath24 . the two pairs @xmath20 and @xmath23 are related by @xmath25 , @xmath26 . the first level of classification of the manifolds @xmath27 is up to homotopy equivalence . in a remarkable paper @xcite james and whitehead succeeded in classifying the manifolds @xmath27 up to homotopy except for the case @xmath28 . we complete this last case by means of standard tools of homotopy theory which were not available in early 1950s when james and whitehead wrote their article . [ hom ] the manifolds @xmath27 and @xmath29 are homotopy equivalent if and only if @xmath30 and @xmath31 . note that since @xmath32 , necessarily @xmath33 if @xmath27 and @xmath34 are homotopy equivalent . the next level , the homeomorphism classification , was studied by tamura @xcite , who gave sufficient conditions for two manifolds @xmath27 and @xmath35 to be homeomorphic . we use basic results in surgery theory to show that , in fact , tamura s conditions are necessary as well . let @xmath36 denote the first pontrjagin class of @xmath27 . [ top ] the manifolds @xmath27 and @xmath35 are homeomorphic if and only if * ( a ) * @xmath27 and @xmath35 are homotopy equivalent and * ( b ) * @xmath37 , or equivalently @xmath38 . the homeomorphism classification is exactly the same as the @xmath39-classification for the manifolds @xmath27 . indeed , we take advantage of this fact in the proof of theorem [ top ] . the diffeomorphism classification of the manifolds @xmath27 is related to the famous article by milnor @xcite , where he proved that the manifolds @xmath40 are exotic homotopy spheres . eells and kuiper @xcite gave a diffeomorphism classification of the homotopy spheres @xmath40 by introducing a new invariant which is known as the @xmath41-invariant . we now recall its definition . assume that @xmath42 is a closed , smooth @xmath43 manifold such that @xmath10 is a @xmath43 boundary , i.e. @xmath44 , where @xmath45 is a closed , smooth @xmath43 manifold . in addition , we require @xmath46 to satisfy the _ @xmath41-condition _ , see @xcite . the @xmath41-condition allows the pull - back of the pontrjagin classes of @xmath46 to @xmath47 . then the @xmath41-invariant is defined as @xmath48\ } \mod 1\ ] ] here @xmath49 $ ] stands for the signature of @xmath46 . the most important feature of the @xmath41-invariant is that it is an invariant of the diffeomorphism type of @xmath10 . we use the @xmath41-invariant to prove the following theorem . [ diff ] the manifolds @xmath27 and @xmath35 are diffeomorphic if and only * ( a ) * @xmath27 and @xmath35 are homeomorphic and * ( b ) * @xmath50 . smooth surgery theory ( see @xcite ) implies that there are exactly 28 different smooth manifolds homotopy equivalent to @xmath27 . a natural question to ask is which manifolds homotopy equivalent to @xmath27 may be represented by manifolds @xmath51 . in a corollary to theorem [ diff ] we give a complete answer to this question . it is a pleasure to acknowledge very helpful discussions with jie wu and his permission to use his thesis @xcite . recently we became aware that d. crowley from indiana university is independently working on a project close to the subject of this paper . we first observe that theorem [ hom ] is proved in theorem 1.7 of @xcite for all cases except for @xmath52 . thus we now consider only the manifolds @xmath53 . first we recall some basic facts from @xcite . let @xmath54 be a sphere bundle with fibre @xmath0 . then there is an exact sequence of homotopy groups : @xmath55 in particular , we have to understand the homomorphism @xmath56 in order to compute the various homology , cohomology and homotopy groups of @xmath57 . in the case of @xmath58 we know that @xmath59 , and by the hurewicz theorem @xmath60 . hence the homomorphism @xmath61 is multiplication by @xmath62 . we assume that the spheres @xmath0 and @xmath1 are oriented by choosing the standard generators @xmath63 and @xmath64 . both generators @xmath65 are given by the identity maps on the spheres . to give an orientation to the manifold @xmath66 one chooses on orientation on the base @xmath1 and on the fiber @xmath0 . here we have that @xmath67 . thus , as pointed out in @xcite , the manifolds @xmath53 have a _ preferred _ orientation which is defined naturally by the standard orientations of the base and fiber spheres . however , the manifolds @xmath53 still may have ( and indeed have ) a self - homotopy map @xmath68 of degree @xmath69 . james & whitehead @xcite give the following definitions . two manifolds @xmath53 and @xmath70 are called _ opposed to each other _ if there exists a homotopy equivalence @xmath71 of degree @xmath69 . a manifold @xmath53 is called _ self - opposed _ if there exists a self - homotopy equivalence @xmath68 of degree @xmath69 . as it is emphasized in @xcite , there is no ambiguity here if @xmath72 or @xmath73 since then every homotopy equivalence @xmath74 must be of degree @xmath75 and this fact is used in @xcite to obtain the homotopy equivalence classification for @xmath72 and @xmath73 . hence in order to complete the classification for @xmath52 we need to show that up to homotopy every homotopy equivalence @xmath76 must be of degree @xmath75 . we prove this fact by showing that every manifold @xmath53 is self - opposed . given a manifold @xmath77 we construct a self - homotopy map @xmath68 of degree @xmath69 by considering @xmath78 as a @xmath79-complex . the cell - structure is described in @xcite . in fact , the @xmath3-skeleton @xmath80 of @xmath78 is the moore space , i.e. @xmath81 , and the space @xmath78 is obtained from @xmath80 by attaching a @xmath82-cell @xmath83 . let @xmath84 be an attaching map for the cell @xmath83 which represents some element @xmath85\in \pi_6(\sigma^2{{\mathbf r}}\p^2)$ ] . consider the long exact sequence of the pair @xmath86 : @xmath87 the above groups and homomorphisms are amongst those computed by jie wu in his thesis @xcite . using his results we obtain the following commutative diagram . @xmath88 here @xmath89 indicates that @xmath90 is a generator of @xmath91 and @xmath92 , @xmath93 , and @xmath94 . in particular , the homomorphism @xmath95 is trivial . let @xmath96 be the element given by the characteristic map @xmath97 , and @xmath63 be the standard generator of @xmath98 . james & whitehead ( * ? ? ? * section 5 ) show that @xmath99 ) = [ \kappa,\iota_3 ] \in \pi_6(\sigma^2{{\mathbf r}}\p^2,s^3)$ ] where @xmath100 was an attaching map for the cell @xmath83 . it is emphasized in ( * ? ? ? * theorem 1.4 . ) that a necessary condition for a manifold @xmath53 to be self - opposed is that _ the whitehead product @xmath101\in \pi_6(\sigma^2{{\mathbf r}}\p^2,s^3)$ ] be of order @xmath62_. this condition clearly holds since @xmath102 . there is also a sufficient condition given in ( * ? ? ? * theorem 1.4 . ) . namely , _ if @xmath103=0 $ ] in the group @xmath104 and @xmath105 is infinite cyclic , then the manifolds @xmath53 are self - opposed . _ this condition may be a source for confusion because the group @xmath105 is not infinite cyclic , but rather @xmath106 , see @xcite . however , james & whitehead also provide a true necessary and sufficient condition for a manifold @xmath53 to be self - opposed in ( * ? ? ? * lemma 4.1 . ) . let @xmath107 be an attaching map for the top cell of @xmath53 . then the following holds ( see ( * ? ? ? * lemma 4.1 ) ) . ( @xmath108 ) * _ a manifold @xmath53 is self - opposed if and only if @xmath109 = i_([\eta,\iota_3])$ ] , where @xmath110 is the generator , and @xmath63 is the standard generator ( as above ) . _ note that @xmath111 = \nu^{\prime}$ ] in our notation , see for example @xcite , and that @xmath92 . thus , to complete the classification , it is enough to prove the following [ la ] let @xmath84 be an attaching map for the top cell of @xmath53 . then @xmath109=i_(\nu^{\prime})$ ] . * proof . * first we show that the whitehead product @xmath101\in \pi_6(\sigma^2{{\mathbf r}}\p^2,s^3)$ ] is non - zero . assume to the contrary that @xmath101=0 $ ] . as @xmath101=j_([\phi])$ ] , the attaching map @xmath112 may then be lifted to a map @xmath113 . this leads to a contradiction as follows . if @xmath114=0 $ ] in @xmath115 , then @xmath116 , and we have a long exact sequence in homotopy : @xmath117 however , @xmath118 , @xmath119 , @xmath120 , and exactness of the above sequence fails . if @xmath114\neq 0 $ ] in @xmath115 then @xmath114=\nu^{\prime}$ ] . let @xmath121 . then we obtain a map @xmath122 which on homology induces an epimorphism @xmath123 and an isomorphism @xmath124 let @xmath125 , @xmath126 , @xmath127 , be generators . using the naturality of the cap product and poincar duality we obtain that @xmath128 hence @xmath129 which fails since @xmath130 . therefore we obtain a contradiction to the existence of the map @xmath131 which implies that @xmath99 ) = [ \kappa,\iota_3 ] \ne 0 $ ] . thus the element @xmath85\in \pi_6(\sigma^2{{\mathbf r}}\p^2)$ ] must be @xmath132 or @xmath133 in both cases @xmath109 = ( \pm 2 \lambda,0 ) = ( 2 \lambda,0)=i_([\eta,\iota_3 ] ) = i_(\nu^{\prime})$ ] . this proves lemma [ la ] . @xmath134 in fact , it is not hard to prove that @xmath85=(\pm\lambda,0)\in \pi_6(\sigma^2{{\mathbf r}}\p^2)$ ] . to do this , it is enough to observe that the the pinching map @xmath135 induces a map in homotopy @xmath136 which takes the element @xmath85 $ ] to zero . it is easy to compute the cohomology of @xmath27 with integer coefficients by using the serre spectral sequence : @xmath137 computing the pontrjagin classes @xcite of the bundles @xmath138 and the total spaces @xmath27 yields @xmath139 where @xmath140 and @xmath141 are generators of @xmath142 and @xmath143 , respectively . we now start with a given homotopy type , i.e. we with an equivalence class of manifolds , all homotopy equivalent to each other . by theorem [ hom ] we know that these manifolds have indices satisfying a certain congruence . in order to use surgery theory to classify these manifolds up to homeomorphism we recall the following definitions based on the work of sullivan , see @xcite for a detailed description . let @xmath78 be a @xmath144-manifold , where @xmath144 stands for the piecewise linear ( @xmath145 ) , topological ( @xmath146 ) , or smooth ( @xmath147 ) category . then @xmath148 , the @xmath144-structure set of @xmath78 , consists of pairs @xmath149 where @xmath150 is a @xmath144-manifold and @xmath151 is a simple homotopy equivalence . two objects @xmath152 and @xmath153 represent the same element in @xmath148 if there exist an h - cobordism @xmath46 with @xmath154 and a homotopy equivalence @xmath155 . moreover we denote by @xmath156 the set of equivalence classes of normal cobordisms . here two normal maps ( or surgery problems ) @xmath157 are normally cobordant if there exists a cobordism @xmath46 with @xmath154 and @xmath158 such that @xmath159 where @xmath160 is a bundle isomorphism . let @xmath161 the following two facts are well - known , see @xcite . \(1 ) @xmath162\,.$ ] \(2 ) _ there is an exact sequence of sets , the structure sequence : _ @xmath163 in our case @xmath164 and the map @xmath165 is an epimorphism in the @xmath146 and @xmath145 category . hence ( 1 ) and ( 2 ) imply that @xmath166\,.\ ] ] we now use the structure of the well - known space @xmath167 . we proceed by first localizing at the prime @xmath168 ; @xmath167 localized at @xmath168 is just a product of eilenberg - maclane spaces . @xmath169 = \prod_{k } k({\mathbf z}_{(2 ) } , 4k ) \times k({\mathbf z}_{2 } , 4k - 2).\ ] ] as there is no torsion in this case we obtain for the homology of @xmath27 with @xmath170 coefficients @xmath171 now for the prime @xmath168 we derive @xmath172\ , ] & \cong & [ m_{m , n}\,,\ ; \prod_{k } k({\mathbf z}_{(2 ) } , 4k ) \times k({\mathbf z}_{2 } , 4k - 2 ) ] \\ \\ & \cong & h^{4}(m_{m , n}\,;{\mathbf z}_{(2 ) } ) \cong { \mathbf z}_{n } \otimes { \mathbf z}_{(2)}\ , . \end{array}\ ] ] [ top / pl ] @xmath173\ , ] \cong [ m_{m , n}^{7}\ , ; g/{{\mathrm p}{\mathrm l}}[2]\,]$ ] proof . * consider the natural map @xmath174 \longrightarrow g/{{\mathrm t}{\mathrm o}{\mathrm p}}[2]$ ] and let @xmath175 be the homotopy fibre of the map @xmath176 . the space @xmath175 is a two - stage postnikov system ( * ? ? ? * theorem 4.8 ) . @xmath177 as @xmath178 , we can lift maps to @xmath179 and obtain @xmath180 \cong [ m_{m , n}^{7}\ , ; k({\mathbf z}_{(2)},4)]\,.\ ] ] thus @xmath181\ , ] \cong h^{4}(m_{m , n}^{7}\,;{\mathbf z}_{(2 ) } ) \cong [ m_{m , n}^{7}\ , ; g/{{\mathrm t}{\mathrm o}{\mathrm p}}[2]\,]\,.$ ] @xmath134 the next step is to localize at primes @xmath182 , i.e. to study @xmath183\,]\,.$ ] by sullivan s theorem ( * ? ? ? * theorem 4.28 ) there is a natural homotopy equivalence @xmath184 \longrightarrow \mbox{bo}[1/2]\,]\,.$ ] hence we obtain that @xmath185\ , ] \cong [ m_{m , n}^{7}\,;\mbox{bo}[1/2]\ , ] \cong \widetilde{\mbox{ko}}^{0}(m_{m , n}^{7 } ) \otimes { \mathbf z}[1/2]\,.\ ] ] we use the atiyah - hirzebruch spectral sequence to compute @xmath186 \cong { \mathbf z}_{n } \otimes { \mathbf z}[1/2 ] \,.$ ] we combine the previous calculations in a cartesian square . @xmath187\ , ] \\ @vvv @vvv \\ \lbrack m_{m , n}^{7}\ , ; g/{{\mathrm p}{\mathrm l}}[1/2]\,\rbrack @ > > > [ m_{m , n}^{7}\ , ; \prod_{j \ge 1 } k({\mathbf q } , 4 j ) ] \end{cd}\ ] ] as @xmath188 we obtain @xmath189 & \cong & [ m_{m , n}^{7}\ , ; g/{{\mathrm p}{\mathrm l}}[2]\ , ] \oplus [ m_{m , n}^{7}\ , ; g/{{\mathrm p}{\mathrm l}}[1/2]\ , ] \\ \\ & \cong & ( { \mathbf z}_{n } \otimes { \mathbf z}_{(2 ) } ) \oplus ( { \mathbf z}_{n } \otimes { \mathbf z}[1/2 ] ) \\ \\ & \cong & { \mathbf z}_{n } \cong h^{4}(m_{m , n}^{7 } ; { \mathbf z})\ , . \end{array}\ ] ] * conclusion : * the previous arguments imply that @xmath190 \cong [ m_{m , n}^{7}\ , ; g/{{\mathrm p}{\mathrm l } } ] \cong { \mathbf z}_{n}\,.\ ] ] note that the structure set @xmath191 in general may contain manifolds that are not sphere bundles over spheres . hence we now define @xmath192 where @xmath193 this subset of @xmath191 has the following properties . [ stop ] let @xmath194 where @xmath195 and @xmath196 . * ( a ) * @xmath197 if @xmath198 ; @xmath199 if @xmath200 ; * ( b ) * every element in @xmath201 can be represented by some manifold @xmath35 . * proof . * part ( a ) follows from a simple counting argument as every fourth element in @xmath202 is identified . statement ( b ) is a consequence of part ( a ) as there are enough manifolds @xmath203 to generate @xmath204 or @xmath205 respectively . if @xmath203 is homotopy equivalent to @xmath206 , then @xmath207 . but we also have that @xmath208 which implies that @xmath209 if @xmath210 and @xmath211 if @xmath212 . @xmath134 the special cases of @xmath213 and @xmath214 can be treated as follows . if @xmath214 and @xmath215 then @xmath216 and we discuss this case in section 4 . if @xmath214 and @xmath217 then @xmath52 and @xmath218 and @xmath219_{(2 ) } \cong h^{4}(m_{m,2};{\mathbf z}_{2 } ) \cong { \mathbf z}_{2}\,.\ ] ] hence there are two homeomorphism classes of manifolds , which the first pontrjagin class can not distinguish as it is always zero . however , by the work of tamura @xcite we know that if @xmath220 , then @xmath221 is homeomorphic to @xmath222 . on the other hand , if @xmath221 is homeomorphic to @xmath222 then they are homotopy equivalent and theorem [ hom ] implies @xmath223 . if @xmath214 and @xmath224 then @xmath225 and @xmath226 . but in this case we obtain from theorem [ hom ] that @xmath227 is homotopy equivalent to @xmath228 if and only if @xmath229 . again the work of tamura @xcite implies that if @xmath230 , then @xmath227 is homeomorphic to @xmath228 . hence for this special case we obtain that @xmath227 is homotopy equivalent to @xmath228 if and only if @xmath227 is homeomorphic to @xmath228 which is the case if and only if @xmath230 . again these manifolds are not distinguished by their first pontrjagin class . as we are only interested in the subset @xmath201 for the homeomorphism classification , theorem [ top ] follows . the work of eells - kuiper @xcite implies that the @xmath41-invariant of equation ( [ inv ] ) determines the differentiable structure of the total spaces @xmath27 . here @xmath231 is computed for any spin coboundary @xmath232 which satisfies condition @xmath41 . hence in order to complete the classification we need to find a suitable coboundary @xmath232 and compute the corresponding @xmath41-invariant . the natural choice for @xmath232 is the associated disk bundle @xmath233 , @xmath234 . this coboundary was already used in @xcite and @xcite . as @xmath232 is a disk bundle over @xmath9 one obtains for the first pontrjagin class @xmath235 and for the signature @xmath236 . hence we obtain for the @xmath41-invariant of the total spaces @xmath27 @xmath237\ } \mod 1 \\ \\ & \equiv & { 1 \over 2^{5 } \cdot 7 } \ \{(n + 2m)^{2 } - 1\ } \ \mod 1\ , . \end{array}\ ] ] as shown in @xcite , the @xmath41-invariant distinguishes the diffeomorphism type in this case and theorem [ diff ] follows . in order to see exactly how many classes of each homotopy type are represented by smooth manifolds of type @xmath27 we introduce the following integer valued function . note that @xmath243 for all @xmath244 shows that there are @xmath245 distinct classes of smooth manifolds @xmath40 . but manifolds of type @xmath40 are homeomorphic to @xmath5 for all @xmath244 , hence we obtain the usual differentiable structure on @xmath5 together with the @xmath6 exotic spheres that can be described as @xmath11- bundles over @xmath9 . * proof of corollary [ difftypes ] . * in each case we first combine the homotopy equivalence condition , @xmath246 , and the homeomorphism condition , @xmath247 , to obtain a restriction on @xmath248 . we then compute @xmath249 using this condition on @xmath248 and calculate the number of values of @xmath249 . this calculation reduces to counting the number of squares @xmath250 modulo @xmath251 or @xmath252 or counting the number of consecutive products @xmath253 modulo @xmath251 or @xmath252 . let @xmath255 . here the homeomorphism condition , @xmath256 , is stronger than the homotopy equivalence condition , @xmath257 . we first set @xmath258 for some @xmath259 . then by ( [ inv2 ] ) @xmath260\nonumber \\ \nonumber \\ & = & \displaystyle -{2^{r-4 } \over 7}\,a\,q\,(m + 2^{r-1}\,a + 2^{r-3}\,a\,q)\end{aligned}\ ] ] recall that @xmath261 hence we need to calculate the number of values of @xmath262 modulo @xmath82 , i.e. @xmath263 \ \mod 7 $ ] . if @xmath265 , we obtain that @xmath266 \mod 7 = \ # \ [ q\,(2^{r-3}\,a\,q + m + 2^{r-1}\,a ) ] \mod 7,\ ] ] as @xmath267 is fixed . hence we are interested in the number of @xmath268 $ ] where @xmath269 is the inverse of @xmath270 modulo @xmath82 . as the inverse of @xmath62 exists modulo @xmath82 , it follows that @xmath271 is always even modulo @xmath82 . hence @xmath272 where @xmath273 . we now count the number of squares modulo @xmath82 and obtain four possible values . the case of @xmath274 reduces to exactly the same calculation as @xmath275 \nonumber \\ \nonumber \\ & = & \!\ ! \displaystyle { 2^{r-4 } \over 7}\,a\,(4 + q)\,(m - 2^{r-3}\,a\,q)\end{aligned}\ ] ] and we can get from ( [ mu1 ] ) to ( [ mu2 ] ) by the translation @xmath276 . for @xmath277 the homotopy equivalence and homeomorphism conditions reduce to @xmath278 and we first set @xmath279 . then @xmath280 \nonumber\\ \nonumber \\ & = & \displaystyle -{1 \over 7}\,a\,q\,(m + 2\,a\,q + 4\,a)\end{aligned}\ ] ] if @xmath281 we obtain one possible value and the first case of ( iii ) in corollary [ difftypes ] follows . if @xmath265 , then @xmath282 \ \mod 7 = \ # \ [ q\,(2\,a\,q + m + 4\,a ) ] \ \mod 7 = \ # \ [ q\,(q + 4\,a^{-1}\,m + 2 ) ] \ \mod 7 = \ # \ [ q\,(q + 2\,(2\,a^{-1}\,m + 1 ) ) ] \ \mod 7 \,.$ ] as in the previous case the calculation reduces to the number of squares modulo @xmath82 and we obtain four possible values . this completes part of the first case of ( i ) in corollary [ difftypes ] . for @xmath283 the values are unchanged as @xmath284 \nonumber\\ \nonumber \\ & = & \displaystyle { 1 \over 7}\,a\,(2 + q)\,(m - 2\,a\,q)\end{aligned}\ ] ] and we can get from ( [ mu3 ] ) to ( [ mu4 ] ) by the translation @xmath285 . for @xmath217 the conditions become @xmath290 and hence we first set @xmath291 . substitution yields @xmath292 if @xmath281 , then @xmath293 for some @xmath294 and @xmath295 . now @xmath296 \ \mod 2 = \ # \ [ q\,(q + 1 + b^{-1}\,m ) ] \ \mod 2 = \ # \ [ q\,(q + 1 + m ) ] \ \mod 2 \,.$ ] if @xmath244 is even , we obtain the number of consecutive products @xmath297 modulo @xmath62 and if @xmath244 is odd , the calculation reduces to counting the number of squares modulo @xmath62 . this implies that @xmath298 if @xmath244 is even with @xmath299 and that @xmath300 if @xmath244 is odd and the same conditions hold . this proves part of the first and second case of ( iv ) . if @xmath265 , then @xmath301 \ \mod 14 = \ # \ [ q\,(q + 1 + a^{-1}\,m ) ] \ \mod 14\,.$ ] hence if @xmath244 is even , then @xmath302 is odd modulo @xmath303 and we calculate the number @xmath297 modulo @xmath303 to obtain four different values . if @xmath244 is odd , then @xmath302 is even modulo @xmath303 and we obtain the number of squares modulo @xmath303 , namely eight different values . this completes part of ( i ) . for @xmath215 the conditions become @xmath306 and @xmath307 if @xmath281 , then @xmath293 and @xmath308 . therefore @xmath309 \ \mod 8 = \ # \ [ q\,(q + 1 + 6\ , b^{-1}\,m ) ] \ \mod 8 = \ # \ [ q\,(q + 1 ) ] \ \mod 8 \,$ ] as @xmath310 is odd modulo @xmath311 . hence we obtain four different values . for @xmath224 the conditions reduce to @xmath290 , i.e. we first set @xmath291 and @xmath319 if @xmath281 , then @xmath293 for some @xmath294 and @xmath320 . therefore @xmath321 \ \mod 2 = \ # \ [ q\,(b\,q + m ) ] \ \mod 2 = \ # \ [ q\,(q + b^{-1}\,m ) ] \ \mod 2 = \ # \ [ q\,(q + m ) ] \ \mod 2 \,$ ] . if @xmath244 is even , we obtain the number of squares modulo @xmath62 , i.e. two different values . if @xmath244 is odd , the calculation reduces to counting the number of consecutive products @xmath322 modulo @xmath62 and we obtain one value . if @xmath265 , then @xmath323 \ \mod 14 = \ # \ [ q\,(q + 2 + a^{-1}\,m ) ] \ \mod 14 \,.$ ] if @xmath244 is even , then @xmath324 is even modulo @xmath303 and we get back to the number of squares modulo @xmath303 , i.e. we obtain eight different values . if @xmath244 is odd , then @xmath324 is odd and the number of @xmath297 modulo @xmath303 leads to @xmath3 different values . m. davis , _ some group actions on homotopy spheres of dimension seven and fifteen _ , amer . j. of math . * 104 * ( 1982 ) 59 - 90 . j. eells and n. kuiper , _ an invariant for certain smooth manifolds _ , annali di math . * 60 * ( 1962 ) 93 - 110 . d. gromoll and w. meyer , _ an exotic sphere with nonnegative sectional curvature _ , ann . of math . * 100 * ( 1974 ) 401 - 406 . k. grove and w. ziller , _ curvature and symmetry of milnor spheres _ , to appear . s.- t. hu , _ homotopy theory _ , academic press ( 1959 ) . i. m. james and j. h. c. whitehead , _ the homotopy theory of spheres bundles over spheres ( ii ) _ , proc . london math . soc . * 5 * ( 1955 ) 148 - 166 . j. milnor , _ on manifolds homeomorphic to the @xmath82-sphere _ , annals of math . * 64 * ( 1956 ) 399 - 405 . i. madsen and j. milgram , _ the classifying spaces for surgery and cobordism of manifolds _ , annals of mathematics studies , princeton university press ( 1979 ) . d. ravenel , _ complex cobordism and stable homotopy groups of spheres _ , academic press ( 1986 ) . i. tamura , _ homeomorphism classification of total spaces of sphere bundles over spheres _ , j. of math . japan * 10 * ( 1958 ) 29 - 43 . h. toda , _ composition methods in homotopy groups of spheres _ , ann . of math . studies no . 49 ( 1962 ) . j. wu , _ on combinatorial description of homotopy groups and the homotopy theory of mod 2 moore spaces _ , ph.d . thesis , department of mathematics , university of rochester , 1995 .	we classify the total spaces of @xmath0-bundles over @xmath1 up to homotopy equivalence , homeomorphism and diffeomorphism . these total spaces have been of interest to both topologists and geometers . they play an important role in the recent work of grove and ziller @xcite , where it is shown that each of these total spaces admits metrics with non - negative sectional curvature .
in this section of the supplementary information , we prove that sampling the distribution @xmath50 given in eq . ( 8) is optimal in that it gives the minimum variance of the born rule estimator of all distributions . our goal is to estimate @xmath89 . we will do so by sampling @xmath52 from some distribution @xmath90 , and compute an estimator given by @xmath91 . the variance @xmath92 of this estimator is given by @xmath93 ^ 2}{p({\boldsymbol{\lambda } } ) } - p^2 \,.\ ] ] to minimize the variance , we choose @xmath90 to minimize the first term in this expression . the distribution @xmath90 that minimizes the variance is @xmath94 . the proof follows directly from the cauchy - schwarz inequality . consider @xmath95 ^ 2}{p({\boldsymbol{\lambda } } ) } \right ) \left ( \sum_{{\boldsymbol{\lambda } } } p({\boldsymbol{\lambda } } ) \right ) \notag \\ & = \sum_{{\boldsymbol{\lambda } } } \frac{[w({\boldsymbol{\lambda}})]^2}{p({\boldsymbol{\lambda}})}\end{aligned}\ ] ] the inequality is saturated for @xmath96 , and therefore this distribution minimizes the variance @xmath92 . with this choice , the variance of the estimator is @xmath97 where @xmath51 . in this section of the supplementary information , we detail ways to exploit an efficiently computable symmetry of the born rule to give a variant of our estimation algorithm . if we replace the quantum circuit and measurement effect with another such that the born rule probability remains the same , then eq . ( 10 ) provides two ( in general ) different estimators for this born rule probability . the rate of convergence of these estimator need not be the same under this symmetry , and so such a variant may provide an advantage . as an example , consider the ` time reversal ' symmetry that exchanges states and measurement effects in a unitary circuit ( with some care taken to appropriately normalise the distributions for states and effects ) . one can define a `` reverse protocol '' which produces a poly precision estimator @xmath98 , provided that the _ total reverse negativity _ of the circuit @xmath99 is polynomially bounded . in general , @xmath100 , as seen from @xmath101 because both @xmath102 and @xmath103 are efficiently computable , one is free to choose the direction of simulation resulting in the faster estimator convergence rate . ( we note that @xmath104 while @xmath105 , which suggests that the reverse protocol would have slower convergence when using a high rank effect . however , in such cases , @xmath106 is in general larger than @xmath107 by a similar factor , cancelling the effect of @xmath42 in the ratio . ) another symmetry of the born rule is the the regrouping of unitaries into different ` elementary ' gates , such as reexpressing @xmath108 as @xmath109 . different groupings can lead to different estimators , as we demonstrate with a simple example using a grouping of two unitaries into one , @xmath110 . we can estimate @xmath111 by sampling trajectories @xmath112 using @xmath113 and @xmath114 , or directly by sampling @xmath115 using @xmath110 as a single step . while both of these methods will produce an unbiased estimator of the born rule , they will not converge at the same rate in general , as a result of the general inequality @xmath116 \neq \frac{{\left\|w_{u}(\lambda_2\|\lambda_0)\right\|}}{{\mathcal{m}}_{u}(\lambda_0 ) } \,.\ ] ] we note that equality holds in the case where @xmath113 and @xmath114 are both nonnegative .	we present a method for estimating the probabilities of outcomes of a quantum circuit using monte carlo sampling techniques applied to a quasiprobability representation . our estimate converges to the true quantum probability at a rate determined by the total negativity in the circuit , using a measure of negativity based on the 1-norm of the quasiprobability . if the negativity grows at most polynomially in the size of the circuit , our estimator converges efficiently . these results highlight the role of negativity as a measure of non - classical resources in quantum computation . estimating the probability of a measurement outcome in a quantum process using only classical methods is a longstanding problem that remains of acute interest today . directly calculating such probabilities using the born rule is inherently inefficient in the size of the quantum system , and efficiently estimating such probabilities for a generic quantum process is expected to be out of reach of classical computers . nonetheless , there are interesting and nontrivial classes of quantum circuits for which we _ can _ efficiently estimate the probabilities of outcomes . the canonical example of such a class is that of stabilizer circuits . such circuits can create highly - entangled states and perform many of the fundamental operations involved in quantum computing ( teleportation , quantum error correction , distillation of magic states ) but the celebrated gottesman - knill theorem allows such circuits to be classically simulated efficiently @xcite . other examples include fermionic linear optics / matchgates @xcite , and some classes of quantum optics @xcite . while these methods may be extended to include bounded numbers of operations outside of the class ( for example , ref . @xcite ) , such extensions generally treat all operations outside of the class on an equal footing ( for example , the cost of adding noisy magic states is the same as adding pure magic states ) and so do not provide any insight into the relative resources of different operations . in this letter , we present a general method for estimating outcome probabilities for quantum circuits using quasiprobability representations . simulation methods based on quasiprobability representations have a long history in physics @xcite , and have recently been used in quantum computation to identify classes of operations that are efficiently simulatable @xcite . our method allows for estimation in circuits wherein the quasiprobabilities may go negative . that is , while making the most efficient use of circuit elements that are represented nonnegatively , it nonetheless provides an unbiased estimator of the true quantum outcome probability regardless of the inclusion of more general elements that are negatively represented . we quantify the performance of this method by providing an upper bound on the rate of convergence of this estimator that scales with a measure of the total amount of negativity in the circuit . _ _ probability estimation.__consider quantum circuits of the following form . the circuit initiates with @xmath0 qudits ( @xmath1-level quantum systems ) in a product state , evolves through a circuit consisting of @xmath2 elementary gates that act nontrivially on at most a fixed number of qudits ( for example , 1- and 2-qudit gates ) , and terminates with a product measurement , i.e. , an independent measurement of each qudit . universal quantum computation can be achieved with circuits of this form . note that we do not include circuits with intermediate measurements and conditional operations based on their outputs ( we return to this consideration in the discussion ) . we aim to estimate the probability of a fixed outcome @xmath3 where @xmath4 denotes the outcome of the measurement on the @xmath5th qudit . ( note that _ estimation _ of the probability of a fixed outcome is distinct from a _ simulation _ as in refs . @xcite , wherein different outcomes are sampled from this distribution . ) a natural benchmark for the precision of an estimator is the precision that can be obtained from sampling the quantum circuit itself . if we had access to a quantum computer that implemented a circuit in this class , then we could use it to estimate the probability of a fixed outcome by computing the observed frequency @xmath6 of outcome @xmath7 over @xmath8 samples . by the hoeffding inequality , @xmath6 will be within @xmath9 of the quantum probability @xmath10 with probability @xmath11 provided the number of samples @xmath12 satisfies @xmath13 this bound implies that for any fixed @xmath14 , the number of samples required to achieve @xmath9 error scales polynomially in @xmath15 . we call estimators satisfying this property _ poly - precision _ estimators . ( we distinguish these from _ exponential - precision _ estimators , defined as estimators for which @xmath12 scales logarithmically in @xmath15 . ) our central results are a classical algorithm that produces a poly precision estimate of a quantum circuit in the above class , and a bound on the efficiency of this algorithm based on a measure of the circuit s negativity in a quasiprobability representation . _ _ quasiprobability representations.__a quasiprobability representation of a qudit over @xmath16 is defined @xcite by a frame @xmath17 and a dual frame @xmath18 , which are ( generally over - complete ) bases for the space of hermitian operators acting on @xmath19 satisfying @xmath20 $ ] for all @xmath21 . the space @xmath16 can be continuous or discrete , and although many quasiprobability representations assume a phase space ( symplectic ) structure on @xmath16 , this is not necessary . we can define quasiprobability distributions on @xmath16 associated with a quantum state @xmath22 , a unitary operator @xmath23 and a measurement effect @xmath24 to be @xmath25 \ , , \notag\\ w_{u}(\lambda'\|\lambda ) & = { \text{tr}}(f(\lambda ' ) u g(\lambda ) u^\dag)\ , , \notag\\ w(e\|\lambda ) & = { \text{tr}}[e g(\lambda ) ] \ , . \label{eq : quasidist}\end{aligned}\ ] ] tensor products of these dual frames gives a dual frame for the product space , and so these definitions extend in the obvious way from a tensor product @xmath26 of @xmath0 qudits to distributions on a phase space @xmath27 . the distribution @xmath28 is real - valued and satisfies @xmath29 , much like a probability distribution , if the frame is normalized using @xmath30 . similarly , the distributions @xmath31 and @xmath32 are normalized like corresponding conditional probabilities . the born rule @xmath33 , which gives the quantum probability for a measurement outcome given the state and process , is reproduced in the quasiprobability representation as would be expected in a probabilistic theory , by @xmath34 this equation follows from the born rule using the definition of the dual frames . importantly , the distributions of a quasiprobability representation will generally take on negative values , and so can not be directly interpreted as probability distributions . the 1-norm of a quasiprobability distribution provides a natural measure of the amount of negativity , i.e. , how much it deviates from a true probability distribution . we define the _ negativity _ @xmath35 of a state @xmath22 as the 1-norm of its quasiprobability representation , @xmath36 ( the _ mana _ of a state using the discrete wigner representation was introduced in ref . @xcite as a measure to bound the resources required for magic state distillation , and defined as the logarithm of the negativity used here . ) analogously , we define the _ negativity _ @xmath37 of a measurement effect @xmath24 to be @xmath38 and the _ point - negativity _ @xmath39 and _ negativity _ @xmath40 of a unitary @xmath23 to be @xmath41 respectively . the negativities for states , unitaries and effects are lower - bounded by 1 , 1 and @xmath42 respectively , with equality if and only if the quasiprobability representation is nonnegative . these negativities will serve as a measure of the cost of each circuit element in our estimator . _ _ estimation procedure.__a quasiprobability representation provides an interpretation of the born rule as the expectation value of a stochastic process . specifically , if the quasiprobability representation of all elements in the circuit are nonnegative , then one may interpret the born rule of eq . as the expected probability of the measurement outcome averaged over a set of trajectories through phase space @xcite . this stochastic interpretation no longer holds if the quasi - probability representation for any of the input states , gates , or measurements is negative . a standard perspective is that , in a quantum description , different trajectories in phase space can be assigned negative weights and can interfere with each other @xcite . monte carlo sampling techniques may still be used , but the key problem is to identify an appropriate distribution to sample trajectories @xmath43 through phase space , where @xmath44 is associated with the preparation and @xmath45 and @xmath46 are associated with the @xmath47th unitary . . becomes @xmath48 with @xmath49 w_{\rho}(\lambda_0 ) \,.\ ] ] using an approach reminiscent of quantum monte carlo methods for fermion systems , we could sample trajectories from a true ( nonnegative ) probability distribution obtained from the absolute value of the quasiprobability , keeping track of the sign of the sampled trajectory . consider the distribution of trajectories given by @xmath50 where @xmath51 measures the negativity of the entire circuit , and we regard @xmath52 as a realization of a random variable @xmath21 . an estimate based on a single realization @xmath52 is given by @xmath53 $ ] , where @xmath54 = \pm 1 $ ] depending on the sign of the input . the expected value of this estimate gives the desired born rule probability @xmath55 \|w({\boldsymbol{\lambda}})\| \\ = \sum_{{\boldsymbol{\lambda } } } w({\boldsymbol{\lambda } } ) = { { \rm pr}}(e\|\rho , u)\ , . \label{eq : unbiased}\end{gathered}\ ] ] note that this estimator minimizes the range of @xmath21 @xcite , and so provides the best bound on the number of samples required to obtain a fixed precision using the hoeffding inequality . sampling from the distribution is also the optimal estimator over the space of trajectories in that it has the smallest variance ( see appendix ) . unfortunately , without any additional structure , this estimator will in general be impractical for two reasons : there is no known efficient method to compute @xmath56 , and sampling from the distribution will in general be inefficient in @xmath0 . to develop an efficient procedure , we sample trajectories @xmath52 following a markov chain , using ( true ) probabilities and conditional probabilities at each timestep . consider an input product state @xmath57 , which has an efficient description @xmath28 in the quasi - probability representation which may be negative . we sample the initial point of the trajectory @xmath44 from the modified distribution @xmath58 . we construct a full trajectory @xmath52 by sampling @xmath59 at each timestep @xmath60 in the circuit from the conditional distribution @xmath61 given by the unitary gate @xmath62 . if the unitary @xmath62 at each timestep @xmath47 of the circuit has an efficient description in the quasi - probability representation , for example it consists of quantum gates acting on a fixed number of systems , then this distribution can be sampled efficiently . we note that this efficiency comes at a cost , as trajectories are no longer sampled from the optimal distribution . an estimate @xmath63 based on a single trajectory @xmath52 of our markov chain protocol is given by @xmath64 \\ \times \textstyle\prod_{l=1}^l \bigl [ { \mathcal{m}}_{u_l}(\lambda_{l-1}){\rm sign}[w_{u_l}(\lambda_l\|\lambda_{l-1 } ) ] \bigr ] w(e\|\lambda_{l } ) \,.\end{gathered}\ ] ] unlike @xmath65 , this estimate is guaranteed to be an efficiently computable function of the sampled path @xmath52 . we note that @xmath63 can lie outside the unit interval , but nonetheless gives an unbiased estimate of the born rule probability , @xmath66 , precisely as in eq . . further , we note that @xmath63 lies in the interval @xmath67 $ ] , where we have defined @xmath68 to be the _ total forward negativity bound _ of the circuit : @xmath69 let @xmath70 be the average of @xmath63 over @xmath8 independent samples of @xmath52 . using the boundedness and unbiasedness properties of @xmath63 , the hoeffding inequality yields an upper bound on the rate of convergence of the average @xmath70 . specifically , @xmath70 will be within @xmath9 of the quantum probability @xmath71 with probability @xmath11 if a total of @xmath72 samples are taken . consequently , if the total forward negativity bound @xmath68 grows at most polynomially with @xmath0 , then our protocol gives an efficient estimate @xmath70 of the quantum probability @xmath73 to within @xmath74 , with an exponentially small failure probability . that is , for circuits with a polynomially - bounded total forward negativity bound , @xmath70 is a poly - precision estimator of the born rule probability and we can sample @xmath70 efficiently in @xmath0 . we note that the total forward negativity bound @xmath75 of is insensitive to the measurement negativity @xmath37 , instead depending only on @xmath76 . any efficiently computable symmetry of the born rule can be used to give a variant on the procedure defined above . the rate of convergence of the estimator need not be symmetric under these born rule symmetries , and so such a variant may provide an advantage . two examples of such symmetries the time reversal symmetry that exchanges states and measurement effects in a unitary circuit , and the regrouping of unitaries into different elementary gates are explored in the appendix . in particular , a variant procedure is presented for which the total negativity bound is insensitive to the negativity of the initial state @xmath77 . _ example : estimation with the discrete wigner function.__the odd-@xmath1 qudit stabilizer subtheory and the associated discrete wigner function provide a canonical example for demonstrating the use of our algorithm ; see also ref . @xcite . using this discrete wigner representation for our estimation algorithm , the nonnegativity of the stabilizer subtheory @xcite ensures that stabilizer states , gates , and rank-1 measurements have negativity @xmath78 and so are `` free '' resources . moreover , due to the existence of nonnegatively represented operations that are not in the stabilizer polytope @xcite , our approach is efficient on a strictly larger set of circuits than those of ref . @xcite . circuits with operations possessing negativity strictly greater than 1 , such as magic states and non - clifford gates , can still be estimated but now come at a cost . provided the total negativity bound grows at most polynomially in @xmath0 , our protocol provides an efficient estimator . as an example , consider a circuit with an input state given by a product state of @xmath79 qutrit magic states @xmath80 , with @xmath81 , together with stabilizer @xmath82 states in a 100-qutrit random clifford circuit , and estimate the probability of measuring @xmath82 on the first qutrit of the output . the total forward negativity bound of this circuit scales exponentially in @xmath79 and consequently the number of samples required to guarantee a fixed precision scales exponentially in @xmath79 by eq . . the results of our numerical simulations , shown in fig . [ fig : randomcliffords ] , indicate that our estimator does indeed converge with an appropriately chosen number of samples . moreover , while the true precision of @xmath70 in our simulations is often orders of magnitude better than the target precision , there are circuits that come close to saturating the target precision , suggesting that our bound can not be substantially improved without further detailed knowledge of the circuit . between the estimated probability and the true probability of the outcome @xmath83 for the first qutrit as a function of the number of magic states @xmath79 . each data point represents a random 100-qutrit clifford circuit with the non - magic states initialized to the @xmath82 state . the number of samples @xmath84 was chosen using eq . with target precision @xmath85 ( indicated by the solid line ) with 95% confidence ( @xmath86 ) , so the number of samples increases exponentially with @xmath79 ( color scale ) . ] _ _ discussion.__our results highlight the role of the total negativity of a circuit as a resource required for a quantum computer to outperform any classical computer . in particular , any circuit element that is represented nonnegatively does not contribute to the total negativity bound and can be viewed as a `` free '' resource within the algorithm . other circuit elements have an associated cost quantified by their negativity , _ unless _ they appear at the final timestep of the algorithm . this latter observation motivates us to exploit the time - reversal and other symmetries of the born rule , seeking to minimize the total forward ( or reverse ) negativity bound . in particular , one could seek equivalent circuits wherein negative operations can be replaced with nonnegative ones by using negative initial states or measurements via gate teleportation . by choosing the forward or reverse procedure as appropriate , the efficiency can be made insensitive to the negativity of these initial states or measurements . it also motivates us to identify quasiprobability representations in which many of the circuit elements of interest are represented nonnegatively . interesting and relevant examples abound , beyond the well - studied qudit discrete wigner function . discrete wigner functions for qubits ( @xmath87 ) can be defined for which all stabilizer states with real coefficients ( rebits ) and all css - preserving unitaries are nonnegatively represented @xcite . the range of quasiprobability representations introduced in ref . @xcite represent discrete subgroups of @xmath88 on a single qubit nonnegatively , but have no nonnegative entangling gates ; as such representations can represent certain non - stabilizer single - qubit states nonnegatively , they may be useful for estimation in circuits for gate synthesis . there is also flexibility in how a quasiprobability representation is defined . for example , a given quasiprobability representation can be modified to describe more states nonnegatively at the expense of a decreasing set of nonnegative measurements , and vice versa , by exploiting the structure of the dual frames . overcomplete frames provides freedom in the choice of dual frame , and the negativity of unitaries and measurements will depend on this choice . as the dual frames formalism itself captures the relationship between quantum states and measurements , there is also freedom in the definition of the quasiprobability representation of unitaries beyond that given by eq . . finally , again using the freedom in the choice of dual for overcomplete frames , it is possible to switch between frames throughout a single circuit . these freedoms can be used to minimize the total negativity bound of the circuit , allowing more efficient estimators . our procedure can be applied to infinite - dimensional hilbert spaces using any of the range of quasiprobability representations with continuous phase spaces developed in the study of quantum optics , by performing an appropriate discretization as in ref . @xcite . in this case , the negativity of distributions is quantified by integrating the absolute value of the distributions over the phase space , and is directly related to the _ volume _ of negativity @xcite . we note that the resulting estimator can be applied to quantum optics experiments including states and measurements with negative wigner function , such as photon number fock states , and so may provide additional insight into the classical simulation cost of boson sampling @xcite . while there exist means to efficiently estimate the outcome probability of a specific linear optics circuit with fock state input and measurement @xcite , our estimation procedure extends these results by providing a general method for estimating outcome probabilities of such linear optical circuits for any input and output together with a bound on the efficiency of this estimation based on the volume of negativity of these states . in addition , our estimation can easily incorporate squeezing , as well as the loss and noise mechanisms common to linear optics experiments . there are two natural ways to extend our results to circuits that include intermediate measurements and conditional operations based on them . first , one could replace the measurement and conditional operation with a coherently - controlled operation , and delay the measurement to the end . we note that such controlled operations can be negative , even if the measurement and classically - controlled operation are both nonnegative . second , our algorithm can be used to directly estimate the probabilities of the intermediate measurements and to sample from them . in this case , the required precision is exponential in the number of intermediate measurements in order to calculate conditional probabilities for subsequent use in the algorithm . thus , in general , both approaches require resources that are exponential in the number of intermediate measurements . finally , our estimation procedure provides insight into the study of operationally meaningful measures of non - classical resources in quantum computation . negativity in a quasiprobability representation has long been used as an indicator of quantum behaviour , but only recently has it been quantified as a resource for quantum computation @xcite . our results provide a related operational meaning of this resource : as a measure that bounds the efficiency of a classical estimation of probabilities . the authors are grateful to j. emerson , c. ferrie , s. flammia , r. jozsa and a. krishna for helpful discussions . this work is supported by the arc via the centre of excellence in engineered quantum systems ( equs ) project number ce110001013 and by the u.s . army research office through grant w911nf-14 - 1 - 0103 . 99 s. aaronson and d. gottesman , * 70 * , 052328 ( 2004 ) . l. g. valiant , siam j. comput . * 31 * , 1229 ( 2002 ) . b. m. terhal and d. p. divincenzo , * 65 * , 032325 ( 2002 ) . s. d. bartlett , b. c. sanders , s. l. braunstein , and k. nemoto , phys . rev . lett . * 88 * , 097904 ( 2002 ) . v. veitch , n. wiebe , c. ferrie , and j. emerson , new j. phys . * 15 * , 013037 ( 2013 ) . l. gurvits , in _ mathematical foundations of computer science _ , lecture notes in computer science * 3618 * , 447 - 458 ( 2005 ) . c. gardiner and p. zoller , _ quantum noise : a handbook of markovian and non - markovian quantum stochastic methods with applications to quantum optics _ ( springer - verlag , berlin , 3rd ed . , 2004 ) . v. veitch , c. ferrie , d. gross , and j. emerson , new j. phys . * 14 * , 113011 ( 2012 ) . a. mari and j. eisert , phys . rev . lett . * 109 * , 230503 ( 2012 ) . d. stahlke , phys . rev . a * 90 * , 022302 ( 2014 ) . c. ferrie and j. emerson , new j. phys . * 11 * , 063040 ( 2009 ) . c. ferrie , rep . . phys . * 74 * , 116001 ( 2011 ) . v. veitch , s. a. h. mousavian , d. gottesman and j. emerson , new j. phys . * 16 * , 013009 ( 2014 ) . c. cormick , e. f. galvao , d. gottesman , j .- p . paz , and a. o. pittenger , phys . rev . a * 73 * , 012301 ( 2006 ) . d. gross , j. math . phys . * 47 * , 122107 ( 2006 ) . n. delfosse , p. allard guerin , j. bian , and r. raussendorf , phys . rev . x * 5 * , 021003 ( 2015 ) . j. j. wallman and s. d. bartlett , * 85 * , 062121 ( 2012 ) . a. kenfack and k. yczkowski , j. opt . b : quantum semiclass . * 6 * , 396 ( 2004 ) . s. aaronson and a. arkhipov , proc . acm symposium on theory of computing , san jose , ca pp . 333342 ( 2011 ) .
sgr a * may be a massive ( @xmath2 ) point - like object dominating the gravitational potential in the inner @xmath3 pc region of the galaxy . this inference is based on the large proper motion of nearby stars ( @xcite ; @xcite ; @xcite ) , the spectrum of sgr a * ( @xcite ) , its low proper motion ( @xmath4 ; @xcite ) , and its unique location ( @xcite ) . the gaseous flows in this region are themselves rather complex , and key constituents appear to be the cluster of mass - losing , blue , luminous stars comprising the irs 16 assemblage , which is located within several arc seconds ( @xmath5 0.04 pc in projection at the distance to the galactic center ) from sgr a. measurements of high outflow velocities associated with ir sources in sgr a west ( @xcite ) and in irs 16 ( @xcite ) , the @xmath6 emission in the circumnuclear disk ( cnd ) from molecular gas being shocked by a nuclear mass outflow ( @xcite ) , broad br@xmath7 , br@xmath8 and he i emission lines ( @xcite ; @xcite ; @xcite ) , and radio continuum observations of irs 7 ( @xcite ) , provide clear evidence of a hypersonic wind , with velocity @xmath9 , a number density @xmath10 , and a total mass loss rate @xmath11 , pervading the inner parsec of the galaxy . many of sgr a s radiative characteristics may be due to its accretion of the irs 16 wind . in the classical bondi - hoyle ( bh ) scenario ( @xcite ) , the mass accretion rate for a uniform hypersonic flow is @xmath12 , in terms of the accretion radius @xmath13 . at the galactic center , with @xmath14 and @xmath15 , we would therefore expect an accretion rate @xmath16 onto the black hole , with a capture radius @xmath17 . since this accretion rate is sub - eddington for a one million solar mass object , the accreting gas is mostly unimpeded by the escaping radiation field and is thus essentially in hydrodynamic free - fall starting at @xmath18 . our initial numerical simulations of this process , assuming a highly simplistic uniform flow past a point mass ( @xcite ; @xcite ) have verified these expectations . on the other hand , the nature of bondi - hoyle accretion onto a point like object also presents somewhat of a challenge in understanding what happens to the gas as it settles down into a planar configuration close to the event horizon . fluctuations beyond the bow shock ( located at @xmath19 ) produce a transient accretion of net angular momentum that ought to result in the formation of a temporary ( albeit small ) disk . the circularization radius is @xmath20 , where @xmath21 is the schwarzschild radius and @xmath22 is the specific angular momentum in units of @xmath23 . our simulations of the bh accretion from a uniform flow suggest that in this configuration @xmath24 . more realistically , the inflow itself carries angular momentum , so that the formation of a disk - like structure at small radii ( i.e. , @xmath25 ) may be difficult to avoid ( @xcite ) . however , the observations do not appear to favor the presence of a _ standard _ @xmath7-disk at small radii ( @xcite ) . the current upper limits on the infrared flux from sgr a * ( @xcite ) suggest that either ( 1 ) the circularized flow does not form an @xmath7-disk , but rather advects most of its dissipated energy through the event horizon ( e.g. , @xcite ) , ( 2 ) the bondi - hoyle flow merges into a massive , fossilized disk , storing most of the deposited matter at large radii ( @xcite ) , or ( 3 ) sgr a * is not a point - like object ( @xcite ) . added to this is the fact that in reality the flow past sgr a * is not likely to be uniform . for example , one might expect many shocks to form as a result of wind - wind collisions within the irs 16 comples , even before the plasma reaches @xmath18 . with this consequent loss of bulk kinetic energy , it would not be surprising to see the black hole accrete at an even larger rate than in the uniform case . the implications for the spectral characteristics of sgr a * , and thus its nature , are significant . we have therefore undertaken the task of simulating the bh accretion from the spherical winds of a distribution of 10 individual point sources located at an average distance of a few @xmath18 from the central object . as we shall see below , the accretion rate depends not only on the distance of the mass - losing star cluster from the accretor but also on the relative spatial distribution of the sources . as suggested by related work involving linear gradients ( @xcite ) , the average value of @xmath22 is also larger than that of the uniform case , exhibiting large temporal fluctuations , but still not as large as one might expect . in this _ letter _ we present the results of the first 3d hydrodynamical calculations of multiple - source stellar winds accreted by a point mass . although we here attempt to model the wind source more realistically than in previous work , we shall still focus on the hydrodynamical aspects of the flow and thus we do not include heating or cooling ( see , e.g. , @xcite ) . nor do we include any relativistic effects . instead , we assume the medium to be an unmagnetized , polytropic gas with pressure @xmath26 where @xmath27 , and @xmath28 are the mass and internal energy densities , respectively , and @xmath29 is the adiabatic index . this value of @xmath8 was chosen so that the sonic point @xmath30 occurs outside the accretor , which typically is modeled with a radius @xmath31 . some earlier work ( e.g. , @xcite ) found that varying @xmath8 affects the flow pattern significantly , particularly at radii much smaller than @xmath18 , while others ( e.g. , @xcite ) found that it does not , particularly at the larger radii considered here . we use the numerical algorithm zeus-3d , a general purpose code for mhd fluids developed at ncsa ( @xcite ) . the calculations are carried out within a cubical domain of solution with @xmath32 active zones geometrically scaled so that the central zones are 1/32 times the size of the outermost zones , closely mimicking the `` multiply nested grids '' arrangement used by other researchers ( e.g. , @xcite ) . this allows for maximal resolution of the accretor within the computer memory limits available while sufficiently resolving the wind sources and minimizing zone - to - zone boundary effects . the total volume is @xmath33 or @xmath34 using a point mass @xmath35 located at the origin and a mach 10 galactic center wind with @xmath36 . the initial density is set to a small value and the velocity is set to zero . the internal energy density is chosen such that the temperature is @xmath37 . free outflow conditions are imposed on the outermost zones and each time step is determined by the courant condition with a courant number of 0.5 . the 10 identical stellar wind sources are modeled by forcing the velocity in 10 subregions of @xmath38 zones to be constant ( at @xmath39 ) while the densities in these subvolumes are set so that the total mass flow into the volume of solution , @xmath40 , is 3@xmath41 . recent observations ( @xcite ) suggest that the irs 16 wind is somewhat colder than the temperature assumed here , with a mach number of @xmath42 30 , and that it may originate from more than 15 sources of varying strength rather than 10 sources of equal strength . this level of detail will be addressed in future work . to gauge the dependence of our results on the source configuration , we here simulate the bh process using two different stellar distributions . for numerical reasons , the sources need to be placed a minimum of 8 zones ( @xmath43 ) apart so no attempt is made here to imitate exactly the distribution within the irs 16 cluster . the _ average _ location for the sources in run 1 is @xmath44 ( with the accretor located at the origin ) and @xmath45 for run 2 . for run 1 , the sources are distributed fairly randomly with x , y , and z being allowed to vary by @xmath46 of the average location . for run 2 , x and y are allowed to be as much as 5 @xmath18 from the average location while z can only vary by @xmath47 . these distributions are chosen to represent the extremes of a spherical stellar distribution versus a more or less planar one . for comparison , irs 16 nw and irs 16 ne , two bright members of the irs 16 cluster , are @xmath48 and @xmath49 away ( in projection ) from sgr a * ( @xcite ) . each simulation is evolved for 2000 years . the wind crossing time is @xmath42 400 years and equilibrium , the point at which the original gas is swept out of the volume of solution and the mass accretion rate stabilizes , appears to be reached within 2 crossing times , or @xmath42 800 years . figure 1 shows a logarithmic negative grey scale image of the density profile for a slice running through the center of the accretor , for run 1 taken 2000 years after the winds are `` turned on '' . the image is 6 @xmath18 on a side with the @xmath50 radius accretor located at the center . one stellar source can be clearly seen to the upper right of the accretor ; the other 9 sources are either not in the plane of the slice or are off to the right . the image spans 4 magnitudes of density with white being @xmath51 and black being @xmath52 . figure 2 is a similar image for run 2 . note the large density fluctuations across the overall region . also , the flow pattern is clearly different for the two configurations . however , once the stellar winds have cleared the region of the original low density gas , both simulations point to an overall average density ( @xmath53 ) in agreement with observations . a substantial difference between these point source simulations and previous uniform calculations is that although there are transient fragmentary shocks extending from the accretor boundary ( see figure 1 ) , there is no large - scale bow shock structure . there are , however , bow shocks around some of the stellar sources as their wind impacts the stronger accumulated wind of the other sources ( see the source in figure 1 ) , and , for run 2 , a semi - stable partial bow shock appears in front of the accretor ( see figure 2 ) . the wind - wind collisions convert kinetic energy into thermal energy with the result that a larger fraction of the gas is captured by the central engine . although this effect is likely to be sensitive to the actual stellar distribution , a larger density of point sources should produce more wind - wind collisions and perhaps further raise the accretion rate . for a wind originating from a single point source located at a distance @xmath54 ( @xmath18 ) from the accretor , @xmath55 is less than @xmath56 due to the divergent flow . specifically , for @xmath57 , @xmath58 . however , wind - wind collisions from multiple sources reduce the effective @xmath39 , thereby increasing the effective @xmath18 so that all of the shocked gas will then accrete and more than compensate for the geometric losses . in figure 3 we present the mass accretion rate , @xmath55 , and the accreted specific angular momentum , @xmath22 , versus time for run 1 , starting 2 crossing times ( @xmath42 800 years ) after the winds are `` turned on '' . figure 4 presents the results for run 2 . the average values for the mass accretion rate once the systems have reached equilibrium are @xmath59 for run 1 and @xmath60 for run 2 . the mass accretion rate shows high frequency temporal fluctuations ( with a period of @xmath61 ) due to the finite numerical resolution of the simulations . the low frequency aperiodic variations ( on the order of @xmath62 in amplitude ) reflect the time dependent nature of the flow . thus , the mass accretion rate onto the central object , and consequently the emission arising from within the accretor boundary , is expected to vary by @xmath63 ( since in some models @xmath64 may vary by as much as @xmath65 ) over the corresponding time scale of @xmath66 years , even though the mass flux from the stellar sources remains constant . similarly , for run 1 , the accreted @xmath22 can vary by @xmath67 over @xmath68 200 years with an average equilibrium value of @xmath69 . for run 2 , @xmath70 with smaller amplitude long term variability . since previous uniform simulations resulted in @xmath71 , it appears that even with a large amount of angular momentum present in the wind , relatively little specific angular momentum is accreted . this is understandable since clumps of gas with a high specific angular momentum do not penetrate within 1 @xmath18 . the variability in the sign of the components of @xmath22 suggests that if an accretion disk forms at all , it dissolves , and reforms ( perhaps ) with an opposite sense of spin on a time scale of @xmath72 years . a variety of accretion scenarios for sgr a * have been proposed over the years ( @xcite ; @xcite ; @xcite ; @xcite ) , each with its own restrictions on @xmath55 and @xmath22 . while the accreted specific angular momentum determined in the present simulations is an order of magnitude too small to support the fossil disk scenario ( since then the energy liberated as the wind impacts the fossil disk should be visible ; @xcite ) , it is still large enough that any standard @xmath7-disk would be easily detectable ( @xcite ) . the advection dominated disk scenario ( @xcite ) , while permitting a large range of values for @xmath22 , requires an accretion rate of @xmath73 or roughly @xmath74 , which is 10 - 20 times smaller than the value derived here . in addition , our @xmath55 is more than @xmath75 times larger than that permitted by the `` mono - energetic '' electron model of sgr a * ( @xcite ) . it appears that additional work is needed to reconcile disk models with the fact that the observed multiple wind sources result in a large mass accretion rate onto the central engine , if its mass is @xmath76 . in view of this , it may not be unreasonable to conjecture that in fact no flattened disk actually forms in sgr a * , but rather that the excess angular momentum is dissipated in a quasi - spherical configuration and that the thermalized energy is then advected inwards through the event horizon before the gas settles onto a plane ( @xcite ) . these simulations suggest that the @xmath77 region of the galaxy , centered on the wind sources , is swept clear of gas , leaving a hot , low density gas filling the central cavity . this is consistent with observations of the region within the cnd ( @xcite ) , and may be acting in concert with other mechanisms to produce the sharp inner edge of the cnd ( e.g. , the abrupt change in gravitational potential ; @xcite ) . additionally , a tongue of hot , dense gas has been observed that connects members of the irs 16 cluster to sgr a * ( @xcite ) . it is worthwhile noting that the images in figures 1 and 2 show ridges of dense gas connecting the sources in the figures to the accretor . this work was partially supported by nasa under grants nagw-2518 and ngt-51637 , and utilized the origin 2000 computer system in friendly user mode at the national center for super computing applications , university of illinois at urbana - champaign . beckert , t. , duschl , w. , mezger , p. , & zylka r. 1996307450 falcke , h. & heinrich , o. 1994292430 genzel , r. , eckart , a. , ott , t. & eisenhauer , f. 1997 , _ m.n.r.a.s . _ , in press . najarro , f. , hillier , d. , kudritzki , r. , krabbe , a. , genzel , r. , lutz , d. , drapatz , s. , & geballe , t. 1994285573 ruffert , m. 1996311817 ruffert , m , & anzer , u. 1995 295108	spectral and kinematic studies suggest that the nonthermal radio source sgr a * , located at the center of the milky way , is a supermassive compact object with a mass @xmath0 . winds from nearby stars , located @xmath1 pc to the east of sgr a * , should , in the absence of any outflow from the putative black hole itself , be accreting onto this object . we report the results of the first 3d bondi - hoyle hydrodynamical numerical simulations of this process under the assumption that the galactic center wind is generated by several different point sources ( here assumed to be 10 pseudo - randomly placed stars ) . our results show that the accretion rate onto the central object can be higher than in the case of a uniform flow since wind - wind shocks dissipate some of the bulk kinetic energy and lead to a higher capture rate for the gas . however , even for this highly non - uniform medium , most of the accreting gas carries with it a relatively low level of specific angular momentum , though large transient fluctuations can occur . additionally , the post - bow - shock focusing of the gas can be substantially different than that for a uniform flow , but it depends strongly on the stellar spatial distribution . we discuss how this affects the morphology of the gas in the inner 0.15 pc of the galaxy and the consequences for accretion disk models of sgr a. 3gcm^-3 3cm^-3 # 1#1 #1#2#3#1 , a&a , # 2 * , # 3 . # 1#2#3#1 , a&a _ ( letters ) _ , * # 2 * , # 3 . # 1#2#3#1 , a&as , # 2 , # 3 # 1#2#3#1 , aj , # 2 , # 3 # 1#2#3#1 , _ apj _ , * # 2 * , # 3 . # 1#2#3#1 , _ apj ( letters ) _ , * # 2 * , # 3 . # 1#2#3#1 , apjs , # 2 , # 3 # 1#2#3#1 , ara&a , # 2 , # 3 # 1#2#3#1 , baas , # 2 , # 3 # 1#2#3#1 , icarus , # 2 , # 3 # 1#2#3#1 , _ mnras _ , * # 2 * , # 3 . # 1#2#3#1 , _ nature _ , * # 2 * , # 3 . # 1#2#3#1 , pasj , # 2 , # 3 # 1#2#3#1 , pasp , # 2 , # 3 # 1#2#3#1 , qjras , # 2 , # 3 # 1#2#3#1 , science , # 2 , # 3 # 1#2#3#1 , soviet astr . , # 2 , # 3 # 1#2#3#1 , soviet astr . lett . , # 2 , # 3 # 1#2#3#4#1 , # 2 , # 3 , # 4
electron correlation effects in narrow d- or f - bands@xcite are manifested in a variety of remarkable macroscopic phenomena , such as metal - insulator transition in transition - metal compounds,@xcite magnetism and high - temperature superconductivity in cuprates,@xcite colossal magnetoresistance in manganates , and non fermi - liquid behavior in transition - metal oxides,@xcite high - t@xmath4 superconductors,@xcite and heavy - fermion metals.@xcite incorporating electronic correlation in essence , the prototypical hubbard model , theoretically studied first in some detail by hubbard,@xcite has been therefore intensively studied in recent years in these contexts . while much attention has also been devoted to the doped region away from half - filling,@xcite motivated by the discovery of high - t@xmath4 superconductivity in the doped cuprates,@xcite the discussion in the following is mainly limited to the half - filled case , focusing only on the magnetic and metal - insulator aspects of the phase diagram . even at half - filling the hubbard model in three dimensions has a fairly rich phase diagram in the interaction - temperature ( @xmath5 ) space , consisting of antiferromagnetic insulating ( afi ) , paramagnetic insulating ( pi ) , and paramagnetic metallic ( pm ) phases.@xcite in addition , in the absence of fermi - surface nesting ( for instance due to next - nearest - neighbor ( nnn ) hopping on a bipartite lattice ) a gapless antiferromagnetic metallic ( afm ) phase also appears possible.@xcite the widely different nature of the constituent phases ( metallic and insulating , magnetic and nonmagnetic ) has made it exceedingly difficult for the entire phase diagram to be studied within a single theoretical framework , and over the years various approaches have been developed to study different parts . the brief account given below of some of the recent developments will also serve to highlight the different physical processes of importance in different sections of the phase diagram . the study of magnetism in correlated electron systems goes back quite some time , to the theory of spin - waves in the band model of ferromagnetic metals.@xcite the correlation effect is implicit in the spin - wave excitation , which can be regarded as a bound state of an electron of one spin with a hole of opposite spin . propelled by the discovery of nearly two - dimensional antiferromagnetism in @xmath6,@xcite the intense activity in recent years in low - dimensional magnetism and spin - fluctuation effects , and more recently in three - dimensional magnetism , has led to significant progress . starting with the afi phase , a major element in this progress has been the unified understanding of this phase in the whole @xmath7 range within a many - body - theoretical framework , thus providing a bridge between the weak - coupling slater spin - density - wave limit and the strong - coupling heisenberg limit of localized spins . for a generalized hubbard model with multiple degenerate orbitals per site , it has been shown diagrammatically that within a systematic inverse - degeneracy expansion , which preserves the spin - rotational symmetry and hence the goldstone mode order by order , identical results are obtained in the strong coupling limit ( @xmath8 ) for all quantum corrections , order by order , as from the linear spin - wave analysis of the equivalent quantum heisenberg antiferromagnet ( qhaf).@xcite various quantities characterizing the afi phase have been evaluated in the whole @xmath7 range , both in two and three dimensions , and shown to interpolate properly between the weak and strong coupling limits . these include the spin - wave ( magnon ) velocity,@xcite magnon energy and density of states,@xcite sublattice magnetization @xmath9,@xcite and the local transverse spin fluctuation magnitude @xmath10.@xcite the rapid suppression ( with increasing @xmath1 ) in the contribution of particle - hole excitations across the charge gap , and the rapid rise in that of the low - energy collective ( magnon ) excitations indicates that an effective spin picture is approximately valid down to surprisingly low @xmath1 values.@xcite for the three - dimensional antiferromagnet , the @xmath1-dependence of the nel temperature @xmath11 , which determines the magnetic phase boundary between the ordered afi phase and the spin disordered pi phase , has also been quantitatively studied in the whole @xmath7 range . a variety of techniques have been employed including functional integral formalism , mainly within the static approximation,@xcite quantum monte carlo methods,@xcite dynamical mean - field theory,@xcite which becomes exact in the limit of infinite dimensions,@xcite onsager reaction field theory,@xcite self - consistent spin - fluctuation theory,@xcite etc . except for the last two , all share the common feature of yielding a @xmath11 which approaches , in the strong coupling limit , the mean - field - theory ( mft ) result for the equivalent spin-1/2 qhaf . this is due to the neglect of the long - wavelength , low - energy magnon excitations , which reduce @xmath11 to nearly 70 % of the mft value within the spin - fluctuation theory,@xcite in excellent agreement with recent qmc results for the qhaf.@xcite the @xmath1-dependence of @xmath11 has been shown to be very closely related to that of the maximum magnon energy @xmath12 which sets the magnon energy scale.@xcite starting from the strong coupling limit and going down in interaction strength , @xmath12 initially increases as @xmath13 , the exchange coupling . however , as the weak coupling limit is approached the falling charge gap ( @xmath14 ) compresses the magnon spectrum , and when @xmath15 the energy scale @xmath12 turns over and starts falling , eventually approaching @xmath14 in the weak coupling limit . we first consider traversing up in temperature , starting in the ordered afi phase in the intermediate coupling regime . with increasing temperature thermal excitation of the low - energy magnons enhances the transverse spin fluctuation @xmath16 , thereby reducing the af order parameter , until at @xmath17 , the sublattice magnetization @xmath18 vanishes and the staggered susceptibility diverges , marking the onset of the pi phase . with further increase in temperature beyond @xmath19 , since no further reduction in the sublattice magnetization is possible , a low - momentum cutoff must be introduced in the sum over magnon modes.@xcite the corresponding length scale sets an upper limit to the magnon wavelength , and therefore determines the spin correlation length @xmath20 such that af order exists only upto length scale @xmath20 . the pi phase is thus characterized by an energy gap , a spin correlation length , and local magnetization @xmath21 whose direction slowly fluctuates in space and time . the twisting between neighboring spins relative to the af alignment increases as @xmath22 with temperature , and this twist activates the hopping term at first order , resulting in a broadening of the two bands , as in the spiral phase studied earlier in the context of the doped hubbard model.@xcite the electronic bandwidth in the low - temperature ordered phase , which is of the order of the exchange energy @xmath13 due to the second - order virtual hopping process , is enhanced due to this twist by order @xmath23 , thus providing a mechanism for coupling between the magnetic ( dis)ordering and the electronic spectrum . for @xmath24 , the correlation length @xmath25 , and the bands are broadened to their maximum extent of order @xmath26 , the free electron half - bandwidth . finally when @xmath27 , there is a crossover into a semiconducting phase . we now consider another sequence traversing down in interaction strength , keeping the temperature fixed . in this case charge fluctuations , which are strongly suppressed in the strong coupling limit , progressively become more important , and when @xmath28 the probability of a site being doubly occupied or vacant becomes appreciable , and the local moment defined by @xmath29 becomes insignificant . double occupancy and vacany also activate the hopping term at first order , broadening the two bands . as the band gap shrinks with decreasing @xmath1 and the bands broaden due to this charge - fluctuation induced hopping , the two bands eventually meet , resulting in the metal - insulator transtion . thus the disappearance of the local moment and the transition into the metallic phase go hand - in - hand , both driven by enhanced charge fluctuations . an explicit demonstration of the existence of an insulating state with two separated bands , _ independent of any magnetic ordering _ , which can undergo a transition with decreasing interaction strength into a metallic phase was first provided by hubbard.@xcite however , the decoupling scheme employed within the equation of motion approach resulted in several drawbacks ( discussed in more detail in section iii ) , such as inappropriate bandwidth , violation of the fermi - liquid theory , and a vanishing critical @xmath30 for the metal - insulator phase transition for arbitrary hopping . of course , magnetic ordering and correlations , and the associated low - energy magnetic excitations were totally neglected . another program , which included the possibility of magnetic ordering , was based on a static random one - body approximation to the hubbard model.@xcite the interaction term was replaced by an effective , self - consistent random potential with statistical correlation to simulate many - body correlation effects . only ising - like magnetic excitations were included and the low - energy magnon excitations were neglected . a continuous magnetic phase transition was obtained between the afi and the pi phases , whereas two crossover temperatures were obtained around which , respectively , the dos at @xmath31 becomes appreciable and the local moments essentially disappear . in recent years the dynamical mean field theory , which is exact in the limit of infinite dimensions , has emerged as a controlled approximation for studying the phase diagram in the whole parameter space . in the limit of infinite dimensions the interaction self energy becomes local , and the quantum many - body problem reduces to the self - consistent solution of a single - site quantum problem of an impurity embedded in an effective medium . by extending this approach to two sites to account for the sublattice structure in the af state , a continuous magnetic phase transition was obtained , as already mentioned above . in addition , when @xmath11 was sufficiently suppressed ( made possible by including a frustration - inducing nnn hopping term ) , a first - order metal - insulator transition line , ending in a second - order critical point was also obtained.@xcite several associated features of the metal - insulator transition in @xmath32 have recently been discussed within the dmft approach , such as volume discontinuities and resistivity changes,@xcite and optical conductivity on both sides of the transition.@xcite the nature of the transition itself continues to be of interest , and a recent study of the changes in the density of states near the transition indicates that the fermi liquid breaks down before the gap opens.@xcite while the afi phase is well understood , and spin fluctuation effects can be studied about the broken - symmetry hf state even in the strong coupling limit , for the magnetically - disordered pi and pm phases a good analytical starting point at the same level is still not available . the decoupling approach presented here provides this for studying the interplay of electronic spectral properties and af correlations in these states . furthermore , the approach is equally applicable to the broken - symmetry state , allowing for a unified description independent of magnetic ordering . the paper is organized as follows . after introducing the equation of motion approach for the single - particle green s function in section ii , and briefly reviewing the hubbard i approximation in section iii , improved decoupling schemes are presented in section iv which take into account af correlations . especially appropriate for the spin - disordered pi phase , the final decoupling procedure in subsection c offers several improvements over the hubbard i approximation . fluctuations are included in section v within a static random approximation and a self - consistent evaluationt of the disorder - averaged self energy within the cpa is carried out . fluctuation effects on the charge gap and the density of states are examined in section vi by studying the coupled , non - linear equations for the self - consistently determined complex self energy . contact with known results for the broken - symmetry state is made in section vii , showing the applicability of the decoupling scheme irrespective of magnetic ordering . deviation from perfect af ordering and effects of short - ranged af spin correlations on the electronic spectrum are studied in section viii . the effects of hole doping on the qualitative nature of the quasiparticle band dispersion , and the transfer of spectral weight between the hubbard bands are briefly studied in section ix . some conclusions are presented in section x. we consider the hamiltonian @xmath33 where the kinetic energy term @xmath34 involves the hopping matrix elements @xmath35 between sites @xmath36 and @xmath37 , and @xmath38 is the local interaction term . in momentum space @xmath39 , where @xmath40 denotes the lattice free - particle energy . we consider the equation of motion approach , in which the time derivative of an operator @xmath41 in the heisenberg representation is obtained from the commutator with the hamiltonian , @xmath42= e^{iht}[\hat{o},h]e^{-iht}$ ] , and apply it to the time - ordered one - particle green s function @xmath43 \ ; .\end{aligned}\ ] ] taking the time derivative and using @xmath44 , the anticommutation property @xmath45 , and the commutator @xmath46 = - \sum_\delta t_\delta \ ; a_{i+\delta\uparrow } + u a_{i\uparrow}n_{i\downarrow}\ ] ] to determine @xmath47 , one obtains @xmath48 which is an equation for @xmath49 in terms of a higher - order correlation function . decoupling the equation of motion at this stage by replacing the operator @xmath50 in the last term by its expectation value results in the hartree - fock approximation @xmath51 where the densities @xmath52 are determined self consistently . this approximation is a reasonable _ starting point _ in the broken - symmetry state , when the long - time average of the local density @xmath50 does not differ substantially from the short - time average . for the af state , basic features such as the correlation effect and the associated electron localization in the strong interaction limit are actually already contained at this level , as evidenced by the charge gap of order @xmath1 and the narrow af bandwidth of order @xmath0 . within a spin - fluctuation approach , wherein transverse fluctuations about the hf state are systematically incorporated , the af state of the half - filled hubbard model has been extensively studied in recent years , including the effects of disorder , impurities , and vacancies.@xcite even the strong coupling limit ( @xmath53 ) is accessible within this formally weak - coupling expansion . this has been explicitly demonstrated by systematically evaluating quantum corrections to various quantities such as the magnon energy , af order parameter , perpendicular spin susceptibility , ground - state energy , spin correlation length.@xcite however , in the spin disordered state this approximation breaks down for strong interaction , when the expectation value of the local density @xmath54 ( the mean density ) differs substantially from the short - time average ( either nearly zero or nearly one ) . in this case the local magnetization strongly fluctuates between 1 and -1 , so that the average local magnetization @xmath55 vanishes , but the local moment @xmath56 defined by @xmath57 remains appreciable . this strong local moment arises from the correlation effect of an electron locally excluding the opposite - spin electron due to the strong electronic repulsion . this provides a clue that the decoupling procedure can be improved by ensuring that the operators which are replaced by their averages involve minimal fluctuations . in the following we describe several such decoupling procedures , all of which are quite distinct from that employed by hubbard.@xcite this approximation , commonly referred to as the hubbard i , is first briefly reviewed here for comparison . in hubbard i the next step was to continue with the equation of motion approach and examine the time derivative of the last term in eq . ( 4 ) involving the self - energy correction , @xmath58 . determining the time derivatives in @xmath59 $ ] from eq . ( 3 ) and the commutator @xmath60 = -\sum_\delta t_\delta \ ; ( a_{i\downarrow}^\dagger a_{i+\delta\downarrow } - a_{i-\delta\downarrow}^\dagger a_{i\downarrow } ) \ ; , \ ] ] and using the anticommutation property @xmath61 , one obtains @xmath62 in order to obtain a closed equation for @xmath63 , the equations were decoupled at this stage by noticing that @xmath64 , and replacing the operators @xmath65 , and @xmath66 in the last three terms by their expectation values in a spin - symmetric , translationally - invariant ground state , with @xmath67 , and @xmath68 , the average density for each spin . fourier transforming to frequency and momentum space using @xmath69 one obtains @xmath70 + u \gamma({\bf k}\omega ) \ ; .\ ] ] substituting for @xmath71/ ( \omega -u)$ ] from above into eq . ( 4 ) , and solving for @xmath72 one obtains the hubbard i result @xmath73 in the half - filled case ( @xmath74 this can be written as , @xmath75 \ ; , \ ] ] where @xmath76 , and @xmath77/2 $ ] are the two band energies . in the strong coupling limit these band energies reduce to @xmath78 and @xmath79 to lowest order , giving a bandwidth of @xmath26 , where @xmath80 is the free - particle bandwidth . thus in going from the non - interacting limit to the strong - coupling limit , the bandwidth decreases only from @xmath80 to @xmath26 , whereas one expects a correlation - narrowed bandwidth of order @xmath0 for the nn hopping model . this is a serious drawback of the hubbard i approximation in the strong correlation limit , and can be traced to the replacement of @xmath50 , a strongly fluctuating quantity , by its average value . another drawback pertains to the absence of a critical interaction strength for the energy gap . for a general free - particle energy @xmath81 , the electronic spectrum is split into two bands for all @xmath1 . also , in general , the integrated spectral weight in each hubbard band is not exactly one - half per spin , although the total for both bands is identically one . this amounts to an inconsistency in the insulating state in that the number of states in @xmath82-space lying below the fermi energy does not exactly yield the number of particles . we adopt a different approach here which removes all of the drawbacks cited above . we continue with eq . ( 4 ) for @xmath83 and consider the second derivative @xmath84 a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \nonumber \\ & - & iu \delta(t)\ ; \langle \psi \| \{n_{i\downarrow } a_{i\uparrow } , a_{j\uparrow}^\dagger \ } \|\psi\rangle .\end{aligned}\ ] ] the derivative of @xmath85 is substituted from eq . ( 4 ) , the derivative @xmath59 $ ] obtained as before from eqs . ( 3 ) and ( 6 ) , and in the last term @xmath61 . putting all this together , and multiplying by @xmath36 , we obtain @xmath86 + \sum_{\delta \delta ' } t_\delta \ ; t_{\delta ' } \ ; g_{i+\delta+\delta ' , j } \nonumber \\ & + & iu \sum_\delta t_\delta \ ; \langle \psi \| { \rm t } \{n_{i\downarrow}(t)+ n_{i+\delta \downarrow}(t ) \ } a_{i+\delta\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \nonumber \\ & - & i u^2 \;\langle \psi \|{\rm t } n_{i\downarrow } ^2 ( t ) a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle + iu \sum_\delta t_\delta \ ; \langle \psi \| { \rm t } \nonumber \\ & & \{a_{i\downarrow}^\dagger ( t ) a_{i+\delta\downarrow}(t ) - a_{i-\delta \downarrow}^\dagger ( t ) a_{i\downarrow}(t ) \ } a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \ ; .\end{aligned}\ ] ] we now proceed further along three different routes by introducing three decoupling schemes termed a , b , and c in the following subsections . schemes a and b are most suitable for hopping terms which only connect sites of opposite and same sublattices respectively . the best features of a and b are then combined in the final scheme c which is applicable for any general hopping . however , as we shall see , we are constrained to the consideration of a bipartite lattice . [ [ section ] ] adding @xmath87 to both sides of eq . ( 13 ) , where @xmath88 is the total density , and substituting for @xmath89 from eq . ( 4 ) , we obtain @xmath90 + \sum_{\delta \delta ' } t_\delta \ ; t_{\delta ' } g_{i+\delta+\delta ' , j } \nonumber \\ & + & iu\sum_\delta t_\delta \ ; \langle \psi \| { \rm t } \{n_{i\downarrow}(t)+ n_{i+\delta \downarrow}(t ) - n \ } a_{i+\delta\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \nonumber \\ & - & iu^2 \ ; \langle \psi \|{\rm t } \{n_{i\downarrow}(t)-n/2 \ } ^2 a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle + iu\sum_\delta t_\delta \ ; \nonumber \\ & & \langle \psi \| { \rm t } \{a_{i\downarrow}^\dagger ( t ) a_{i+\delta\downarrow}(t ) - a_{i-\delta \downarrow}^\dagger ( t ) a_{i\downarrow}(t ) \ } a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \ ; . \nonumber \\\end{aligned}\ ] ] in order to decouple this equation for @xmath49 involving several higher - order correlation functions we now replace the operators @xmath91 , @xmath92 , and @xmath66 in the last three terms by their ground - state expectation values , again in the spin - symmetric , translationally - invariant ground state , with @xmath93 , the average density for each spin . at half - filling ( @xmath94 ) , we have @xmath95 , a c - number , so that no approximation is involved in this replacement . furthermore , in the strong coupling limit , when double occupancy and vacancy are both prohibited , the total charge density operator @xmath96 , the unit operator , so that the term @xmath97 is also related to the second moment of the local magnetization @xmath98 . in general , @xmath99 fluctuates between the two values @xmath100 , depending on whether the site is occupied or empty . the average @xmath101 vanishes in the paramagnetic state when @xmath102 . more importantly , it also vanishes in the presence of af correlations , provided the sites @xmath36 and @xmath37 are in opposite sublattices , in which case @xmath103 . this scheme is therefore most appropriate when the hopping term @xmath35 connects sites @xmath36 and @xmath37 of opposite sublattices , as in the nn hopping case . we take @xmath67 , as in hubbard i. this is obviously so if translational symmetry is present , but also in presence of af correlations , as both terms are off - diagonal in the sublattice basis , and of equal magnitude . after fourier transformation to frequency and momentum space , and introducing a hartree shift in the energy axis through the transformation @xmath104 , we finally obtain @xmath105 \ ; , \ ] ] where @xmath106 and @xmath107 . the energy spectrum thus splits into two bands , independently of whether there is long - range af order or not , with an energy gap of @xmath108 . furthermore , the bandwidth is correctly of order @xmath0 in the strong coupling limit . [ [ section-1 ] ] an alternate decoupling scheme is more appropriate when the hopping term involves sites in the same sublattice . adding to both sides of eq . ( 13 ) , the terms @xmath109 , and substituting for @xmath110 from eq . ( 4 ) , we obtain @xmath111 \nonumber \\ & + & iu\sum_\delta t_\delta \ ; \langle \psi \| { \rm t } \{n_{i\downarrow}(t)- n_{i+\delta \downarrow}(t ) \ } a_{i+\delta\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \nonumber \\ & - & iu^2\ ; \langle \psi \|{\rm t } \{n_{i\downarrow}(t ) -n/2 \ } ^2 a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle + iu\sum_\delta t_\delta \ ; \nonumber \\ & & \langle \psi \| { \rm t } \{a_{i\downarrow}^\dagger ( t ) a_{i+\delta\downarrow}(t ) - a_{i-\delta \downarrow}^\dagger ( t ) a_{i\downarrow}(t ) \ } a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \ ; .\end{aligned}\ ] ] we notice that on the right - hand side the second - order hopping term @xmath112 present in eq . ( 15 ) has exactly cancelled . again , as before , the operators are replaced by their expectation values . we note that we can safely set @xmath113 , not only in the symmetric paramagnetic state , but even in the presence of af correlations , provided that sites @xmath36 and @xmath37 are in the same sublattice . fourier transformation leads to @xmath114 with the same hartree shift employed through the transformation @xmath115 , we obtain @xmath116 \ ; .\end{aligned}\ ] ] thus @xmath117 causes the free - particle spectrum to be split into two bands , with no correlation effect on the bandwidth . the two bands overlap when the band separation @xmath108 is smaller than the free particle bandwidth , but a gap opens up beyond a certain critical interaction strength . hence for a general @xmath81 , there is a non - zero critical interaction strength . however , the number of states in each band is precisely 1/2 per spin . [ [ section-2 ] ] finally , we consider a third scheme which combines the best features of a and b. we recall the replacement of operators by their expectation values in the decoupling procedures , and focus on the operators @xmath118 and @xmath119 in eqs . ( 14 ) and ( 16 ) , particularly in view of af correlations in the ground state . for @xmath120 connecting sites of opposite sublattices , the short - time average @xmath121 is weakly fluctuating , but not so when sites @xmath36 and @xmath37 are in the same sublattice . therefore , scheme a is suitable when hopping is limited to opposite sublattices . on the other hand , the short - time average @xmath122 is weakly fluctuating when @xmath120 connects sites of the same sublattice , so that scheme b is suitable when hopping is limited to the same sublattice . this immediately suggests that for general hopping both schemes can be further improved upon by taking an intermediate measure . for this purpose we are constrained to the consideration of a bipartite lattice , for which the hopping terms can be divided into two groups , connecting sites of opposite and same sublattices . we therefore write , @xmath123 where sites @xmath124 and @xmath125 refer to neighbors of site @xmath36 in the opposite and same sublattices respectively , and furthermore @xmath126 and @xmath127 are the two associated free - particle energies . the intermediate measure which leads to an improvement over both schemes a and b is simply effected by including only those additional terms of scheme b which correspond to the same sublattice . thus , we add to both sides of eq . ( 13 ) the terms @xmath128 , where the sum over neighbors @xmath129 in the last three terms are restricted to sites in the _ same _ sublattice . as before , we obtain after substituting for @xmath130 from eq . ( 4 ) , @xmath131 + \sum_{\lambda,\lambda ' } t_\lambda \ ; t_{\lambda ' } g_{i+\lambda+\lambda',j } \nonumber \\ & + & iu\sum_\mu t_\mu \ ; \langle \psi \| { \rm t } \{n_{i\downarrow}(t)- n_{i+\mu \downarrow}(t ) \ } a_{i+\mu\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \nonumber \\ & + & iu\sum_\lambda t_\lambda \ ; \langle \psi \| { \rm t } \{n_{i\downarrow}(t ) + n_{i+\lambda \downarrow}(t ) -n \ } a_{i+\lambda\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \nonumber \\ & - & iu^2\ ; \langle \psi \|{\rm t } \{n_{i\downarrow}(t ) -n/2 \ } ^2 a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle + iu\sum_\delta t_\delta \ ; \nonumber \\ & & \langle \psi \| { \rm t } \{a_{i\downarrow}^\dagger ( t ) a_{i+\delta\downarrow}(t ) - a_{i-\delta \downarrow}^\dagger ( t ) a_{i\downarrow}(t ) \ } a_{i\uparrow}(t ) a_{j\uparrow}^\dagger ( 0 ) \|\psi\rangle \ ; .\end{aligned}\ ] ] the cancellation of the second - order hopping terms , referred to earlier below eq . ( 16 ) , is only partial now , and only terms involving opposite sublattice hopping finally survive . again replacing the operators with their expectation values as before , with @xmath132 , we obtain after fourier transformation @xmath133 which , after the same hartree shift @xmath115 , yields @xmath134\end{aligned}\ ] ] where @xmath135 is associated with the opposite sublattice hopping . results of both schemes a and b are recovered by taking appropriate limits . for only opposite sublattice hopping @xmath136 , so that eq . ( 15 ) is obtained , whereas for only same sublatttice hopping @xmath137 , which yields eq . ( 18 ) . in general , when hopping involving same sublattice sites ( e.g. nnn hopping ) is present , a finite @xmath117 is required for the band gap to appear , yielding a critical interaction strength which separates the metallic and insulating phases . quite generally , @xmath138 , the free - particle band energy associated with opposite - sublattice hopping , changes sign under the transformation @xmath139 , while the band energy @xmath140 does not . therefore , the @xmath141-space is split into two degenerate zones . for evaluation of the integrated spectral weight , the @xmath141-sum therefore covers the whole @xmath141 space . as @xmath138 is odd , the integrated spectral weight in each band is identically one - half per spin . in the previous section , the equation of motion for @xmath142 was decoupled by replacing the operators @xmath143 , @xmath144 and @xmath145 by their ( vanishing ) expectation values in the spin - symmetric , translationally - invariant ground state . if @xmath146 represents these operators , then @xmath147 and @xmath148 are in general non - vanishing due to fluctuations and correlations present in the ground state . in the following we approximately account for these fluctuations by replacing the operator @xmath146 by static , independently random terms @xmath149 distributed across the lattice with @xmath150 and @xmath151 determined self - consistently from @xmath148 . when the above replacements are made , we see from eq . ( 20 ) that the charge / potential fluctuation terms @xmath152 and @xmath153 couple with the hopping term , leading to charge fluctuation induced hopping . on the other hand , the hopping induced fluctuation term @xmath154 is purely imaginary and couples with the local interaction term yielding a hopping induced relaxation . in eq . ( 20 ) the operators @xmath143 and @xmath144 were replaced ( in the decoupling procedure ) by their expectation values , which vanish in the paramagnetic state at half filling ( @xmath94 ) , and also when af correlations are present . as shown below this approximation essentially amounts to a neglect of local charge fluctuations and is valid in the low - temperature , strong - coupling limit , when double occupancy and vacancy are strongly suppressed . in the strong coupling limit we have a spin picture of the ground state , and for arbitrary orientation of the local spin direction , the ground state expectation values @xmath155 and @xmath156 remain invariant due to the af correlations between neighbouring spins . if the spin at site @xmath36 is inclined at angle @xmath157 from the @xmath158 direction , then @xmath159 , and @xmath160 . hence the operators @xmath161 and @xmath162 essentially behave like c - numbers , and the above replacements are valid . even the strong quantum spin fluctuations , which significantly reduce the af order parameter in two dimensions , do not substantially alter this picture as the major spin - fluctuation contribution arises from long - wavelength magnon modes . however , for weak correlation and/or high temperature , when double occupancy is not strongly penalized , the local densities @xmath163 and @xmath164 _ independently _ fluctuate between zero and one at short time scales , and therefore @xmath165 and @xmath166 strongly fluctuate between minus one and plus one . in the following we account for these fluctuations within a static random approximation , together with disorder averaging at the cpa level . a more complete theory which includes the spatial and temporal correlations in the fluctuations will be developed later by considering the appropriate two - body green s functions in eq . ( 20 ) within systematic approximations like the rpa . even in the absence of charge fluctuations , @xmath165 can fluctuate if ferromagnetic correlations are present in the ( paramagnetic ) ground state . similarly these fluctuations are also present if the af correlations are only short - ranged , so that the neighbouring spins are twisted from perfect af alignment . this is further discussed in section viii . we note that the energy scale @xmath167 of these fluctuation terms is intermediate between the local @xmath168 scale and the second - order hopping scale @xmath169 in eq . the effect of these order-@xmath167 fluctuation terms is therefore to renormalize the bandwidth by order @xmath170 in the strong coupling limit , as opposed to the order @xmath0 renormalization by the second - order virtual hopping process . thus the charge - fluctuation and spin - twisting processes activate the hopping term at first order . for simplicity , we consider the nn hopping case , so that the sites @xmath36 and @xmath37 are in opposite sublattices . we assume that @xmath1 times the density fluctuation term @xmath171 , which we denote by @xmath172 , is a random potential distributed independently across the lattice . the average of this random potential term vanishes , but its second moment @xmath173 reflects a measure of fluctuations in the system . at half filling ( @xmath94 ) , the long - time average yields the fluctuation magnitude @xmath174 , which has the following limits . in the strong coupling limit when double occupancy is strongly suppressed and af correlations are present , we have @xmath175 . at the other extreme , in the non - interacting limit , when a site may be independently occupied or vacant with equal probability of one - half , we have @xmath176 . therefore , the fluctuation magnitude lies in the range @xmath177 . the random potential term @xmath178 is treated at the cpa level , and the disorder - averaged green s function is obtained in the following . after the same decoupling approximation made earlier , we rewrite eq . ( 14 ) in a matrix representation ( constructed from the site basis ) as @xmath179 { \bf g}(\omega)=\omega { \bf 1 } + { \bf t } \ ; , \ ] ] where the matrix @xmath180 is associated with the nn hopping term and has matrix elements @xmath181 , whereas @xmath182 has matrix elements @xmath183 , the random potential term . for further convenience we introduce another green s function @xmath184 through the relation @xmath185 $ ] , so that @xmath184 obeys the equation , @xmath186{\bf g}(\omega)= { \bf 1 } \ ; .\ ] ] the usual cpa treatment of the random potential term , involving perturbative expansion followed by disorder averaging , leads to the following result at the level of rainbow diagrams , @xmath187 where , @xmath188 is the disorder - averaged self energy , to be determined self consistently . for the purpose of disorder averaging , we assume for simplicity that the random potential terms @xmath178 are independently random for each pair of sites @xmath36 and @xmath37 , so that @xmath189 , where the second moment @xmath190 is a measure of the potential disorder . this essentially amounts to a neglect of spatial correlations in the fluctuations . in this case there are two terms in the disorder self energy @xmath191 the second term @xmath192 , however , vanishes because @xmath193 is identically zero . this follows because @xmath194 , given by the fourier transform of eq . ( 25 ) , is even under the transformation @xmath195 , while the other term @xmath196 is odd , and therefore the @xmath197 summation over the whole brillouin zone yields the vanishing result . the disorder self energy is therefore diagonal in the site basis , and hence momentum independent . furthermore , the sum @xmath198 in the equation for @xmath199 yields @xmath200 , the number of nearest neighbours , so that the self energy involves the combination @xmath201 . these features of the local self energy are as in the dmft where the hopping term is taken to be @xmath202 , so that @xmath203 is finite in the limit of infinite dimensions.@xcite finally , after fourier transformation of eq . ( 25 ) and of @xmath204 $ ] , we obtain the following result for the full green s function @xmath205 where the disorder - averaged self energy is obtained self consistently from @xmath206 the above provides a self - consistent scheme for determining the green s function when fluctuations are included . in the following we quantitatively examine the fluctuation effects on the spectral properties , in particular the charge gap and the density of states . thermal contribution to charge and spin fluctuations brings in temperature into the problem , and makes the charge gap sensitive to temperature , leading to the possibility of a temperature - driven metal - insulator transition . temporal and spatial correlations in fluctuations are neglected in this treatment . the present study is nonetheless useful because when these correlations are included by considering the appropriate two - body green s functions ( having the form @xmath207 ) , the structure of the resulting equation for @xmath49 will be similar to eq . ( 28 ) . the purely imaginary hopping induced fluctuation term @xmath208 couples diagonally to the green s function . therefore , eq . ( 23 ) will contain , in addition , a diagonal term @xmath209 , where @xmath210 . again assuming that @xmath211 are independently random , and are uncorrelated with the @xmath178 , within the cpa we obtain another diagonal self energy with a negative coefficient which will simply renormalize the @xmath212 defined earlier . a qualitative criterion for the metal - insulator transition is obtained by considering the competition between the kinetic energy term ( bandwidth @xmath80 ) which delocalizes the electrons making the system metallic , and the correlation term ( coulomb barrier @xmath1 ) which localizes the electrons making the system insulating . we will first make a simple analysis of the effective charge gap when fluctuations are included to their full extent . we shall see that the qualitative criterion @xmath213 for the vanishing of the gap , and therefore for the metal - insulator transition , arises in a very natural way . we have seen earlier that when fluctuations are allowed to their full extent , @xmath214 , and therefore the disorder strength @xmath215 . considering the disorder self energy from eq . ( 29 ) to the lowest order , we have for @xmath216 @xmath217 where the result at the right is obtained by neglecting the free - electron energy @xmath218 in comparison to @xmath219 , for the purpose of obtaining an estimate . now from eq . ( 28 ) the charge gap vanishes when @xmath220 , which leads to the following estimate for the critical value @xmath30 for the metal - insulator transition , @xmath221 which is of the order of the free - electron band width . within the approximations used , for @xmath222 the energy gap never closes , and the system remains insulating . we now obtain the self - consistent solution for the self energy @xmath223 from eq . ( 29 ) . as @xmath223 is complex in general , it is convenient to introduce a complex function @xmath224 by equating @xmath225 in eq . ( 29 ) the self - consistency equation for @xmath223 may be recast in terms of @xmath226 ; if @xmath227 denotes the prefactor , eq . ( 29 ) then reads @xmath228 , the real and imaginary components of which are @xmath229 we now discuss the solution for the above two coupled , non - linear equations in @xmath230 and @xmath231 , where the parameter @xmath232 incorporates the fluctuation magnitude . we first consider the solution with @xmath233 , which corresponds to a vanishing imaginary part of the self energy and the density of states , and therefore to the energy gap region . since @xmath234 decreases monotonically with positive @xmath230 , for a given @xmath63 there exists a physically significant solution @xmath235 of the equation @xmath236 . for @xmath237 , we have @xmath238 , therefore eq . ( 34 ) has a solution only if @xmath233 , whereas for @xmath239 a solution exists only for @xmath240 . thus , @xmath241 is the point where the self energy changes from purely real to complex , and therefore corresponds to the band - edge energy , as further discussed below . there is another special value @xmath242 , this one having to do with eq . the right - hand - side term @xmath243 diverges in both the limits @xmath244 and @xmath245 , and therefore it must have a minimum at some intermediate value @xmath242 , as shown in fig . 1 from an explicit calculation . since @xmath246 , this minimum corresponds to a maximum value @xmath247 beyond which no real solution for @xmath248 exists . it is interesting to note that these two values @xmath241 and @xmath249 are actually _ identical _ because their parent equations @xmath250 and @xmath251 are so in view of the relation @xmath252 . the significance of this coincidence is that the imaginary part of the self energy grows continuously , so that there is no discontinuity in the density of states , and therefore when the band gap vanishes the resulting metal - insulator transition is also continuous . solving for real @xmath248 from eq . ( 33 ) for a given @xmath237 and @xmath233 , we thus have the solution to the eqs . ( 33 ) , ( 34 ) . if @xmath253 is the solution corresponding to @xmath254 , then since @xmath255 , we have @xmath256 , beyond which no real solution for @xmath248 exists , as discussed earlier . therefore @xmath253 is an upperbound for solutions with @xmath233 , for which the imaginary part of the self energy and the density of states vanish . this implies that @xmath253 is the band - edge energy , and @xmath257 is the energy gap region , with the energy gap @xmath258 . a convenient way to obtain @xmath253 , and hence the band gap @xmath259 is from the plot of @xmath243 vs. @xmath230 in the range @xmath260 . the variation of the band gap so obtained with the fluctuation magnitude @xmath261 in the range @xmath262 is shown for two and three dimensions in figs . 2 and 3 respectively for different values of @xmath1 . there exists a critical interaction strength @xmath30 , such that for @xmath263 with increasing @xmath261 the energy gap vanishes at some critical value @xmath264 , whereas for @xmath222 the band gap remains finite and the system remains insulating . in two and three dimensions we obtain the critical interaction strengths @xmath265 and @xmath266 respectively , ( @xmath267 and @xmath268 ) . with increasing fluctuation strength @xmath63 the solution @xmath241 of the equation @xmath236 increases , as also the right - hand side of eq . ( 33 ) , so that the corresponding @xmath253 decreases until at some critical value @xmath269 we finally have @xmath270 , and the left - hand side of eq . ( 33 ) can not increase any further . this critical fluctuation strength at which @xmath253 and hence the band gap vanish marks the metal - insulator transition . for @xmath271 a solution of eq . ( 33 ) exists only if @xmath240 , which leads to a finite density of states at @xmath272 . it is important to note that @xmath231 and the density of states increases continuously with @xmath273 , signifying a continuous metal - insulator transition . while the range @xmath274 corresponds to the energy - gap region with @xmath233 , the other side @xmath239 necessarily goes with @xmath240 , and therefore corresponds to the band region wherein the self energy is complex . it is also clear from eq . ( 33 ) that for @xmath275 a solution exists only if @xmath230 is negative , so that the band region corresponds to @xmath276 . to solve eqs . ( 33 ) , ( 34 ) for @xmath226 for some fixed @xmath63 , we first solve for @xmath231 from eq . ( 34 ) for a given @xmath230 , and then solve for a real @xmath248 from eq . ( 33 ) with this pair @xmath226 . by varying @xmath230 between a sufficiently negative value and @xmath277 , the full range of solutions are thus obtained for @xmath278 . from these self - consistent solutions for @xmath226 the complex self energy is then obtained from @xmath279 . we now discuss the evaluation of the density of states using this complex self energy @xmath223 . for this purpose we rewrite eq . ( 28 ) for the green s function as @xmath280\ ] ] where @xmath281 . writing @xmath282 and @xmath283 @xmath284^{1/4 } \nonumber \\ \theta({\bf k},\omega ) & = & \frac{1}{2}\tan ^{-1 } \frac{\sigma_i(\omega ) } { e_{\bf k}^2 + \sigma_r(\omega ) } \end{aligned}\ ] ] we obtain @xmath285 , in terms of which the density of states is finally obtained from @xmath286 \ ; .\ ] ] here we have made use of the transformation property of @xmath81 which changes sign under the transformation @xmath139 while @xmath287 does not , so that when summed over all @xmath141 the @xmath81 term in eq . ( 35 ) yields a vanishing contribution . the results for the density of states for a simple cubic lattice and @xmath288 are shown in fig . 4 for several values of the fluctuation strength @xmath261 . due to the fluctuation - induced states in the gap , the band gap decreases with @xmath261 , and at a critical value @xmath289 , the band gap vanishes and the system undergoes a continuous metal - insulator transition . so far we have been concerned with the spin - symmetric state . however , the present approach provides for a description of the broken - symmetry state as well , and in this section we briefly make contact with the known results for the af state with respect to the af band energies and amplitudes . we consider the nn hopping case , and proceed with the decoupling approach a. in the af state the density @xmath163 has values @xmath290 and @xmath291 on the two sublattice sites , where @xmath9 is the sublattice magnetization . translational symmetry therefore exists only within the sublattice basis . after employing the same decoupling procedure as before leading to eq . ( 15 ) , fourier transformation within the sublattice basis leads to @xmath292 \ ; .\ ] ] the quasiparticle band energies in the af state are thus given by @xmath293 , same as for the symmetric case . to make contact with the hf results for the af state , we make the following approximation consistent with the hf scheme @xmath294 in this case @xmath295 , so that @xmath296 , which now involves the sublattice magnetization . with this approximation , ( 38 ) simplifies to @xmath297 \ ; , \ ] ] with @xmath298 . this is precisely of the same form as obtained in the hf description of the af state.@xcite consequently , the expressions for the af - state electron amplitudes are identical , resulting in the same self - consistency condition @xmath299 , which determines the sublattice magnetization @xmath9 . af correlations have already been considered while introducing the decoupling schemes a , b , and c in section iv . here we examine the influence of short - range af ordering on such aspects of the electronic spectrum as the charge gap , the density of states etc . in the strong correlation limit and for half - filling ( @xmath94 ) , when charge fluctuations are absent , the expectation value @xmath300 vanishes only for perfect af ordering when @xmath301 . however , for @xmath302 , when the spin ordering is not perfect , the finite spin correlation length @xmath20 results in a twisting of neighbouring spins by @xmath303 on the average . if at site @xmath37 the spin is pointing up , but at site @xmath36 the spin instead of pointing down is twisted by a small angle @xmath157 , then @xmath304 , so that @xmath305 is slightly reduced from 1 . whereas if the spin at site @xmath36 is pointing down , but at the neighboring site @xmath37 the spin instead of pointing up is twisted by the small angle @xmath157 , then @xmath305 is slightly increased from 1 . this results in a twist - induced fluctuation term arising from the expectation value @xmath300 which fluctuates between @xmath306 and @xmath307 , so that @xmath308 this twist - induced fluctuation term couples with the hopping term at first order in eq . ( 14 ) , so that the quasiparticle band energies are modified by an order-@xmath170 term . this feature is similar to that in the spiral state of the doped hubbard model , where the twisting of neighbouring spins activates the hopping term at first order and broadens the bands symmetrically.@xcite the pulled - down states are filled whereas the pushed - up states are empty due to doping , resulting in stabilization of the spiral state . this twist - induced fluctuation term will also contribute to the overall fluctuation strength discussed in section v , with the disorder strength @xmath63 now including a term related to the inverse spin - correlation length . therefore with increasing temperature the enhanced magnon excitation and the spin - fluctuation induced spin disordering will also contribute to states in the gap and the band - gap reduction . thus in general , both charge and spin fluctuations contribute towards an effective disorder strength @xmath63 which will determine the overall electronic spectral properties . in the strong coupling limit , doping with holes or electrons results in transfer of spectral weight between the two hubbard bands . consider the atomic limit ( @xmath309 ) where this feature is most pronounced and easily understood . a site with a single electron of spin @xmath310 contributes two states at energies @xmath311 ( occupied ) and @xmath312 ( unoccupied ) . when a spin-@xmath310 hole is introduced by removing the spin-@xmath310 electron , the spectrum changes to both states at energy @xmath313 , because electrons of either spin can now go into the empty site with zero energy . extending to a collection of sites , if @xmath314 is the hole concentration , then the density of states ( both spins ) at energies @xmath315 and @xmath1 are @xmath316 and @xmath317 respectively , so that the density of unoccupied states in the lower energy level is @xmath318 . finite hopping introduces a spread in the hole energy of order @xmath170 , so that this picture of the transfer of spectral weight remains valid as long as the band separation @xmath319 is much bigger than the bandwidth @xmath320 . in order to examine the effect of hole doping on the spectral weights , we consider eq . ( 14 ) appropriate for the nn hopping case , and focus on the correlation function term @xmath321 , which reduces to the single - particle green s function in the half - filled case @xmath94 . for a total particle density @xmath322 , corresponding to a hole concentration @xmath314 , the term @xmath323 reduces to @xmath324 , which now contains an operator term for finite doping . however , the doping - induced term @xmath325 can be written in terms of @xmath326 from eq . ( 4 ) as @xmath327 \ ; , \end{aligned}\ ] ] so that the correlation function term under consideration can be exactly obtained in terms of the single - particle green s function @xmath49 . decoupling the equation for @xmath49 as before in section iv a , we obtain after fourier transformation , @xmath328 g({\bf k}\omega ) \nonumber \\ & = & \omega - nu/2 + \epsilon_{\bf k } - x u \ ; .\end{aligned}\ ] ] after simplification and the hartree shift @xmath329 , we obtain : @xmath330 where as before @xmath331 . the quasiparticle band energies are therefore given by @xmath332 , which in the strong coupling limit reduce to @xmath333 $ ] , when only terms upto first order in @xmath334 are retained . this indicates that doping leads to a qualitative change in the nature of the band itself , with the dispersion changing from @xmath335 of order @xmath0 to @xmath336 , resulting in an effective doping - induced hopping strength and bandwidth of order @xmath3 . this is again very different from the hubbard i `` rigid - band '' result wherein there is only a slight doping - induced modification of the bandwidth , but no qualitative change in the nature of the band . we examine the loss of integrated spectral weight in the upper band due to doping . integrating the spectral function @xmath337 over the upper band , and summing over both spins , we obtain @xmath338 evaluating the momentum sum in the strong coupling limit by retaining terms only upto order @xmath339 , we obtain , @xmath340 upto order @xmath170 the integrated spectral weight in the upper band is therefore simply reduced to @xmath317 by doping , as in the atomic limit . similarly the weight in the lower band is enhanced to @xmath316 . in conclusion , the new decoupling scheme is appropriate for both the spin - symmetric ( paramagnetic metallic and insulating ) phases as well as the broken - symmetry af phase . although restricted to bipartite lattices , the decoupling approximations involve only weakly fluctuating operators , resulting in several improvements over the hubbard i scheme . independent of magnetic ordering , the scheme yields , at the lowest order , the correct strong - coupling bandwidth of order @xmath0 , a non - zero critical interaction strength ( above which the band gap opens ) only if same - sublattice hopping ( e.g. , nnn hopping ) is also present , and the correct quasiparticle spectral weights ( of 1/2 per spin ) in each band for the half - filled model . charge and spin fluctuations , which are manifested in double occupancy and vacancy of sites and spin twisting due to short - range af ordering , were studied within a static , random approximation . the self - consistent , cpa - level evaluation of the disorder - averaged self energy showed a fluctuation - induced band - broadening and consequent decrease in the band gap . the critical interaction strength , above which the band gap remains finite even when the fluctuations are included to their full extent , was obtained as @xmath267 and @xmath268 in two and three dimensions respectively . the solution of the non - linear self - consistent equation for the disorder self energy shows that with increasing fluctuation strength the band gap shrinks to zero continuously and the density of states @xmath2 between the bands also grows continuously . the monotonic dependence of fluctuation strength on temperature translates into a continuous metal - insulator transition . the fluctuation magnitude as well as its dynamics can be determined from the appropriate correlation function , and this is presently under investigation . the dynamics is relevant in view of the switching of the spectral function between the two hubbard bands , and the possibility of a motionally narrowed central peak in the density of states when the hubbard switching rate becomes comparable with the energy separation between the bands . quantum monte carlo simulations do show a central peak in the vicinity of the metal - insulator transition.@xcite with doping a quantitatively correct transfer of spectral weight between the two hubbard bands in the strong - coupling limit was obtained at the lowest - order level in the decoupling scheme . furthermore , in contrast to the hubbard i approximation , a qualitative change in the nature of the band itself was obtained , with an effective doping - induced hopping strength and bandwidth of order @xmath3 . extension of the equation of motion approach to the transverse spin propagator to obtain the low - energy spin - fluctuation spectrum will provide a magnetic description of the spin - symmetric state , and enable one to approach the magnetic phase boundary from the paramagnetic side . this , as well as correlations and dynamics of fluctuations , and their impact on the electronic and magnetic spectrum , are presently under investigation . the author is grateful to dieter vollhardt for helpful correspondence . this work was supported in part by research grant ( no . sp / s2/m-25/95 ) from the department of science and technology , india . g. a. thomas , d. h. rapkine , s. a. carter , a. j. millis , t. f. rosenbaum , p. metcalf , and j. m. honig , phys . lett . * 73 * , 1529 ( 1994 ) . j. schlipf , m. jarrell , p. g. j. van dongen , n. blmer , s. kehrein , th . pruschke , and d. vollhardt , report no . cond - mat/9902267 .	a new decoupling scheme is developed for the hubbard model which provides a unified description of the spin - symmetric ( paramagnetic metallic and insulating ) phases as well as the broken - symmetry afi phase . independent of magnetic ordering , the scheme yields , in the lowest order , the correct strong - coupling bandwidth of order @xmath0 ( for the nn hopping model ) and band gap of order @xmath1 , a non - zero critical interaction strength ( above which the band gap opens ) only if same - sublattice hopping ( e.g. , nnn hopping ) is also present , and the correct integrated spectral weights ( 1/2 per spin ) in each band for the half - filled model . the effects of charge and spin fluctuations , including spin twisting due to finite spin correlation length , are investigated within a static , random approximation . a self - consistent evaluation of the disorder - averaged self energy within the cpa is carried out numerically . fluctuations activate the hopping term at first order resulting in band - broadening , and the consequent decrease in the band gap with fluctuation strength is obtained for several @xmath1 values in two and three dimensions . we find that the band gap shrinks to zero continuously , and subsequently the density of states @xmath2 between the bands grows continuously , leading to a continuous metal - insulator transition . for finite doping there is transfer of spectral weight between the hubbard bands , and a qualitative change in the nature of the quasiparticle band dispersion , with an effective doping - induced hopping strength and bandwidth of order @xmath3 . 2
let @xmath5 be a bipartite graph on @xmath0 vertices with the vertex decomposition @xmath6 , that is , each edge in @xmath2 has one vertex in @xmath7 and the other in @xmath8 . an _ alternating labeling _ @xmath9 is a bijection @xmath10 such that for two adjacent vertices @xmath11 and @xmath12 we have that @xmath13 . another way to phrase this condition is that every vertex @xmath14 in @xmath7 is a local minimum of the bijection @xmath9 and every vertex @xmath15 in @xmath8 is a local maximum . define the euler number @xmath16 to be the number of alternating labelings @xmath9 of the vertices of the graph . two examples are in order . first , for the path @xmath17 on @xmath0 vertices the euler number @xmath18 is the number of alternating permutations , that is , the classical euler number @xmath19 . second , for a cycle @xmath20 of even length @xmath0 we have that @xmath21 . since there are @xmath22 possible positions for the largest label @xmath0 , the labeling reduces to the path @xmath23 . observe that we can not drop the condition that the graph @xmath2 is bipartite , since the labeling can not alternate along an odd cycle . alternatively , for a non - bipartite graph @xmath2 let @xmath24 . observe that the definition of the euler number is independent of the order of @xmath7 and @xmath8 . we also have the trivial lower bound @xmath25 by assigning @xmath8 the @xmath26 largest labels . equality in ( [ inequality_lower_bound ] ) is only obtained for the complete bipartite graphs . moreover , extending the classic `` multiplication theorem '' due to macmahon ( * ? ? ? * article 159 ) , for the disjoint union of two graphs @xmath2 and @xmath27 we have @xmath28 where @xmath2 and @xmath27 have @xmath29 respectively @xmath0 vertices . our interest is to study subgraphs of the cartesian product of two graphs . for graphs @xmath2 and @xmath27 , and @xmath30 a subset of vertices of @xmath2 , define the product @xmath31 as the graph on the vertex set @xmath32 where two vertices @xmath33 and @xmath34 are adjacent in @xmath31 if either @xmath35 and @xmath36 is adjacent to @xmath37 , or @xmath14 is adjacent to @xmath15 and @xmath38 . the cartesian product @xmath39 is obtained as a special case of the product @xmath31 with @xmath40 . ( 40,10)(0,0 ) ( 0,0)(10,0)5(0,0 ) ( 0,10 ) ( 0,0)(0,1)10 ( 0,10)(1,0)40 for the general problem we obtain the following asymptotics . let @xmath2 be a bipartite graph on @xmath0 vertices and @xmath30 a non - empty subset of the vertices of the graph @xmath2 . then there exist three positive real numbers @xmath41 , @xmath42 and @xmath43 such that @xmath44 and @xmath45 [ theorem_asymptotic ] for a bipartite graph @xmath46 on @xmath0 vertices define two subsets @xmath47 and @xmath48 of the @xmath0-dimensional unit cube @xmath49^{v}$ ] in @xmath0-dimensional euclidean space @xmath50 by @xmath51^{v } \:\ : : \:\ : x_{u } + x_{v } \leq 1 \mbox { for } \{u , v\ } \in e \ } , \\ y & = & \ { \vec{x } \in [ 0,1]^{v } \:\ : : \:\ : x_{u } \leq x_{v } \mbox { for } u \in v_{1},\ , v \in v_{2},\ , \{u , v\ } \in e \ } .\end{aligned}\ ] ] the two subsets @xmath47 and @xmath48 have the same volume , which is given by the euler number of the graph @xmath2 divided by @xmath52 . by reflecting the set @xmath48 over all of the hyperplanes of the form @xmath53 where @xmath12 we obtain the set @xmath47 . hence their volumes agree . by cutting the @xmath0-dimensional cube with the hyperplanes @xmath54 for all @xmath55 we obtain @xmath52 simplices of the same volume . each simplex corresponds to a permutation by reading the order of the coordinates of a point in its interior . the set @xmath48 is the union of a subcollection of these simplices corresponding to an alternating labeling of the graph @xmath2 . let the @xmath56-function @xmath57 be defined on the set @xmath58 by @xmath59 define the operator @xmath60 on @xmath61 by @xmath62(\vec{x } ) = \int_{\vec{y } \in x } \chi(\vec{x } , \vec{y } ) \cdot f(\vec{y } ) \ : d\vec{y } .\ ] ] since @xmath57 is symmetric , that is , @xmath63 , the operator @xmath60 is a self - adjoint hilbert - schmidt operator . thus we conclude that the spectrum of @xmath60 is real and discrete with @xmath64 as the only possible accumulation point . furthermore , all the eigenvalues and eigenfunctions of the operator are real . since the eigenvalues @xmath41 lie in the closed interval @xmath65 $ ] . hence , there is a largest eigenvalue in absolute value . moreover , the eigenfunctions form a complete orthogonal set . let @xmath66 denote the constant function with value @xmath67 on set @xmath48 . now we have for a bipartite graph @xmath2 on @xmath0 vertices and @xmath30 a subset of the vertices of the graph @xmath2 , @xmath68\right\rangle } .\ ] ] expanding the inner product and each of the @xmath69 applications of the operator @xmath60 , we have that @xmath70\right\rangle } & = & \int_{\vec{x}_{1 } } t^{m-1}[{{\mathbf 1}}](\vec{x}_{1 } ) \ : d\vec{x}_{1 } \nonumber \\ & = & \int_{\vec{x}_{1 } } \int_{\vec{x}_{2 } } \chi(\vec{x}_{1},\vec{x}_{2 } ) \cdot t^{m-2}[{{\mathbf 1}}](\vec{x}_{2 } ) \ : d\vec{x}_{2 } d\vec{x}_{1 } \nonumber \\ & \vdots & \nonumber \\ & = & \int_{\vec{x}_{1 } } \int_{\vec{x}_{2 } } \cdots \int_{\vec{x}_{m } } \chi(\vec{x}_{1},\vec{x}_{2 } ) \cdot \chi(\vec{x}_{2},\vec{x}_{3 } ) \cdots \chi(\vec{x}_{m-1},\vec{x}_{m } ) \ : d\vec{x}_{m } \cdots d\vec{x}_{2 } d\vec{x}_{1 } . \label{equation_m_n}\end{aligned}\ ] ] let @xmath71 be the vector @xmath72 . then the integral in equation ( [ equation_m_n ] ) is over all of the @xmath73 variables @xmath74 with the boundary condition that ( i ) @xmath75 , ( ii ) @xmath76 for @xmath77 and @xmath78 , and ( iii ) @xmath79 for @xmath80 an edge in @xmath2 . these inequalities describe exactly the set @xmath81 and hence the integral is given by the ratio @xmath82 . let @xmath2 be a bipartite graph on @xmath0 vertices and @xmath30 a non - empty subset of the vertices of the graph @xmath2 . then we have @xmath83 where the eigenvalues of the operator @xmath60 are @xmath84 and @xmath85 is the eigenfunction associated to the eigenvalue @xmath86 . [ theorem_main ] expand the function @xmath66 in terms of eigenfunctions : @xmath87 apply @xmath88 and take the inner product with @xmath66 and the result follows . when the set @xmath30 is empty , theorem [ theorem_main ] is trivial . in that case @xmath89 is the disjoint union of @xmath29 copies of @xmath2 . using ( [ equation_macmahon ] ) we have that @xmath90 when the graph @xmath2 consists of a singleton vertex and @xmath30 consists of this vertex , then the product @xmath91 is exactly the path on @xmath29 vertices , and its euler number is the classical @xmath92 . in this case the operator @xmath60 is given by @xmath62(x ) = \int_{0}^{1-x } f(z ) \ : dz .\ ] ] this operator has eigenvalues @xmath93 where @xmath94 and eigenfunctions . calculating @xmath95 and @xmath96 we obtain the following classical asymptotic expansion for the euler number @xmath97 where @xmath98 , that is , @xmath99 . see ( * ? ? ? * section 4 ) . [ example_classical_euler_number ] let @xmath100 be the subspace of @xmath61 consisting of the functions only depending on the variables @xmath101 where @xmath14 belongs to @xmath30 . in the case when @xmath30 is the vertex set of the graph @xmath2 the space @xmath100 is @xmath61 . the following result applies to the case when @xmath30 is strictly contained in the vertex set of @xmath2 . the image of the operator @xmath60 is contained in subspace @xmath100 . hence all the eigenfunctions associated to non - zero eigenvalues belong to @xmath100 . [ proposition_subspace ] for a vertex @xmath15 not in the set @xmath30 , observe that the function @xmath102 does not depend on the variable @xmath103 . hence when integrating @xmath104 over all @xmath105 the resulting function @xmath106 $ ] does not depend on @xmath103 , that is , @xmath106 $ ] belongs to the space @xmath100 . the second statement follows from the defining relation for eigenfunctions . the frobenius - perron result applies to matrices , that is , linear operators on a finite - dimensional vector space . an operator version of frobenius - perron was discovered by kren and rutman @xcite . we present a specialized version of their result . let @xmath107 be a measurable space . recall that two functions in @xmath108 are considered the same if they differ on a set of measure @xmath64 . we call a function @xmath109 non - negative if @xmath110 for almost all @xmath111 . similarly , we call the function @xmath112 positive if @xmath113 for almost all @xmath111 . an operator @xmath114 on @xmath108 is _ positivity improving _ if for all non - negative but non - zero functions @xmath112 the function @xmath115 $ ] is positive . let @xmath114 be an operator @xmath108 such that there is a positive integer @xmath116 so that @xmath117 is positivity improving . then the largest eigenvalue @xmath41 ( in modulus ) of @xmath114 is real , positive and simple . moreover , the associated eigenfunction @xmath118 is a positive function on @xmath107 . applying kren - rutman to our operator @xmath60 , we have the operator @xmath119 is positivity improving . the largest eigenvalue ( in absolute value ) @xmath41 of the operator @xmath60 is real , positive and simple . furthermore , the associated eigenfunction @xmath118 is positive . let @xmath112 be a non - negative , non - zero function in @xmath61 . by the definition of the operator @xmath60 we have that in a neighborhood of @xmath64 the function @xmath106 $ ] has a positive support . by applying the operator @xmath60 again we obtain that every point in the interior of @xmath48 takes a positive value in the function @xmath120 $ ] . the remainder of the proposition follows from kren - rutman . by letting @xmath41 be the largest eigenvalue of the operator @xmath60 and letting @xmath42 be a bound on the next largest eigenvalue such that @xmath44 , the result follows . @xmath121 we now turn our attention to the comb graph . see figure [ figure_comb ] . recall that the comb graph is defined by the product @xmath122 . in this case the space @xmath47 is the triangle @xmath123 however , following proposition [ proposition_subspace ] in order to find the eigenvalue and eigenfunctions of @xmath60 it is enough to consider the subspace @xmath100 of @xmath61 consisting of functions depending only on the variable @xmath124 . observe that @xmath100 inherits the inner product @xmath125 moreover , the operator @xmath60 is given by @xmath62(x ) = \int_{0}^{1-x } ( 1-z ) \cdot f(z ) \ : dz . \label{eq : comb : integral}\ ] ] the next step is to find all of the eigenvalues and eigenfunctions of the operator @xmath60 , that is , functions @xmath126 so that @xmath127(x)=\int_{0}^{1-x } ( 1-z ) \cdot f(z ) \ : dz . \label{eq : comb : eig : fun}\ ] ] we convert this integral equation into a differential equation by differentiating to obtain @xmath128 to convert this into an ordinary differential equation ( ode ) , we define @xmath129 and thus @xmath130 together with the boundary conditions @xmath131 which come from the integral equation and the algebraic relationship between @xmath112 and @xmath132 , this linear system is equivalent to the original integral equation . the condition that @xmath133 is our choice of normalization for the eigenfunctions . the only solution which has @xmath134 is identically zero , thus this normalization is valid . we proceed to solve this ode numerically . first , we solve the system from 0 to 1/2 for various values of @xmath41 and find the difference @xmath135 . this is plotted in figure [ fig : lambda:2 ] . the roots of this plot correspond to eigenvalues and we find them numerically . for each eigenvalue @xmath41 the functions @xmath112 and @xmath132 are found and @xmath126 over the unit interval is reconstructed . the eigenfunctions corresponding to the largest ( in absolute value ) eigenvalues are plotted in figure [ fig : eig : fun:2 ] . the resulting norms and constants @xmath136 are tabulated in table [ tab : num : vals:2 ] . .the values of @xmath41 , @xmath137 , @xmath138 , and @xmath136 for the first four eigenfunctions shown in figure [ fig : eig : fun:2 ] . the constant @xmath136 is the ratio @xmath139 for the @xmath140 eigenfunction . [ cols="<,<,<,<",options="header " , ] the resulting predictions for the euler numbers are shown in the last column of table [ tab : e_n ] . another graph to investigate is the product with the even cycle @xmath141 , that is , @xmath142 . we conjecture that the resulting euler number is asymptotically a constant times the associated euler number for the product with a path , that is , @xmath143 as @xmath29 tends to infinity and @xmath43 is a positive constant less than @xmath67 for @xmath30 non - empty . does the eigenfunction @xmath118 corresponding to the largest eigenvalue @xmath41 carry information about the distribution of entries in the first copy of @xmath2 in an alternating labeling of @xmath91 ? more specifically , in the case of alternating @xmath144 by @xmath29 arrays , let @xmath145 be the number of alternating arrays where the first column consists of the two entries @xmath146 and @xmath147 , where @xmath148 . is the integer @xmath145 approximated by @xmath149 where @xmath43 is the appropriate constant and @xmath132 is the first eigenfunction displayed in figure [ fig : eigen : fun ] ? these techniques for obtaining the asymptotic behavior of the euler numbers can be used for other classes of graphs as well . see for instance the graph @xmath150 in figure [ figure_hexagons ] , which is built by gluing hexagons together . although theorem [ theorem_asymptotic ] does not directly apply to this class of graphs , one can extend the theory to obtain the same asymptotic result . hence we have @xmath151 the essential question remaining is can the associated eigenvalue problem be solved explicitly . ( 100,30)(0,0 ) ( 0,0)(20,0)2(0,10 ) ( 0,20 ) ( 10,0 ) ( 10,30 ) ( 20,10 ) ( 20,20 ) ( 0,10)(0,1)10 ( 0,10)(1,-1)10 ( 0,20)(1,1)10 ( 10,0)(1,1)10 ( 10,30)(1,-1)10 ( 20,10)(0,1)10 ( 80,0)(20,0)1(0,10 ) ( 0,20 ) ( 10,0 ) ( 10,30 ) ( 20,10 ) ( 20,20 ) ( 0,10)(0,1)10 ( 0,10)(1,-1)10 ( 0,20)(1,1)10 ( 10,0)(1,1)10 ( 10,30)(1,-1)10 ( 20,10)(0,1)10 ( 40,10)(1,-1)8 ( 40,20)(1,1)8 ( 80,10)(-1,-1)8 ( 80,20)(-1,1)8 ( 55,15 ) ( 60,15 ) ( 65,15 ) keeping @xmath0 fixed we know that @xmath152 for a constant @xmath153 and the largest eigenvalue @xmath154 . can anything be determined about the sequence @xmath154 ? what can be said about the asymptotics of the euler number @xmath155 as @xmath29 tends to infinity ? a different direction is to study the descent number of directed graphs ( digraphs ) . for a digraph @xmath46 on @xmath0 vertices define its _ descent number _ to be the number of labelings @xmath9 of the vertices with @xmath67 through @xmath0 such that for each directed edge @xmath156 we have that @xmath13 . if the digraph contains a directed cycle then the descent number is zero . for an acyclic digraph ( digraphs without directed cycles ) the descent number is strictly positive . the classical descent set statistics for permutations is obtained be looking at orientations of the path . by gluing directed graphs together , one obtains classes of graphs whose asymptotics of the descent number is natural to study via linear operators and their eigenvalues . the technique of translating a combinatorial problem into a problem of studying an operator and its spectrum was also explored in @xcite , where consecutive pattern avoiding in permutations were studied . finally , we end with a purely enumerative question for trees ( connected graphs without cycles ) . for a tree @xmath60 on @xmath0 vertices the classical euler number @xmath19 is a lower bound for @xmath157 , that is , @xmath158 furthermore , equality only holds when the tree @xmath60 is the path @xmath17 . the first author thanks bob strichartz and the department of mathematics at mit where this paper was completed . the first author was partially supported by national security agency grant h98230 - 06 - 1 - 0072 . linear operators leaving invariant a cone in a banach space . _ uspehi matem . nauk ( n.s . ) _ * 3 * ( 1948 ) , no . 1 ( 23 ) , 395 . english translation in _ amer . math . translation _ * 1950 * ( 1950 ) , no . 26 , 1128 . _ r. ehrenborg , department of mathematics , university of kentucky , lexington , ky 40506 - 0027 _ , + [email protected] + _ y. farjoun , department of mathematics , mit , cambridge , ma 02139 - 4307 _ , + [email protected] .	we define the euler number of a bipartite graph on @xmath0 vertices to be the number of labelings of the vertices with @xmath1 such that the vertices alternate in being local maxima and local minima . we reformulate the problem of computing the euler number of certain subgraphs of the cartesian product of a graph @xmath2 with the path @xmath3 in terms of self adjoint operators . the asymptotic expansion of the euler number is given in terms of the eigenvalues of the associated operator . for two classes of graphs , the comb graphs and the cartesian product @xmath4 , we numerically solve the eigenvalue problem .
a number of recent observational studies have revealed a new population of red , massive galaxies at redshift @xmath24 ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , utilizing near - infrared ( ir ) wavelengths which are relatively less affected by dust extinction . at the same time , a number of studies focused on the assembly of stellar mass density at high redshift by comparing observational data and semi - analytic models of galaxy formation ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? these works argued that the hierarchical structure formation theory may have difficulty in accounting for sufficient early star formation . these concerns grew with the mounting evidence for high redshift galaxy formation including the discovery of extremely red objects ( eros ) at @xmath25 ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , sub - millimeter galaxies at @xmath26 ( e.g. * ? ? ? * ; * ? ? ? * ) , lyman break galaxies ( lbgs ) at @xmath17 ( e.g. * ? ? ? * ) , and galaxies at @xmath27 , detected either by their lyman-@xmath28 emission ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) or by their optical to near - ir colors ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? we now face the important question as to whether this evidence for high - redshift galaxy formation is consistent with the concordance @xmath1cdm model . the redshift range around @xmath29 is a particularly interesting epoch also for another reason . the redshift interval @xmath30 has long been known as the ` redshift desert ' @xcite for galaxy surveys , because there was no large volume limited sample of galaxies available in this regime until very recently . this is because strong emission lines from h ii regions of galaxies , such as [ o ii ] @xmath313727 , [ o iii ] @xmath32 4959 , 5007 , h@xmath28 and h@xmath33 , redshift out of optical wavelengths above 9300 . these spectral features are necessary to easily identify galaxy redshifts , allowing ground - based telescopes to benefit from the low - night sky background and high atmospheric transmission in the wavelengths range 4000 9000 . however , recently @xcite and @xcite have introduced new techniques for exploring the ` redshift desert ' , making it possible to identify a large number of galaxies efficiently with the help of a color selection criteria in the color - color plane of @xmath34 vs. @xmath8 . in this technique , galaxies at @xmath35 are located photometrically from the mild drop in the @xmath36 filter owing to the ly-@xmath28 forest opacity , and galaxies at @xmath37 are recognized from the lack of a break in their observed - frame optical spectra . the large sample of galaxies identified by these authors at @xmath0 makes it now possible to study galaxy formation and evolution for over 10 gyrs of cosmic time , from redshift @xmath6 to @xmath38 , without a significant gap . we note that the epoch around @xmath0 is particularly important for understanding galaxy evolution because at around this time , the number density of quasi - stellar objects ( qsos ) peaked ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and the ultra - violet ( uv ) luminosity density began its decline by about an order of magnitude from @xmath24 to @xmath38 ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? these recent observational studies of galaxies at @xmath0 , both in the uv and near - ir wavelengths , imply a range of novel tests for the hierarchical structure formation theory . in this study , we analyze the properties of massive galaxies at @xmath0 formed in state - of - the - art cosmological hydrodynamic simulations of the @xmath1cdm model , and we compare them with observations . in an earlier recent study @xcite , we argued that , based on two different types of hydrodynamic simulations ( eulerian tvd and sph ) and the theoretical model of @xcite ( hereafter h&s model ) , the predicted cosmic star formation rate ( sfr ) density peaks at @xmath39 , with a _ higher _ stellar mass density at @xmath6 than suggested by current observations ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , in contrast to some claims to the contrary @xcite . we also compared our results with those from the updated semi - analytic models of @xcite , @xcite , and @xcite , and found that our simulations and the h&s model predicts an earlier peak of the sfr density and a faster development of stellar mass density compared to these semi - analytic models . it is then interesting to examine what our simulations predict for the properties of massive galaxies at @xmath24 . in this paper , we analyze for this purpose the same set of hydrodynamic simulations that was used in @xcite , with a special focus on the most massive galaxies at this epoch . in section [ sec : simulation ] , we briefly describe the simulations that we use . in section [ sec : method ] , we summarize our method for computing spectra of simulated galaxies both in the rest - frame and the observed - frame . in section [ sec : colcol ] , we show the color - color diagrams and color - magnitude diagrams of simulated galaxies . in section [ sec : mstar ] we discuss the stellar masses and number density of galaxies at @xmath0 . we then investigate the observed - frame @xmath40-band luminosity function and the rest - frame @xmath20-band luminosity function in section [ sec : lf ] , followed by an analysis of the star formation histories in section [ sec : sfhistory ] . finally , we summarize and discuss the implications of our work in section [ sec : discussion ] . we will discuss results from two different types of cosmological hydrodynamic simulations . both approaches include `` standard '' physics such as radiative cooling / heating , star formation , and supernova ( sn ) feedback , although the details of the models and the parameter choices vary considerably . one set of simulations was performed using an eulerian approach , which employed a particle - mesh method for the gravity and the total variation diminishing ( tvd ) method @xcite for the hydrodynamics , both with a fixed mesh . the treatment of the radiative cooling and heating is described in @xcite in detail . the structure of the code is similar to that of @xcite , but the code has significantly improved over the years with additional input physics . it has been used for a variety of studies , including the evolution of the intergalactic medium @xcite , damped lyman-@xmath28 absorbers @xcite , and galaxy formation ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? our other simulations were done using the lagrangian smoothed particle hydrodynamics ( sph ) technique . we use an updated version of the code @xcite , which uses an ` entropy conserving ' formulation @xcite of sph to mitigate problems with energy / entropy conservation ( e.g. * ? ? ? * ) and overcooling . the code also employed a subresolution multiphase model of the interstellar medium to describe self - regulated star formation which includes a phenomenological model for feedback by galactic winds @xcite , and the impact of a uniform ionizing radiation field @xcite . this approach has been used to study the evolution of the cosmic sfr @xcite , damped lyman-@xmath28 absorbers @xcite , lyman - break galaxies @xcite , disk formation @xcite , emission from the intergalactic medium @xcite , and the detectability of high redshift galaxies @xcite . in both codes , at each time - step , some fraction of gas is converted into star particles in the regions that satisfy a set of star formation criteria ( e.g. significantly overdense , jeans unstable , fast cooling , converging gas flow ) . upon their creation , the star particles are tagged with physical parameters such as mass , formation time , and metallicity . after forming , they interact with dark matter and gas only gravitationally as collisionless particles . more specifically , in the tvd simulation , the star formation rate is formulated as @xmath41 where @xmath42 is the star formation efficiency and @xmath43 is the star formation time - scale . in the tvd n864l22 run , @xmath44 and @xmath45 were used , where @xmath46 is the local dynamical time - scale owing to gravity . at each time - step , a part of the gas in a cell is converted to a star particle according to equation ( [ eq : tvd ] ) , provided the gas is ( a ) moderately overdense ( @xmath47 ) , ( b ) jeans unstable ( @xmath48 ) , ( c ) cooling fast ( @xmath49 ) , and ( d ) the flow is converging into the cell ( @xmath50 ) . on the other hand , the sph simulations analyzed in this study parameterized the star formation rate as @xmath51 where @xmath33 is the mass fraction of short - lived stars that instantly die as supernovae ( taken to be @xmath52 ) . the star formation time - scale @xmath43 is again taken to be proportional to the local dynamical time of the gas : @xmath53 , where the value of @xmath54 is chosen to match the @xcite law . in the model of @xcite , this parameter simultaneously determines a threshold density @xmath55 , above which a multiphase structure of the gas , and hence star formation , is allowed to develop . this physical density is @xmath56 for the simulations in this study , corresponding to a comoving baryonic overdensity of @xmath57 at @xmath38 . ( see * ? ? ? * for a description of how @xmath55 is determined self - consistently within the model . ) note that the density @xmath58 that determines the star formation rate in equation ( [ eq : sph ] ) is actually the average density of cold clouds determined by a simplified equilibrium model of a multiphase interstellar medium , rather than the total gas density . however , in star - forming regions , the density is so high that the mean density in cold gas always dominates , being close to the total gas density . we will further discuss the differences between our star formation recipes in the two codes in some more detail in section [ sec : discussion ] . the cosmological parameters adopted in the simulations are intended to be consistent with recent observational determinations ( e.g. * ? ? ? * ) , as summarized in table [ table : simulation ] where we list the most important numerical parameters of our primary runs . while there are many similarities in the physical treatment between the two approaches ( mesh tvd and sph ) , we see that the tvd simulation has somewhat higher mass resolution , and the sph somewhat higher spatial resolution . in that sense , the two approaches are complementary and results found in common are expected to be robust . we begin our analysis by identifying galaxies in the simulations as groups of star particles . since the two simulations use inherently different methods for solving hydrodynamics , we employ slightly different methods for locating galaxies in the two types of simulations , each developed to suit the particular simulation method well . because the stellar groups of individual galaxies are typically isolated , there is little ambiguity in their identification , such that systematic differences owing to group finding methods are largely negligible . for the tvd simulation , we use a version of the algorithm hop @xcite to identify groups of star particles . this grouping method consists of two discrete steps : in the first step , the code computes the overdensity at the location of each particle , and in a second step , it then merges identified density peaks based on a set of overdensity threshold parameters specified by the user . there are three threshold parameters : a peak overdensity @xmath59 above which the density peak is identified as a candidate galaxy , an outer overdensity @xmath60 which defines the outer contour of groups , and finally a saddle point overdensity @xmath61 which determines whether two density peaks within the outer density contour should be merged . @xcite suggest using @xmath62 for identifying dark matter halos in n - body simulations . we applied the algorithm to the star particles in our simulation using various combinations of these parameters , and found that the results do not depend very much on the detailed choice . this reflects the higher concentration and isolation of star particle groups as compared to the dark matter , making group finding typically straightforward . however , the stellar peak overdensities relative to the mean are so high that the saddle - overdensity parameter in hop is unimportant , and the small galaxies associated with a large central galaxy at the center of a large dark matter halo are often not well separated by this algorithm . therefore , we decided not to apply the second step of the algorithm . instead , we merge two density peaks identified by the first step only if they are within two neighboring cells . by visually inspecting the particle distribution of very massive dark matter halos , we found that this method succeeds in separating some of the small galaxies embedded in a large dark matter halo especially at the outskirts of the halo , and helps to somewhat mitigate the ` overmerging problem ' in high density regions . however , even with this revised method , the central massive galaxy remains afterwards , and the addition of a small number of galaxies to the faint - end of the luminosity function does not change the qualitative feature of our results . clearly more work is needed to improve this situation on the grouping in the future , for example , using an improved grouping method such as voboz @xcite . for the sph simulations , we employ the same method as in @xcite , where we identified isolated groups of star particles using a simplified variant of the subfind algorithm proposed by @xcite . in short , we first compute an adaptively smoothed baryonic density field for all star and gas particles . this allows us to robustly identify centers of individual galaxies as isolated density peaks . we then find the full extent of these galaxies by processing the gas and star particles in order of declining density , adding particles one by one to the galaxies . if all of the 32 nearest neighbors of a particle have lower density , this particle is considered to be a new galaxy ` seed ' . otherwise , the particle is attached to the galaxy that its nearest denser neighbor already belongs to . if the two nearest denser neighbors belong to different galaxies , and one of these galaxies has less than 32 particles , these galaxies are merged . if the two nearest denser neighbors belong to different galaxies and both of these galaxies have more than 32 particles , then the particle is attached to the larger group of the two , leaving the other one intact . finally , we restrict the set of gas particles processed in this way to those particles which have at least a density of @xmath63 times the threshold density for star formation , i.e. @xmath64 ( see * ? ? ? * for details on how the star formation threshold is determined ) . note that we are here not interested in the gaseous components of galaxies we only include gas particles because they make the method more robust owing to the improved sampling of the baryonic density field . we found that the above method robustly links star particles that belong to the same isolated galaxy in sph simulations . a simpler friends - of - friends algorithm with a small linking length can achieve a similar result , but the particular choice one needs to make for the linking length in this method represents a problematic compromise , either leading to artificial merging of galaxies if it is too large , or to loss of star particles that went astray from the dense galactic core , if it is too small . note that , unlike in the detection of dark matter substructures , no gravitational unbinding algorithm is needed to define the groups of stars that make up the galaxies formed in the simulations . we consider only galaxies with at least 32 particles ( star and gas particles combined ) in our subsequent analysis . for both simulations , each star particle is tagged with its mass , formation time , and metallicity of the gas particle that it was spawned from . based on these three tags , we compute the emission from each star particle , and co - add the flux from all particles for a given galaxy to obtain the spectrum of the simulated galaxy . we use the population synthesis model by @xcite which assumes a @xcite initial mass function ( imf ) within a mass range of @xmath65\,{{\rm m}_{\odot}}$ ] , as recommended by @xcite . spectral properties obtained with this imf are very similar to those obtained using the @xcite imf , but the @xcite imf provides a better fit to counts of low - mass stars and brown dwarfs in the galactic disk @xcite . we use the high resolution version of the spectral library of @xcite which contains 221 spectra describing the spectral evolution of a simple stellar population from @xmath66 to @xmath67 gyr over 6900 wavelength points in the range of 91 - 160 @xmath68 m . once the intrinsic spectrum is computed , we apply the @xcite extinction law with three different values of @xmath69 in order to investigate the impact of extinction . these values span the range of extinction estimated from observations of lbgs at @xmath6 @xcite . using the spectra computed in this manner , we then derive rest - frame colors and luminosity functions of the simulated galaxies . to obtain the spectra in the observed frame , we redshift the spectra and apply absorption by the intergalactic medium ( igm ) following the prescription of @xcite . once the redshifted spectra in the observed frame are obtained , we convolve them with different filter functions , including @xmath2 @xcite and standard johnson bands , and compute the magnitudes in both ab and vega systems . apparent @xmath36 , @xmath70 , @xmath40 magnitudes are computed in the ab system to compare our results with @xcite and @xcite , while the rest - frame v - band magnitude is computed in the vega system . in figure [ colcol.eps ] , we show the color - color diagrams of simulated galaxies at @xmath0 in the observed - frame @xmath71 vs. @xmath8 plane , for both sph ( top 2 panels ) and tvd ( bottom right panel ) simulations . only galaxies brighter than @xmath72 are shown in order to match the magnitude limit of the sample of @xcite . we plot different symbols for three different values of calzetti extinction : @xmath73 ( blue dots ) , 0.15 ( green crosses ) , and 0.3 ( red open squares ) . as the level of extinction by dust is increased from @xmath73 to 0.3 , the measured points move towards the upper right corner of each panel . this behavior is expected for a conventional star - forming galaxy spectrum , as demonstrated in figure 2 of @xcite . the color - selection criteria used by @xcite and @xcite is shown by the long - dashed boxes . the upper ( lower ) box corresponds to their bx ( bm ) criteria for selecting galaxies at @xmath74 ( @xmath75 ) . the number density of all the galaxies shown ( i.e. @xmath4 ) is given in each panel for @xmath73 , 0.15 , 0.30 from top to bottom , respectively . it is encouraging to see that in all the panels most of the simulated galaxies actually satisfy the observational color selection criteria for the case of @xmath3 . this suggests that the simulated galaxies have realistic uv colors compared to the observed ones . simulated galaxies with no extinction tend to be too blue compared to the color - selection criteria . in the sph g6 run , the distribution is wider than in the sph d5 and tvd n864l22 runs , owing to the larger number of galaxies in a larger simulation box , and the distribution actually extends beyond the color selection boundaries . the spatial number density of galaxies with @xmath4 is about @xmath76 for the sph g6 run . we will discuss the number density of galaxies with various cuts in figure [ nden.eps ] in more detail . we repeated the same calculation with the sloan digital sky survey s @xmath77 filter set , and found that the distribution of points hardly changes , even when we exchange @xmath78 with @xmath79 , and @xmath8 with @xmath80 . this means that searches for galaxies in the redshift range @xmath30 can also substitute the sloan filters @xmath81 when applying a color - selection following the ` bx ' and ` bm ' criteria adopted by @xcite and @xcite . in figure [ colmag.eps ] , we show the simulated galaxies at @xmath0 in the plane of apparent @xmath40 magnitude and @xmath8 color . again , we use three different symbols for three different values of extinction : @xmath73 ( blue dots ) , 0.15 ( green crosses ) , and 0.3 ( red open squares ) . the vertical long - dashed line and the arrow indicate the magnitude limit of @xmath72 used by @xcite . the horizontal line at @xmath82 corresponds to the upper limit of the color - selection box of @xcite and @xcite . we see that most of the galaxies brighter than @xmath72 automatically satisfy the criterion @xmath83 , and almost no galaxies with @xmath4 fall out of the region for @xmath3 . there is a significant population of dim ( @xmath84 ) galaxies with @xmath85 . we will see below that these are low - mass galaxies with stellar masses @xmath86 . we will discuss the amount of stellar mass contained in the galaxies that satisfy the color selection criteria in section [ sec : mstar ] . in figure [ rmag_mstar.eps ] , we show the apparent @xmath87 magnitude vs. stellar mass of simulated galaxies at redshift @xmath0 , for both sph ( top 2 panels ) and tvd ( bottom right panel ) simulations . the three different symbols correspond to three different values of extinction : @xmath73 ( blue dots ) , 0.15 ( green crosses ) , and 0.3 ( red open squares ) . the vertical long - dashed line and the arrow indicate the magnitude limit of @xmath72 used by @xcite . from this figure , we see that there are many galaxies with stellar masses larger than @xmath88 and @xmath4 at @xmath0 in our simulations . in the sph g6 run , the masses of the most luminous galaxies are substantially larger than those in the d5 run . but this is primarily a result of a finite box size effect ; the brightest galaxies have a very low space - density , and so they are simply not found in simulations with too small a volume . in the sph g6 run , the most massive galaxies at @xmath0 have masses @xmath89 with a number density of @xmath90 . such galaxies are hence not commonly found in a comoving box - size of @xmath91 , in fact , only one such galaxy exists in the d5 run . cosmic variance hence significantly affects the resulting number density , and one ideally needs a simulation box larger than @xmath92 to obtain a reliable estimate of the number density of such massive galaxies . we note however that a couple of such massive galaxies are also found in the tvd n864l22 run , which is presumably in part a result of overmerging owing to the relatively large cell size compared to the gravitational softening length of the d5 run . in figure [ nden.eps ] , we summarize the results on the cumulative number densities of galaxies , either measured above a certain threshold value of stellar mass [ panel ( a ) ] , or below a threshold value of observed - frame @xmath40-band magnitude [ panel ( b ) ] , or rest - frame @xmath20-band magnitude [ panel ( c ) ] . in panel ( a ) , we see that the results of different runs agree reasonably well in the threshold mass range of @xmath93 , but differ somewhat at the lower and higher mass thresholds . at the low - mass end , the lower resolution of the g6 run results in a lower number density than in the d5 run . at the high - mass end , the smaller box - size limits the number density in the d5 run compared to the higher value in the g6 run . the result of the tvd run falls in between those of d5 and g6 at the low - mass end , but gives larger values relative to the sph runs at the high - mass end , probably owing to the overmerging problem . a similar level of agreement can be seen in panel ( c ) , where the rest - frame magnitude is used for the @xmath94-axis , but in panel ( b ) , comparatively large differences are present partly owing to the larger stretch in the abscissa . we find that the amount of stellar mass contained in the galaxies that satisfy the color - selection criteria is in general far from being close to the total stellar mass . owing to the limited box - size and resolution effects , it is somewhat difficult to obtain an accurate estimate of the mass fraction above some magnitude limit . however , we list the stellar mass fractions as a reference for future work . in the g6 run , which has comparatively low mass resolution , the stellar mass fraction contained in galaxies that pass the criteria ( i.e. @xmath4 & color - selection ) account for only ( 60 , 65 , 42)% of all the stars , for extinctions of @xmath95 , respectively . these fractions are relatively large compared to other runs , because , as we will see in section [ sec : lf ] , the luminosity function in the g6 run starts to fall short of galaxies near @xmath96 compared to the d5 run owing to its limited resolution , and the g6 run lacks low - mass faint galaxies with @xmath19 . on the other hand , in the d5 run , the stellar mass fraction contained in galaxies that pass the criteria ( i.e. @xmath4 & color - selection ) account for only ( 32 , 17 , 2)% , and ( 16 , 13 , 1)% of the stars in the tvd n864l22 run , for extinctions of @xmath95 , respectively . these fractions are smaller compared to those of the g6 run , because the d5 and tvd n864l22 runs lack massive bright galaxies owing to their smaller box sizes . if we relax the criteria from ` @xmath4 & color - selection ' to only @xmath4 , then the corresponding stellar mass fractions increase to ( 75 , 65 , 44)% for g6 , ( 32 , 17 , 3)% for d5 , and ( 46 , 34 , 29)% for tvd n864l22 . in summary , we expect that the true stellar mass fraction that satisfies the criteria lies somewhere between the results of g6 and d5 ( and tvd n864l22 ) runs , which should be roughly @xmath97 . this implies that the current surveys with a magnitude limit of @xmath96 at best account for half of the stellar mass in the universe . @xcite reached a similar conclusion based on different theoretical arguments that compared the results of our hydrodynamic simulations and the theoretical model of @xcite with near - ir observations of galaxies ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . @xcite and @xcite suggested that the unaccounted - for contribution to the stellar mass might be hidden in a red population of galaxies . our simulation suggests that the missed stellar masses are mostly contained in the fainter ( @xmath19 ) and redder galaxies . but note that we have not considered large values of extinction with @xmath98 . in figure [ mstar_col.eps ] , we show the observed - frame @xmath8 color vs. stellar mass of simulated galaxies at @xmath0 for both sph ( top 2 panels ) and tvd ( bottom right panel ) simulations . here , only the case of @xmath3 is shown . in the top two panels for the sph runs , the red open crosses show the galaxies that are brighter than @xmath72 , and the blue dots show the rest of the galaxies . the discrete stripes seen at low - mass end of the distribution for the sph runs are due to the discreteness of the star particles in the simulation . for the tvd run , there is a strong concentration of points at @xmath99 , which we will discuss in detail in section [ sec : sfhistory ] . an interesting point to note in this figure is that the bluest galaxies at @xmath0 are not necessarily the most massive ones . rather , the bluest galaxies with colors @xmath100 are galaxies with somewhat lower mass in the range @xmath101 . the most massive galaxies with @xmath102 have slightly redder colors of @xmath103 owing to their underlying old stellar population . in the tvd simulation , there are a couple of galaxies with @xmath104 and @xmath105 , which are absent in the sph simulations . these galaxies are dominated by the old stellar population because they have completed their star formation before @xmath17 as we will see in section [ sec : sfhistory ] . figure [ lf_r.eps ] shows the observed - frame @xmath40-band luminosity function for all simulated galaxies at @xmath0 in both sph ( top 2 panels ) and the tvd ( bottom left panel ) simulations . in these 3 panels , blue long - dashed , green solid , and red dot - short - dashed lines correspond to @xmath73 , 0.15 , 0.30 , respectively . the vertical dotted lines indicate the magnitude limit of @xmath72 used by @xcite . in the bottom right panel , a combined result is shown with the sph d5 ( green solid line ) , g6 ( red long - dashed line ) , and tvd ( blue dot - dashed line ) results overplotted for the case of @xmath3 . for comparison , schechter functions with the following parameters ( simply chosen by eyeball fit ) are shown as short - dashed lines : @xmath106 for sph d5 & g6 runs ( top two panels ) and the combined results ( bottom right panel ) , and @xmath107 for tvd ( bottom left ) , respectively . the bottom right panel of figure [ lf_r.eps ] gives a relative comparison between the different runs . we see that the sph g6 run contains a larger number of luminous galaxies with @xmath108 compared to the other runs , an effect that results from its larger box - size . on the other hand , the g6 run lacks fainter galaxies at @xmath109 compared to d5 run owing to its limited mass resolution . the sph g6 run hence covers the bright - end of the schechter function with @xmath110 , while the d5 run covers better the faint - end of the luminosity function with a slope of @xmath111 . the tvd n864l22 run agrees with the d5 results at the bright - end near @xmath112 , but the deviation from the d5 result starts to become somewhat large at @xmath113 . we will discuss a number of possible reasons for these differences between the sph and tvd runs in section [ sec : discussion ] . in figure [ lf_v.eps ] , we show the rest - frame @xmath20-band luminosity function of all simulated galaxies at @xmath0 for both sph ( top 2 panels ) and tvd ( bottom left panel ) simulations . in these three panels , blue long - dashed , green solid , and red dot - short - dashed lines correspond to extinctions of @xmath73 , 0.15 , 0.30 , respectively . in the bottom right panel , the combined result is shown with the sph d5 ( green solid line ) , g6 ( red long - dashed ) , and tvd ( blue dot - dashed ) runs overplotted for the case of @xmath3 . for reference , the magnitude limit of @xmath114 ( for @xmath115 ) for the survey of @xmath6 lbgs by @xcite is indicated by the vertical dotted line and the arrow . for comparison , we also include schechter function fits with parameters ( @xmath116 as short - dashed curves in all panels . these schechter parameters are the same as the ones that @xcite fitted to their observational data , except that the characteristic magnitude @xmath117 is 0.5 magnitudes brighter . the same amount of brightening is seen in the g6 run from @xmath6 ( see fig . 5 of @xcite has to be added to obtain correct ab magnitudes . the corrected version of the paper will be posted at astro - ph/0311295 . the rest - frame @xmath20-band results were not affected by this error . ] ) to @xmath0 . so the g6 run with @xmath3 describes the bright - end of the luminosity function at @xmath0 and @xmath6 quite well . in the bottom left panel for the tvd run , another schechter function with a much shallower faint - end slope and parameters @xmath118 is also shown with a magenta long - dash - short - dashed line for comparison . one can see that the tvd run contains more fainter galaxies with @xmath119 than the sph runs . this could be due to the higher baryonic mass resolution and the additional small - scale power in the tvd run compared with the sph runs as we will discuss more in section [ sec : sfhistory ] . we can compute the expected characteristic magnitude @xmath120 of @xmath0 galaxies from the observational estimate of @xmath120 at @xmath6 . the difference in the luminosity distance between @xmath6 and @xmath0 corresponds to about one magnitude for our adopted cosmology , and we saw in the previous paragraph that the rest - frame @xmath20-band luminosity function brightened by @xmath121 magnitude from @xmath6 to @xmath0 . therefore , we expect that @xmath120 should be brighter by about 1.5 magnitudes at @xmath0 compared to @xmath6 . @xcite reported @xmath122 for the observed - frame @xmath40-band luminosity function of the @xmath6 galaxies . therefore we obtain @xmath123 as the expected value for the @xmath0 galaxies . the result of the sph g6 run shown in fig . [ lf_r.eps ] is reasonably close to this expected value . it is also clear that the box size of the d5 and tvd runs are too small to reliably sample the brightest galaxies at @xmath0 . this is also seen in fig . [ lf_v.eps ] as a lack of brightest galaxies at @xmath124 in the d5 and tvd runs . this is because the brightest galaxies form in rare high density peaks which are at the knots of the large - scale filaments , and one needs a sufficiently large volume to sample a fair number of such bright galaxies reliably . since the comoving correlation length of lbgs at @xmath6 are @xmath125 @xcite , one needs a box size larger than at least @xmath126 in order to have a fair sample of lbgs at @xmath6 . the simulated volume of the g6 run is @xmath127 comoving , and one of the face of the simulation box corresponds to 2.5 @xmath128 at @xmath6 . the coverage area of the lbg survey of @xcite at @xmath6 is about 0.4 @xmath128 , and the depth of their survey is @xmath129 which corresponds to @xmath130 comoving for our adopted flat-@xmath1 cosmology . therefore the volumes of the g6 simulation and the steidel et al . survey are comparable . in figures [ sfhist_d5.eps ] , [ sfhist_g6.eps ] , and [ sfhist_tvd.eps ] , we show the star formation ( sf ) histories of massive galaxies at @xmath0 as a function of cosmic age , for the sph d5 , g6 , and tvd runs , respectively . in these figures , panels @xmath131 & @xmath132 show results for the two most massive galaxies in each simulation box , while panels @xmath133 & @xmath134 are for the two reddest galaxies among those with @xmath16 . in the right side of each panel , the following quantities for the case of @xmath3 are listed from top to bottom : stellar mass in units of @xmath135 , rest - frame v - band magnitude , observed - frame @xmath40-band magnitude , and @xmath8 color . in panel @xmath136 , the cumulative star formation history of the galaxies in panels @xmath137 is shown . the results of the sph d5 and g6 runs ( fig . [ sfhist_d5.eps ] & [ sfhist_g6.eps ] ) suggest that the most massive galaxies have almost continuously formed stars with a rate exceeding @xmath10 over a few gyrs from @xmath11 to @xmath0 ( note that @xmath138 ) . half of their stellar mass was already assembled by @xmath139 . the reddest galaxies ( panels @xmath133 & @xmath134 ) have colors of @xmath140 , and their star formation has become less active after @xmath6 , with the bulk of their stellar mass created before @xmath6 as can be seen in the panel @xmath136 . the tvd simulation ( fig . [ sfhist_tvd.eps ] ) on the other hand suggests a more sporadic star formation history than the sph simulations , and the two most massive galaxies in the simulation box experienced massive star formation between @xmath6 and @xmath0 that exceeded @xmath141 for brief periods . the peak star formation rate of the most massive galaxy in the g6 run also reaches @xmath141 occasionally . in the real universe , such violent starburst activity is found in high - redshift quasars and submm sources where the existence of dense , warm , and massive molecular clouds of @xmath142 is inferred from the observed rotational emission line of co ( e.g. * ? ? ? * ; * ? ? ? the number density of quasar host galaxies with such a violent star formation activity is not well - constrained yet , but it probably is not very large because the space density of quasars with @xmath143 is smaller than @xmath144 ( e.g. * ? ? ? the space density of submm sources is also similar : @xmath145 at @xmath24 @xcite . these values are much smaller than the space density of lyman break galaxies @xmath146 @xcite . finding 3 such massive starbursts at @xmath0 in our tvd simulation box of @xmath147 results in a number density of @xmath148 , which may be too high compared to the observed value . it is also possible that the two galaxies shown in figure [ sfhist_tvd.eps ] are affected by the overmerging problem which tends to merge galaxies in very high density regions owing to a lack of spatial resolution . if these two massive galaxies were broken up into several less massive galaxies , then the @xmath141 of star formation would be distributed to a few hundreds @xmath149 for each galaxy . the two reddest massive galaxies in the tvd run ( fig . [ sfhist_tvd.eps]c & [ sfhist_tvd.eps]d ) have cumulative star formation histories which are similar to those of the sph runs , with half of their stellar mass already assembled by @xmath12 . they have completed their star formation by @xmath17 and have been quiet from @xmath6 to @xmath0 . because they are dominated by an old stellar population , they have very red colors of @xmath104 and show up in the upper right corner of the tvd ( bottom right ) panel in fig . [ colcol.eps ] . another way of looking at the star formation history of galaxies is to look at the distribution of formation times of stellar particles in the simulation for a given set of galaxies as a whole ; i.e. we here analyze the combined star formation history for a particular class of galaxies divided by their stellar mass . in figure [ age_dist.eps ] , we show the mass - weighted probability distribution of stellar masses as a function of formation time ( @xmath150 corresponds to the big bang ) , for different samples of galaxies . to this end , we have split the total galaxy sample into 3 categories by stellar mass at @xmath0 : @xmath16 ( top panel ) , @xmath151 ( middle panel ) , and @xmath152 ( bottom panel ) . different line types correspond to different simulations : red long - dashed ( sph g6 run ) , blue solid ( sph d5 run ) , and black dot - dashed ( tvd n864l22 run ) . the vertical dotted lines indicate four different epochs , namely @xmath0 , 3 , 4 , & 5 , as indicated in the top panel . in the two top panels for massive galaxies , the distribution roughly agrees between the different runs , except for the large bump in the tvd run close to @xmath0 , which is caused by starbursts in a few massive galaxies . also note that the distribution of g6 run is slightly more bumpy owing to lack of mass resolution compared to other runs . in the bottom panel for the less massive galaxies , the distribution is very different between the 3 runs . in the tvd run , the bulk of stars in low - mass galaxies forms before @xmath139 , whereas the opposite is true for the sph g6 run . these low - mass galaxies that formed very early on appeared as a strong concentration of points at @xmath153 in fig . [ mstar_col.eps ] for the tvd n864l22 run . the result for the sph d5 appears to be intermediate between the two . this could perhaps be caused by the additional small - scale power in the tvd run compared with the sph runs . the initial mean inter - particle separations of both dark matter and gas particles in the two sph runs are comoving 104 and @xmath154 for the d5 and g6 run , respectively , but the mean inter - particle separation of dark matter particles in the tvd run is comoving @xmath155 , with a baryonic cell size of @xmath156 . therefore , at very early times , the tvd run should be able to resolve smaller matter fluctuations than the two sph runs . such an effect could explain the earlier formation epoch of low - mass galaxies in the tvd run compared to the two sph runs , provided it is not compensated by the comparatively lower gravitational force resolution ( roughly 2 cells ) of the tvd particle - mesh method . we have used two different types of hydrodynamic cosmological simulations ( eulerian tvd and sph ) to study the properties of massive galaxies at @xmath0 in a @xmath1cdm universe , with particular emphasis on an observationally inspired selection based on the @xmath2 filter set . the simulated galaxies at @xmath0 satisfy well the color - selection criteria proposed by @xcite and @xcite when we assume @xcite extinction with @xmath3 . however , we find that the fraction of stellar mass contained in galaxies that pass the color - selection criteria could be as low as 50% of the total . the number density of simulated galaxies brighter than @xmath4 at @xmath0 is about @xmath157 for @xmath3 in the most representative run ( sph g6 run ) , roughly one order of magnitude larger than that of lyman break galaxies at @xmath6 . the increase of the number density of bright galaxies can be viewed as a consequence of the ongoing hierarchical build - up of larger and brighter galaxies with ever more stellar mass . the most massive galaxies at @xmath0 have stellar masses of @xmath7 , and their observed - frame @xmath8 colors lie in the range @xmath9 provided that the median extinction is @xmath158 . they have been continuously forming stars with a rate exceeding @xmath10 over a few gyrs from @xmath11 to @xmath0 . typically , half of their stellar mass was already assembled by @xmath12 . the bluest galaxies with @xmath13 at @xmath0 are somewhat less massive , with @xmath14 , and are less dominated by old stellar populations . our study suggests that the majority of the most massive galaxies at @xmath0 could be detected at rest - frame uv wavelengths contrary to some recent claims based on near - ir studies ( e.g. * ? ? ? * ; * ? ? ? * ) of galaxies at the same epoch , provided the median extinction is in the range @xmath18 as indicated by the surveys of lyman break galaxies at @xmath6 . we plan to extend our analysis to the near - ir pass - bands in subsequent work . we find that these massive galaxies reside in the highest overdensity regions in the simulations at @xmath0 . while it is encouraging to see that the two very different types of hydrodynamic simulation methods give a similar picture for the overall properties of massive galaxies at @xmath0 , we have also seen some notable differences in the results of the sph and tvd simulations throughout the paper . for example , the star formation history of galaxies in the tvd simulation is clearly more sporadic or ` bursty ' than that in the sph simulations . this probably owes to the differences in resolution and the details of how star formation is modeled in the two codes , as we described in section [ sec : simulation ] . while the dependence of the star formation rate on local gas density is essentially the same for the tvd and sph simulations , the different numerical coefficients effectively multiplying the local density on the right hand sides of equations ( [ eq : tvd ] ) and ( [ eq : sph ] ) produce different temporal smoothing of the local star formation rates and the parameters adopted by the tvd code leads to a more `` bursty '' output ( @xmath159 ) , although not to a significantly different total star formation rate . at this point we simply must await a more detailed theory or be guided by observations in choosing between the two approaches . another difference we saw is that the faint - end slope of the rest - frame @xmath20-band luminosity function : the tvd simulation gives @xmath160 , similar to what is observed locally @xmath161 ( e.g. * ? ? ? * ) , but the sph d5 and g6 runs give much steeper slope of @xmath162 . the question of the nature of the faint - end slope of the luminosity function is one that we have not resolved at this point . the difference may owe to the difference in the treatment of feedback , but @xcite found a flat faint - end slope of @xmath161 in a high - resolution softened lagrangian hydrodynamic ( slh ) simulation @xcite even when the code did not include sn feedback , suggesting that the photoionization of the gas in low - mass halos by the uv background radiation field is responsible for suppressing the formation of low - mass galaxies ( see also * ? ? ? * ; * ? ? ? * ; * ? ? ? thus the theoretical expectation is not fully clear at present , and it will be very interesting to see what is indicated by upcoming observational programs of deeper high redshift galaxies and future theoretical studies . the present study shows that despite a considerably different numerical methodology and large differences in the assumed model for star formation , two independent hydrodynamic codes predict the formation of rather massive galaxies at @xmath163 in the @xmath164cdm cosmology , with properties that are broadly consistent with current observations . the results do not suggest that hierarchical galaxy formation fails to be able to account for these objects . on the contrary , it predicts that significant additional mass in galaxies will be found as the observational selection criteria are broadened . we thank scott burles , hsiao - wen chen , masataka fukugita , max pettini , alice shapley , and rob simcoe for enlightening discussions . we also thank nick gnedin for refereeing the paper and for constructive comments on the manuscript . this work was supported in part by nsf grants aci 96 - 19019 , ast 00 - 71019 , ast 02 - 06299 , and ast 03 - 07690 , and nasa atp grants nag5 - 12140 , nag5 - 13292 , and nag5 - 13381 . the sph simulations were performed at the center for parallel astrophysical computing at harvard - smithsonian center for astrophysics . the tvd simulations were performed at the national center for supercomputing applications ( ncsa ) . dickinson , m. , stern , d. , giavalisco , m. , ferguson , h. c. , tsvetanov , z. , chornock , r. , cristiani , s. , dawson , s. , dey , a. , filippenko , a. v. , moustakas , l. a. , nonino , m. , papovich , c. , ravindranath , s. , riess , a. , rosati , p. , spinrad , h. , vanzella , e. , et al . 2003 , apj , 600 , l99 glazebrook , k. , abraham , r. , mccarthy , p. , savaglio , s. , chen , h .- w . , crampton , d. , murowinski , r. , jorgensen , i. , roth , k. , hook , i. , marzke , r. , & carlberg , r. 2004 , nature , in press ( astro - ph/0401037 ) van dokkum , p. , franx , m. , schreiber , n. f. , illingworth , g. , daddi , e. , knudsen , k. k. , labbe , i. , moorwood , a. , rix , h .- w . , rottgering , h. , rudnick , g. , trujillo , i. , van der werf , p. , van der wel , a. , van starkenburg , l. , & wuyts , s. 2004 , apj , in press ( astro - ph/0404471 ) cccccc tvd : n864l22@xmath165 & 22.0 & @xmath166 & @xmath167 & @xmath168 & 25.5sph : d5@xmath169 & 33.75 & @xmath170 & @xmath171 & @xmath172 & 4.2sph : g6@xmath169 & 100.0 & @xmath173 & @xmath174 & @xmath175 & 5.3 [ table : simulation ]	we study the properties of galaxies at redshift @xmath0 in a @xmath1 cold dark matter ( @xmath1cdm ) universe , using two different types of hydrodynamic simulation methods eulerian tvd and smoothed particle hydrodynamics ( sph ) and a spectrophotometric analysis in the @xmath2 filter set . the simulated galaxies at @xmath0 satisfy the color - selection criteria proposed by @xcite and @xcite when we assume calzetti extinction with @xmath3 . we find that the number density of simulated galaxies brighter than @xmath4 at @xmath0 is about @xmath5 for @xmath3 in our most representative run , roughly one order of magnitude larger than that of lyman break galaxies at @xmath6 . the most massive galaxies at @xmath0 have stellar masses @xmath7 , and their observed - frame @xmath8 colors lie in the range @xmath9 . they typically have been continuously forming stars with a rate exceeding @xmath10 over a few gyrs from @xmath11 to @xmath0 , although the tvd simulation indicates a more sporadic star formation history than the sph simulations . of order half of their stellar mass was already assembled by @xmath12 . the bluest galaxies with colors @xmath13 at @xmath0 are somewhat less massive , with @xmath14 , and lack a prominent old stellar population . on the other hand , the reddest massive galaxies at @xmath0 with @xmath15 and @xmath16 completed the build - up of their stellar mass by @xmath17 . interestingly , our study indicates that the majority of the most massive galaxies at @xmath0 should be detectable at rest - frame ultra - violet wavelengths , contrary to some recent claims made on the basis of near - infrared studies of galaxies at the same epoch , provided the median extinction is less than @xmath18 as indicated by surveys of lyman break galaxies at @xmath6 . however , our results also suggest that the fraction of stellar mass contained in galaxies that pass the color - selection criteria used by @xcite could be as low as 50% of the total stellar mass in the universe at @xmath0 . our simulations imply that the missing stellar mass is contained in fainter ( @xmath19 ) and intrinsically redder galaxies . the bright end of the rest - frame @xmath20-band luminosity function of @xmath0 galaxies can be characterized by a schechter function with parameters @xmath21 , while the tvd simulation suggests a flatter faint - end slope of @xmath22 . a comparison with @xmath6 shows that the rest - frame @xmath20-band luminosity function has brightened by about 0.5 magnitude from @xmath6 to @xmath0 without a significant change in the shape . our results do _ not _ imply that hierarchical galaxy formation fails to account for the massive galaxies at @xmath23 .
let us begin by introducing notation . a message word consisting of @xmath10 bits is encoded in an @xmath11-bit long codeword , @xmath12 . in the case of binary , linear coding a convenient representation of the code is given by @xmath13 constraints , often called parity checks or simply checks . formally , @xmath14 with @xmath15 , is one of the @xmath16 codewords if and only if @xmath17 for all checks @xmath18 , where @xmath19 if the bit @xmath20 contributes the check @xmath21 . the relation between bits and checks ( we use @xmath19 and @xmath22 interchangeably ) is often described in terms of the @xmath23 parity - check matrix @xmath24 consisting of ones and zeros : @xmath25 if @xmath19 and @xmath26 otherwise . a bipartite graph representation of @xmath27 , with bits marked as circles , checks marked as squares , and edges corresponding to the respective nonzero elements of @xmath27 , is called the tanner graph of the code . for an ldpc code @xmath27 is sparse , i.e. most of the entries are zeros . transmitted through a noisy channel , a codeword gets corrupted due to the channel noise , so that the channel output is @xmath28 . even though information about the original codeword is lost at the receiver , one still possesses the full probabilistic information about the channel , i.e. the conditional probability , @xmath29 , for a codeword @xmath30 to be a pre - image for the output word @xmath31 , is known . in the case of independent noise samples , the full conditional probability can be decomposed into the product , @xmath32 . a convenient characteristic of the channel output at a bit is the so - called log - likelihood , @xmath33/2 $ ] . the decoding goal is to infer the original message from the received output , @xmath31 . ml decoding ( which generally requires an exponentially large number , @xmath34 , of steps ) corresponds to finding the most probable transmitted codeword given @xmath31 . belief propagation ( bp ) decoding @xcite constitutes a fast ( linear in @xmath35 ) yet generally approximate alternative to ml . as shown in @xcite the set of equations describing bp becomes exactly equivalent to the so - called symbol maximum - a - posteriori ( map ) decoding in the loop - free approximation ( a similar construction in physics is known as the bethe - tree approximation @xcite ) , while in the low - noise limit , i.e. in the limit of very large snr , @xmath36 , ml and map become indistinguishable and the bp algorithm reduces to the min - sum algorithm : @xmath37 \min_{j\neq i}^{j\in\beta } \big\| \eta^{(n)}_{j\beta } \big\| , \label{min_sum}\ ] ] where the message field @xmath38 is defined on the edge that connects bit @xmath20 and check @xmath21 at the @xmath39-th step of the iterative procedure and @xmath40 . the result of decoding is determined by a - posteriori log - likelihood , @xmath41 , defined by the right - hand - side of eq . ( [ min_sum ] ) with the restriction @xmath42 dropped . the ber measuring probability of errors at a given bit @xmath20 becomes @xmath43 where @xmath44 if @xmath45 and @xmath46 otherwise ; @xmath47 is assumed for the input ( since in a symmetric channel the ber is invariant with respect to the choice of the input codeword ) . when the ber is small ( snr is large ) the integral over output configurations @xmath31 in eq . ( [ ber ] ) is approximated by : @xmath48 where @xmath49 are the special instanton configurations of the output maximizing @xmath50 under the error - surface condition , @xmath51 ; @xmath52 combines combinatorial and phase - volume factors ( the latter one accounts for what physicists call fluctuations around the respective instanton ) . individual instanton contributions into the rhs of eq . ( [ bi ] ) decrease significantly with increasing snr . thus at large snr only instanton(s ) with the highest @xmath53 is ( are ) relevant . for the common model of the additive white gaussian noise ( awgn ) channel , @xmath54 finding the instanton , @xmath55 , turns into minimizing the length @xmath56 with respect to the unit vector in the noise space @xmath57 , where @xmath56 measures the distance from the zero - noise point to the point on the error surface corresponding to @xmath57 . this task of the instanton analysis for the awgn channel was discussed in detail in @xcite . in @xcite we developed a numerical scheme where the value of the length @xmath58 for any given unit vector @xmath57 was found by the bisection method . the minimum of @xmath56 was found by a downhill simplex method also called `` amoeba '' @xcite , with accurately tailored ( for better convergence ) annealing . ( note , that even previously , the numerical instanton method was successfully verified in @xcite against analytical results in the loop - free case . ) to demonstrate the utility of this method we chose in @xcite the example of the @xmath2 ldpc code of @xcite . ( the parity check matrix of the code is shown in fig . s1 of @xcite . ) the code includes @xmath59 bits and @xmath60 checks . each bit is connected to three checks while any check is connected to five bits . the minimal hamming distance of the code is @xmath61 , i.e. for @xmath1 , and if the decoding is ml , ber becomes @xmath62 in the gaussian channel . iterative decoding is suboptimal , thus respective error - floor asymptotics become @xmath63 . the numerical , and subsequent theoretical , analyses of @xcite suggest that the instantons , as well as the respective effective distance , @xmath64 , do depend on the number of iterations . focusing primarily on the already nontrivial case of four iterations , we showed in @xcite that the three minimal weight ( largest probability ) instantons have effective lengths , @xmath65 , @xmath66 and @xmath67 respectively . these instantons were found as the result of multiple attempts at instanton - amoeba minimization . the remarkable integer / rational structure of the instantons found numerically by instanton - amoeba for the awgn channel admits a clear theoretical explanation discussed in details in @xcite . we have already suggested in @xcite that the instanton analysis ( in both numerical and theoretical parts ) is actually generic , and is thus applicable to wide range of different codes and channels . to illustrate this last point we focus in this manuscript on analysis of the generalized additive white laplacian noise ( awln ) channel : @xmath68 where @xmath0 is the signal - to - noise - ratio ( snr ) and @xmath21 is the regularization parameter . ( we are mainly interested in the @xmath69 , thus @xmath21 is introduced primarily for the purpose of accurately regularizing / resolving the singularity at @xmath70 . ) if the detected signal at a bit is @xmath71 , the respective log - likelihood at the bit is defined as @xmath72= \biggl.\frac{\sqrt{(2-\xi)^2+\alpha^2}-\sqrt{\xi^2+\alpha^2}}{2 } \biggr\|_{\alpha\to 0}\to \biggl\ { \begin{array}{cc } -1 , & \xi\geq 2 ; \\ 1-\xi , & 2\geq \xi\geq 0;\\ + 1 , & 0\geq \xi.\end{array}\biggr . , \label{h1}\ ] ] where one chooses to measure log - likelihoods in the awln channel in units of snr , @xmath0 ( and not in units of the snr squared that was natural choice in the awgn channel ) . in the laplacian channel ( @xmath73 ) , finding the instanton means to minimize @xmath74 with respect to the unit vector @xmath75 in the @xmath11-dimensional space , rather then minimizing @xmath56 that was the case for the gaussian channel . notice , that just from this fact one finds absolutely no reason to expect that instantons for gaussian and laplacian channels are in any way related to each other . the contribution of an instanton to the ber for @xmath1 , characterized by the effective distance @xmath4 , is estimated as @xmath76 in the case of the laplacian channel . in this section we describe a theoretical approach to the instanton analysis . this analysis , based on the computational tree construction of wiberg @xcite , was discussed in @xcite for the gaussian channel . here we explain how the analysis can be modified to describe instantons in the case of the awln channel . we start with the universal , i.e. channel insensitive , part of the construction . the computational tree is built by unwrapping the tanner graph of a given code into a tree from a bit for which we would like to determine the probability of error . ( this bit will be called erroneous bit . ) the number of generations in the tree is equal to the number of min - sum iterations . as observed in @xcite , the result of decoding at the erroneous bit of the original code is exactly equal to the decoding result in the tree center . it should be noted that once log - likelihoods representing an instanton are distributed on the tree , one can verify directly ( by propagating messages from the leaves to the tree center ) that the algorithm produces zero a - posteriori log - likelihood at the tree center . any check node processes messages coming from the tree periphery in the following way : ( i ) the message with the smallest absolute value ( one assumes no degeneracy ) is passed , ( ii ) the source bit of the smallest message is `` colored '' , and ( iii ) the sign of the product of inputs is assigned to the outcome . at any bit that lies on the colored leaves - to - center path the incoming messages are summed up . the initial messages at any bit of the tree are log - likelihoods and , therefore , the result obtained in the tree center is a linear combination of the log - likelihoods with integer coefficients . the integer @xmath77 corresponding to bit @xmath20 of the original graph is the sum of the signatures over all colored replicas of @xmath20 on the computational tree . therefore the condition at the tree center becomes @xmath78 . so far the discussion has been generic . let us now adapt this generic construction to the case of the generalized awln channel described by eq . ( [ lap1 ] ) . returning from the computational tree to the original graph and maximizing the integrand of eq . ( [ ber ] ) with the condition , @xmath78 , enforced we arrive at the following expressions for the effective length : @xmath79 where @xmath80 is the lagrange multiplier enforcing the zero a - posteriori log - likelihood condition at the tree center . minimizing eq . ( [ s1 ] ) with respect to @xmath81 one derives @xmath82 expressing @xmath81 in terms of @xmath77 and @xmath80 involves solving a cubic equation . note also that , first , the left - hand - side of eq . ( [ var1 ] ) is monotone with respect to @xmath81 in the interval of @xmath83 , and , second , the expression becomes @xmath84 as @xmath69 . consider the domain of @xmath85 and @xmath86 . then eq . ( [ var1 ] ) becomes @xmath87 one observes that either of the two possibilities is realized at @xmath69 : ( 1 ) @xmath88 and @xmath89 ; or ( 2 ) @xmath90 then @xmath91 . one should also add to these two possible cases a third one ( that is obviously not explained by eq . ( [ var2 ] ) ) where @xmath92 lies exactly on the border of the monotone interval , i.e. at @xmath93 and @xmath94 . one finds that structurally an instanton consists of three types of colored bits corresponding to @xmath95 , @xmath96 and @xmath93 respectively . one also finds that for all the colored bits with @xmath96 all the respective @xmath77 approach the same limit , @xmath97 , even though respective @xmath81 may be distinct . this results in the following form of the zero a - posteriori log - likelihood condition at the tree center @xmath98 where `` @xmath99 '' is the set of colored bits on the original graph with @xmath100 ; @xmath101 is the total number of the colored bits on the computational tree ; @xmath102 is the number of marginal , @xmath93 , bits on the computational tree ; @xmath103 is the number of replicas found on the computational tree for a bit @xmath20 that belongs to the @xmath99 ( this is the same number for all @xmath104 ) ; @xmath105 is the number of distinct marginal ( i.e. @xmath93 ) bits on the original graph . ( [ qc ] ) represents the major theoretical result of our analysis . it explains the rational origin of the effective length and shows how the effective length depends on the set of integers , @xmath106 and @xmath105 carrying the coding / decoding specific information . notice also a remarkable common feature of the instantons . there is actually a strong degeneracy here if the number of colored bits with @xmath96 is two or larger : the effective length depends only on @xmath107 , while otherwise , and modulo the requirement @xmath100 , the @xmath81 fields can be chosen arbitrarily . therefore , to estimate the ber corresponding to the given set of integers , @xmath108 , thus explaining a continuous family of instantons rather then an individual instanton , one should also account for the degree of degeneracy ( volume of the respective part of the phase space ) , @xmath109 $ ] . our estimate shows that the @xmath110-terms give sub - leading ( i.e. non - exponential in @xmath0 ) corrections to the major ( effective length ) factors . so far we have not discussed how to find the discrete values @xmath111 and @xmath105 that are obviously dependent on the explicit structure of the ldpc code considered . to solve this problem we adopt the instanton - amoeba approach of @xcite . we consider the @xmath2 ldpc code of @xcite as an example . the minimal hamming distance of the code is @xmath112 , i.e. for @xmath1 , and if the decoding is ml , the ber becomes @xmath113 in the case of the laplacian channel . the ber is higher , @xmath114 with @xmath115 , if one decodes iteratively . for the min - sum decoding with @xmath116 iterations we found that three minimal length instantons are characterized by @xmath117 , @xmath118 and @xmath119 respectively . these instantons were disclosed as the result of multiple attempts at instanton - amoeba numerical minimization described above . the colored parts of the tanner graph of the code and the respective parts of the computational tree for the three configurations are shown in fig . 1 : the three panels ( a , b , c ) correspond to the three instantons showing minimal effective lengths . each panel consists of two diagrams showing the relevant ( colored ) part of the tanner graph and respective ( four iterations deep ) part of the computational tree . bits / circles are shown with numbers correspondent to the ordering of the bits explained in @xcite . the shadow bit is the erroneous one , i.e. it is the bit whose a - posteriori log - likelihood is exactly zero on the @xmath116 step of the min - sum decoding . according to our theoretical analysis detailed above ( that is also confirmed numerically in detail ) there are three types of colored bits / cirlces with @xmath95 , @xmath96 and @xmath93 , shown in fig . 1 in white , green and red respectively . the white bits on the computational tree that are not numbered ( and respective bits of the tanner graph simply not shown in the figure ) can be chosen arbitrarily with the only requirement that they are distinct from any bits shown numbered in the figure . code and respective computational tree with four iterations are shown . numbers marking the bits are introduced in accordance with the convention described in @xcite . the color coding used for bits is white for @xmath95 , green for @xmath96 and red for @xmath93 . see detailed explanations in the text.,width=384 ] configuration ( a ) shown in fig . 1a consists of @xmath120 green bits with @xmath96 ( numbers @xmath121 and @xmath122 ) each appearing @xmath123 times on the computational tree , and @xmath120 red bits with @xmath93 ( numbers @xmath124 and @xmath125 ) each appearing @xmath126 times on the computational tree . the remaining @xmath125 bits carry @xmath95 noise . therefore , the respective integers corresponding to this instanton are @xmath127 , @xmath128 , @xmath129 , @xmath130 , thus resulting according to eq . ( [ qc ] ) in @xmath131 . this effective distance was found with numerical precision by the instanton - amoeba method . the degeneracy in this family of instantons is one parametric : @xmath132 and @xmath133 . here , the erroneous bit , i.e. the bit with zero a - posteriori log - likelihood ( marked striped in the fig . 1a ) is the white one . configuration ( b ) consists of three green bits , with @xmath134 , appearing four times each on the computational tree , and three red bits , with @xmath93 , appearing @xmath135 and @xmath136 times respectively . for this instanton one finds , @xmath137 , @xmath138 , and @xmath139 ( @xmath101 is always @xmath140 for the four iterations decoder ) so that @xmath141 . the erroneous bit ( number @xmath142 ) is white . one bit , numbered @xmath143 , is special here . even though this bit has many replicas on the computational tree , thus potentially , it would be advantageous to have it green carrying the non - zero value of the noise , self - consistency strictly requires that @xmath144 . there are at least two reasons for this . first of all , if bit @xmath143 turns green making @xmath145 one of its replicas on the computational tree ( the one marked pale and adjusted to the red check / square in fig . 1b ) screens bit @xmath146 ( positioned next to the red check ) in the sense that this screened bit will not contribute to @xmath147 . consequently , @xmath147 becomes @xmath136 and not @xmath126 , thus making @xmath102 smaller and @xmath148 larger . second , if @xmath149 then one of the @xmath123 replicas of the bit @xmath143 contributes a - posteriori log - likelihood at the tree center with `` @xmath150 '' sign and @xmath151 becomes equal to @xmath152 , rather than @xmath123 . this number is smaller than @xmath138 thus leading to the undesired effective length increase . finally , the resulting degeneracy in the instanton family is two - parametric ( corresponding to appearance of three , and not four , green bits ) : @xmath153 and @xmath154 . configuration ( c ) has the same effective length as configuration ( b ) , @xmath155 , even though it is very different structurally . the ( c ) instanton has no green bits , thus it is non - degenerate and only one special configuration of @xmath156 fields is realized . there are four red bits , @xmath157 and @xmath158 , appearing on the computational tree @xmath159 and @xmath136 times respectively . the erroneous bit , @xmath142 , is red . the integers are @xmath160 , @xmath161 and @xmath162 , thus according to eq . ( [ qc ] ) @xmath163 . , for the @xmath2 code with four iterations of the min - sum decoding over the laplacian channel . solid and dashed lines show slopes ( defined upto a constant shift in the log - lin plot ) correspondent to the instanton with the shortest effective distance , @xmath131 , and to the ideal ml decoding with the hamming distance , @xmath164 , respectively . filled and empty circles show results of direct monte - carlo simulations for @xmath116 and @xmath165 iterations respectively . , width=384 ] in @xcite we have argued , following the logic of @xcite , that each instanton is equidistant from some number of pseudo - codewords , i.e. codewords on the respective computational tree . this observation is obviously generic and thus it is applicable as well to the case of the laplacian channel considered in the manuscript . for any of the instantons discussed above and illustrated in fig . 1 there exists a respective pair of pseudo - codewords . ( as argued in @xcite an instanton may be degenerate corresponding to a triple or in principle to even larger set of pseudo - codewords . this degeneracy , that was found present in the awgn channel , was not observed here for the awln channel . ) the first pseudo - codeword in a pair is just the all @xmath166 codeword ( @xmath166 sits at every bit ) . the second pseudo - codeword in a pair can be introduced according to the following rule : put @xmath167 at any colored ( i.e. white , green or red ) bit and @xmath166 at any uncolored ( thus not shown in fig . obviously this choice of the pseudo - codewords pairs is unambiguous . indeed the white bits that are not numbered can be chosen arbitrarily ( within appropriate bounds described above ) . moreover , the ambiguity in choosing the second pseudo - codeword ( containing @xmath167 bits ) is even stronger . indeed , this second pseudo - codeword can consists only of two @xmath167 bits ( all other bits will carry @xmath166 ) : one @xmath167 should be positioned at any red bit ( with @xmath93 ) from the last generation of the computational tree and another @xmath167 is placed on any uncolored bit sharing a check with the red bit . this special form of degeneracy is due to the fact that log - likelihoods assigned to any red bit and to an uncolored bit adjusted to the red one are the same in absolute value but opposite in sign . the analysis of the previous section suggests that the leading @xmath36 asymptotic for ber ( and for the frame - error - rate as well ) is governed by the instanton with the lowest effective length found , i.e. @xmath169 $ ] . we have checked this prediction against direct monte - carlo ( mc ) simulations and found very good agreement already in the range accessible for the mc , @xmath170 . see fig . 2 . we observed that the actual behavior of ber is well described by the instanton not only in the asymptotic regime of highest snr but also in the regime of moderate snr where there is no a - priori reason to expect the instanton approximation to work so well . our further discussion is to eliminate this point suggesting a plausible explanation for this surprising generality of the instanton asymptotic . let us first clarify why the validity of the instanton asymptotic at the moderate values of snr shown in fig . 2 is surprising . indeed , according to eq . ( [ lap1 ] ) the average value of the noise configuration length , @xmath171 , unconstrained by the requirement to have zero a - posteriori log - likelihood at a bit , is @xmath172 . this means that even at @xmath173 ( the largest snr shown in fig . 2 ) , where the error probability is already small , @xmath174 , the typical length of noise realization is still essentially larger then the respective instanton prediction : @xmath175 . therefore , naively one expects the instanton to work well at @xmath176 , where @xmath177 , while according to the mc results shown in fig . 2 the instanton asymptotic sets already at @xmath178 , where @xmath179 . ( notice , that the situation for the gaussian channel is similar , see fig . s2 of @xcite . there the typical @xmath180 is @xmath181 , resulting in @xmath182 , while the respective instanton value is @xmath183 . ) our approach to explaining the validity of the instanton asymptotic already at the moderate nsr is of the reverse engineering type : we first formalize what the monte - carlo results suggest and then present a plausible explanation for this phenomenon . one useful object is the probability distribution function of the channel noise length ( fully unconstrained ) , that gets the following forms for the gaussian and laplacian channels respectively : @xmath184^{-1 } \exp(-l \cdot s)$ ] and @xmath185^{-1 } \exp(-l^2 \cdot s^2/2)$ ] . typical noise configuration forms a `` spherical '' layer with `` radius '' @xmath186 in the @xmath11-dimensional noise space ( or @xmath187 for the gaussian channel ) . spherical " layer becomes thinner as @xmath11 grows . obviously @xmath188 does depend on the channel but it does not depend on the decoding scheme . another useful object , that is decoding sensitive , is the ( cumulative ) distribution function @xmath189 of the length for the points positioned exactly at the error - surface . estimation for fer is related to the area of spherical layer that lies outside of the error surface : @xmath190 . strictly speaking this relation is exact for the gaussian channel , but it is only qualitatively right ( up to a @xmath191 coefficient accounting accurately for the phase factor ) in the laplacian channel case . the integral @xmath190 can be viewed as the laplace transform . deducing from the monte carlo simulations that the instanton asymptotic is valid , @xmath192 ( or @xmath193 for the gaussian channel ) one derives by the inverse laplace transform @xmath194 ( or @xmath195 for the gaussian channel ) . we suggest that the special dependence of @xmath196 on @xmath197 , deduced from the mc simulations , corresponds to an @xmath198-dimensional area of the part of error - surface with lengths @xmath197 or less , @xmath199 , where @xmath200 is the respective line element . given that the instanton configuration is extremal one assumes that , @xmath201 , that results exactly in the right expression for @xmath189 in the gaussian channel . we ought to assume that in the laplacian channel case expansion of @xmath197 with respect to @xmath202 about @xmath203 , is of another type , @xmath204 , thus confirming the @xmath189 dependence on @xmath197 observed in the laplacian channel simulations . the set of tasks formulated in the introductory section was accomplished in sections [ sec : theo]-[sec : amoeba ] . we have shown for the example of the laplacian channel that the instanton - amoeba optimization scheme , introduced in @xcite and tested in @xcite for the gaussian channel , is computationally efficient . we extended the theoretical analysis of @xcite , based on the computational tree approach of @xcite , explaining the rational structure of instantons in the laplacian channel . the instantons are shown to be different for different channels considered with the same coding / decoding scheme . even though the fact that effective lengths differ for different channels was already demonstrated in @xcite for the example of the binary symmetric and binary erasure channels , this manuscript additionally proved that not only the effective weights but also the instanton configurations themselves were different structurally for different channels . we conclude by noting that the observations made in this manuscript , extending and complementing our previous works @xcite , virtually solve in a straightforward way the problem of the generic error - floor analysis . ( no sampling in the configurational space , e.g. of the kind used in @xcite , was required ) . we intentionally choose the well known @xmath2 code ( used routinely for testing ) to demonstrate the exciting opportunities the instanton - amoeba approach has to offer . our next goal is to apply our scheme to a variety of other ( e.g. longer ) codes , decoding schemes and channels .	we analyze the performance of low - density - parity - check codes in the error - floor domain where the signal - to - noise - ratio , @xmath0 , is large , @xmath1 . we describe how the instanton method of theoretical physics , recently adapted to coding theory , solves the problem of characterizing the error - floor domain in the laplacian channel . an example of the @xmath2 ldpc code with four iterations ( each iteration consisting of two semi - steps : from bits - to - checks and from checks - to - bits ) of the min - sum decoding is discussed . a generalized computational tree analysis is devised to explain the rational structure of the leading instantons . the asymptotic for the symbol bit - error - rate in the error - floor domain is comprised of individual instanton contributions , each estimated as @xmath3 , where the effective distances , @xmath4 , of the the leading instantons are @xmath5 and @xmath6 respectively . ( the hamming distance of the code is @xmath7 . ) the analysis shows that the instantons are distinctly different from the ones found for the same coding / decoding scheme performing over the gaussian channel . we validate instanton results against direct simulations and offer an explanation for remarkable performance of the instanton approximation not only in the extremal , @xmath8 , limit but also at the moderate @xmath0 values of practical interest . a novel exciting era has begun in coding theory with the discovery of low - density - parity - check ( ldpc ) @xcite and turbo @xcite codes . these codes are special , not only because they can approach the virtually error - free transmission limit , but mainly because a computationally efficient iterative decoding scheme is readily available . when operating at moderate noise values these approximate decoding algorithms show an unprecedented ability to correct errors , a remarkable feature that has attracted a lot of attention @xcite . however , it was also shown that the approximate algorithms are incapable of matching the performance of maximum - likelihood ( ml ) decoding beyond the so - called error - floor threshold found at higher signal - to - noise - ratios ( snr ) . the importance of error - floor , i.e. highest snr , analysis was recognized in the early stages of the turbo code revolution @xcite , and it soon became apparent that ldpc codes are also not immune to the error - floor deficiency @xcite . to estimate the error - floor asymptotics in modern high - quality systems is a notoriously difficult task because direct numerical methods , e.g. monte carlo , can not be used to determine bit - error - rate ( ber ) below @xmath9 . the main approaches to the error - floor analysis problem proposed to date include : ( i ) a heuristic approach of the importance sampling type @xcite , utilizing theoretical considerations developed for a typical randomly constructed ldpc code performing over the binary - erasure channel @xcite , and ( ii ) deriving lower bounds for ber @xcite . recently , we ( in collaboration with v. chernyak and b. vasic ) have also proposed a physics inspired approach that is capable of a computationally tractable analysis of the error floor phenomenon @xcite . an efficient numerical scheme was proposed , which was ab - initio by construction , i.e. the optimization scheme required no additional assumptions ( e.g. no sampling ) . this numerical optimization scheme , called the instanton - amoeba scheme , which plays a central role in our analysis , was shown to be accurate at producing configurations whose validity , for actual optimal noise configurations , can be verified theoretically . finally , the instanton - amoeba scheme , introduced in @xcite , is also generic , in that there are no restrictions related to the type of decoding or the channel . to illustrate the last , most important point , we complement the analysis of @xcite that focused on the white gaussian symmetric channel , by using the new results for the white laplacian channel explained in this manuscript . our choice of the laplacian channel is arbitrary , e.g. influenced by its relevance to the description of fiber - optics communication channels . ( amplifier noise , known to have gaussian statistics , contributes additively to the electric signal carrying the information . however , the intensity , the electric field squared , is detected on the receiver end . therefore , the resulting transition probability , characterizing the noise in the fiber optics channel as the whole , shows exponential tails . see @xcite for a discussion of these and other statistical errors in fiber optics channels . ) our goal is to demonstrate using the example of the laplacian channel that * the numerical optimization approach to finding the most damaging configuration of the noise , instanton - amoeba , is computationally efficient . * subsequent theoretical analysis of the iterative decoding instanton based on the notion of the computational tree is channel specific , but it is also generalizable , i.e. the theoretical scheme can be modified to explain the rational structure of instantons for any , in particular laplacian , channel . * instanton configurations for different channels are different . thus , information on the error - floor analysis available for a channel ( say , for the gaussian channel ) does not allow quantitative description of the error - floor in another channel ( say , in the laplacian channel ) . this manuscript is organized as follows . we describe the problem of error - floor analysis , introduce the laplacian channel and briefly review the results of @xcite in section [ sec : intro ] . our theoretical analysis , generalizing the computational tree approach of @xcite , is detailed in section [ sec : theo ] . section [ sec : amoeba ] explains results of numerical instanton - amoeba evaluation . here we present the three most important ( for @xmath1 ) instantons , for example of the min - sum decoding with four iterations performing over the @xmath2 ldpc code described in @xcite . we also discuss here how the numerically found instanton is rationalized / explained theoretically . the comparison of monte carlo simulations and instanton prediction is discussed in section [ sec : moderate ] . some final remarks and comments are presented in section [ sec : concl ] concluding the manuscript .
the production of single top quark events via the electroweak interaction was reported in 2009 by the d0 @xcite and cdf @xcite collaborations . a measurement of the single top quark production cross section provides a direct measurement of the the quark mixing matrix element @xmath3 @xcite . it also serves as a probe of the @xmath5 coupling @xcite and is sensitive to several models of new physics @xcite . single top quark production proceeds via the @xmath1-channel exchange of a virtual @xmath6 boson between a light quark line and a heavy quark line ( @xmath7 ) and the @xmath2-channel production and decay of a virtual @xmath6 boson ( @xmath8 ) , shown in fig . [ fig : feynman ] . -channel single top quark production and ( b ) @xmath1-channel production , including the top quark decay.,width=377 ] here we present an updated measurement of the combined single top quark production cross section using 5.4 fb@xmath0 of data collected by the d0 experiment at the tevatron proton - antiproton collider at fermilab @xcite . the @xmath2-channel and @xmath1-channel modes are measured separately and the first observation of @xmath1-channel production is reported @xcite . the single top quark final state consists of a lepton ( electron or muon ) and a neutrino from the @xmath6 boson decay , a @xmath9 quark from the top quark decay and an additional light quark ( @xmath1-channel ) or @xmath9 quark ( @xmath2-channel ) . the event selection for all analysis channels requires large @xmath10 and two to four jets , at least one of which is @xmath9-tagged . the backgrounds to this signature are dominated by @xmath6 bosons produced in association with jets ( @xmath6+jets ) , with smaller contributions from @xmath11 pairs . multijet events also contribute to the background when a jet is misidentified as an isolated electron or a heavy - flavor quark decay results in an isolated lepton . diboson ( @xmath12 , @xmath13 , @xmath14 ) and @xmath15+jets contribute smaller backgrounds . in total , about 8,000 events are selected with an expected sm signal of about 400 events . in addition to the single top analysis sample , several cross - check samples are defined : a without any @xmath9-tag requirements ; a sample containing exactly two jets and sum of transverse momenta of all objects @xmath16 gev ; and a sample containing exactly four jets and requiring @xmath17 gev . figure [ fig : variables ] shows a comparison between the data and the signal+background model for several discriminating variables used in the analysis in the different samples . , ( b ) @xmath6 boson transverse mass , ( c ) light quark jet pseudorapidity multiplied by lepton charge , ( d f ) reconstructed top quark mass . subfigures ( a ) and ( b ) are before @xmath9-tagging , ( c ) and ( d ) are after @xmath9-tagging , ( e ) is a control sample dominated by @xmath6+jets , and ( f ) is a control sample dominated by @xmath11 pair events . the hatched bands show the @xmath18 uncertainty on the background prediction for distributions obtained after @xmath9-jet identification ( c f ) . , title="fig:",width=226 ] , ( b ) @xmath6 boson transverse mass , ( c ) light quark jet pseudorapidity multiplied by lepton charge , ( d f ) reconstructed top quark mass . subfigures ( a ) and ( b ) are before @xmath9-tagging , ( c ) and ( d ) are after @xmath9-tagging , ( e ) is a control sample dominated by @xmath6+jets , and ( f ) is a control sample dominated by @xmath11 pair events . the hatched bands show the @xmath18 uncertainty on the background prediction for distributions obtained after @xmath9-jet identification ( c f ) . , title="fig:",width=226 ] , ( b ) @xmath6 boson transverse mass , ( c ) light quark jet pseudorapidity multiplied by lepton charge , ( d f ) reconstructed top quark mass . subfigures ( a ) and ( b ) are before @xmath9-tagging , ( c ) and ( d ) are after @xmath9-tagging , ( e ) is a control sample dominated by @xmath6+jets , and ( f ) is a control sample dominated by @xmath11 pair events . the hatched bands show the @xmath18 uncertainty on the background prediction for distributions obtained after @xmath9-jet identification ( c f ) . , title="fig:",width=226 ] , ( b ) @xmath6 boson transverse mass , ( c ) light quark jet pseudorapidity multiplied by lepton charge , ( d f ) reconstructed top quark mass . subfigures ( a ) and ( b ) are before @xmath9-tagging , ( c ) and ( d ) are after @xmath9-tagging , ( e ) is a control sample dominated by @xmath6+jets , and ( f ) is a control sample dominated by @xmath11 pair events . the hatched bands show the @xmath18 uncertainty on the background prediction for distributions obtained after @xmath9-jet identification ( c f ) . , title="fig:",width=226 ] , ( b ) @xmath6 boson transverse mass , ( c ) light quark jet pseudorapidity multiplied by lepton charge , ( d f ) reconstructed top quark mass . subfigures ( a ) and ( b ) are before @xmath9-tagging , ( c ) and ( d ) are after @xmath9-tagging , ( e ) is a control sample dominated by @xmath6+jets , and ( f ) is a control sample dominated by @xmath11 pair events . the hatched bands show the @xmath18 uncertainty on the background prediction for distributions obtained after @xmath9-jet identification ( c f ) . , title="fig:",width=226 ] , ( b ) @xmath6 boson transverse mass , ( c ) light quark jet pseudorapidity multiplied by lepton charge , ( d f ) reconstructed top quark mass . subfigures ( a ) and ( b ) are before @xmath9-tagging , ( c ) and ( d ) are after @xmath9-tagging , ( e ) is a control sample dominated by @xmath6+jets , and ( f ) is a control sample dominated by @xmath11 pair events . the hatched bands show the @xmath18 uncertainty on the background prediction for distributions obtained after @xmath9-jet identification ( c f ) . , title="fig:",width=226 ] the expected single top quark signal is small compared to the statistical data uncertainty and the uncertainty on the background prediction . thus , multivariate analysis methods are employed to extract the single top quark signal . several different multivariate methods ( boosted decision trees , bayesian neural networks and a neuro - evolution network ) are trained and then combined into one filter . this training is done separately for the combined single top analysis and for the dedicated @xmath1-channel and @xmath2-channel analyses . the resulting discriminants are shown in fig . [ fig : discriminants ] . -channel , ( c , d ) @xmath1-channel , and ( e , f ) combined single top production . the full distributions are shown in ( a , c , e ) and the signal region is shown in ( b , d , f ) . the @xmath8 , @xmath7 , and @xmath8+@xmath7 contributions are normalized to their measured cross sections . the hatched bands show the @xmath18 uncertainty on the background prediction.,title="fig:",width=226 ] -channel , ( c , d ) @xmath1-channel , and ( e , f ) combined single top production . the full distributions are shown in ( a , c , e ) and the signal region is shown in ( b , d , f ) . the @xmath8 , @xmath7 , and @xmath8+@xmath7 contributions are normalized to their measured cross sections . the hatched bands show the @xmath18 uncertainty on the background prediction.,title="fig:",width=226 ] -channel , ( c , d ) @xmath1-channel , and ( e , f ) combined single top production . the full distributions are shown in ( a , c , e ) and the signal region is shown in ( b , d , f ) . the @xmath8 , @xmath7 , and @xmath8+@xmath7 contributions are normalized to their measured cross sections . the hatched bands show the @xmath18 uncertainty on the background prediction.,title="fig:",width=226 ] -channel , ( c , d ) @xmath1-channel , and ( e , f ) combined single top production . the full distributions are shown in ( a , c , e ) and the signal region is shown in ( b , d , f ) . the @xmath8 , @xmath7 , and @xmath8+@xmath7 contributions are normalized to their measured cross sections . the hatched bands show the @xmath18 uncertainty on the background prediction.,title="fig:",width=226 ] -channel , ( c , d ) @xmath1-channel , and ( e , f ) combined single top production . the full distributions are shown in ( a , c , e ) and the signal region is shown in ( b , d , f ) . the @xmath8 , @xmath7 , and @xmath8+@xmath7 contributions are normalized to their measured cross sections . the hatched bands show the @xmath18 uncertainty on the background prediction.,title="fig:",width=226 ] -channel , ( c , d ) @xmath1-channel , and ( e , f ) combined single top production . the full distributions are shown in ( a , c , e ) and the signal region is shown in ( b , d , f ) . the @xmath8 , @xmath7 , and @xmath8+@xmath7 contributions are normalized to their measured cross sections . the hatched bands show the @xmath18 uncertainty on the background prediction.,title="fig:",width=226 ] the cross section is measured using a bayesian approach , including all systematic uncertainties and their correlations @xcite . the expected and observed cross section measurements are given in table [ tab : xsections ] . the separate @xmath1-channel and @xmath2-channel cross sections are extracted assuming the sm contribution for the @xmath2-channel and @xmath1-channel , respectively . the @xmath1-channel signal is observed with a significance of more than five standard deviations , and the combined single top measurement is the most precise measurement to date . .expected and observed cross sections in pb for @xmath8 , @xmath7 , and @xmath8+@xmath7 production . all results assume a top quark mass of @xmath19 and the expected cross section normalize the single top signal to its theoretical prediction @xcite . [ cols="<,^,^ " , ] the single top quark production cross section is proportional to the ckm matrix element @xmath20 , allowing for a @xmath3 measurement without having to assume three quark generations or ckm matrix unitarity . we only assume that sm sources for single top quark production and that top quarks decay exclusively to @xmath21 , as well as that the @xmath5 interaction is cp - conserving and of the @xmath22 type . we set a lower limit at the 95% confidence level of @xmath23 . the @xmath1-channel final state is sensitive to flavor - changing neutral currents ( fcnc ) via quark - gluon couplings . in fcnc interactions a gluon vertex couples an up quark ( @xmath24 ) or a charm quark ( @xmath25 ) to the top quark . the rate for these events at the tevatron is large because the initial state contains two light quarks . the d0 collaboration searched for fcnc interactions in 2.3 fb@xmath0 @xcite , using the single top quark observation analysis sample @xcite . a bayesian neural network ( bnn ) is used to separate the fcnc signal from the large backgrounds . the bnn output distribution is shown in fig . [ fig : fcnc ] . the observed limits on the fcnc couplings are @xmath26 tev@xmath0 and @xmath27 tev@xmath0 , without making assumptions about the @xmath25 and @xmath24 couplings , respectively . these can be translated into limits on top quark decay branching fractions , which are @xmath28 and @xmath29 . , title="fig:",width=226 ] , title="fig:",width=226 ] models of new physics that introduce additional symmetries predict the existence of additional heavy charged bosons , generally called @xmath4 . the d0 collaboration has performed a search for @xmath4 boson production in 2.3 fb@xmath0 of data @xcite , using the same event selection and background modeling as the single top quark observation analysis @xcite . this analysis explores the full parameter space of both sm - like left - handed couplings ( @xmath30 ) and right - handed couplings ( @xmath31 ) and a mixture of both for the @xmath4 boson . for each @xmath4 boson mass under consideration , a boosted decision tree ( bdt ) is trained to separate the @xmath4 boson signal from the backgrounds . the bdt output for one mass is shown in fig . [ fig : wprime](a ) . the resulting limits on the @xmath4 boson mass as a function of the two couplings are shown in fig . [ fig : wprime](b ) . for sm - like left - handed @xmath4 couplings , the limit on the @xmath4 boson mass is @xmath32 gev . for purely right - handed couplings it depends on whether a right - handed neutrino ( @xmath33 ) exists and is lighter than the @xmath4 boson , the limit is @xmath34 gev for @xmath35 and @xmath36 gev for @xmath37 . if both left - handed and right - handed couplings are present , the limit is @xmath38 gev . boson search for a @xmath4 boson mass of 850 gev , including the expected @xmath4 boson signal with mixed couplings . ( b ) contours of 95% c.l . lower limits on @xmath39 in the ( al , ar ) plane . [ fig : wprime],title="fig:",width=226 ] boson search for a @xmath4 boson mass of 850 gev , including the expected @xmath4 boson signal with mixed couplings . ( b ) contours of 95% c.l . lower limits on @xmath39 in the ( al , ar ) plane . [ fig : wprime],title="fig:",width=226 ] the d0 experiment at the tevatron has presented updates on the measurement of single top quark production . besides a more precise combined single top quark production cross section measurement , the @xmath1-channel mode was isolated for the first time . all of the measurements are consistent with the sm expectation . several new physics scenarios have been explored in the single top quark final state and limits have been set on on flavor - changing neutral currents and on new heavy boson @xmath4 production . 99 v. m. abazov _ et al . _ ( d0 collaboration ) , phys . lett . * 103 * , 092001 ( 2009 ) . t. aaltonen _ et al . _ ( cdf collaboration ) , phys . 103 * , 092002 ( 2009 ) . jikia and s.r . slabospitsky , phys . b * 295 * , 136 ( 1992 ) . chen , f. larios , and c. p. yuan , phys . b * 631 * , 126 ( 2005 ) . e. boos , l. dudko , and t. ohl , eur . j. c * 11 * , 473 ( 1999 ) a.p . heinson , a.s . belyaev , and e.e . boos , phys . d * 56 * , 3114 ( 1997 ) . abazov _ et al . _ ( d0 collaboration ) , phys . lett . * 101 * , 221801 ( 2008 ) . v. m. abazov _ et al . _ ( d0 collaboration ) , phys . lett . * 102 * , 092002 ( 2009 ) . t. tait and c .- p . yuan , phys . d * 63 * , 014018 ( 2001 ) . v. m. abazov _ et al . _ ( d0 collaboration ) , _ submitted to phys . b _ , [ arxiv:1108.3091 [ hep - ex ] ] ( 2011 ) . v. m. abazov _ et al . _ ( d0 collaboration ) , phys . d * 78 * , 012005 ( 2008 ) . n. kidonakis , phys . d * 74 * , 114012 ( 2006 ) . v. m. abazov _ et al . _ ( d0 collaboration ) , phys . lett . * b693 * , 81 - 87 ( 2010 ) . [ arxiv:1006.3575 [ hep - ex ] ] .	updates of electroweak single top quark production measurements by the d0 collaboration are presented using 5.4 fb@xmath0 of proton - antiproton collision data from the tevatron at fermilab . measurements of the @xmath1-channel , @xmath2-channel and combined single top quark production cross section are presented , including an updated lower limit on the ckm matrix element @xmath3 . also reported are results from searches for gluon - quark flavor - changing neutral currents and @xmath4 boson production .
diffuse , synchrotron radio sources with no obvious optical counterparts have been detected in several clusters of galaxies . they provide evidence for the existence of @xmath3 mpc scale distribution of magnetic fields and relativistic particles in the intra - cluster medium ( icm ) . these sources are broadly classified into two types- radio halos and radio relics . radio halos are unpolarized ( @xmath4 ) , have a regular morphology and are centrally located in galaxy clusters like the hot gas emitting thermal x - rays ; radio relics are elongated , single or double arc - like , located at the peripheries of clusters and show strong polarization ( @xmath5 ) ( feretti & giovannini 1996 ; giovannini et al 2002 ; ferrari et al 2008 ) . these sources have steep spectra ( @xmath6 , @xmath7 ) and are @xmath8 - 1 mpc in size . halos and relics have typical surface brightnesses of @xmath9mjy arcmin@xmath10 at 1.4 ghz . the origins of relativistic electrons and large scale magnetic fields in the icm are still unclear . the relativistic electron population in the icm could either be injected by radio galaxies and stellar feedback or electrons in the icm accelerated in shocks . the relativistic ( lorentz factor @xmath11 ) population of electrons can not diffuse to distances of @xmath3 mpc in their short radiative lifetimes of @xmath12yr in magnetic fields @xmath13 g . this leads to the requirement of in - situ re - acceleration of electrons in the icm to produce a mpc scale radio halo . there are two classes of models which are considered - ( a ) re - acceleration of electrons by turbulence in the icm ( reviews by brunetti 2003 , 2004 ; sarazin 2004 ; fujita et al 2003 ; brunetti et al 2004 ; brunetti & lazarian 2007 ) and ( b ) the secondary electron models(dennison 1980 ; blasi & colafrancesco 1999 ; dolag & ensslin 2000 ) . in the secondary electron models , the protons in the icm that are accelerated to relativistic energies in shocks lead to the production of relativistic electrons and positrons by inelastic p - p collisions . the secondary electrons are produced continuously in the cluster by the protons which have lifetimes comparable to the age of the cluster and thus cluster wide emission can be explained without encountering the problem of short radiative lifetimes of relativistic electrons . the gamma ray flux from the icm predicted by these models has not been detected so far . this is a serious drawback of these models . the two models considered for relics are ( a ) diffusive shock acceleration of thermal electrons by fermi i process ( ensslin et al 1998 ; roettiger et al 1999 ) or of fossil relativistic electrons ( markevitch et al 2005 ) and ( b ) revival of fossil radio cocoons by adiabatic compression due to shocks in the icm ( ensslin & gopal krishna 2001 , hereafter eg01 ; ensslin & bruggen 2002 ; hoeft et al . 2004 ) . large scale structure formation simulations show that structure formation , accretion and mergers drive shocks in the icm ( miniati et al . 2000 ; ryu et al . 2003 ; pfrommer et al . sub - structures in the x - ray emission from cluster cores , imaged in high resolution by _ chandra _ and xmm - newton , have indicated non - virialized states of galaxy clusters ( eg . mcnamara et al . 2002 ; markevitch et al . the gravitational binding energy released in cluster - cluster mergers is @xmath14 erg and thus if only @xmath15 of it is utilised in accelerating particles , sources like the radio halos ( @xmath16 erg ) could be produced . there is increasing evidence that radio halos and relics occur in clusters which have undergone a merger event in the last @xmath17 gyr ( feretti et al . 2006 ) . in a few clusters such as a2256 ( briel & henry 1994 ) and a1367 ( donnelly et al . 1998 ) it has been possible to detect cluster merger shocks as temperature jumps in the x - ray emitting icm . these clusters are also hosts to radio relics which have been proposed to have formed by acceleration of electrons in the shock structures of the cosmological large - scale matter flows ( ensslin et al 1998 ) . the shocks or turbulence in the icm , which are considered responsible for the acceleration of electrons , would leave signatures in the resulting spectrum of the accelerated particles . spectral index is a tracer of the age of the relativistic plasma and hence its distribution over the halo / relic can reveal sites of ongoing reacceleration . relics showing a trend of gradual flattening of spectral index from the edge near the cluster core to the edge farther from the cluster core could indicate relation to outgoing merger shocks . in the case of the double relics in a1240 ( bonafede et al . 2009 ) and the relic in a521 ( giacintucci et al . 2008 ) this possibility based on morphologies of the relics , the spectral index maps and the polarization has been considered . patches of steep / flat spectral index could indicate complex process of re - acceleration such as turbulent reacceleration . radial steepening of spectral index of halos could also be related to the cluster scale magnetic field distribution ( feretti et al 2004 ) . the correspondence between x - ray sub - structure or temperature map and the spectral index maps can reveal properties of the relativistic electron population . in order to carry out such studies of halos and relics , multi - frequency observations are necessary . radio halos and relics have been studied mainly at 1.4 ghz ( giovannini et al 1999 ; govoni et al 2001 ; bacchi et al 2003(hereafter ba03 ) ; clarke et al 2006 ) . observations at frequencies other than @xmath18ghz are required to study the spectra of halos and relics and to compare with model predictions . recently a low frequency halo has been detected in the cluster a521 by brunetti et al . ( 2008 ) in the maps at 235 , 330 and 610 mhz . this halo was missed at higher frequencies ( @xmath19 ghz ) due to cutoff in the spectrum . this detection emphasises the importance of low frequency observations and provides an example of a class of halos which will be detected only at low frequencies . the extended nature and steep spectra make halos and relics ideal targets for low frequency ( @xmath20 ghz ) observations . in recent years , the giant metrewave radio telescope ( gmrt ) has provided imaging opportunities at frequencies of 610 , 330 , 235 and 150 mhz with resolutions of @xmath21 , @xmath22 , @xmath23 and @xmath24 respectively . a few attempts have been made to image halos and relics with the gmrt . the radio halo in galaxy cluster a3562 has been successfully imaged at 235 mhz by giacintucci et al ( 2005 ) . brunetti et al . ( 2008 ) have discovered the low frequency halo in a521 with the gmrt . a search for radio halos and relics in galaxy clusters , which resulted in the detection of 10 new radio halos and 3 relics , has been carried out at 610 mhz with the gmrt ( venturi et al 2008 ) . at low frequencies , the primary beams are larger compared to those at higher frequencies and thus a larger field can be imaged in a single pointing . considering the fact that relics occur at peripheries of clusters , it is essential that a large field around the cluster centre be imaged . for example , in a2255 , which is well known for the presence of a halo and relic , new relics at a distance of @xmath25 mpc ( @xmath26 ) from the cluster centre were discovered in 350 mhz observations ( pizzo et al 2008 ) . with the hybrid configuration of the gmrt , imaging of structures of angular sizes @xmath27 with a resolution of @xmath28 at 150 mhz and with a sensitivity of @xmath31.5 - 3 mjybeam@xmath29 is possible . we selected halos and relics for low frequency observations with the gmrt . haloes / relics which had extents of @xmath30 at 1.4 ghz were selected so that the coverage at short baselines available at the gmrt is sufficient to image the full extents of these sources . further , sources with expected @xmath31 surface brightness at 150 mhz of @xmath32mjybeam@xmath29 were selected . with our previous experience of the dynamic range limitation of @xmath3 1500 in the gmrt images , we avoided those clusters which had radio sources @xmath33 jy within the hpbw of primary beam ( @xmath34 ) at 150 mhz . as a first phase of imaging galaxy clusters at low frequencies , three galaxy clusters a2255 , a2256 and a754 were observed at 150 mhz with the gmrt . galaxy cluster a2255 is host to a radio halo and a relic ( feretti et al . 1997 ) and two relics about 2 mpc from the cluster centre ( pizzo et al 2008 ) . the radio halo of a2255 is polarized ( @xmath5 , govoni et al 2005 ) . in a2256 a radio halo and a relic have been detected ( kim 1999 ; clarke et al 2006 ; brentjens 2008 ) . in a754 two components of diffuse radio emission ( kassim et al . 2001 , hereafter ka01 ; ba03 ) have been detected but the nature of these are still unclear . the x - ray images of these clusters show sub - structures which indicate on - going mergers ( sun et al 2002 ; markevitch et al 2003 ; sakelliou et al 2006 ) . these are of particular interest to investigate the connection between the mergers and the radio halos / relics . apart from being hosts to interesting extended diffuse radio sources , these clusters are complex and dynamically active . in this paper we present our results for the galaxy cluster a754 . a754 is a rich galaxy cluster at a redshift of 0.0542 @xcite , with an x - ray luminosity of @xmath35 @xmath36 erg s@xmath29 measured in the asca energy band of 0.5 - 10 kev ( henriksen & markevitch 1996 ) and a temperature of @xmath37kev ( wu et al . 1998 ; govoni et al 2004 ) . non - virialized nature of a754 has been inferred from the morphologies of x - ray emissions in the images made with rosat ( henry & briel 1995 ) , _ chandra _ ( markevitch et al 2003 ) and xmm - newton ( henry 2004 ) . the projected distribution of optical galaxies has been reported to be bimodal and oriented along the east - west axis ; the mean redshifts of the two galaxy concentrations being identical within errors ( fabricant et al 1986 ; zabludoff & zaritsky 1995 , hereafter zz95 ) . the peak of x - ray emission in a754 is not spatially coincident with either of the two major galaxy concentrations ( zz95 ) . a merger of two sub - clusters approximately in the plane of the sky along the east - west axis has been suggested ( zz95 ) . henriksen et al . ( 1996 ) using _ asca _ data proposed a non - head - on collision of subclusters in a754 . roettiger et al ( 1998 ) , using 3d mhd / n - body simulations , have proposed explanations for most of the observed x - ray properties of a754 by considering an off - axis merger between 2 smaller clusters of mass ratio less than 2.5:1 . markevitch et al . ( 2003 ) , based on the temperature map obtained using the _ chandra _ data on a754 , have suggested a complex geometry of merger involving more than 2 subclusters or a cool gas cloud sloshing independently from its former host subcluster . fusco - femiano et al ( 2003 ) have reported a 3.2@xmath38 detection of excess x - ray emission at energies above 45 kev with the _ bepposax _ which is possibly due to the inverse - compton emission by the relativistic plasma in the icm of a754 . absence of cooling flow ( white et al 1997 ; peres et al 1998 ) in a754 is another property indicating the on - going dynamical activity in the cluster . the diffuse radio emission in a754 has been under investigation for more than 25 years . diffuse radio emission in a754 at 2.7 ghz ( wielebinski et al 1977 ) and at 408 mhz ( mills , hunstead & skellern 1978 ) have been reported . the diffuse emission could not be distinguished from the blend of discrete sources due to poor resolutions of @xmath39 in those images . the presence of diffuse halo source in a754 at 610 mhz was also suggested by harris et al . ( 1980 ) based on the deficit in the flux recovered from the emission of discrete sources compared to the total flux reported in earlier low frequency observations . attempts of investigating the diffuse radio emission at 330 mhz in a 40 minute observation with the vla in b and c arrays ( giovannini , tordi & feretti 1999 ; giovannini & feretti 2000 ) failed to detect the diffuse emission due to lack of uv - coverage . the presence of diffuse emission in a754 has been confirmed by ka01 in the images made from 3 hr observations at each of 330 and 74 mhz with the vla - c array . in the deepest image of a754 at 1.3 ghz with the vla - d array , ba03 have confirmed the presence of two components in the diffuse radio emission . + in this paper we use @xmath40 , @xmath41 and a value of 73 km s@xmath29mpc@xmath29 for the hubble constant , @xmath42 . this implies a distance scale of 61.9 kpc arcmin@xmath29 at the redshift of a754 . gmrt is an aperture synthesis radio telescope consisting of 30 dish antennas , each 45 m in diameter . the array has 14 dishes placed in an area of 1 km@xmath43 , forming a central compact array , providing the uv- coverage at short baselines(@xmath44 m ) . remaining 16 antennas are spread along the 3 arms of a y - shape to a maximum distance of @xmath45 km . gmrt observations of the a754 region at 150 mhz for a total of @xmath46hr were conducted on 17 sep . 2007 . high resolution ( hires ) mode of the gmrt correlator provided frequency resolution of 31.25 khz over a bandwidth of 8 mhz . this mode was used to minimize the effects of bandwidth smearing and for better excision of radio frequency interference(rfi ) . standard sources 3c48 and 3c147 were used for the flux and the bandpass calibration . the source 0837 - 198 was used as the gain and phase calibrator . data at 150 mhz with the gmrt were adversely affected by rfi . an rfi affecting about 10 consecutive frequency channels and drifting over 2 mhz in frequency during the total 5 hr of observation was found along with other rfis which affected certain channels for limited amount of time . the aips ( astronomical image processing system ) task spflg was used to remove channels affected by rfi . a careful examination of frequency channels in all baselines was carried out to prevent loss of any good data while removing the rfi affected data . about 35% data were lost due to rfi . polyhedron imaging with 66 facets was carried out to account for the non - coplanarity of the incoming wavefront within the large primary beam of @xmath34 . hydra a , a strong radio source , @xmath47 from the phase centre was imaged to remove the effect of its sidelobes from the field of interest . the fwhm of the primary beam of the gmrt at 150 mhz is @xmath48 which corresponds to @xmath49 km of ionosphere at a height of @xmath50 km from the surface of the earth . the passage through the ionosphere of the incoming wavefront leads to an excess ionospheric phase . this excess phase , if constant over the entire field of view , can be solved for using self - calibration . this is the assumption of isoplanicity . isoplanicity could be a poor assumption for the gmrt at frequencies lower than 325 mhz due to long baselines and large field of view ( rao 2003 ) . spatial and temporal variations in the electron density in the ionosphere result in variations in excess phase within the primary beam . this leads to variations in the positions of sources with time . furthermore , at any given time the different positions in the primary beam can experience different position shifts . this is non - isoplanicity . one self calibration solution may not work for the entire primary beam under such conditions . we made images of 10 min data selected at the beginning and at the end of the observation to check if there was any effect of non - isoplanicity . positions of the sources in the entire primary beam in both the 10 min images were consistent with each other and also with those in the nrao vla sky survey(nvss ) catalogue within @xmath31 error of @xmath51 . self calibration worked equally well for all the sources in the entire primary beam . no effects of non - isoplanicity were seen . + data at 330 mhz from the gmrt archive were also edited and calibrated using aips . at 330 mhz @xmath5 data were lost due to rfi . polyhedron imaging was carried out with 32 facets . standard steps of self calibration were performed . image with a resolution of @xmath52 was produced . we used the vla d - array data from the archive at 1363 mhz to make an image with a resolution of @xmath53 . these images using the archival data were used to estimate the contributions of unresolved sources to the diffuse emission . images at 74 and 330 mhz(ka01 ) with resolutions of @xmath54 and @xmath55 , respectively , were provided by n. kassim and the image with a resolution of @xmath56 at 1363 mhz(ba03 ) was provided by g. giovannini . these images were used to study the diffuse emission . details of the data are given in table 1 . gmrt image with the highest possible resolution of @xmath57 was made using the 150 mhz data . while this high resolution image showed the detailed morphologies of several cluster radio galaxies , images with lower resolutions were produced at 150 mhz to detect the diffuse halo emission . for comparison with the 1363 mhz image , we convolved the image at 150 mhz with a beam of @xmath56 ; rms of 5.5 mjybeam@xmath29 was achieved . images at 150 and 1363 mhz of the a754 region where the diffuse emission is detected are shown in fig . 1a and 1b respectively . in fig . 2a and 2b , images at 330 and 74 mhz(kassim et al . 2001 ) , respectively , are shown . in each of these images the locations of diffuse emission detected at 150 mhz are marked by rectangles . we use labels for discrete sources in the a754 field as used in fig . 3 of ba03 . for estimating the flux densities of discrete sources we have made use of images with resolutions of @xmath58 and @xmath59 at 1363 and 330 mhz respectively . the source s6 in fig . 1b is not detected at 150 mhz . it has flux densities of 33.0 and 7.0 mjy at 330 and 1363 mhz respectively which imply a spectral index of 1.0 . the non - detection of s6 above the @xmath60 level of 16.5 mjy at 150 mhz indicates that the spectrum has turned over . the non - detection of s5 ( 2.5 mjy at 1363 mhz ) at 330 and 150 mhz is consistent with a spectral index of 0.8 . flux density of s1 at 1363 mhz is 3.6 mjy and the non - detection at 330 and 150 mhz implies a spectrum flatter than 0.8 . the details of the flux densities of each blob of diffuse emission marked by rectangle in fig . 1a are given in table 2 . the west relic in the 74 mhz image of ka01 at ra=09h07m25s dec=-09d36m52s is not detected in any of the 150 , 330 and 1363 mhz images . if the west relic is real then the non - detection of it at 150 mhz implies a spectral index between 74 and 150 mhz steeper than 1.8 . the integrated spectra of each of the blobs 1 , 2 , 3 and 4 with 3@xmath38 error bars are plotted in fig . the flux densities of the blobs at each of the frequencies were estimated using rectangles of the same dimension as shown in fig . the @xmath31 error on the total flux density at 150 and 330 mhz is @xmath61 and at 1363 mhz is @xmath62 . in the case where the emission was not detected , the total flux density recovered in the corresponding box is reported as an estimate of upper limit . these upperlimits are plotted for blobs 1 , 2 and 3 at 330 mhz and for blob 1 at 1.4 ghz . in the case of blob 3 , the discrete source s6 contributed 4.8 mjy to the box shown . this value has been subtracted from the total flux density in box 3 to get the estimate of the flux density of the blob 3 . the locations of the diffuse radio emission are marked by boxes on the x - ray image ( xmm - newton , 0.8 - 2 kev , henry et al . ( 2004 ) ) of a754 ( fig . the locations of optical subclusters ( zz95 ) are marked by star symbols . the proposed direction of merger of 2 sub - clusters of galaxies in a754 is along the line joining the two optical clumps ( zz95 ) . the peak of the x - ray emission is elongated perpendicular to the proposed direction of merger and so is blob 4 emission but @xmath63 kpc from the x - ray peak . there is no clear indication of a shock in the x - ray images of the region where the relic emission is located . the blob 4 is cospatial with the western clump of optical galaxies . based on the comparison of temperature map and radio image of a754 , govoni et al ( 2004 ) show a clear anti - correlation between the location of the diffuse radio sources and the regions of hottest x - ray emitting gas in a754 . the blob 4 in fig . 1a is consistent with the halo marked in the fig.1 of fusco - femiano et al ( 2003 ) . however , we shall refer to the blob 4 as a relic due to its elongated morphology and the location away from the peak of the x - ray emission . a spectral index of @xmath64 was estimated using a linear fit to the spectrum . the extent of the relic projected in the plane of the sky is @xmath65 kpc@xmath43 . the volume of the relic , estimated assuming a cylindrical geometry with the smaller side of the box as the diameter , is @xmath66 mpc@xmath67 . the equipartition magnetic field of @xmath68 g was computed using standard formulae ( pacholczyk 1970 ) , integrating the radio emission between 10 mhz and 10 ghz , with a radio spectral index @xmath69 , and assuming equal energy density in protons and electrons , and a volume filling factor of 1 . we have tried to understand the diffuse emission in a754 in the context of the adiabatic compression model ( eg01 ) . radio galaxies deposit relativistic plasma to the icm through jets . after the jets turn off , apart from the @xmath70 work done while expanding in the icm , the relativistic plasma in the radio lobes loses energy by synchrotron and inverse compton cooling on timescales of @xmath12yr . this plasma , if compressed adiabatically by shock waves in the icm , can emit detectable radio emission . this idea of reviving fossil radio lobes by adiabatic compression is discussed in eg01 in detail . the salient features of this model are described here . the rate of change of momentum , ( @xmath71 ) of relativistic electron is proportional to the magnetic energy density ( synchrotron losses ) , the energy density of the cosmic microwave background radiation ( inverse compton losses ) and the adiabatic gain or loss due to the change in the volume ( @xmath72 ) of the radio plasma . a ratio of the initial volume of the radio cocoon ( @xmath73 ) and the volume at time @xmath74 ( @xmath75 ) , called the compression ratio is defined as @xmath76 . the change in the volume is approximated to be a power law in time , @xmath77 . adiabatic expansion of a spherical volume ( radius @xmath78 to @xmath79 ) of ultra - relativistic particles of energy @xmath80 leads to the scaling of energy as @xmath81 ( eqn . 11.30 , longair 1981 ) and thus of energy density as @xmath82 . assuming isotropic adiabatic expansion of magnetized plasma , the magnetic field energy density ( @xmath83 ) scales as @xmath84 . the evolution of the radio plasma is divided into five discrete phases - ( 0 ) injection , ( 1 ) expansion , ( 2 ) lurking , ( 3 ) flashing and ( 4 ) fading . _ phase 0 _ ( injection ) : the radio galaxy is active . a large expanding volume is filled with the radio plasma by the jets . the typical timescale over which this phase lasts is @xmath85 gyr ( eg01 ; alexander & leahy 1987 ) . this expansion is supersonic with respect to the external medium with an index @xmath86 , with the assumption that there is no gas density gradient in the vicinity of the radio galaxy ( eg01 ; kaiser & alexander 1997 ) . _ phase 1 _ ( expansion ) : the jets of the radio galaxy are off ; the radio cocoon expands in the surrounding medium due to higher internal pressure compared to the surrounding icm . the expansion is considered sedov - like ( @xmath87 ) throughout this phase for simplicity ( eg01 ) . _ phase 2 _ ( lurking ) : the radio cocoon reaches pressure equilibrium with the icm and thus the volume of the plasma remains constant . _ phase 3 _ ( flashing ) : if a shock compresses a lurking radio cocoon , there is enhanced radio emission due to adiabatic compression of the magnetic field . _ phase 4 _ ( fading ) : the compressed cocoon continues to lose energy by synchrotron and inverse compton losses and fades . each phase is characterised by its duration ( @xmath88 ) , the timescale of expansion ( @xmath89 ) and an index ( @xmath90 ) that determines the rate of expansion / compression of the radio cocoon . the compression ratio during a phase is related to these parameters by the relation , @xmath91 . the synchrotron emission in the _ i@xmath92 _ phase at a given frequency @xmath93 is given by , @xmath94 where @xmath95 is the characteristic frequency , given by , @xmath96 and @xmath97 is the dimensionless spectral emissivity of a mono - energetic electron in an isotropically oriented magnetic field ( ensslin et al 1999 ) and @xmath98 is the electron spectrum in the _ i@xmath92 _ phase resulting from an initial power - law distribution . further , the five phases are discussed under three scenarios . scenario a : cocoon at cluster centre- the external pressure provided by the icm is large and thus the pressure inside the radio cocoon is considered to be only twice the external pressure . the lurking phase lasts only for 0.1 gyr ; the cocoon will not be able to emit detectable radio emission after compression if it is older . scenario b : cocoon at cluster periphery- the freshly injected plasma is over pressured compared to the surrounding medium by a factor @xmath44 . pressure equilibrium with the surrounding medium is achieved faster . the plasma can be revived by shock compression even after 1 gyr . scenario c : cocoon near an active radio galaxy(smoking gun)- a radio galaxy which is possibly the source of the relativistic electrons is visible near the relic . this case is illustrated by the example of the relic in coma by eg01 ( see eg01 for details ) . in a754 the relic ( blob 4 ) has been detected at all the 3 frequencies . the model has been used to obtain a spectrum which fits the observed spectrum of the relic best . scenario a was considered best suited for a754 since the relic is located in a region where the x - ray emission has been detected ( henry et al 2004 ) . the steps carried out to arrive at the model spectra from the adiabatic compression model for the relic are described below . the radio plasma was assumed to be in @xmath99 phase and the estimated magnetic field ( @xmath100 ) and volume ( @xmath72 ) of the relic were used . using compression ratio ( @xmath101 ) for the @xmath99 phase , back calculation was carried out to find the @xmath100 and @xmath72 for @xmath102 phase successively till phase 0 . the spectrum was obtained for each of the phases . it was checked whether the model spectrum in the @xmath99 phase , in which the relic was assumed to be in , was a good fit to the integrated spectrum of the relic . if the spectrum was a bad fit to the observed spectrum , the plasma was assumed to be in another phase and the same steps were carried out . in a case where the relic was assumed to be in phase 1 , the corresponding model spectrum was flatter than could fit the observed data . in another case where the relic was assumed to be in phase 2 , the spectrum was steeper . a phase intermediate to 1 and 2 was tried and the best fit to the observed spectrum was obtained . this intermediate phase is referred as `` phase @xmath103 '' . the model spectra are shown in fig . these spectra were constructed using the parameters listed in table 3 . the observed values of the flux density of blob 4 at 150 , 330 and 1363 mhz with the @xmath104 error bars are plotted along with the model spectra . the initial energy distribution of the relativistic electrons was assumed to be a power law with a spectral index of 2.6 ( @xmath105 ) . the flux density at 150 mhz was used to normalize the model spectrum . a good fit was obtained by considering the relic to be in flashing phase ( phase 3 ) too . the volume in the injection phase ( phase 0 ) required in this case to produce the observed size of the relic after compression by shock is @xmath106 mpc@xmath67 . this implies that a volume of @xmath106mpc@xmath67 must have been filled with radio plasma by cluster radio galaxies forming a gigantic cocoon . considering the typical size of a radio galaxy to be 100 kpc ( middle value in fig . 9 of blundell et al 1999 ) and assuming it to fill a sphere of that radius with relativistic plasma , at least @xmath107 such sources are required to be active within a period of @xmath108gyr , to fill a giant cocoon of volume 0.15 mpc@xmath67 . such a situation is improbable in any typical galaxy cluster . in imaging extended emission at radio frequencies crucial roles are played by the uv - coverage , the sensitivity and the field of view . these vary with the observing frequency and the instrument used . further , the spectra of halos and relics are not the same over the extent of the source ; there can be isolated regions of steep and flat spectral indices . due to this interplay between the different aspects of the observational limitations and the intrinsic properties of the diffuse emission , multi - frequency observations of halos and relics are necessary to obtain maximum information and to make further inferences . in the case of a754 , blob 1 has been detected only at 150 mhz whereas blob 4 is detected at all of 150 , 330 and 1363 mhz , but better at 1363 mhz . the spectra of blobs 1 , 2 and 3 are steeper than that of blob 4 . the framework of adiabatic compression model ( eg01 ) has been used to explain the spectrum of the blob 4 in a754 . it was found that , the relic ( blob 4 ) can be considered as a cocoon of a radio galaxy lurking for @xmath109 yr in the icm . no shock compression was required to produce the spectrum of the relic . the consideration of blob 4 as a compressed radio cocoon ( flashing phase ) led to an unrealistic situation of the cluster having a gigantic ( @xmath106 mpc@xmath67 ) radio cocoon in the past , which appears unlikely . thus , merger shocks need not be invoked to explain the observed properties of the relic . the flux densities at 150 , 330 and 1363 mhz imply spectral indices steeper than 1.5 for blobs 1 , 2 and 3 . based on the result for blob 4 from the adiabatic compression model , these blobs of diffuse emission are likely to be radio cocoons lurking in the icm for more than @xmath110yr . in a few clusters that have relics , merger shocks have been considered to be responsible for accelerating electrons ( eg . a3376 , bagchi et al . according to the numerical model of roettiger et al ( 1998 ) in a754 , a large subcluster has crossed the main cluster from east to west . the 4 diffuse blobs in a754 lie along the east - west axis which is approximately the proposed axis of merger ( zz95 ) . the proposed location of the eastern shock wave is coincident with the x - ray peak and is close to blobs 1 and 2 . the steep spectral indices of blobs 1 and 2 could be explained in the fermi i acceleration scenario by a low mach number shock ( mach number @xmath111 for a spectral index steeper than 2.0 ; blandford & eichler 1987 ) . the western shock has been proposed to be at a location @xmath112 mpc from the x - ray peak ( roettiger et al . 1998 ; ka01 ) . the blob 4 which is in the western region is only 0.7 mpc from the x - ray peak and is not coincident with the proposed location of the shock . according to the numerical model the shocks crossed the cluster more than 0.3 gyr ago which is longer than the radiative lifetime of the electrons observed at 150 mhz . in a754 , no other relics have been detected in the primary beam at 150 mhz . sensitive polarization study of the diffuse emission has not been carried out to find other signatures of shocks such as alignment of magnetic field in the plane of the shock . upperlimits on polarized flux density of @xmath113 and @xmath114 have been obtained by ba03 on the western and eastern diffuse emission respectively in a754 . mergers can create turbulence in the icm which could then accelerate particles . it has been proposed that fermi acceleration powered by turbulence in the icm can accelerate particles to relativistic energies ( brunetti et al 2008 and references therein ) if the turbulence lasts for more than @xmath110yr . particles in the icm , through resonant scattering off the turbulent waves , can be stochastically accelerated to relativistic energies ( melrose 1980 ) . the relation between mergers and turbulent acceleration of particles is complex and the details are not clear . the steep spectrum with a cutoff at high frequencies is a signature of turbulent reacceleration ( kuo et al 2003 ; brunetti et al 2004 ; cassano et al 2005 ) . a halo has been detected in a521 at low frequencies such as 235 , 330 and 610 mhz ( brunetti et al . 2008 ) having spectral index of @xmath25 . a high frequency spectral cutoff is suspected by the authors for this halo making it undetectable at 1.4 ghz with the reported sensitivity of their observation . turbulent reacceleration model has been invoked to explain the low frequency halo . it is likely that many such diffuse features having cutoff at high frequencies could be lurking in clusters but have not been detected since no low frequency observations have been carried out . the non - detection of blob 1 at 330 and 1363 mhz indicates cutoff in its spectrum . this implies the possibility of turbulent reacceleration . the case of blob 4 , not requiring shock to explain the radio emission , is also consistent with the picture of turbulent reacceleration . another possibility is that the shocks accelerated the protons in the icm which produced secondary electrons and positrons by hadronic collisions . these secondary particles then produced the observed radio - halo emission . in this case the radio emission can last long after the passage of shock . this scenario will be tested with the detection of gamma ray flux expected from the decay of neutral pions after the p - p collisions ( ensslin et al . 1997 ; colafrancesco & blasi 1998 ; blasi 1998 ; dolag & ensslin 2000 ) . we have presented a gmrt 150 mhz image of the a754 region . we detected 4 blobs of diffuse emission at 150 mhz with a resolution of @xmath115 . we made high resolution images using the archival data at 330 ( gmrt ) and 1363 mhz ( vla ) to estimate flux densities of discrete sources and used the published images of a754 at 74 , 330 and 1363 mhz(vla ) to compare the diffuse features . the blob 1 is detected only at 150 mhz . the blob 4 , is detected at all the three frequencies . the blob 4 , lies along the axis of merger in a754 and on the far side of the x - ray peak ; thus we refer to it as a relic . within the framework of the adiabatic compression model ( eg01 ) , we found that the blob 4 can be explained as an ageing cocoon of an old radio galaxy in the cluster . the spectra of blobs 1 , 2 and 3 of diffuse emission are steeper and thus can be considered as radio cocoons older than the relic . not requiring shock to explain the radio emission of blob 4 , is also consistent with the picture of turbulent reacceleration . we thank biman nath and dipankar bhattacharya for critical comments on the manuscript . we thank g. giovannini ( 1363 mhz ) and n. kassim ( 74 and 330 mhz ) for providing the images of a754 . giant metrewave radio telescope is run by the national centre for radio astrophysics of the tata institute of fundamental research . the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc . alexander , p. , leahy , j. p. 1987 , , 225 , 1 bacchi , m. , feretti , l. , giovannini , g. , & govoni , f. 2003 , a&a , 400 , 465 bagchi , j. , durret , f. , neto , g. b. l. , paul , s. 2006 , sci , 314 , 791 blandford , r. , & eichler , d. 1987 , phy . rep . , 154 , 1 blasi , p. , & colafrancesco , s. 1999 , astropart . , 12 , 169 blundell , k. m. , rawlings , s. , willot , c. j. , kassim , n. e. & perley , r. 2002 , , 46 , 75 bonafede , a. , giovannini , g. , feretti , l. , govoni , f. , & murgia , m. 2009 , a&a , 494 , 429 brentjens , m. a. 2008 , a&a , 489 , 69 briel , u. g. , & henry , j. p. 1994 , nature , 372 , 439 brunetti , g. 2003 , matter and energy in clusters of galaxies , asp conference proceedings , 301 , ed . stuart b. & chorng - yuan h. , astronomical society of the pacific san francisco , 349 brunetti , g. , giacintucci , s. , cassano , r. , lane , w. , dallacasa , d. , venturi , t. , kassim , n. e. , setti , g. , cotton , w. d. & markevitch , m. 2008 , nature , 455 , 944 brunetti , g. 2004 , outskirts of galaxy clusters : intense life in the suburbs , ed . diaferio , a. , iau colloquium , 195 , 148 brunetti , g. , blasi , p. , cassano , r. , & gabici , s. 2004 , , 350 , 1174 brunetti , g. & lazarian , a. 2007 , , 378 , 245 cassano , r. , & brunetti , g. 2005 , , 357 , 1313 clarke , t. e. & ensslin , t. 2006 , , 131 , 2900 colafrancesco , s. & blasi , p. 1998 , astroparticle phy . , 9 , 227 dennison , b. 1980 , , 239 , 93 dolag , k. , & ensslin , t. a. 2000 , a&a , 362 , 151 donnelly , r. h. , markevitch , m. , forman , w. , jones , c. , david , l. p. , churazov , e . , & gilfanov , m. 1998 , , 500 , 138 ensslin , t. a. , biermann , p. l. , klein , u. , & kohle , s. 1998 , a&a , 332 , 395 ensslin , t. a. & bruggen , m. 2002 , , 331 , 1011 ensslin , t. a. , & gopal - krishna 2001 , a&a , 366 , 26 ensslin , t. a. , lieu , r. , biermann , p. l. 1999 , a&a , 344 , 409 fabricant , d. , beers , t. c. , geller , m. j. , gorenstein , p. , huchra , j. p. , & kurtz , m. j. 1986 , , 308 , 530 feretti , l. 2006 , proceedings of the xlist rencontres de moriond , xxvith astrophysics moriond meeting : `` from dark halos to light '' , ed . l.tresse , s. maurogordato and j. tran thanh van feretti , l. , boehringer , h. , giovannini , g. , & neumann , d. 1997 , a&a , 317 , 432 feretti , l. & giovannini , g. 1996 , iau symp . 175 , extragalactic radio sources , ed . r. d. ekers , c. fanti , and l. padrielli ( dordrecht : kluwer ) , 333 feretti , l. , orru , e. , brunetti , g. , giovannini , g. , kassim , n. , & setti , g. 2004 , a&a , 423 , 111 ferrari , c. , govoni , f. , schindler , s. , bykov , a. m. & rephaeli , y. 2008 , ssrv , 134 , 93 fujita , y. , takizawa , m. & sarazin , c. l. 2003 , , 584 , 190 fusco - femiano , r. , orlandini , m. , de grandi , s. , molendi , s. , feretti , l. , giovannini , g. , bacchi , m. , govoni , f. 2003 , a&a , 398 , 441 giacintucci , s. , venturi , t. , brunetti , g. , bardelli , s. , dallacasa , d. , ettori , s. , finoguenov , a. , rao , a. p. , & zucca , e. 2005 , a&a , 440 , 867 giacintucci , s. , venturi , t. , macario , g. , dallacasa , d. , brunetti , g. , markevitch , m. , cassano , r. , bardelli , s. , & athreya , r. 2008 , a&a , 486 , 347 giovannini , g. , & feretti , l. 2000 , newastr , 5 , 335 giovannini , g. & feretti , l. 2002 , merging processes in galaxy clusters , ed . l. feretti , i.m . gioia , g. giovannini , assl ( dordrecht : kluwer ) , 272 , 197 giovannini , g. , tordi , m. , & feretti , l. 1999 , , 4 , 141 govoni , f. , feretti , l. , giovannini , g. , bohhringer , h. , reiprich , t. h. , & murgia , m. 2001 , a&a , 376 , 803 govoni , f. , markevitch , m. , vikhlinin , a. , vanspeybroeck , l. , feretti , l. , & giovannini , g. 2004 , , 605 , 695 govoni , f. , murgia , m. , feretti , l. , giovannini , g. , dallacasa , d. , & taylor , g. b. 2005 , a&a , 430 , 5 harris , d. e. , pineda , f. j. , delvaille , j. p. , schnopper , h. w. , costain , c. h. , & strom , r. g. 1980 , a&a , 90 , 283 henriksen , mark j. & markevitch , maxim l. 1996 , , 466 , 79 henry , j. p. , briel , u. g. , 1995 , , 443 , 9 henry , j. p. , finoguenov , a. , & briel , u. g. 2004 , , 615 , 181 hoeft , m. , bruggen , m. , & yepes , g. 2004 , , 347 , 389 kaiser , c. r. , alexander , p. 1997 , , 286 , 215 kassim , n. e. et al . , 2001 , , 559 , 785 kim , k. -t . 1999 , jkas , 32,75 kuo , p. -h . , hwang , c. -y . , & ip , w. -h . 2003 , , 594 , 732 longair , m. s. 1981 , high energy astrophysics ( cambridge university ) markevitch , m. , govoni , f. , brunetti , g. , & jerius , d. 2005 , , 627 , 733 markevitch , m. , mazzotta , p. , vikhlinin , a. , burke , d. , butt , y. , david , l. , donnelly , h. , forman , w. r. , harris , d. , kim , d .- w . , virani , s. , vrtilek , j. 2003 , , 586 , 19 mcnamara , b. r. 2002 , the high energy universe at sharp focus : chandra science , asp conf . proc . , 262 , ed . schlegel , e. m. & vrtilek , s. d. san francisco : astronomical society of the pacific , 351 melrose , d. b. 1980 , plasma astrophysics ( gordon & breach ) miniati , f. , ryu , d. , kang , h. , jones , t. w. , cen , r. , & ostriker , j. 2000 , , 542 , 608 mills , b. y. , hunstead , r. w. , & skellern , d. j. 1978 , , 185 , 51 pacholczyk , a. g. 1970 , radio astrophysics ( freeman : san francisco ) peres , c. b. , fabian , a. c. , edge , a. c. , allen , s. w. , johnstone , r. m. , & white , d. a. 1998 , , 298 , 416 pfrommer , c. , springel , v. , ensslin , t. a. , & jubelgas , m. 2006 , , 367 , 113 pizzo , r. f. , de bruyn , a. g. , feretti , l. , & govoni , f. 2008 , a&a , 481 , 91 rao a. p. 2003 , low frequency radio astronomy , ed . chengalur , j. n. , gupta , y. , & dwarakanath , k. s.(http:@xmath116www.gmrt.ncra.tifr.res.in/gmrt@xmath117hpage/ user / doc / weblf / lfra / index.html ) roettiger , k. , burns , j. o. , & stone , j. m. 1999 , , 518 , 603 roettiger , k. , stone , j. m. , & mushotzky , r. f. 1998 , , 493 , 62 ryu , d. , & kang , h. 2003 , jkas , 36 , 105 sakelliou , i. , & ponman , t. j. , 2006 , , 367 , 1409 sarazin , c. l. 2004 , jkas , 37 , 433 strubble , m. f. & rood , h. j. 1999 , 125 , 35s sun , m. , murray , s. s. , markevitch , m. , & vikhlinin , a. 2002 , , 565 , 867 venturi , t. , giacintucci , s. , dallacasa , d. , cassano , r. , brunetti , g. , bardelli , s. , & setti , g. 2008 , a&a , 484 , 327 white , d. a. , jones , c. , & forman , w. 1997 , , 292 , 419 wielebinski , r. , waldthausen , h. , kronberg , p. p. , & haslam , c. g. t. 1977 , nature , 266 , 239 wu , x .- p . , fang , l .- z . , & xu , w. , 1998 , a&a , 338 , 813 zabludoff , a. i. , & zaritsky , d. 1995 , , 447 , 21 lccccc frequency & telescope & date&time on&synthesized & rms + ( mhz ) & & of obsn.&source(min)&beam&(mjybeam@xmath29 ) + 74(ka01)&vla - c&21 mar . 2000&180&@xmath54&200 + 150 & gmrt & 17 sep . 2007 & 225 & @xmath118 & 2.6 + 330 & gmrt & 23 jun . 2005&150&@xmath119 & 0.6 + 330(ka01 ) & vla - c & 21 mar . 2000&180&@xmath55 & 5.3 + 1363(ba03 ) & vla - d & 25 sep . 2000 & 180 & @xmath120 & 0.1 + lcccccllll & & @xmath121 & extent + & ( 150 mhz ) & ( 330 mhz)&(1363 mhz ) & & ( kpc@xmath43 ) + blob 1 & 174 & @xmath12228 & @xmath1222.0 & 2.0 & 240@xmath123 200 + blob 2 & 188 & @xmath12238 & 6.5 & 1.5 & 260@xmath123 160 + blob 3 & 340 & @xmath122130 & 5.0 & 1.9 & 280@xmath123 540 + blob 4 & 540 & 230 & 23.0 & 1.4&350@xmath123 400 + llllll & @xmath88 & @xmath89 & @xmath90 & @xmath72 & @xmath100 + & ( gyr ) & ( gyr ) & & ( mpc@xmath67 ) & ( @xmath124 g ) + phase 0 & 0 & 0.015 & 1.8 & 0.02 & 1.0 + phase 1 & 0.0054 & 0.01 & 1.2 & 0.04 & 0.7 + phase 1.5 & 0.09 & @xmath125 & 0 & 0.04 & 0.7 + phase 2 & 0.1 & @xmath125 & 0 & 0.04 & 0.7 + phase 3 & 0.069 & -0.11 & 2.0 & 0.006 & 2.42 + phase 4 & 0.02 & @xmath125 & 0 & 0.006 & 2.42 +	we present a low frequency study of the diffuse radio emission in the galaxy cluster a754 . we present new 150 mhz image of the galaxy cluster a754 made with the giant metrewave radio telescope ( gmrt ) and discuss the detection of 4 diffuse features . we compare the 150 mhz image with the images at 74 , 330 and 1363 mhz ; one new diffuse feature is detected . the flux density upperlimits at 330 and 1363 mhz imply a synchrotron spectral index , @xmath0 , ( @xmath1 ) for the new feature . the west relic detected at 74 mhz ( kassim et al 2001 ) is not detected at 150 mhz and is thus consistent with its non - detection at 1363 mhz ( bacchi et al 2003 ) and 330 mhz(kassim et al 2001 ) . integrated spectra of all the diffuse features are presented . the fourth diffuse feature is located along the proposed merger axis ( zabludoff et al 1995 ) in a754 and 0.7 mpc away from the peak of x - ray emission ; we refer to it as a relic . we have made use of the framework of adiabatic compression model ( ensslin & gopal - krishna 2001 ) to obtain spectra . we show that the spectrum of the fourth diffuse feature is consistent with that of a cocoon of a radio galaxy lurking for about @xmath2 yr ; no shock compression is required . the other three diffuse emission have spectra steeper than 1.5 and could be cocoons lurking for longer time . we discuss other possibilities such as shocks and turbulent reacceleration being responsible for the diffuse emission in a754 .
the goal of this paper is to elaborate on a kick mechanism first discussed by @xcite that may explain pulsar velocities greater than @xmath1 km@xmath2s@xmath3 . there have been a number of studies that have compiled and modelled the velocities of pulsars . although they disagree on whether the distribution is indeed bimodal , they agree that a significant number of pulsars are travelling faster than can be attributed to neutrino kicks . the analysis of @xcite favours a bimodal velocity distribution with peaks at 90 km@xmath2s@xmath3 and 500 km@xmath2s@xmath3 with @xmath4 of pulsars travelling at speeds greater than 1000 km@xmath2s@xmath3 . alternatively @xcite and @xcite both predict a single peaked distribution with an average velocity of @xmath5 km@xmath2s@xmath3 , but point out that the faster pulsars b2011 + 38 and b2224 + 64 have speeds of @xmath6 km@xmath2s@xmath3 . large velocities are unambiguously confirmed with the model independent measurement of pulsar b1508 + 55 moving at @xmath7 km@xmath2s@xmath3 @xcite . currently no mechanism exists that can reliably kick the star hard enough to reach these velocities . asymmetric explosions can only reach 200 km@xmath2s@xmath3 @xcite , and asymmetric neutrino emission is plagued by the problem that at temperatures high enough to produce the kick the neutrinos are trapped inside the star @xcite . alterations of the neutrino model that take into account only a thin shell of neutrinos require large temperatures and huge surface magnetic fields . the kick mechanism we will discuss relies on the existence of topological vector currents of the form described by @xcite , @xmath8 where @xmath9 and @xmath10 are the chemical potential of the right and left - handed electrons , and @xmath11 is the magnetic flux . there are three requirements for topological vector currents to be present : an imbalance in left and right - handed particles @xmath12 , degenerate matter @xmath13 , and the presence of a background magnetic field @xmath14 . all of these are present in neutron and quark stars . the weak interaction , by which the star attains equilibrium , violates parity ; particles created in this environment are primarily left - handed . the interior of the star is very dense , @xmath15 mev , and cold , @xmath16 mev , such that the degeneracy condition @xmath13 is met , and neutron stars are known to have huge surface magnetic fields , @xmath17 g. interested readers are encouraged to read @xcite . kicks caused by the topological current are very similar to neutrino kicks . the helicity of the electrons and the single spin state in the lowest landau level allow electrons to travel with a preferred direction through the neutron star . the difference between the two mechanisms is the linear momentum that the particle transfers to the star . neutrinos have an energy close to the temperature of the star , while the electrons , which are degenerate , have an energy close to their fermi momentum . we will discuss this critical difference in section [ difference ] . if the electrons carried by the current can transfer their momentum into space either by being ejected or by radiating photons the current could push the star like a rocket . in typical neutron stars this is unlikely because the crust ( the region where @xmath18 ) is thought to be about 1 km thick . once it reaches this thick crust , the current will likely travel along it and be reabsorbed into the bulk of the star . but if the crust is very thin , the electrons may leave the system or emit photons that will carry their momentum to space . the crust for bare quark stars is thought to be about 1000 fm ( * ? ? ? * see also recent developments by @xcite ) . with this in mind we conjecture that stars with very large kicks , @xmath19 km@xmath2s@xmath3 , are quark stars and that slow moving stars , @xmath20 km@xmath2s@xmath3 , are kicked by some other means , such as asymmetric explosions or neutrino emission , and are typical neutron stars . confirmation of this would provide an elegant way to discriminate between neutron stars and quark stars . the kick will likely only occur in quark stars that are bare or have very thin crusts . we will consider only the existence of light quarks in the star , which attain equilibrium through the quark direct urca processes , @xmath21 where the neutrino terms only appear in the final state of the interaction . the direct urca processes in quark matter do not have the same fermi momentum restrictions as in neutron matter and are thus not suppressed . these processes are at equilibrium when , @xmath22 in this balance we ignore the neutrino chemical potential . in discussing the kick we will focus on temperatures where the star makes the transition from being opaque to neutrinos to being transparent . we will refer to this temperature range as the translucent regime . in the opaque regime the neutrino chemical potential is nonzero . as the star cools the mean free path of the neutrinos becomes large enough that all neutrinos created may freely leave the star . this is precisely the temperature at which the topological current begins to flow and creates a kick . in discussing these temperatures there is the concern that the neutrino chemical potential can not be ignored . however , once in the transparent regime the neutrino chemical potential dissipates much faster than the time scale of the kick , thus we can safely ignore it . the quark matter must also be electrically neutral , @xmath23 where @xmath24 is the charge density and @xmath25 are the number densities of the quarks and electrons . we will consider the simplest model where the quark masses are set to zero , thus their fermi momenta are equal to their fermi energies . unfortunately , this approximation requires the electron density to be zero to satisfy equation ( [ eq : charge_balance ] ) , thus removing an essential element from the star : electrons . the strange quark mass is quite large and the number density should be described using nonrelativistic fermi statistics , which readmits elections into the mix to conserve charge . we will follow @xcite by assuming that the quarks are all massless and an electron chemical potential is present to capture the contribution of the mass of the strange quark . the fermi momentum of the quark species can be written in terms of the baryon number density , denoted @xmath26 , as @xmath27 for typical densities in the core of the neutron star the fermi momentum is @xmath28 mev , which is greater than the strange quark mass . neglecting the mass of the strange quark is a crude approximation but works well for this calculation . we can obtain the electron fermi momentum using the fraction of electrons to baryons , @xmath29 , which yields @xmath30 because @xmath31 , we find that @xmath32 . the typical value for the electron fraction is @xmath33 . this value is not critical in determining the strength of the kick because we will find that the dependance on @xmath34 cancels . however , as we will discuss in section [ difference ] , the value of the fermi momentum @xmath34 when compared to the core temperature of the star @xmath35 is critical in understanding how topological currents generate large kicks . in order to properly determine the size of a kick we need to understand how the core temperature of the quark star evolves with time . unfortunately , kicks are likely to occur right after the birth of the star during the most poorly understood stage of cooling . for insight into the cooling curve of a quark star we follow the work in two papers . the initial cooling of the star is described in @xcite , which focuses on neutrino diffusion through the star and thermal cooling . the star then cools until the neutrinos can escape the quark star and the cooling moves into a purely radiative regime as discussed in @xcite . the part of the cooling curve between these two well defined mechanisms constitutes the translucent regime . the quark direct urca processes , which are weak interactions , create both the neutrinos responsible for cooling the star and the parity that drives the current . the topological current only flows when the mean free path of the electron with respect to the weak interaction is larger than the neutron star . as they take part in the same processes , this is also when neutrinos start to escape the star . the mean free path of the weak interaction for an electron in quark matter has been calculated in @xcite and is @xmath36 equation ( [ eq : l_quark ] ) tells us that for a star of radius @xmath37 km the current starts to flow at @xmath38 k. at this point on the cooling curve the temperature changes on the scale of seconds . . , width=317 ] it is not clear how to model the short but critical cooling period that occurs in the translucent regime , as it is a very complicated process , but we assume it must make a smooth transition from the primarily thermal cooling of the core found in @xcite to the neutrino cooling found in @xcite . because we are interested in an estimate of the kick and not the exact details of the cooling we will interpolate between these two cooling curves by assuming the transition is smooth and exponential . we will use the cooling curve from figure 2a of @xcite that models the cooling of the core of the star . we then piece it together using an exponential decay to cool the star to @xmath39 k and match the initialization of the cooling curve given in figure 2 of @xcite . by requiring that the first derivatives are equal it is possible to find a unique exponential decay that patches these cooling curves together . figure [ figure_cooling ] plots this patched region that describes the translucent region of the cooling curve . it is possible that the cooling in the translucent regime is quite different from the exponential patch we propose . the severity of the exponential cooling provides us with a reasonable guess for the lower bound of the kick . slower cooling in the translucent regime would result in larger kicks . the topological current runs along the magnetic field of the star and consists of electrons created in the core of the star travelling to the surface . once at the surface these electrons transfer net momentum to the star . we assume that due to the thin crust of the quark star all of the electron s momentum is able to be transferred . the electrons may be physically ejected from the star or the momentum may be transferred by some radiative process such as bremsstrahlung . the size of the kick is directly related to the strength of the current , which is related to the creation rate of left - handed electrons in the quark star and the population of electrons in the lowest landau level . the calculations regarding the strength of the current in this paragraph are detailed in @xcite . we provide only a brief illustrative overview here . the direct urca processes for quarks create electrons at a rate per unit volume @xmath40 where @xmath41 and @xmath42 is a dimensionless integral . the helicity of the electrons @xmath43 and the fraction of electrons in the lowest landau level are contained in the parameter , @xmath44 the total number current for electrons reaching the surface of the star is given by , @xmath45 where @xmath46 g is the critical magnetic field , @xmath35 is the core temperature of the star , and @xmath47 is nuclear density . the typical density for quark matter is @xmath48 but could easily be higher . because the current originates in the core of the star , we are interested in the strength of the internal magnetic field . the dipolar component of the vela pulsar at its surface is about @xmath49 g @xcite but the field in the bulk of the star is likely much stronger based on virial theorem arguments in @xcite . the scalar virial theorem for the star states that @xmath50 where @xmath51 is the total rotational kinetic energy , @xmath52 is the internal energy , @xmath53 is the gravitational energy , and @xmath54 is the magnetic energy . because the kinetic energies are positive , the magnetic field energy can be at most equal to the gravitational energy , @xmath55 , which yield possible core fields of @xmath56 g. this is an extremely large field and is unlikely as it is a strict upper bound . arguments by @xcite based on glitch timing of the vela pulsar may put minimums on the core field depending on whether the protons are superconductors or not . as pointed out in @xcite , currents along type - ii vortices are likely to destroy superconductivity . @xcite also point out that the presence of topological currents could similarly destroy superconductivity in neutron stars , thus the non - superfluid estimate for the internal field from @xcite is appropriate . in very dense matter they find minimum interior fields of @xmath57 g. based on this we choose a value of the core magnetic field to be @xmath58 . the current , equation ( [ eq : current_quark ] ) , gives a number rate of electrons reaching the surface of the star . to get the momentum transferred per second we simply multiply by the momentum each electron carries , @xmath34 , which is large due to the degeneracy and is given by equation ( [ eqn : k_e ] ) . the acceleration for a @xmath59 star is then , @xmath60 we see that all the factors of @xmath34 cancel because @xmath61 contains a factor of @xmath62 . the kick is independent of the fermi momentum of the electrons , but as we will discuss later this does not mean the fermi momentum is unimportant . as seen in figure [ figure_kick ] , the star quickly reaches a speed of @xmath63 which is big enough to account for the large kicks seen in many pulsars . this result relies on the internal magnetic field of the star being two orders of magnitude larger than what we observe on the surface of most pulsars . we have also neglected the dissipation of electrons in the very thin crust of the quark star . as plotted , the entire kick seems to happen very quickly , but the current keeps running throughout the star s life . with a large internal magnetic field the mechanism can account for kicks seen in young pulsars such as vela . but because the kick is constantly running , pulsars with smaller internal magnetic fields will eventually attain very large speeds very late in life . it is important to understand the differences between neutrino kicks and topological kicks because superficially they seem very similar . in both cases a particle is created through the weak interaction with a favoured helicity ( left - handed ) , but for simplicity we will assume in this section that they are all left - handed . ] and the large magnetic field forces the electrons into landau levels . the lowest landau level only accepts a spin down electron , whereas all the rest allow both spin up and spin down states . combining the preferred helicity of the particles with the spin state asymmetry in the lowest landau level means that the neutrinos and electrons now travel with a preferred direction within the star . if the star is cool enough , these particles can escape the star without decaying . the neutrino does not participate in qed interactions and the electrons propagate due to the current , which is allowed because parity is a symmetry of qed . the particles that exit the star provide the linear momentum required to push the star . the electrons and neutrinos that contribute towards the kick are created at the same rate @xmath64 , given by equation ( [ eq : rate_quark ] ) . we are only interested in interactions that create particles in the lowest landau level so this rate is suppressed by a factor to account for the population of the lowest landau level . the momentum transferred to provide the kick is simply the momentum per particle that escapes the star . on the surface of the star , where the chemical potential is low , boltzmann statistics are used to model the occupation of the lowest landau level . this is what is seen in many neutrino kick models . this is because neutrinos require such hot temperatures to produce kicks that only a thin shell near the edge of the star is considered to contribute to the kick . in the core of the star , where the electrons are very degenerate , we must use fermi statistics . this predicts an occupation of @xmath65 in the lowest landau level in the interior of the star . at core temperatures below @xmath66 k the neutrinos can escape the star and the topological current is active . the flux of particles contributing to both the electron kick and the neutrino kick is about the same @xmath67 . the difference between the two mechanisms comes from the momentum that the relevant particle carries . the neutrinos are created thermally and the typical momentum of a neutrino is equal to the temperature of the star @xmath51 . the momentum of the electrons comes from the large chemical potential , @xmath68 mev . the momentum transfer per unit time for neutrinos dependance we expect for a calculation involving neutrino luminosity . ] is @xmath69 and for electrons is @xmath70 . when the kick starts the star has a temperature of only @xmath71 mev . the electron kick is stronger than the neutrino kick by a factor of @xmath72 initially , when the star is very hot , the electron kick is two orders of magnitude stronger than the neutrino kick . furthermore , as the star cools the neutrino kick gets even weaker , while the electrons continue to have a momentum dictated by their chemical potential . this why electrons generate larger kicks than neutrinos . at the beginning of the star s life the energy from the kick does not contribute to the cooling of the star , but later in life the current could become the dominant cooling mechanism . this may account for the colder than normal temperatures seen in the vela pulsar . the reason is that only a small fraction of the electrons created in the neutron star actually escape due to the low occupation of the lowest landau levels . as discussed earlier the estimate for the neutrino kick has a similar suppression factor , but all of the neutrinos actually escape the star , even those that do not contribute toward a neutrino kick . the electrons only propagate because the asymmetry in the lowest landau level allows the helicity states to propagate out of the star . those electrons that do not contribute toward the kick are trapped inside the star . the helicity states that reach the surface due to the current are the only electrons that contribute to the cooling of the star . the neutrinos cool the star with a luminosity @xmath73 where the electrons cool the star with an energy current ( luminosity ) of @xmath74 . the ratio of electron cooling to neutrino cooling is @xmath75 at the star s birth the electrons cool the star at about 1/100 the rate of neutrino cooling . as the star cools , eventually @xmath76 and the current becomes the dominant cooling mechanism of the star . this transition occurs at @xmath77 k. the current is an additional cooling mechanism to consider in stars that have cooled below @xmath0 k. this may explain the unusually cold surface temperature of @xmath78 k that is observed @xcite in the vela pulsar . we have found that the topological currents introduced in @xcite can generate kicks greater than 1000 km@xmath2s@xmath3 . this may explain large pulsar kicks that are otherwise unattainable using traditional kick mechanisms . topological kicks occur only if the electrons that make up the topological current can transfer their momentum through the star s crust . because neutron stars have thick crusts , it is unlikely that they will be kicked due to this mechanism , but rather by mechanisms that generate weak kicks . in contrast , bare quark stars have very thin crusts that may allow the electrons to escape . because of this distinction we conjecture that many of the fastest pulsars are quark stars . in calculating the kick velocity we assumed that the cooling curve of the quarks star must make a smooth transition from the primarily thermal cooling of the core found in @xcite to the neutrino cooling found in @xcite . this patched cooling curve allowed us to integrate over time the momentum transferred from the topological current and obtain the strength of the kick . we have also demonstrated that the energy lost due to the current does not change the cooling of the star during the kick . but , after the star cools below @xmath0 k the topological current becomes the dominant cooling mechanism rather than neutrino emission . the magnitude of the kick also depends on the strength of the magnetic field inside the star . the virial theorem puts a limit on the internal field of the star that is much higher than the surface field . to generate the kick we used a field strength of @xmath79 this is a reasonable estimate for pulsars that have a surface magnetic field of @xmath80 g such as the vela pulsar . this work was supported by the national science and engineering council of canada . the authors thank jean - franois caron and george hicks for their useful comments and corrections .	we show that kicks generated by topological currents may be responsible for the large velocities seen in a number of pulsars . the majority of the kick builds up within the first second of the star s birth and generates a force about two orders of magnitude larger than a neutrino kick in the same temperature and magnetic field regime . because of the nature of the topological currents the star s cooling is not affected until it reaches @xmath0 k ; thereafter the current replaces neutrino emission as the dominant cooling process . a requirement for the kick to occur is a suitably thin crust on the star ; this leads us to speculate that pulsars with large kicks are quark stars and those with small kicks are neutron stars . if true this would be an elegant way to distinguish quark stars from neutron stars . [ firstpage ] neutron star quark star kicks
a two - level laser is a source of coherent or chaotic light emitted by two - level atoms inside a cavity coupled to a vacuum reservoir via a single - port mirror . in one model of such a laser , two - level atoms initially in the upper level are injected at a constant rate into a cavity and removed after they have decayed due to spontaneous emission [ 1,2 ] . in another model the two - level atoms available in a cavity are pumped to the upper level by some convenient means such as electron bombardment [ 2,3 ] . there has been a considerable interest to study the quantum properties of the light generated by a two - level laser [ 4 - 10 ] . it is found that the light generated by this laser operating well above threshold is coherent and the light generated by the same laser operating below threshold is chaotic . in the quantum theory of a laser , one usually takes into consideration the interaction of the atoms inside the cavity with the vacuum reservoir outside the cavity . there may be some justification for the possibility of such interaction for a laser with an open cavity into which and from which atoms are injected and removed . however , there can not be any valid justification for leaving open the laser cavity in which the available atoms are pumped to the upper level by means of electron bombardment . therefore , the aforementioned interaction is not feasible for a laser in which the atoms available in a closed cavity are pumped to the upper level by electron bombardment . we seek here to analyze the quantum properties of the light emitted by two - level atoms available in a closed cavity and pumped to the upper level at a constant rate . thus taking into account the interaction of the two - level atoms with a resonant cavity mode and the damping of the cavity mode by a vacuum reservoir , we obtain the photon statistics , the quadrature variance , and the power spectrum for the light emitted by the atoms . we carry out this analysis by putting the noise operators associated with the vacuum reservoir in normal order and without considering the interaction of the two - level atoms with the vacuum reservoir outside the cavity . we consider here the case in which @xmath0 two - level atoms are available in a closed cavity . then the interaction of the cavity mode with one of the atoms can be described at resonance by the hamiltonian @xmath1 where @xmath2 is a lowering atomic operator , @xmath3 is the annihilation operator for the cavity mode , and @xmath4 is the coupling constant between the atom and the cavity mode . we assume that the laser cavity is coupled to a vacuum reservoir via a single - port mirror . in addition , we carry out our calculation by putting the noise operators associated with the vacuum reservoir in normal order . thus the noise operators will not have any effect on the dynamics of the cavity mode operators . we can then drop the noise operator and write the quantum langevin equation for the operator @xmath3 as @xmath5,\ ] ] where @xmath6 is the cavity damping constant . therefore , with the aid of ( [ 1 ] ) , we readily find @xmath7 furthermore , employing the relation @xmath8\rangle\ ] ] along with eq . ( [ 1 ] ) , one readily obtains @xmath9 @xmath10 @xmath11 where @xmath12 @xmath13 we see that eqs . ( [ 6 ] ) , ( [ 7 ] ) , and ( [ 8 ] ) are nonlinear differential equations and hence it is not possible to obtain exact time - dependent solutions of these equations . thus applying the large - time approximation scheme [ 11 ] , we obtain from eq . ( [ 4 ] ) the approximately valid relation @xmath14 now substitution of ( [ 11 ] ) ( with the time argument suppressed ) into the aforementioned equations yields @xmath15 @xmath16 @xmath17 where @xmath18 we wish to call the parameter defined by eq . ( [ 15 ] ) the stimulated emission decay constant . based on the definition of this decay constant , we infer that an atom in the upper level and inside a closed cavity emits a photon due to its interaction with the cavity light . we certainly identify this process to be stimulated photon emission . in order to include the contribution of all the atoms to the dynamics of the two - level laser , we sum eqs . ( [ 12 ] ) , ( [ 13 ] ) , and ( [ 14 ] ) over the @xmath0 two - level atoms , so that @xmath19 @xmath20 @xmath21 in which @xmath22 @xmath23 @xmath24 with the operators @xmath25 and @xmath26 representing the number of atoms in the upper and lower levels . furthermore , with the aid of the identity @xmath27 we see that @xmath28 in addition , using the definition given by ( [ 2 ] ) and setting for any @xmath29 @xmath30 we have @xmath31 we therefore observe that @xmath32 in which @xmath33 following the same procedure , one can also establish that @xmath34 with @xmath35 in the presence of @xmath0 two - level atoms , we rewrite eq . ( [ 4 ] ) as @xmath36 we now proceed to determine the value of the constant @xmath37 . thus applying the steady - state solution of eq . ( [ 4 ] ) , we easily find @xmath38_{k}={\gamma_{c}\over\kappa}(\hat{\eta}_{b}^{k}-\hat{\eta}_{a}^{k})\ ] ] and on summing over all atoms , we have @xmath39={\gamma_{c}\over\kappa}(\hat{n}_{b}-\hat{n}_{a}),\ ] ] where @xmath40=\sum_{k=1}^{n}[\hat{b},\hat{b}^{\dagger}]_{k}\ ] ] stands for the commutator of @xmath3 and @xmath41 when the cavity mode is interacting with all the @xmath0 two - level atoms . on the other hand , using the steady - state solution of eq . ( [ 30 ] ) , one can easily establish that @xmath42=n\bigg({2\lambda\over\kappa}\bigg)^{2}(\hat{n}_{b}-\hat{n}_{a}).\ ] ] hence from eqs . ( [ 32 ] ) and ( [ 34 ] ) , we get @xmath43 and on account of this result , eq . ( [ 30 ] ) can be written as @xmath44 the two - level atoms available in the cavity are pumped from the lower to the upper level by means of electron bombardment . the pumping process must certainly affect the time evolution of the atomic operators . hence we take into account the effect of the pumping process on the time evolution of the operators @xmath45 and @xmath46 by rewriting eqs . ( [ 17 ] ) and ( [ 18 ] ) as @xmath47 and @xmath48 where @xmath49 is the rate at which a single atom is pumped to the upper level . now taking into account ( [ 23 ] ) , one can put eq . ( [ 37 ] ) in the form @xmath50 we immediately see that the steady - state solution of this equation is @xmath51 and the steady - state solution of eq . ( [ 38 ] ) turns out to be @xmath52 finally , we seek to determine the effect of the pumping process on the dynamics of the atomic operator @xmath53 . to this end , we rewrite eq . ( [ 16 ] ) as @xmath54 where @xmath55 is a noise operator with a vanishing mean and @xmath56 is a parameter whose value remains to be fixed . employing the relation @xmath57 along with eqs . ( [ 42 ] ) and ( [ 26 ] ) , we get @xmath58 now comparison of eqs . ( [ 39 ] ) and ( [ 44 ] ) shows that @xmath59 and @xmath60 this result implies that @xmath61 following a similar procedure , one can also easily establish that @xmath62 it is certainly interesting to consider different regimes of the laser operation . we then wish to call the regime of laser operation with more atoms in the upper level than in the lower level above threshold , and the regime of laser operation with equal number of atoms in the upper and lower levels threshold . thus inspection of eq . ( [ 41 ] ) shows that for the laser operating at threshold @xmath63 and for the laser operating above threshold @xmath64 . applying the solution of eq . ( [ 42 ] ) and considering the atoms to be initially in the lower level , we easily find @xmath65 moreover , the expectation value of the solution of eq . ( [ 36 ] ) is expressible as @xmath66 on account of ( [ 49 ] ) and the assumption that the cavity light is initially in a vacuum state , eq . ( [ 50 ] ) reduces to @xmath67 we note from eqs . ( [ 36 ] ) and ( [ 51 ] ) that @xmath3 is a gaussian variable with zero mean . using the steady - state solution of eq . ( [ 36 ] ) , @xmath68 we get @xmath69 so that in view of ( [ 40 ] ) , the mean photon number has at steady state the form @xmath70 we note that for the two - level laser operating well above threshold @xmath71 , eq . ( [ 54 ] ) reduces to @xmath72 and for the same laser operating at threshold @xmath73 , we have @xmath74 furthermore , the variance of the photon number for the cavity light can be written as @xmath75 and recalling that @xmath3 is a gaussian variable with zero mean , we readily get @xmath76 employing eq . ( [ 52 ] ) and taking into account ( [ 28 ] ) , we find @xmath77 thus with the aid of ( [ 58 ] ) along with ( [ 53 ] ) and ( [ 59 ] ) , we arrive at @xmath78 we see from eq . ( [ 40 ] ) that for the two - level laser operating well above threshold @xmath79 so that on account of this and ( [ 23 ] ) , we arrive at @xmath80 therefore , for @xmath81 , the variance of the photon number turns out to be @xmath82 this represents the normally - ordered variance of the photon number for coherent light . on the other hand , for the same laser operating at threshold , we see that the variance of the photon number is @xmath83 which represents the normally - ordered variance of the photon number for chaotic light . we next wish to calculate the quadrature variance for the plus and minus quadrature operators defined by @xmath84 and @xmath85 it can be readily established that @xmath86=2i{\gamma_{c}\over\kappa}(\hat{n}_{a}-\hat{n}_{b}).\ ] ] thus on account of this result , we see that @xmath87 moreover , taking into account ( [ 53 ] ) and ( [ 59 ] ) , one can easily verify that @xmath88 now in view of eq . ( [ 61 ] ) and ( [ 62 ] ) , we see that for the laser operating well above threshold @xmath89 with @xmath90 is given by eq . ( [ 55 ] ) . in addition , for the laser operating at threshold , we have @xmath91 in which @xmath90 is given by ( [ 56 ] ) . this represents the normally - ordered quadrature variance for chaotic light . we define coherent light to be a light mode in which the uncertainties in the two quadratures are equal and satisfy the minimum uncertainty relation . on account of eq . ( [ 68 ] ) , we note that for the laser operating well above threshold , @xmath92 . hence on the basis of this result and eq . ( [ 70 ] ) , we assert that the light generated by the two - level laser operating well above threshold is coherent . we have seen that the quadrature variance for this case takes the relatively small value described by eq . ( [ 70 ] ) . however , this value markedly differs from the corresponding value for an ideal coherent light . this must be due to the fact that the coherent light generated by the two - level laser is represented by operators with vanishing mean . perhaps it is also worth mentioning that whenever the laser coherent light is used as a driving or pump mode , we may replace the operators representing this light mode by the square root of the steady - state mean photon number . such replacement transforms the laser coherent light into an ideal coherent light . it is also interesting to consider the power spectrum of the cavity light . the power spectrum of a single - mode light with central frequency @xmath93 is expressible as @xmath94 upon integrating both sides of eq . ( [ 72 ] ) over @xmath95 , we readily get @xmath96 in which @xmath90 is the steady - state mean photon number . from this result , we observe that @xmath97 is the steady - state mean photon number in the interval between @xmath95 and @xmath98 [ 12 ] . we now proceed to calculate the two - time correlation function that appears in eq . ( [ 72 ] ) for the cavity light . to this end , we realize that the solution of eq . ( [ 36 ] ) can be written as @xmath99 on the other hand , the solution of eq . ( [ 42 ] ) is expressible as @xmath100 so that on introducing this into eq . ( [ 74 ] ) , we have @xmath101 a plot of eq . ( [ 86 ] ) for @xmath102 and @xmath103 . thus on carrying out the first integration , we arrive at @xmath104\nonumber\\ & & + { g\over\sqrt{n}}e^{-\kappa\tau/2}\int_{0}^{\tau}d\tau'\int_{0}^{\tau'}d\tau '' e^{((\kappa-\eta)\tau'+\eta\tau'')/2 } \hat{f}(t+\tau'').\end{aligned}\ ] ] now multiplying both sides on the left by @xmath105 and taking the expectation value of the resulting equation , we have @xmath106\nonumber\\ & & \hspace{-10mm}+{g\over\sqrt{n}}e^{-\kappa\tau/2}\int_{0}^{\tau}d\tau'\int_{0}^{\tau'}d\tau''e^{((\kappa-\eta)\tau'+\eta\tau'')/2 } \langle\hat{b}^{\dagger}(t)\hat{f}(t+\tau'')\rangle.\end{aligned}\ ] ] furthermore , applying the adjoint of eq . ( [ 52 ] ) and taking into account ( [ 26 ] ) , we have @xmath107\nonumber\\ & & \hspace{-10mm}+{\gamma_{c}\over 2n}e^{-\kappa\tau/2}\int_{0}^{\tau}d\tau'\int_{0}^{\tau'}d\tau''e^{((\kappa-\eta)\tau'+\eta\tau'')/2 } \times\langle\hat{m}_{a}^{\dagger}(t)\hat{f}(t+\tau'')\rangle,\end{aligned}\ ] ] so that in view of the fact that @xmath108 there follows @xmath109 finally , on combining ( [ 81 ] ) with ( [ 72 ] ) and carrying out the integration , we readily arrive at @xmath110^{2}}}\bigg ] -{\eta\overline{n}\over{\kappa-\eta}}\bigg[{\kappa/2\pi\over{(\omega-\omega_{0})^{2 } + [ \kappa/2]^{2}}}\bigg].\end{aligned}\ ] ] we realize that the mean photon number in the interval between @xmath111 and @xmath112 is expressible as @xmath113 in which @xmath114 . therefore , upon substituting ( [ 82 ] ) into eq . ( [ 83 ] ) and carrying out the integration , applying the relation @xmath115 we arrive at @xmath116 where @xmath117 is given by @xmath118 from the plot in fig . 1 , we easily find @xmath119 , @xmath120 , @xmath121 . then combination of these results with eq . ( [ 85 ] ) yields @xmath122 , @xmath123 , @xmath124 . we have carried out our analysis by putting the noise operators associated with the vacuum reservoir in normal order and applying the large - time approximation scheme . the procedure of normal ordering the noise operators renders the vacuum reservoir to be a noiseless physical entity . we uphold the viewpoint that the notion of a noiseless vacuum reservoir would turn out to be compatible with observation . based on the definition of the stimulated emission decay constant , we infer that an atom in the upper level and inside a closed cavity emits a photon due to its interaction with the cavity light . we certainly identify this process to be stimulated emission . we have found that for the two - level laser operating well above threshold , the uncertainties in the plus and minus quadratures are equal and satisfy the minimum uncertainty relation . in view of this , we have identified the light generated by the laser operating well above threshold to be coherent . the quadrature variance of the laser coherent light is markedly different from that of an ideal coherent light . this must be due to the fact that the coherent light generated by the two - level laser is represented by operators with vanishing mean . perhaps it is also worth mentioning that whenever the laser coherent light is used as a driving or pump mode , we may replace the operators representing this light mode by the square root of the steady - state mean photon number . such replacement transforms the laser coherent light into an ideal coherent light . moreover , we have seen that the light generated by the two - level laser when operating at threshold is chaotic and a large part of the total mean photon number is confined in a relatively small frequency interval . [ 1 ] d.f . walls and j.g . milburn , quantum optics ( springer - verlag , berlin , 1994).m.o . scully and m.s . zubairy , quantum optics ( cambridge university press , cambridge , 1997).r . loudon , the quantum theory of light ( oxford university press , new york , 2000).m.o . scully and w.e . lamb jr . , phys . 208 , 159 ( 1967).y.k . wang and w.e . lamb jr . , phys . a 8 , 866 ( 1973).k.j . mcneil and d.f . walls , j. phys . a 8 , 104 ( 1975).f.t . arecchi , m. asdente , and a.m. ricca , phys . a 14 , 383 ( 1976).c . benkert and o.m . scully , phys . a 41 , 2756 ( 1990).r . pike and s. sarkar , the quantum theory of radiation ( oxford university press , new york , 1996).l . davidovich , rev . 68 , 127 ( 1997).fesseha kassahun , opt . 284 , 1357 ( 2011).fesseha kassahun , fundamentals of quantum optics ( lulu press inc . , north carolina , 2010 ) .	we analyze the quantum properties of the light generated by a two - level laser in which the two - level atoms available in a closed cavity are pumped to the upper level by means of electron bombardment . we consider the case in which the two - level laser is coupled to a vacuum reservoir via a single - port mirror and seek to carry out our analysis by putting the noise operators associated with the vacuum reservoir in normal order . it is found that the two - level laser generates coherent light when operating well above threshold and chaotic light when operating at threshold . moreover , we have established that a large part of the total mean photon number is confined in a relatively small frequency interval . this is a revised version of a paper published in opt . commun . ( 284 , 1357 , 2011 ) . keywords : photon statistics , quadrature variance , power spectrum
clathrates are materials with oversized polyhedral cages and guest atoms loosely bound inside . type - i clathrates are based on the @xmath5 structure , where @xmath6 stands for the cage atom and @xmath7 is the guest atom ( ba , eu , sr etc . ) . group iv clathrates with cages formed by ga , ge , sn and si atoms are particularly well studied examples . recent investigations of such intermetallic clathrates showed excellent thermoelectric properties@xcite . this is especially interesting for the group iv elements given their great potential in the modern semiconductor industry @xcite . research on the framework and caged atomic motions has also shown anomalous vibrational properties and their connection to the electronic behavior @xcite . understanding the guest atom vibration modes , often called `` rattling '' , has been considered to be one of the most important ways to reveal the essence of these phenomena . many methods including raman scattering @xcite , inelastic neutron scattering @xcite , optical conductivity @xcite , nmr relaxation @xcite and theoretical calculations @xcite have been reported to analyze the rattling atoms in clathrates using different models . during the last few years , ultra - low lattice thermal conductivity ( @xmath8 ) and glasslike thermal behavior have been discovered in particular in type - i ba@xmath2ga@xmath3sn@xmath4 @xcite . the rattling of guest atoms in the larger of its two structural cages has been confirmed to have close connection to those properties @xcite . anharmonic oscillators have also been used as trial models for the rattling phonons to analyze the thermoelectric properties and nmr relaxation behavior of other similar materials @xcite . in this paper , we discuss the nmr relaxation behavior and guest atomic motion of ba@xmath2ga@xmath3sn@xmath4 clathrates . a raman process involving local vibrational modes , which is responsible for the nmr quadrupolar relaxation , will be discussed . simple one and two dimensional anharmonic potentials will be introduced to model the anharmonic local oscillators . the shape and energy levels of these potential wells will be obtained by matching the simulations to the nmr experimental results . the clathrate samples were prepared using the self - flux method , following a technique reported previously @xcite . because of the existence of a type - l / type - viii dimorphism in ba@xmath2ga@xmath3sn@xmath4 , a carefully controlled annealing process is needed during the sample making procedure . for type - i ba@xmath2ga@xmath3sn@xmath4 , the pure elements were mixed together based on the nominal composition followed by an initial arc melting in argon environment . annealing in an evacuated quartz tube at 900 @xmath9c for 50 hours was then applied , followed by a controlled slow cooling to 500 @xmath9c in 80 hours @xcite . x - ray diffraction ( xrd ) measurements were performed on a bruker d8 x - ray powder diffractometer , and wavelength dispersion spectroscopy ( wds ) measurements were done in a cameca sx50 spectrometer . rietveld refinements of the xrd results were performed using expgui @xcite , and the result confirmed the composition and structure of type - i ba@xmath2ga@xmath3sn@xmath4 with no type - viii reflections detected and with 1% ( per mol ) ba(ga , sn)@xmath10 minority phase obtained in the fit @xcite . @xmath1ga and @xmath0ga nmr spin - lattice relaxation rates at the central transition frequency under 8.8 t from 4.2k to 295 k. inset : @xmath1ga nmr lineshapes at temperatures 4.2 k , 77 k and 295 k , scaled proportional to 1/@xmath11 . ] nmr experiments were carried out under external magnetic fields of 8.8 t and 7 t in a temperature range from 4.2 k to 295 k using a pulse spectrometer and a homemade multi - temperature detecting probe . the nuclei measured are @xmath1ga and @xmath0ga with different gyromagnetic ratios @xmath12 and quadrupole moments q , where @xmath13 rad / s g@xmath14 , @xmath15 rad / s g@xmath14 ( ref . ) , @xmath16 @xmath17 , and @xmath18 @xmath17 ( ref . ) . the inset to fig.1 shows the central portion of the @xmath1ga nmr lineshapes at three temperatures under 8.8 t . no significant change in average shift vs temperature has been observed in the lineshape mapping . the small changes at the base of the lineshape vs. temperature are due to unreacted ga metal . the @xmath1ga lineshape is a superposition of two close peaks , which are due to different sites of the framework atoms @xcite . here we consider only the behavior at the center of the resonance , which is due to a superposition of different local configurations . the weighted center shift of this resonance is about 0.033% at 295 k. for comparison , the @xmath0ga lineshape under the same conditions , has a weighted shift of 0.023% . the shift includes magnetic and quadrupole terms , which can be expressed as @xmath19 , where @xmath20 stands for the magnetic shift and @xmath21 is the quadrupole moment of the nucleus . from the observed field - dependence we extracted @xmath22 as the center - of - mass magnetic shift and a negative quadrupole shift . in this case @xmath20 is mainly a knight shift due to conduction electrons , with some contribution due to chemical shifts . note that we did not observe any significant change in @xmath20 , such as those observed in na - si type ii clathrates @xcite , indicating excitations involving sharp electronic features in the system . as reported previously we have also performed a structural analysis using @xmath23-@xmath24 calculations of the low - temperature nmr shifts for this sample , modeling in particular the first - order quadrupole broadening at the base of the lineshape according to the distribution of ga framework occupation . this agreed with the experimental lineshape quite well @xcite . nmr spin - lattice relaxation measurements were performed at the central transition frequency at the center of the lineshape . the relaxation time , @xmath25 , is a fitted value based on a magnetic relaxation mechanism using a standard multi - exponential function for recovery of the central transition @xcite . the quadrupole relaxation process entails a different relaxation function , which however leads only to an overall scaling of the @xmath25 , and does not affect any of the dynamical fitting parameters described below . the signal is also a superposition of different framework sites , however we fitted to a single average @xmath25 as parameter . 1 shows the resulting rates for both @xmath1ga and @xmath0ga under a field of 8.8 t. a clear peak at a temperature around 10 k can be observed for both nuclei . the isotopic ratio , @xmath26 , under 8.8 t is shown in the inset of fig . 2 . according to hyperfine relaxation theory @xcite , if the relaxation mechanism contains only a magnetic part , @xmath25 should be inversely proportional to @xmath27 which gives @xmath28 , while if the quadrupolar relaxation is in control , @xmath25 should be inversely proportional to @xmath29 which gives @xmath30 . the experimental ratio for our sample is consistent over a wide temperature range and is close to the quadrupole moment ratio . thus , the relaxation is mainly controlled by the quadrupole mechanism , indicating that lattice vibrations are the most important contribution . as our experimental data are a mixture of magnetic and quadrupole parts , it is necessary to separate them for further investigations . the corresponding relationships are @xcite , @xmath31 where @xmath25 is the overall experimental relaxation time while @xmath32 and @xmath33 represent the magnetic and quadrupole parts . according to eq . ( 1 ) and ( 2 ) , the relaxation rates were separated into two contributions as shown in fig . , the result confirms the dominant role of the quadrupole relaxation rate . at higher temperatures , as values of @xmath34 do not change much , therefore the sample appears to follow a korringa - like relation @xcite , which would normally indicate the influence of metallic electrons if @xmath35 were magnetic . however a recent model @xcite for relaxation dominated by anharmonic localized vibrations indicates such behavior as a high - temperature limit , along with a low - temperature peak much as observed here . our later simulation based on this anharmonic model will be compared with the quadrupolar relaxation rates we have separated . separated @xmath25 relaxation rates for @xmath1ga and @xmath0ga : @xmath0ga - quadrupole ( diamonds , as labeled ) , @xmath1ga - quadrupole ( circles , as labeled ) , @xmath1ga - magnetic ( squares ) , @xmath0ga - magnetic ( triangles ) . inset : isotopic ratio of overall rates under 8.8 t , with limits for pure quadrupolar / magnetic relaxation indicated . ] from refinements of x - ray diffraction spectra for ba@xmath2ga@xmath3sn@xmath4 , the guest ba(2 ) atom has location probability concentrated near four equivalent off - center positions with off - center dynamic displacements around 0.4 @xcite . our first principles calculations @xcite also gave similar values for the static displacement of the ba atom due to cage asymmetry . a 1-d double well potential was also introduced by dahm and ueda to analyze this kind of problem , and has shown good agreement for the pyrochlore case @xcite . to model our data , therefore we use the hamiltonian , @xmath36 where @xmath37 , @xmath38 , and @xmath39 are the mass , momentum and spatial coordinate of the guest atom ba @xcite . an effective localized phonon frequency , @xmath40 , and thermal average of @xmath41 were introduced in a self - consistent quasiharmonic approximation giving @xmath42 , where @xmath43 and the relationship between @xmath40 and @xmath11 is given by , @xmath44 where @xmath45 , and @xmath46 is a dimensionless anharmonicity factor . as the relaxation is dominated by the quadrupole term , a two - phonon raman process can be used to describe the nmr relaxation . this can be expressed @xcite as @xmath47n(\omega)\end{aligned}\ ] ] with @xmath48 where @xmath49 is the second spatial derivative of the electric field gradient , @xmath50 is the nuclear lamor frequency , @xmath51 is the phonon spectral function , @xmath52 is the bose function , @xmath53 is the retarded phonon propagator , @xmath54 is a phonon damping rate and @xmath55 is the renormalized phonon frequency determined by the phonon self - energy , @xmath56re@xmath57 . here , we assume the real part of the phonon self energy , re@xmath57 , to be temperature independent as assumed in [ 17 ] . by carefully choosing parameters , we obtained a good fit to our data as shown in fig . the matching results clearly show that the spin - lattice relaxation mechanism can be explained by the rattling phonon model . quadrupole nmr relaxation rate for @xmath1ga compared with the fitted 1-d anharmonic model ( main plot , solid curve ) and simplified 2-d model ( inset , solid curve ) . ] the corresponding values for the parameters are @xmath58 k , @xmath59 , @xmath60 k and @xmath61 k. the potential well is given by the calculated expression , @xmath62 where @xmath63 is in j with @xmath39 given in meters . also from equations ( 4 ) and ( 5 ) , when @xmath64 k , @xmath65 @xmath66 11 thz and @xmath67 @xmath66 0.12 . since in 2d @xmath68 , this corresponds to a rms guest atom displacement of 0.17 , which is not far from the values reported previously @xcite . solving the schrdinger equation numerically , the energy levels of this double well potential can be calculated as shown in fig . the energy difference between the lowest two states , @xmath69 @xmath66 30 k , is much smaller than those for higher energy levels , which agrees with recent reported results from other methods @xcite , but with a larger @xmath69 . fitted 1-d double well potential and its energy levels . ] this model can also be extended to a simplified 2-d potential by using @xmath70 in equation ( 4 ) . we correspondingly modified the relationship in equation ( 5 ) . then , following the same procedure , we obtain a fitting similar to the 1-d model . the result is shown in the inset of fig . 3 , with the fitted values @xmath58 k , @xmath71 , @xmath60 k and @xmath61 k. the corresponding potential is @xmath72 , where @xmath73 is in j with @xmath74 given in meters . the average displacement is still 0.17 , which indicates a consistency of the model compared with the 1-d case . compared to previous ga nmr results for sr@xmath2ga@xmath3ge@xmath4 , also identified to behave as an anharmonic rattler system @xcite , it seems initially surprising that the @xmath75 in sr@xmath2ga@xmath3ge@xmath4 does not show a similar phonon - dominated behavior but instead follows a korringa law quite closely for several decades of temperature @xcite . however , a previous report for sr - ge clathrates used density functional theory to extract potential well parameters for sr in the large cage @xcite , giving a 2d anharmonic potential much like the model used here . the resulting potential has a very small quadratic term , but also a much smaller anharmonicity parameter @xmath76 than found here , and inserting the calculated parameters into the relaxation theory described above yields a smaller quadrupole contribution to @xmath75 , rising slowly with temperature without exhibiting a peak as in fig . 3 . also in the analysis reported for the elastic response of eu- and sr - filled ge clathrates @xcite a four - well potential was used to model the vibrational response for eu , but for sr a harmonic einstein oscillator model provided satisfactory agreement . thus it is consistent that the @xmath75 in sr@xmath2ga@xmath3ge@xmath4 is dominated by interactions with the charge carriers , while the quadrupole - dominated peak observed in ba@xmath77ga@xmath3sn@xmath4 is indicative of the much larger anharmonicity for rattler atoms . in our measurements , we prepared a second sample of type - i ba@xmath77ga@xmath3sn@xmath4 in the same way , whose @xmath78 exhibits a low - temperature maximum that is nearly identical to that of fig . 1 . in fitting to the model oscillator potential , the position of the @xmath78 peak is particularly sensitive to @xmath79 , which is close to the spacing of the two lowest levels in fig . the ability to model this behavior with a single set of parameters attests to the lack of irregularity of the cage potential , despite the presence of quasi - random framework substitution . this is apparently due to the sn - based cage size , providing space for the relatively unconstrained motion of the ba(2 ) atoms without allowing for a permanent distortion @xcite . thus , type - i ba@xmath77ga@xmath3sn@xmath4 can be viewed as possessing a more or less uniform array of strongly anharmonic local oscillators . the nmr relaxation times are particularly sensitive to the low - frequency anharmonic motion of these atoms , and thus provide an excellent probe for this behavior . in conclusion , type - i ba@xmath77ga@xmath3sn@xmath4 nmr lineshapes and spin - lattice relaxation rates indicated the presence of a strong quadrupole relaxation mechanism . analysis showed this behavior to be due to a strongly anharmonic rattler - type motion of the caged ba atoms . fitting using a 1-d double well potential with strong anharmonicity showed good agreement with the experimental data , which offers a good explanation for the rattling behavior and the relaxation mechanism .	atomic motion of guest atoms inside semiconducting clathrate cages is considered as an important source for the glasslike thermal behavior.@xmath0ga and @xmath1ga nuclear magnetic resonance ( nmr ) studies on type - i ba@xmath2ga@xmath3sn@xmath4 show a clear low temperature relaxation peak attributed to the influence of ba rattling dynamics on the framework - atom resonance , with a quadrupolar relaxation mechanism as the leading contribution . the data are analyzed using a two - phonon raman process , according to a recent theory involving localized anharmonic oscillators . excellent agreement is obtained using this model , with the parameters corresponding to a uniform array of localized oscillators with very large anharmonicity .
blazars represent a small subset of the most enigmatic class of radio - loud active galactic nuclei ( agn ) , exhibiting strong variability at all wavelengths of the whole electromagnetic ( em ) spectrum , strong polarization from radio to optical wavelengths , and usually core dominated radio structures . the radiation of blazars at all wavelengths is predominantly nonthermal . this class includes bl lacertae ( bl lac ) objects and flat - spectrum radio quasars ( fsrqs ) . bl lacs show largely featureless optical continuum . in a unified model of the radio - loud agn based on the angle between the line of sight and the emitted jet from the source , blazars jet make angle of @xmath2 10@xmath3 from the line of sight ( urry & padovani 1995 ) . the radiation emitted by the plasma , with bulk relativistic motion in the jet oriented at small viewing angles , is affected by relativistic beaming , which in turn implies a shortening of time scales by a factor @xmath4 , where @xmath5 is the doppler factor . from observations of blazars , it is known that they vary on the diverse time scales . variability time scales of blazars can be broadly divided into 3 classes viz . intra - day variability ( idv ) or micro - variability , short term outbursts and long term trends . significant variations in flux of a few tenths of magnitude over the course of a day or less is often known as idv ( wagner & witzel 1995 ) . short term outburst and long term trends can have time scales range from few weeks to several months and several months to years , respectively . in last about two decades , variability of blazars in radio to optical bands on diverse time scales have been reported in a large number of papers ( e.g. miller et al . 1989 ; courvoisier , et al . 1995 ; heidt & wagner 1996 ; takalo et al . 1996 ; sillanp@xmath6 et al . 1996a , 1996b ; bai , et al . 1998 , 1999 ; fan et al . 1998 , 2002 , 2007 ; xie , et al . 2002a ; gupta et al . 2004 ; ciprini et al . 2003 , 2007 and references therein ) . we selected a sample of a dozen blazars which are prominent candidates for simultaneous multi - wavelength observing campaigns in their outburst phase . the motivation of the present work was to observe these blazars in search for idv , short term variability and also find out if there is any one being in the outburst state . blazar emission mechanism in the outburst state and detected idv is strongly supported by the jet based models of radio - loud agn . in general , blazar emission in the outburst state is nonthermal doppler boosted emission from jets ( blandford & rees 1978 ; marscher & gear 1985 ; marscher et al . 1992 , hughes et al . there are other models of agn that can explain the idv in any type of agn are optical flares , disturbances or hot spots on the accretion disk surrounding the black hole of the agn ( mangalam & wiita 1993 and references therein ) . models based on the instabilities on accretion disc are mainly supported blazars idv when the blazar is in the low - state . when a blazar is in the low - state , any contribution from the jets if at all present , is very weak . recently , it is noticed that , in the low luminosity agn , accretion disk is radiatively inefficient ( chiaberge et al . 2006 ; capetti et al . 2007 ) . so , there will be an alternative way to explain the idv in the low - state of blazars , in which a weak jet emission will be responsible for the idv . with this motivation we recently carried out optical photometric observations of the nine blazars : s2 0109@xmath1224 , ao 0235@xmath1164 , s5 0716@xmath1714 , pks 0735@xmath1178 , oj 287 , on 231 , 3c 279 and 1es 2344@xmath1514 in r passband and 3c 454.3 in v and r passbands . the paper is arranged as follows : section 2 describes observations and data analysis method , in section 3 we mentioned our results , discussion and conclusion of the present work is reported in section 4 . the photometric observations of the blazars s2 0109@xmath1224 , ao 0235@xmath1164 , s5 0716@xmath1714 , pks 0735@xmath1178 , oj 287 , on 231 , 3c 279 and 1es 2344@xmath1514 in johnson r passband and 3c 454.3 in johnson v and r passbands were carried out using the ccd detector ( 1024 pixels @xmath7 1024 pixels ) mounted at f/13.3 cassegrain focus of the 1.02 meter rc ( ritchey chertien ) optical telescope at yunnan observatory , kunming , china . these blazars were observed during october 27 , 2006 to march 20 , 2007 in 13 observing nights . each pixel of the ccd detector projected on the sky corresponds to 0.38 arcsec in both the dimensions . the entire ccd chip covers @xmath2 6.5 @xmath7 6.5 arcmin@xmath8 of the sky . the read out noise and gain of the ccd were 3.9 electrons and 4.0 electrons / adu respectively . throughout the observing run , the typical seeing was @xmath2 1.8 arcsec ranging from 1.5 to 2.0 arcsec . several bias frames were taken intermittently in each observing night and twilight sky flats were taken in v and r passbands . detail photometric observation log is given in the table 1 . for each observing night we generated a master bias by taking median of all bias frames in the night . master bias frame is subtracted from all the target image frames and flat field image frames of the night . then master v and r passband flat field image frames were generated for the night by taking median of all flat field images in v and r passband respectively . image processing or pre processing ( bias subtraction , flat - fielding and cosmic rays removal ) were done using standard routines in iraf ( image reduction and analysis facility ) software . photometric reduction or processing of the data were performed by aperture photometric technique using daophot ii ( dominian astronomical observatory photometry ) software ( stetson 1987 ) . aperture photometry was carried out with concentric apertures of radii 8 , 11 , 15 and 20 pixels . the data reduced with different aperture radii were found to be in good agreement . however , it was noticed that the best signal to noise ratio was obtained with the aperture radius of 15 pixels ( 5.7 arcsec slightly more than 3 @xmath7 of typical fwhm ) . in our observations of all nine blazars , local standard stars were also always present in the observed image frames . we always observed three or more local standard stars in our every target blazar field . the magnitude of standard stars which we observed in the field of our observed blazars are given in table 2 . these standard stars were used to check the non variable characteristic of standard stars and then transformed the instrumental magnitude of the blazars to the standard magnitude . finally we selected two standard stars from each blazar field and plotted their differential instrumental magnitude light curve by adding arbitrary offset for clarity in the plot with the light curves of blazars ( calibrated magnitude ) . photometric data of our blazars observations are provided in table 4224 , s5 0716@xmath1714 , pks 0735@xmath1178 , on 231 and 1es 2344@xmath1514 . data of the blazars ao 0235@xmath1164 , oj 287 , 3c 279 and 3c 454.3 will be stored in the webt ( http://www.to.astro.it/blazars/webt ) archive . for questions regarding their avalability , please contact the webt president massimo villata ( [email protected] ) ] . using aperture photometry of blazar and few standard stars in the blazar field , we determined the differential instrumental magnitude of blazar @xmath0 standard star a , blazar @xmath0 standard star b and standard star a @xmath0 standard star b. finally we selected star a or b to do calibration of the blazar data . we determined observational scatter from blazar @xmath0 standard star a @xmath9 ( bl @xmath0 star a ) , blazar @xmath0 standard star b @xmath9 ( bl @xmath0 star b ) and standard star a @xmath0 standard star b @xmath9 ( star a @xmath0 star b ) . the variability of target blazar is investigated by using variability parameter c , introduced by romero et al . ( 1999 ) . the variability parameter c is expressed as the average of c@xmath10 and c@xmath11 @xmath12 if c @xmath13 2.57 , the confidence limit of variability is 99% . the value of c by using the two standard stars for all the nine blazars for different observing nights are reported in table 3 . intra - day variability amplitude defined as ( heidt & wagner 1996 ) @xmath14 where a@xmath15 and a@xmath16 are the maximum and minimum magnitude in the calibrated light curve of the blazar . @xmath9 is the averaged measurement error of the blazar light curve . * s2 0109@xmath1224 * s2 0109@xmath1224 is a bl lac object which has featureless optical spectrum . the host galaxy of the source was unresolved in ntt observations ( falomo 1996 ) and also in k passband observations using ukirt ( wright et al . 1998 ) . falomo ( 1996 ) suggested that the redshift of the blazar z @xmath17 0.4 based on its optical appearance , and assuming [email protected] for it , a value similar to that characterizing some galaxies located at few arcseconds north - east of the object . the blazar has shown optical flux variability on diverse time scales ( ciprini et al . 2003 ) . the degree of optical polarization varied between 10% to 30% ( takalo 1991 ; valtaoja et al . 1993 ) . we observed s2 0109@xmath1224 on january 11 , 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star d - star c1 ) with arbitrary offset are displayed in the bottom panel of fig . 6 . from the figure it is clear that the source has not shown idv on one night of our observations . we also performed the idv detection test described in the variability detection criterion section and found that the value of c is 1.07 , which also confirms the non detection of idv in our observations . ciprini et al . ( 2003 ) have done the multi band optical monitoring of the source for seven years ( 1994 @xmath0 2002 ) and also plotted its b passband historical light curve ( 1906 @xmath0 2002 ) . 1994 @xmath0 2002 light curves of the source show the brightest and faintest states of the source in r passband @xmath2 13.2 and @xmath2 16.5 magnitudes . in the present observations , we observed the source @xmath19 1.0 magnitude fainter than its brightest state and @xmath19 2.3 magnitude brighter than its faintest state . therefore , it appears likely that we have observed the source in its pre / post outburst state . * ao 0235@xmath1164 * ao 0235@xmath1164 at ( redshift z = 0.94 ) is classified as a bl lac object by spinrad & smith ( 1975 ) . this blazar has been extensively observed in the recent past from radio to x - ray bands and has shown variations in all those bands on diverse time scales ( ghosh & soundararajaperumal 1995 ; heidt & wagner 1996 ; romero et al . 1997 , 2000 ; webb et al . 2000 ; raiteri et al . 2001 ; padovani et al . 2004 ; and references therein ) . it is one of the blazars which has shown very high polarization p@xmath20 44% and p@xmath21 36.4% ( impey et al . 1982 ; stickel et al . 1993 ; fan & lin 1999 ) . using 25 years ( 1975 - 2000 ) of radio and optical data , raiteri et al . ( 2001 ) predicated that the blazar should show a possible correlated periodic radio and optical outburst on the time scale of [email protected] years that was expected in february - march 2004 . an intense webt ( whole earth blazar telescope ) multi - wavelength observing campaign during 2003@xmath02005 was organized and the predicted out burst could not be detected ( raiteri et al . 2005 ; 2006a ; 2006b ) . raiteri et al . ( 2006b ) did analysis of the long term optical light curves and reported that the optical outburst may have a longer time scale @xmath2 8 years . we observed ao 0235@xmath1164 on january 09 and 10 , and march 19 , 2007 in r passband . the light curve of the blazar and differential instrumental magnitude ( star 1 - star 3 ) with an arbitrary offset are displayed in fig . 1 . from the figure it is seen that the source has shown idv on january 10 and march 19 , 2007 . we also performed the idv detection test and found that the value of c for january 09 , january 10 and march 19 are 2.00 , 4.14 and 3.73 respectively which confirms that the source has shown idv in the observations of january 10 and march 19 . the variability amplitude on january 10 and march 19 are 13.7% and 9.5% respectively . day to day variations are distinctly visible in figs . 1 and 7 . the filled circles show the day average magnitude of ao 0235@xmath1164 and open circles show the day average of differential magnitude of two standard stars ( star 1 and star 3 ) offset by the same arbitrary constant on all the three observing nights . fig . 7 also shows the variability trend and if the source has reached even brighter state than january 10 , it will be before march 19 , 2007 . from variability trend from observations show the source was still in outburst phase on march 19 . most interestingly , if we compare our figs . 1 and 7 to the long term light curve ( 1990 - 2005 ) of ( fig . 4 top panel of raiteri et al . 2006b ) , we found that we have possibly detected the brightest state ( outburst ) of the source after around mid 1999 ( previous optical outburst state ) . it is interesting that the interval between the last optical outburst in 1999 and the present detected possible outburst state is about 7.5 years , which is consistent with the prediction made by ( raiteri et al . our observations of the source was also the part of the late 2006 to march 2007 multi - wavelength campaign by webt . the multi - wavelength campaign results show the source was in the outburst state in this period ( raiteri et al . 2007b ) . * s5 0716@xmath1714 * s5 0716@xmath1714 is one of the brightest bl lac objects . the optical continuum is so featureless that all attempts made to determine its redshift have failed . the non detection of host galaxy first set a lower limit of z @xmath13 0.3 ( wagner et al . 1996 ) and very recently z @xmath13 0.52 ( sbarufatti et al . 2005 ) . this object is among one of the most well studied bl lac objects for variability studies in the whole em spectrum on diverse time scales ( e.g. wagner et al . 1990 ; heidt & wagner 1996 ; ghisellini et al . 1997 ; villata et al . 2000 ; raiteri et al . 2003 ; pian et al . 2005 ; bach et al . 2005 , 2006 ; nesci et al . 2005 ; ostorero et al . 2006 ; montagni et al . 2006 ; foschini et al . 2006 ; wu et al . 2007 ; fan et al . 2007 and references therein ) . wagner & witzel ( 1995 ) reported that the duty cycle of the source is 1 which implies that the source is almost always in the active state . first optical polarization study of this source was done by takalo et al . they found that the source shows high polarization @xmath2 20% and intra - day polarization variability upto 3.5% . fan et al . ( 1997a ) reported even much higher optical polarization up to 29% in the source . since 1994 , people started observing this source intensely in optical bands . there are 4 major optical outbursts reported in the source , these are : at the beginning of 1995 , in late 1997 , in the fall of 2001 and in march 2004 ( raiteri et al . 2003 ; foschini et al . these 4 outbursts give a possible period of long term variability of @xmath2 [email protected] years . we observed s5 0716@xmath1714 on january 10 , 12 ; february 23 ; march 19 and 20 ; 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star 5 - star 3 ) with different arbitrary offset are displayed in different panels in fig . 2 . from fig . 2 , it is clear that the source has shown idv on 2 nights ( january 12 and february 23 , 2007 ) out of 5 nights of observations . we also performed the idv detection test and found that the value of c for january 10 , january 12 , february 23 , march 19 and march 20 , 2007 are 1.80 , 2.86 , 3.88 , 0.95 and 1.56 respectively , which also confirms that the source has shown idv on january 12 and february 23 . the variability amplitude on january 12 and february 23 , 2007 are 6.3% and 5.2% respectively . day to day variations are also distinctly visible in the figs . 2 and 7 . in fig . 7 , filled circles show the day average magnitude of s5 0716@xmath1714 and open circles show the day average differential magnitude of star 5 and star 3 with the same arbitrary constant added for all 5 observing nights . 7 shows clearly the day to day variations in the source and also shows the variability trend in the source . the source was at r @xmath23 12.58 mag on january 12 , 2007 which is comparable to the earlier outburst magnitude in r passband ( @xmath2 12.75 mag in beginning of 1995 ; 12.6 mag in late 1997 ; 12.55 mag in the fall of 2001 ) ( raiteri et al . so , we have possibly detected the brightest state of the outburst phase in the source on january 12 , 2007 . if the source has reached even brighter state than january 12 , it will be before february 23 , 2007 . if we compare the detected possible outburst in the present observations , it gives a period of the present outburst is after @xmath2 2.8 years from recent previous outburst in march 2004 . it is consistent from the possible long term period of the source [email protected] years . * pks 0735@xmath1178 * pks 0735@xmath1178 is an optically bright , highly variable source and classified as a bl lac object ( carswell et al . 1974 ) . the host galaxy of the source is unresolved in optical imaging . in the optical spectrum of pks 0735@xmath1178 , carswell et al . ( 1974 ) did not find any emission line , but they noticed two sharp absorption features at 3981@xmath24 and 3991@xmath24 which they identified with the mg ii @xmath252798 doublet at redshift z @xmath13 0.424 which sets a lower limit of its redshift . very recently imaging redshift z = 0.424 is reported for the source using hst snapshot image ( sbarufatti et al . the source has been extensively studied in radio and optical bands in search for variability . in search for long term variability , in about a century long optical data using jurkevich method , periodic components of [email protected] years were found ( fan et al . 1997b ; qian & tao 2004 ) . in a recent paper , ciprini et al . ( 2007 ) has reported several short possible periods . there are several attempts to search for optical idv and day to day variation in the source in last about one and half decades . optical idv upto 0.5 mag has been reported in the source on some occasions ( e.g. xie et al . 1992 ; massaro et al . 1995 ; fan et al . 1997b ; zhang et al . pks 0735@xmath1178 has shown high degree of optical and near - infrared polarizations which varies from 1% to 30% ( e.g mead et al . 1990 ; takalo 1991 ; takalo et al . 1992 ; valtaoja et al . 1991 , 1993 ; tommasi et al . 2001 ) . we observed pks 0735@xmath1178 on january 11 , 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star c4 - star c1 ) with arbitrary offset are displayed in the second panel from the bottom in fig . 6 . from the figure it is clear that the source has shown idv on one night of our observations . we also performed the idv detection test and found that the value of c is 4.00 , which also confirms that the source has shown idv on one night of our observations . the observed variability amplitude is 4.9% . ciprini et al . ( 2007 ) have done the multi band optical monitoring of the source for more than ten years ( 1993 @xmath0 2004 ) and also plotted its b passband historical light curve ( 1906 @xmath0 2004 ) . 1993 @xmath0 2004 light curve of the source shows the brightest and faintest states of the source in r passband @xmath2 14.0 and @xmath2 17.0 magnitudes . in the present observations , we observed the source @xmath19 1.7 magnitude fainter than the brightest state and @xmath19 1.3 magnitude brighter than its faintest state . therefore , it appears likely that we have observed the source in pre / post outburst state . * oj 287 * blazar oj 287 at redshift z = 0.306 is one of the most extensively observed extragalactic object . in the long term optical light curve , it exhibits a @xmath2 12 year periodic outburst with a double - peaked maxima in its flux variations ( takalo et al . 1994 ; pursimo et al . sillanp@xmath6 et al . ( 1988 ) explained the periodic outburst in the blazar with a binary black hole model . in radio to optical bands , the source has been studied for flux and polarization variability on diverse time scales by several groups ( e.g. jorstad et al . 2007 ; hovatta et al . 2007 ; zhang et al . 2007 ; bach et al . 2007 ; wu et al . 2006 ; fan et al . 2006 and references therein ) . takalo et al . ( 1994 ) have reviewed its observational properties from radio to x - ray bands . we observed oj 287 on october 29 , november 17 , november 18 , 2006 ; and january 10 , 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star 10 - star 11 ) with same arbitrary offset are displayed in fig . 3 . from the figure it is clear that the source has not shown idv on any of the four nights of our observations . we also performed the idv detection test and found that the value of c for october 29 , november 17 , november 18 , 2006 ; and january 10 , 2007 are 0.72 , 1.00 , 0.73 and 1.07 respectively , which also confirms that the source has not shown idv on any of the 4 nights of our observations . in fig . 7 , filled circles show the day average magnitude of oj 287 and open circles show the differential magnitude of star 10 and star 11 with the same arbitrary offset on all the 4 nights of observations . from fig . 7 , day to day variation in the source is clearly visible . pursimo et al . ( 2000 ) have done the multi band optical monitoring of the source for about five years ( 1993 @xmath0 1998 ) and also plotted its v passband historical light curve for about a century long observations . 1993 @xmath0 1998 light curve of the source shows the brightest and faintest states of the source in the r passband @xmath2 13.5 and @xmath2 16.5 magnitudes . in the present observations , we observed the source @xmath2 14.6 @xmath0 15.0 mag in r passaband which is @xmath19 1.0 magnitude fainter than the brightest state and @xmath19 1.5 magnitude brighter than its faintest state . therefore , it appears likely that we have observed the source in pre / post outburst state . our observations of the source were also the part of november 2006 multi - wavelength campaign by webt . * on 231 * on 231 at redshift z = 0.102 is classified as a bl lac object ( browne 1971 ) . historical b passband light curve of the source ( 1935 - 1997 ) was plotted by tosti et al . there are not many attempts to study the variability behavior of the source . source has shown optical flux variations on diverse time scales ranging from few hours to several years ( smith & nair 1995 ; xie et al . 1992 ) . we observed on 231 on january 11 , 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star d - star a ) with arbitrary offset are displayed in second panel from top in fig . 6 . from the figure it is clear that the source has not shown idv on one night of our observations . we also performed the idv detection test and found that the value of c is 1.06 , which also confirms that the source has not shown idv on one night of our observations . tosti et al . ( 1998 ) have done the multi band optical monitoring of the source for three years ( 1994 @xmath0 1997 ) in its great outburst state . 1994 @xmath0 1997 light curve of the source shows the brightest and faintest states in r passband @xmath19 13.5 and @xmath19 15.0 magnitudes . in the present observations , we observed the source at r @xmath2 15.0 magnitude . therefore , it appears likely that we have observed the source in the low - state which is comparable to the faintest state observed by tosti et al . ( 1998 ) . * 3c 279 * 3c 279 is an optical violent variable quasar and has shown large optical variation of @xmath26b @xmath17 6.7 magnitude from archival plates of harvard collection ( eachus & liller 1975 ) . the most rapid optical variation @xmath26v = 1.17 magnitude was reported in 40 min on may 22 , 1996 ( xie et al . 1999 ) . on another occasion the source brightened by @xmath2 2.0 magnitude within 24 hours ( webb et al . 1990 ) . a large amplitude short term optical variability of @xmath26 r = 0.91 magnitude in 49 days ( april to june 2001 ) has been reported earlier by ( xie et al . it has shown very high optical polarization p = [email protected] % ( scarpa & falomo 1997 ) . the source has also shown optical polarization variability 17% to 8% in three nights and intra - day polarization is also detected ( andruchow et al . 2003 ) . using 27 years of near - infrared k band observations , fan ( 1999 ) has reported that the source has shown 4.59 magnitude variations and a strong period of [email protected] years . we observed 3c 279 on january 12 and february 23 , 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star 8 - star 1 ) with different arbitrary offset are displayed in fig . the top and bottom panels of fig . 4 show the observations on january 12 and february 23 , 2007 respectively . from the figure it is clear that the source has not shown idv on both the nights ( january 12 , and february 23 , 2007 ) of observations . we also performed the idv detection test and found that the value of c for january 12 and february 23 , 2007 are 1.0 and 1.0 respectively , which also confirms that the source has not shown idv on both the nights of our observations . day to day variations are distinctly visible in the fig . 4 . in fig . 7 , filled circles show the day average magnitude of 3c 279 and open circles show the differential magnitude of star 8 and star 1 with same arbitrary offset on both the nights of observations . fig . 7 also shows the variability trend and if the source has reached even brighter state than january 12 , it was before february 23 , 2007 . so , we have possibly observed the source in the outburst phase . we have detected large amplitude @xmath27r = 1.5 magnitude variation in the source in the time span of 42 days which is much larger variation with previous @xmath26r = 0.91 magnitude in 49 days ( april to june 2001 ) reported by ( xie et al . our observations of the source was also the part of april 2006 @xmath0 february 2007 multi - wavelength observing campaign by webt in the outburst state of the source . * 3c 454.3 * the blazar 3c 454.3 is a flat spectrum radio quasar at redshift z = 0.859 . long term optical and radio observations ( 1966 @xmath0 2005 ) were displayed by villata et al . they detected an unprecedented optical outburst in ( 2004 @xmath0 2005 ) which lasted about a year in their multi wavelength observing campaign by webt . to further see the behavior of the source , two more multi - wavelength observing campaigns were organized by webt ( oct . 2005 @xmath0 june 2007 ; july 2007 - ongoing ) . our observations of the source was also the part of oct . 2005 @xmath0 june 2007 multi - wavelength observing campaign by webt . we observed 3c 454.3 during oct . 27 - 30 , 2006 in v and r passbands and january 9 - 10 , 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star 3 - star 4 ) with same arbitrary offset for all 6 nights observations in r passband are plotted in the panels in the bottom row of the figure 5 . v passband light curves for all 4 nights observations with same arbitrary offset are plotted in the panels in the top row of the figure 5 . we performed the idv detection test and found that the value of c in r passband 2.46 , 2.64 , 2.42 , 2.08 , 3.83 and 1.77 for observations on october 27 , october 28 , october 29 , october 30 , 2006 ; january 09 and january 10 , 2007 respectively . the value of c shows that source has shown idv in r passband on 2 nights . the variability amplitude on october 28 , 2006 and january 09 2007 were 5.1% and 17.1% respectively . the value of c in v passband 2.05 , 1.79 , 3.00 and 0.50 were obtained for observations on october 27 , october 28 , october 29 and october 30 , 2006 respectively . the value of c shows that source has shown idv in v passband on one night . the variability amplitude on october 29 , 2006 was 11.6% . the present observations of the source was in its post outburst phase ( raiteri et al . 2007a ) . * 1es 2344 + 514 * 1es 2344 + 514 is classified as a bl lac object based on lack of optical emission lines and caii ` break strength ' less than 25% ( perlman et al . 1996 ) . it was first detected in tev @xmath28rays ( @xmath13 350 gev ) by the whipple group ( catanese et al . 1998 ) and discovered among one of the first five tev detected blazars . first attempt of its optical idv and short term variability was made by xie et al . ( 2002a ) in 6 observing nights during january 2000 to january 2001 . they detect small amplitude idv and short term variability in their observations . we observed 1es 2344 + 514 on january 12 , 2007 in r passband . the light curves of the blazar and differential instrumental magnitude ( star c2 - star c1 ) with arbitrary offset are displayed in the top panel in fig . 6 . from the figure it is clear that the source has not shown idv on one night of our observations . we also performed the idv detection test and found that the value of c is 1.44 , which also confirms the non detection of idv in source in one night of our observations . there is not much optical data available for the source , so , it is difficult to know in which state was the source during our observations . this source will remain one of our prime target to search for optical variability on diverse time scales in near future observations . from our observations , our data favor the hypothesis that three blazars : ao 0235@xmath1164 , s5 0716@xmath1714 and 3c 279 were in the outburst state ; one blazar : 3c 454.3 was in the post outburst state ; three blazars : s2 0109@xmath1224 , pks 0735@xmath1178 and oj 287 were in the pre / post outburst state ; one blazar : on 231 was in the low - state ; and the state of one blazar : 1es 2344@xmath1514 was not known because there is not much optical data available for the blazar . the available statistics is not sufficient to draw strong conclusions . detected outburst phase in ao 0235@xmath1164 possibly confirms the @xmath2 8 years period in optical bands which was predicted by raiteri et al . ( 2006b ) . ao 0235@xmath1164 has shown idv on two night out of 3 nights of observations . one night in which we did not detected any idv , observations duration were very short and no idv could be detected ( if it has really occurred on that night ) . idv can be broadly divided into intrinsic and extrinsic classes . extrinsic idv is caused by refractive interstellar scintillation and only relevant in the low - frequency radio observations . another extrinsic cause of idv is gravitational micro - lensing . optical imaging and spectroscopic observations of the source have revealed foreground absorbing systems at z = 0.524 and z = 0.851 ( cohen et al . 1987 ; nilsson et al . the flux of the source is contaminated and absorbed by foreground absorbing systems , the stars of which can act as gravitational micro - lenses . the reported optical idv in ao 0235@xmath1164 is well suited by the gravitational micro - lensing . day to day variations are also noticed in our observations in the blazar . out of 5 nights observations of s5 0716@xmath1714 , idv is detected in 2 nights of observations . day to day variations are distinctly visible in all 5 nights of observations which confirms the variability detection on short time scale . detected outburst possibly confirm the long term periodic variation time scale of [email protected] years . idv reported in s5 0716@xmath1714 is intrinsic one . the dominantly fundamental model for intrinsic variability is shocks propagating through the jet ( blandford & k@xmath29nigl 1979 ; marscher & gear 1985 ) . models to explain intrinsic idv are based on a relativistic shock propagating down a jet and interacting with irregularities in the flow ( marscher et al . 1992 ) or relativistic shocks changing directions in the jet ( nesci et al . another related model involves non - axisymmetric bubbles carried outward in relativistic magnetized jets ( camenzind & krockenberger 1992 ) . the idv reported here in s5 0716@xmath1714 can be explained by any one of these models . out of 2 nights observations of 3c 279 , idv is not detected . day to day variations are shown in our two nights of observations of the source which confirms the variability detection on short time scale . we detected short term variation of @xmath2 1.5 magnitude in r passband in 42 days which is much larger than the earlier detected short term variation of 0.91 magnitude in 49 days ( xie et al . when we observed 3c 279 , the source was in outburst / high - state . in the densely sampled light curves in search for idv in the source at outburst / high - state has not shown any idv . no idv detection in 3c 279 shows the jet emission do not have any irregularities in the jet flow , relativistic shocks directions have also not changed from the line of sight , no non - axisymmetric bubbles were carried outward in the relativistic magnetized jets . two other blazars : pks 0735@xmath1178 and 3c 454.3 have shown idv in our observations . 3c 454.3 was in the post outburst phase and pks 0735@xmath1178 was in the pre / post outburst phase during our observations . the idv detected in these two blazars is well suited for shock - in - jet models described above . in 3c 454.3 , we have also noticed the short term variability . no idv was detected in s2 0109@xmath1224 , oj 287 , on 231 and 1es 2344@xmath1514 in our observations . in oj 287 , we have noticed the short term variability . we thankfully acknowledge the critical comments by the referee which helped us to improve the paper significantly . we are thankful to prof . w. yuan for careful reading of the manuscript . acg and jmb gratefully acknowledge the financial support from the national natural science foundation of china ( grant nos . nsfc 10533050 and nsfc 10573030 ) . jhf s work is supported by the national natural science foundation of china ( grant nos . nsfc 10573005 and nsfc 10633010 ) . andruchow , i. , cellone , s. a. , romero , g. e. , et al . 2003 , a&a , 409 , 857 bach , u. , krichbaum , t. p. , ros , e. , et al . 2005 , a&a , 433 , 815 bach , u. , krichbaum , t. p. , kraus , a. , et al . 2006 , a&a , 452 , 83 bach , u. , raiteri , c. m. , villata , m. , et al . 2007 , a&a , 464 , 175 bai , j. m. , xie , g. z. , li , k. h. , et al . 1998 , a&as , 132 , 83 bai , j. m. , xie , g. z. , li , k. h. , et al . 1999 , a&as , 136 , 455 blandford , r. d. , & k@xmath29nigl , a. 1979 , apj , 232 , 34 blandford , r. d. , & rees , m. 1978 , physica scripta , 17 , 265 browne , i. w. a. 1971 , nature , 231 , 515 capetti , a. , axon , d. j. , chiaberge , m. , et al . 2007 , a&a , 471 , 137 carswell , r. f. , strittmatter , p. a. , williams , r. e. et al . 1974 , apj , 190 , l101 camenzind , m. & krockenberger , m. 1992 , a&a , 255 , 59 catanese , m. , akerlof , c. w. , badran , h. m. , et al . 1998 , apj , 501 , 616 chiaberge , m. , gilli , r. , macchetto , f. d. et al . 2006 , apj , 651 , 728 ciprini , s. , takalo , l. o. , tosti , g. , et al . 2007 , a&a , 467 , 465 ciprini , s. , tosti , g. , raiteri , c. m. , et al . 2003 , a&a , 400 , 487 cohen , r. d. , smith , h. e. , junkkarinen , v. t. , & burbidge , e. m. 1987 , apj , 318 , 577 courvoisier , t. j .- l , blecha , a. , bouchet , p. et al . 1995 , apj , 438 , 108 eachus , l. a. & liller , w. 1975 , apj , 200 , l61 fan , j. h. , cheng , k. s. , zhang , l. , & liu , c. h. 1997a , a&a , 327 , 947 fan , j. h. , xie , g. z. , lin , r. g. , et al . 1997b , a&as , 125 , 525 fan , j. h. , xie , g. z. , pecontal , e. , et al . 1998 , apj , 507 , 173 fan , j. h. 1999 , mnras , 308 , 1032 fan , j. h. & lin , r. g. 1999 , apjs , 121 , 131 fan , j. h. , lin , r. g. , xie , g. z. , et al . 2002 , a&a , 381 , 1 fan , j. h. , hua , t .- x . , yuan , y .- h . , et al . 2006 , pasj , 58 , 945 fan , j. h. , liu , y. , yuan , y. h. , et al . 2007 , a&a , 462 , 547 falomo , r. 1996 , mnras , 283 , 241 fiorucci , m. , tosti , g. , & rizzi , n. , 1998 , pasp , 110 , 105 fiorucci , m. & tosti , g. , 1996 , a&as , 116 , 403 foschini , l. , tagliaferri , g. , pian , e. , et al . 2006 , a&a , 455 , 871 ghisellini , g. , villata , m. , raiteri , c. m. , et al . 1997 , a&a , 327 , 61 ghosh , k. k. , & soundararajaperumal , s. 1995 , apjs , 100 , 37 gupta , a. c. , banerjee , d. p. k. , ashok , n. m. , & joshi , u. c. 2004 , a&a , 422 , 505 heidt , j. , & wagner , s. j. , 1996 , a&a , 305 , 42 hovatta , t. , tornikoski , m. , lainela , m. , et al . 2007 , a&a , 469 , 899 hughes , p. a. , aller , h. d. & aller , m. f. 1992 , apj , 396 , 469 impey , c. d. , brand , p. w. j. l. & tapia , s. 1982 , mnras , 198 , 11 jorstad , s. g. , marscher , a. p. , stevens , j. a. , et al . 2007 , aj , 134 , 799 mangalam , a. v. , & wiita , p. j. , 1993 , apj , 406 , 420 marscher , a. p. , & gear , w. k. 1985 , apj , 298 , 114 marscher , a. p. , gear , w. k. & travis , j. p. , 1992 , in variability of blazars , ed . e. valtaoja , m. valtonen , cambridge university press , p. 85 massaro , e , nesci , r. , parola , g. c. et al . 1995 , a&a , 299 , 339 mead , a. r. g. , ballard , k. r. , brand , p. w. , et al . 1990 , a&as , 83 , 183 miller , h. r. , carini , m. t. , & goodrich , b. d. 1989 , nature , 337 , 627 montagni , f. , maselli , a. , massaro , e. , et al . 2006 , a&a , 451 , 435 nesci , r. , massaro , e. , rossi , c. , et al . 2005 , aj , 130 , 1466 nilsson , k. , charles , p. a. , pursimo , t. , et al . 1996 , a&a , 314 , 754 ostorero , l. , wagner , s. j. , gracia , j. , et al . 2006 , a&a , 451 , 797 padovani , p. , costamante , l. , giommi , p. , et al . 2004 , mnras , 347 , 1282 perlman , e. s. , stocke , j. t. , schachter , j. f. , et al . 1996 , apjs , 104 , 251 pian , e. , foschini , l. , beckmann , v. , et al . 2005 , a&a , 429 , 427 pursimo , t. , takalo , l. , sillanp@xmath6 , a. , et al . 2000 , a&as , 146 , 141 qian , b. & tao , j. 2004 , pasp , 116 , 161 raiteri , c. m. , villata , m. , aller , h. d. , et al . 2001 , a&a , 377 , 396 raiteri , c. m. , villata , m. , tosti , g. , et al . 2003 , a&a , 402 , 151 raiteri , c. m. , villata , m. , ibrahimov , m. a. , et al . 2005 , a&a , 438 , 39 raiteri , c. m. , villata , m. , kadler , m. , et al . 2006a , a&a , 452 , 845 raiteri , c. m. , villata , m. , kadler , m. , et al . 2006b , a&a , 459 , 731 raiteri , c. m. , villata , m. , larionov , v. m. , et al . 2007a , a&a , 473 , 819 raiteri , c. m. , villata , m. , larionov , v. m. , et al . 2007b , a&a , accepted romero , g. e. , combi , j. a. , benaglia , p. , et al . 1997 , a&a , 326 , 77 romero , g. e. , cellone , s. a. , & combi , j. a. 1999 , a&as , 135 , 477 romero , g. e. , cellone , s. a. , & combi , j. a. 2000 , a&a , 360 , l47 sbarufatti , b. , treves , a. , & falomo , r. , 2005 , apj , 635 , 173 scarpa , r. & falomo , r. 1997 , a&a , 325 , 109 sillanp@xmath6 , a. , haarala , s. , valtonen , m. j. , et al . 1988 , apj , 325 , 628 sillanp@xmath6 , a. , takalo , l. o , pursimo , t. , et al . 1996a , a&a , 305 , 17l sillanp@xmath6 , a. , takalo , l. o , pursimo , t. , et al . 1996b , a&a , 315 , 13l smith , p. s. , balonek , t. j. , heckert , p. a. , et al . 1985 , aj , 90 , 1184 smith , a. g. & nair , a. d. 1995 , pasp , 107 , 863 spinrad , h. , & smith , h. e. , 1975 , apj , 201 , 275 stetson , p. b. 1987 , pasp , 99 , 191 stickel , m. , fried , j. m. , & k@xmath30hr , h. 1993 , a&as , 98 , 393 takalo , l. o. 1991 , a&as , 90 , 161 takalo , l. o. , sillanp@xmath6 , a. , nilsson , k. , et al . 1992 , a&as , 94 , 37 takalo , l. o. , sillanp@xmath6 , a. , nilsson , k. , 1994 , a&as , 107 , 497 takalo , l. o. , sillanp@xmath6 , a. , pursimo , t. , et al . 1996 , a&as , 120 , 313 tommasi , l. , palazzi , e. , pian , e. , et al . 2001 , a&a , 376 , 51 tosti , g. , fiorucci , m. , luciani , m. , et al . 1998 , a&as , 130 , 109 urry , c. m. , & padovani , p. 1995 , pasp , 107 , 803 valtaoja , l. , karttunen , h. , efimov , y. s. , shakhovskoi , n. m. 1993 , a&a , 278 , 371 valtaoja , l. , sillanp@xmath6 , a. , valtaoja , e. et al . 1991 , aj , 101 , 78 villata , m. , raiteri , c. m. , ghisellini , g. et al . 1997 , a&as , 121 , 119 villata , m. , raiteri , c. m. , lanteri , l. , et al . 1998 , a&as , 130 , 305 villata , m. , mattox , j. r. , massaro , e. , et al . 2000 , a&a , 363 , 108 villata , m. , raiteri , c. m. , balonek , t. j. , et al . 2006 , a&a , 453 , 817 wagner , s. j. , sanchez - pons , f. , quirrenbach , a. , & witzel , a. 1990 , a&a , 235 , l1 wagner , s. j. , & witzel , a. 1995 , ara&a , 33 , 163 wagner , s. j. , witzel , a. , & heidt , j. et al . 1996 , aj , 111 , 2187 webb , j. r. , carini , m. t. , clements , s. , et al . 1990 , aj , 100 , 1452 webb , j. r. , howard , e. , benitez , e. , et al . 2000 , aj , 120 , 41 wright , s. c. , mchardy , i. m. , abraham , r. g. , et al . 1998 , mnras , 296 , 961 wu , j. , zhou , x. , wu , x .- b . , et al . 2006 , aj , 132 , 1256 wu , j. , zhou , x. , ma , j. , et al . 2007 , aj , 133 , 1599 xie , g. z. , li , h. k. , liu , f. k. , et al . 1992 , apjs , 80 , 683 xie , g. z. , li , h. k. , zhang , x. et al . 1999 , apj , 522 , 846 xie , g. z. , zhou , s. b. , dai , b. z. , et al . 2002a , mnras , 329 , 689 xie , g. z. , liang , e. w. , zhou , s. b. , et al . 2002b , mnras , 334 , 459 zhang , x. , zhang , l. , zhou , g. , et al . 2004 , aj , 128 , 1929 zhang , x. , zhao , g. , zheng , y. g. , et al . 2007 , aj , 133 , 1995 cccccccc s2 0109@xmath1224 & 01 12 05.7 & @xmath122 44 39.2 & @xmath170.4 & 11.01.2007 & r & 80 & 3.00 + ao 0235@xmath1164 & 02 38 38.9 & @xmath116 36 59.3 & 0.94 & 09.01.2007 & r & 4 & 0.26 + & & & & 10.01.2007 & r & 79 & 3.04 + & & & & 19.03.2007 & r & 6 & 0.40 + s5 0716@xmath1714 & 07 21 53.5 & @xmath171 20 36.4 & 0.3 ? , 0.52 ? & 10.01.2007 & r & 116 & 3.03 + & & & & 12.01.2007 & r & 185 & 3.56 + & & & & 23.02.2007 & r & 60 & 1.00 + & & & & 19.03.2007 & r & 82 & 1.58 + & & & & 20.03.2007 & r & 195 & 4.24 + pks 0735@xmath1178 & 07 38 07.4 & @xmath117 42 19.0 & 0.424 & 11.01.2007 & r & 90 & 3.88 + oj 287 & 08 54 48.9 & @xmath120 06 31.0 & 0.306 & 29.10.2006 & r & 15 & 1.14 + & & & & 17.11.2006 & r & 18 & 1.18 + & & & & 18.11.2006 & r & 16 & 1.06 + & & & & 10.01.2007 & r & 49 & 1.79 + on 231 & 12:21:31.7 & @xmath128:13:58.5 & 0.102 & 11.01.2007 & r & 72 & 3.24 + 3c 279 & 12 56 11.2 & @xmath005 47 21.5 & 0.5362 & 12.01.2007 & r & 208 & 4.00 + & & & & 23.02.2007 & r & 100 & 2.35 + 3c 454.3 & 22 53 57.7 & @xmath116 08 54.0 & 0.859 & 27.10.2006 & r , v & 15,9 & 1.59 + & & & & 28.10.2006 & r , v & 15,8 & 1.27 + & & & & 29.10.2006 & r , v & 15,10 & 1.76 + & & & & 30.10.2006 & r , v & 14,2 & 1.10 + & & & & 09.01.2007 & r & 11 & 0.68 + & & & & 10.01.2007 & r & 7 & 0.54 + 1es 2344@xmath1514 & 23 47 04.8 & @xmath151 42 17.9 & 0.044 & 12.01.2007 & r & 53 & 1.89 + cccc s2 0109@xmath1224 & i & 12.11 ( 0.04 ) & 1 + & c1 & 14.72 ( 0.06 ) & 1 + & d & 14.09 ( 0.05 ) & 1 + & e & 14.94 ( 0.05 ) & 1 + ao 0235@xmath1164 & 1 & 12.69 ( 0.02 ) & 2 + & 2 & 12.23 ( 0.02 ) & 2 + & 3 & 12.48 ( 0.03 ) & 2 + & 6 & 13.64 ( 0.04 ) & 3 + & c1 & 14.23 ( 0.05 ) & 3 + s5 0716@xmath1714 & 2 & 11.12 ( 0.01 ) & 4 + & 3 & 12.06 ( 0.01 ) & 4 + & 5 & 13.18 ( 0.01 ) & 4 + & 6 & 13.26 ( 0.01 ) & 4 + & 8 & 13.79 ( 0.02 ) & 4 + pks 0735@xmath1178 & c & 13.85 ( 0.04 ) & 5 + & c1 & 12.89 ( 0.04 ) & 5 + & c2 & 12.79 ( 0.04 ) & 5 + & c4 & 13.80 ( 0.04 ) & 5 + & c7 & 14.70 ( 0.06 ) & 5 + oj 287 & c1 & 15.50 ( 0.07 ) & 6 + & c2 & 15.66 ( 0.08 ) & 6 + & 4 & 13.74 ( 0.04 ) & 6 + & 10 & 14.34 ( 0.05 ) & 6 + & 11 & 14.65 ( 0.05 ) & 6 + on 231 & a & 11.72 ( 0.04 ) & 6 + & c1 & 16.03 ( 0.10 ) & 6 + & d & 13.86 ( 0.04 ) & 6 + 3c 279 & 1 & 12.05 ( 0.02 ) & 7 + & 5 & 15.47 ( 0.02 ) & 7 + & 8 & 14.44 ( 0.00 ) & 8 + 3c 454.3 & 1 & 13.15 ( 0.02 ) & 7 + & 2 & 13.19 ( 0.02 ) & 7 + & 3 & 14.00 ( 0.02 ) & 7 + & 4 & 14.79 ( 0.02 ) & 7 + & 5 & 14.83 ( 0.02 ) & 7 + & 6 & 14.83 ( 0.03 ) & 7 + & 7 & 14.94 ( 0.02 ) & 7 + & 8 & 15.34 ( 0.02 ) & 7 + 1es 2344@xmath1514 & c1 & 12.25 ( 0.04 ) & 3 + & c2 & 14.20 ( 0.05 ) & 3 + & c2 & 15.40 ( 0.08 ) & 3 + blazar & star & v magnitude & reference@xmath31 + & no . & ( error ) & + 3c 454.3 & 1 & 13.71 ( 0.02 ) & 7 + & 2 & 13.80 ( 0.02 ) & 7 + & 3 & 14.44 ( 0.02 ) & 7 + & 4 & 15.21 ( 0.02 ) & 7 + & 5 & 15.30 ( 0.02 ) & 7 + & 6 & 15.34 ( 0.03 ) & 7 + & 7 & 15.74 ( 0.02 ) & 7 + & 8 & 15.94 ( 0.02 ) & 7 + ccccccccccc 11.01.2007 & s2 0109@xmath1224 & 11.36 @xmath0 14.35 & r & 80 & 0.007 & 0.008 & 0.007 & nv & 1.07 & + 09.01.2007 & ao 0235@xmath1164 & 12.60 @xmath0 12.86 & r & 4 & 0.007 & 0.009 & 0.004 & nv & 2.00 & + 10.01.2007 & & 13.10 @xmath0 16.14 & r & 79 & 0.029 & 0.029 & 0.007 & v & 4.14 & 13.7 + 19.03.2007 & & 11.96 @xmath0 12.36 & r & 6 & 0.040 & 0.042 & 0.011 & v & 3.73 & 9.5 + 10.01.2007 & s5 0716@xmath1714 & 16.34 @xmath0 19.37 & r & 116 & 0.009 & 0.009 & 0.005 & nv & 1.80 & + 12.01.2007 & & 13.92 @xmath0 17.71 & r & 185 & 0.019 & 0.021 & 0.007 & v & 2.86 & 6.3 + 23.02.2007 & & 12.22 @xmath0 13.22 & r & 60 & 0.016 & 0.015 & 0.004 & v & 3.88 & 5.2 + 19.03.2007 & & 12.60 @xmath0 14.18 & r & 82 & 0.006 & 0.013 & 0.010 & nv & 0.95 & + 20.03.2007 & & 11.99 @xmath0 16.25 & r & 195 & 0.023 & 0.027 & 0.016 & nv & 1.56 & + 11.01.2007 & pks 0735@xmath1178 & 14.57 @xmath0 18.45 & r & 90 & 0.012 & 0.012 & 0.003 & v & 4.00 & 4.9 + 29.10.2006 & oj 287 & 19.58 @xmath0 20.72 & r & 15 & 0.027 & 0.044 & 0.049 & nv & 0.72 & + 17.11.2006 & & 21.58 @xmath0 22.76 & r & 18 & 0.006 & 0.008 & 0.007 & nv & 1.00 & + 18.11.2006 & & 21.69 @xmath0 22.75 & r & 16 & 0.013 & 0.028 & 0.028 & nv & 0.73 & + 10.01.2007 & & 19.61 @xmath0 21.40 & r & 49 & 0.016 & 0.016 & 0.015 & nv & 1.07 & + 11.01.2007 & on 231 & 18.93 @xmath0 22.17 & r & 72 & 0.010 & 0.009 & 0.009 & nv & 1.06 & + 12.01.2007 & 3c 279 & 18.96 @xmath0 22.96 & r & 208 & 0.008 & 0.010 & 0.009 & nv & 1.00 & + 23.02.2007 & & 17.82 @xmath0 20.17 & r & 100 & 0.005 & 0.009 & 0.007 & nv & 1.00 & + 27.10.2006 & 3c 454.3 & 11.76 @xmath0 13.10 & r & 15 & 0.036 & 0.028 & 0.013 & nv & 2.46 & + & & 12.05 @xmath0 13.35 & v & 9 & 0.022 & 0.023 & 0.011 & nv & 2.05 & + 28.10.2006 & & 15.06 @xmath0 16.33 & r & 15 & 0.018 & 0.019 & 0.007 & v & 2.64 & 5.1 + & & 15.44 @xmath0 16.11 & v & 8 & 0.041 & 0.034 & 0.021 & nv & 1.79 & + 29.10.2006 & & 11.15 @xmath0 12.54 & r & 15 & 0.014 & 0.015 & 0.006 & nv & 2.42 & + & & 11.44 @xmath0 12.90 & v & 10 & 0.042 & 0.036 & 0.013 & v & 3.00 & 11.6 + 30.10.2006 & & 11.56 @xmath0 12.66 & r & 14 & 0.013 & 0.012 & 0.006 & nv & 2.08 & + & & 11.86 @xmath0 11.95 & v & 2 & 0.013 & 0.000 & 0.013 & nv & 0.50 & + 09.01.2007 & & 11.62 @xmath0 12.30 & r & 11 & 0.059 & 0.056 & 0.015 & v & 3.83 & 17.1 + 10.01.2007 & & 11.95 @xmath0 12.49 & r & 7 & 0.054 & 0.077 & 0.037 & nv & 1.77 & + 12.01.2007 & 1es 2344@xmath1514 & 11.61 @xmath0 13.50 & r & 53 & 0.010 & 0.013 & 0.008 & nv & 1.44 & + cc blazar name & date + s2 0109@xmath1224 & 11.01.2007 + ut & magnitude + 11.35889 & 14.148 + 11.39917 & 14.155 + 11.43583 & 14.142 + 11.47250 & 14.145 + 11.51083 & 14.139 + 11.54833 & 14.144 + 11.58556 & 14.138 + 11.62194 & 14.140 + 11.65917 & 14.140 + 11.69639 & 14.144 +	we selected a sample of a dozen blazars which are the prime candidates for simultaneous multi - wavelength observing campaigns in their outburst phase . we searched for optical outbursts , intra - day variability and short term variability in these blazars . we carried out optical photometric monitoring of nine of these blazars in 13 observing nights during our observing run october 27 , 2006 @xmath0 march 20 , 2007 by using the 1.02 meter optical telescope equipped with ccd detector and bvri johnson broad band filters at yunnan astronomical observatory , kunming , china . from our observations , our data favor the hypothesis that three blazars : ao 0235@xmath1164 , s5 0716@xmath1714 and 3c 279 were in the outburst state ; one blazar : 3c 454.3 was in the post outburst state ; three blazars : s2 0109@xmath1224 , pks 0735@xmath1178 and oj 287 were in the pre / post outburst state ; one blazar : on 231 was in the low - state ; and the state of one blazar : 1es 2344@xmath1514 was not known because there is not much optical data available for the blazar to compare with our observations . we observed densely sampled 1534 image frames of these nine blazars . out of three nights of observations of ao 0235@xmath1164 , intra - day variability was detected in two nights . out of five nights of observations of s5 0716@xmath1714 , intra - day variability was detected in two nights . in one night of observations of pks 0735@xmath1178 , intra - day variability was detected . out of six nights of observations of 3c 454.3 , intra - day variability was detected in three nights . no intra - day variability was detected in s2 0109@xmath1224 , oj 287 , on 231 , 3c 279 and 1es 2344@xmath1514 in their 1 , 4 , 1 , 2 and 1 nights of observations respectively . ao 0235@xmath1164 , s5 0716@xmath1714 , oj 287 , 3c 279 and 3c 454.3 were observed in more than one night and short term variations in all these blazars were also noticed . from our observations and the available data , we found that the predicted optical outburst with the time interval of @xmath2 8 years in ao 0235@xmath1164 and @xmath2 3 years in s5 0716@xmath1714 have possibly occurred .
the @xmath5 k superconductor sr@xmath0ruo@xmath1 @xcite is widely believed to be a rare example of a spin - triplet cooper paired system @xcite . unlike heavy fermion materials which are possible candidates for spin triplet pairing ( such as upt@xmath6 ) there are no @xmath7 or @xmath8 heavy elements in the unit cell , and so the effects of spin - orbit coupling should be relatively weak . in this respect paring in sr@xmath0ruo@xmath1 should be closely analogous to the case of superfluid @xmath9he . one should expect that weak or moderate @xmath10-fields would be able to rotate the d - vector order parameter and induce domain walls , textures and other topological defects . as in the case of @xmath11 , the experimental observation of such d - vector rotations would be a decisive test of the pairing symmetry . systematic study of such rotations as a function of various physical parameters would both confirm the pairing symmetry and also place strong constraints on the pairing mechanism and microscopic hamiltonian parameters . one of the strongest pieces of evidence for spin triplet cooper pairing in sr@xmath0ruo@xmath1 was the observation that the electronic spin susceptibility @xmath12 remained constant below @xmath13 for magnetic fields in the a - b plane @xcite . this observation is inconsistent with a spin - singlet pairing state ( @xmath14 or @xmath15-wave ) for which @xmath16 , where @xmath17 is the normal state pauli spin susceptibility and @xmath18 is the yoshida function . on the other hand the observations would be immediately consistent with a chiral symmetry paring state of the form @xmath19 which below we shall refer to as pairing state @xmath20 . for such a pairing state is is well known @xcite that the spin susceptibility has the following tensor form @xmath21 and so the susceptibility tensor is constant as a function of @xmath22 for field directions perpendicular to the @xmath23 vector . the @xmath24 knight shift experiments performed in ( ab ) plane magnetic fields @xcite and neutron scattering experiments @xcite both observed a constant susceptibility below @xmath13 , a result which _ uniquely _ points to a chiral triplet paring state with @xmath25 . more recently murakawa _ @xcite have measured the @xmath26ru - knight shift of sr@xmath0ruo@xmath1 in a superconducting state under the influence of magnetic field parallel to the c - axis . in contradiction to expectations from eqs . ( [ eq : chiral - a ] ) and ( [ eq : chi - a ] ) they found that its value is _ also unchanged from the normal state value below _ they remarked that this result would be consistent with eq . ( [ eq : chi - a ] ) only if it is assumed that the @xmath23 vector rotates away from the @xmath27 direction under the influence of the external field . such a rotation of the @xmath23 vector is well known in the case of superfluid @xmath9he - a @xcite , where to minimize free energy @xmath23 orients itself perpendicular to the external field unless pinned by surface effects . however , in sr@xmath0ruo@xmath1 it is expected that spin - orbit coupling would fix the chiral state @xmath23 vector to the crystal c - axis @xcite . _ @xcite were able to measure @xmath12 down to fields as low as @xmath3 mt at @xmath28 mk , showing that it was equal to @xmath17 to within experimental accuracy at all temperatures and fields measured below @xmath29 ( @xmath30 mt for c - axis fields ) . the implication of this result , that @xmath23 vector rotation must occur at below @xmath3 mt , therefore places very strong constraints on the strength of pinning by the spin - orbit coupling . in this work we examine the combined effects of spin - orbit coupling and c - axis magnetic field on the chiral state of sr@xmath0ruo@xmath1 . we show that a realistic physical model , which we have previously shown to be consistent with a wide range of experimental data , allows us to find reasonable parameter ranges where the chiral state eq . ( [ eq : chiral - a ] ) is stable in zero field , but where a transition to another state occurs even for fields of order @xmath3 mt or smaller . the minimum energy states in finite c - axis field are not simply the chiral state with @xmath23 vector rotated to the @xmath4 plane , but rather are non - chiral pairing states of the form @xmath31 on the @xmath32 fermi surface sheet ( which below we refer to as @xmath33 and @xmath34 respectively ) . the spin susceptibility for either of these states is of the form @xcite @xmath35 corresponding to constant spin susceptibility for c - axis fields consistent with the results of murakawa _ et al . _ @xcite . since these states are symmetry distinct from the chiral state eq . ( [ eq : chiral - a ] ) the transition from one to another should occur at a finite external field , @xmath36 , and it should be accompanied by a finite entropy change @xmath37 . below we map out a generic phase diagram for the transitions between pairing states of sr@xmath0ruo@xmath1 in a c - axis field , assuming parameter values consistent with the murakawa s _ et al . _ experiments @xcite . we also estimate the entropy change associated with the transition , and comment on its experimental observability . our preliminary results have been presented in @xcite . as the pairing mechanism operating in the strontium ruthenate superconductor is not known @xcite it is important to understand the experiments on the basis of semi - phenomenological models @xcite . specifically we write the effective pairing hamiltonian as @xmath38 where @xmath39 and @xmath40 refer to the three ru @xmath41 orbitals @xmath42 , @xmath43 and @xmath44 , and @xmath45 and @xmath46 label the sites of a body centered tetragonal lattice . the hopping integrals @xmath47 and site energies @xmath48 were fitted to reproduce the experimentally determined fermi surface @xcite . @xmath49 is the effective ru 4d spin - orbit coupling parameter , and the effective hubbard parameters @xmath50 are generally both spin @xcite and orbital dependent @xcite . our approach is based on self consistent solution of the bogolubov - degennes equations for the model ( [ hubbard ] ) in the spin triplet channel for each possible symmetry distinct order parameter . we have shown elsewhere @xcite that a minimal realistic model requires two hubbard @xmath51 parameters : @xmath52 for nearest neighbor in - plane interactions between ru @xmath53 orbitals ( corresponding to the @xmath32 fermi surface sheet ) and @xmath54 for out of plane nearest neighbor interactions between ru @xmath55 and @xmath56 orbitals ( corresponding to the @xmath57 and @xmath58 fermi surface sheets ) . when these two parameters are chosen to be spin independent constants ( for @xmath59 ) and to give the experimental @xmath60k on all three @xmath57 , @xmath58 and @xmath32 fermi surface sheets , then the calculated specific heat , thermal conductivity and @xmath4 plane penetration depth are in very good agreement with experiments @xcite . figure ( [ fig : chi ] ) shows the susceptibilities calculated within the model for the ( a ) and ( b ) states with the symmetries given by eqs . [ eq : chiral - a ] and [ eq : deltab ] . for the chiral state ( a ) the triplet d - vector is defined in all 3 bands as @xmath61 , with @xmath62 denoting contributions from different orbitals which are given by @xmath63 for the ru @xmath64 orbitals and , @xmath65for @xmath66 corresponding to the ru @xmath55 and @xmath56 orbitals @xcite . the susceptibility results , for the chiral ( a ) state show that the dependence of @xmath67 for c - axis b - fields is is relatively structureless . the relevant yoshida function @xmath18 is equal to unity at @xmath68 and drops essentially linearly to zero at @xmath69 , consistent with the expectation from eq.[eq : chi - a ] and the existence of the horizontal line node in the gap on @xmath57 and @xmath58 @xcite . in contrast , we show in fig . [ fig : chi ] that for the non - chiral ( b ) pairing state the susceptibility for a - axis b - fields decreases approximately linearly towards @xmath70 at @xmath69 , again consistent with the expectations of eq . [ eq : chi - bc ] . these results confirm that in spite of very complicated character of the order parameter ( see ref . @xcite ) having zeros on the @xmath57 and @xmath58 sheets of the fermi surface , the overall behavior of the calculated susceptibility tensor is consistent with the expectations from superfluid @xmath71 @xcite . in zero magnetic field , but in the presence of non - zero spin - orbit interaction @xmath49 the model predicts that the ground state is the chiral state @xmath20 with d - vector along the crystal c - axis , provided that the spin - orbit coupling leads to a small spin dependence of the effective pairing interaction @xcite . choosing @xmath72 about 1% larger than @xmath73 is sufficient to stabilize the chiral state even for large spin orbit coupling . in contrast , for spin - independent interactions , @xmath74 , the alternative pairing states @xmath33 and @xmath34 ( @xmath75 and @xmath76 @xcite ) are the ground states for any value of @xmath77 ( @xmath78 ) @xcite . in fig . [ fig2 ] we show the free energies of the @xmath20 @xmath33 and @xmath34 symmetry pairing states as a function of external magnetic field both for @xmath79 and @xmath80 . the model parameters were chosen to make the chiral state @xmath20 stable at zero field . for the field in ab - plane the free energy of the chiral state @xmath20 increases with field more slowly than for the @xmath33 and @xmath34 states . this means that in large in - plane fields the @xmath20 state becomes relatively more stable than @xmath33 or @xmath34 . on the other hand , in an c - axis field the chiral phase increases its free energy much faster than the other phases , until at a certain critical field , @xmath36 , the @xmath33 or @xmath34 solutions become more stable . therefore at the field @xmath36 we expect a `` spin flop '' type phase transition from a d - vector oriented along the @xmath2-axis to one where the d - vector lies in the @xmath4 plane . for the parameter values used in fig . [ fig2 ] this critical field is @xmath81 mt . to one where the d - vector lies in the @xmath4 plane . for the parameter values used in this prediction is consistent with what one expects from the analogous superfluid state in @xmath11-a . in bulk @xmath11 there is no preferred symmetry direction in space and so the d - vector simply rotates continuously to remain perpendicular to the applied field . but in thin films the d - vector is pinned , and only rotates at a finite critical field , known as the freedericksz transition @xcite . in the case of sr@xmath0ruo@xmath1 the transition is not simply a rotation of the chiral d - vector but is also a transition from a chiral to non - chiral pairing state of different symmetry . the two distinct solutions shown in fig . [ fig2 ] ( note that @xmath33 and @xmath34 are essentially degenerate ) have different entropies and hence this freedericksz - like spin - flop is a first order thermodynamic phase transition . one can also ask whether the specific case of fig [ fig2 ] is typical for a general set of model parameters . in our effective hubbard model the spin - orbit interaction enters the hamiltonian both directly _ via _ @xmath49 in eq . ( [ hubbard ] ) , and also in the spin dependent pairing potential @xmath82 . these are not really independent , since a full theory , such as a spin fluctuation model @xcite , would include both effects on the same footing . in fig . [ fig3 ] we show a part of the full phase diagram of our model @xcite for a specific choice of the parameters @xmath49 and @xmath82 . the inset shows the effect of the c - axis magnetic field in shifting the phase boundary between the chiral state @xmath20 and the alternative @xmath33 and @xmath34 states . at any point in the phase diagram where @xmath20 is stable for zero field , there is a definite critical field for which the spin flop to @xmath33 or @xmath34 takes place . however it is only a small region of the phase diagram , close to the @xmath83 phase boundary , where the transition takes place in fields as small as @xmath84 mt . the implications of the murakawa _ et al . _ experiment @xcite are that the parameters do indeed lie in this region . the alternative , is that the spin flop only takes place at a much larger field , and thus would never be observed in any c - axis field below @xmath85 mt , in contradiction to the simplest interpretation of murakawa _ et al . _ s result . if the relevant parameters do allow the d - vector rotation to occur in a relatively low field , then we expect a phase diagram similar to fig . this hypothesis makes a definite and clear testable prediction , namely the existence of a new first order phase boundary in the @xmath86 plane , which has not yet been observed . a phase diagram of the type shown in fig . [ fig4 ] is required from the experiments @xcite . the nqr experiments @xcite indicate the state with d - vector lying in the ( ab ) plane is realized in a c - axis magnetic field of order @xmath87 mt and larger , and that this d - vector orientation must be present at these fields for all temperatures below @xmath13 . on the other hand the nonzero kerr signal obtained at fields around @xmath88 mt @xcite ( see the points in fig . ( 3a ) of this reference and the corresponding discussion in the text ) is a clear indication that at these fields the chiral state is stable . accepting the usual interpretations of these two experiments one is forced to accept the general structure of the phase diagram as in fig . [ fig4 ] as the simplest picture which is consistent with both experiments , with a possible margin of uncertainty of about @xmath88 mt in the position of the @xmath89 transition line . an obvious question is why has the predicted phase transition in fig . [ fig4 ] not been observed experimentally ? experimental data @xcite show multiple superconducting phases at the magnetic field parallel to the ( ab ) plane , but none have been reported for fields along the c - axis . the answer to this question we suggest is that the phases in question have essentially identical values of the quasi - particle energy gap , @xmath90 on the fermi surface , and therefore to a very high level of accuracy they have essentially identical entropy . this is because to leading order only the direction of @xmath91 changes at each point on the fermi surface , and not its magnitude . to confirm this we have calculated the entropy change along the transition line @xmath87 mt and this is shown in the fig . the total change in entropy is very small , of order @xmath92 per formula unit , suggesting that it might not have been detected in specific heat experiments . the entropy change in fig . [ fig5 ] corresponds to a latent heat of at most @xmath93mj / mol , compared to the zero field specific heat jump @xmath94 of about @xmath95mj / k@xmath96mol at @xmath60k @xcite . assuming that lattice strain or inhomogeneity leads to some smearing of the ideal first order phase transition this small entropy change could easily be masked by the experimental noise we should note that in the present calculations we have included the influence of the magnetic field in the zeeman energies only and neglected the orbital effects altogether . this means that the vortex contribution to the condensation energy of both the ( a ) and ( b ) phases is assumed to be the same . there is no reason to expect a significantly different vortex lattice response in the different phases , since the only changes are in the direction of @xmath91 on the fermi surface and not in its magnitude @xmath90 . as mentioned above , the experiments @xcite indicate that the region where the chiral @xmath20 state is stable is restricted to the low field part of the @xmath86 plane below a critical field which must be in the range of about @xmath97 mt . this places a number of constraints on the possible parameter values which we can use in our model hamiltonian . using our model hamiltonian , eq . [ hubbard ] , we can ask what are the possible values of the spin - dependent pairing interaction needed to qualitatively describe both experiments . in our model this is the ratio @xmath98 for opposite spin pairing compared to equal spin pairing . the other model parameters are fixed beforehand , including the known band structure and the values of interaction parameters @xmath52 and @xmath54 which were previously fitted to other experiments @xcite . for simplicity we have taken the spin - orbit coupling to have a fixed value , @xmath99 , where @xmath100 is the @xmath32 band in - plane nearest neighbor hopping integral in our tight binding band structure fit . with all other parameters fixed , we then calculate the minimum spin dependent interaction enhancement @xmath82 required to stabilize the chiral state in magnetic fields of up to @xmath87 mt . it turns out that this minimal value of @xmath82 changes with temperature , as shown in fig . [ fig_upr ] . the temperature dependence of @xmath82 shown in fig . [ fig_upr ] arises because the condensation energies calculated for a given value of @xmath82 for @xmath20 and @xmath33 states cross as function of temperature as shown in fig . [ fig_fab ] . for a small value of @xmath82 for which the @xmath33 state will be stable at low temperatures , we find that this state is destabilized relative to the chiral state closer to @xmath13 . let us stress that the model we are using is , in principle at least , valid for arbitrarily large spin - orbit coupling . however , the parametrization of the fermi surface which we have used ceases to be valid for @xmath101 . therefore in order not to refit the normal state band structure for every value of @xmath49 we have limited the calculations to small values of @xmath49 . in fact most of the calculations presented in this work have been done for @xmath102 . this restricted choice does not limit the validity of our conclusions , because for every value of the spin - orbit coupling parameter , @xmath49 , it is possible to find a corresponding value of @xmath82 which will stabilize the chiral state at @xmath103 as discussed earlier ( _ c.f . _ fig . [ fig4 ] ) . close enough to this boundary of stability a c - axis magnetic field will rotate the d - vector and stabilize one of the other four states @xcite independently of the sign or magnitude of @xmath49 . this minimal value of @xmath82 depends not only on @xmath49 and @xmath10 but also on temperature , as shown in fig . [ fig_upr ] . however , this uncertainty in the value of @xmath82 precludes an unique theoretical determination of the phase diagram on the @xmath86 plane without additional assumptions . assuming a constant ( _ i.e. _ temperature independent ) value of @xmath82 which leads to stabilization of the chiral phase at the lowest temperatures would lead to a h - t phase diagram with the chiral phase occupying most of it . the non - chiral state would be limited to low temperature - high field corner . other assumptions about the temperature dependence of @xmath82 could produce different topology phase diagrams . rather than try to explore all these possibilities we simply assume that the boundary between the stable phases at low c - axis fields is given by a line at approximately @xmath104 mt ( shown in fig . [ fig2 ] ) as dictated by experiment . from this assumption we have then calculated the corresponding temperature dependence of the minimal anisotropy @xmath82 which is necessary to stabilize the a - phase along the line ( as shown in fig . [ fig_upr ] ) . we find this approach more direct than trying to determine the expected temperature dependence of @xmath82 _ a priori _ from a spin fluctuation feedback mechanisms @xcite stabilizing chiral phase . we note that our effective pairing hubbard model does not really include the full self - consistent effects of spin fluctuation feedback @xcite , which could make the effective pairing interactions @xmath51 and/or @xmath105 temperature dependent . however , the predicted phase transition places severe constraints on the superconducting pairing mechanism when one takes into account a sizable spin - orbit coupling and the detrimental effect of the c - axis b - field on the chiral phase . to summarize , we have explored the role played by spin - orbit coupling in determining the superconducting states of strontium ruthenate in the presence of c - axis and ab - plane oriented magnetic fields . we showed that the d - vector rotation can provide a consistent interpretation of both the nqr and kerr effect data , provided the spin - orbit coupling constant @xmath49 is small enough . unfortunately , lda estimates of @xmath49 are significantly larger than the values we have assumed in the calculations presented here@xcite . a large value of spin - orbit coupling has previously been cited as evidence against the d - vector rotation picture @xcite . despite these objections , it is clear from our results that even for large values of @xmath49 the transition field @xmath36 for d - vector rotation , in fig . [ fig2 ] , can lie within experimental constraints @xcite , but only if the pairing interaction spin anisotropy @xmath82 is such as to make the free energy difference between ( a ) and ( b ) phases finely balanced so that a small field of order @xmath3 mt is sufficient to cause the d - vector rotation . we have shown that for any value of @xmath49 there is corresponding anisotropy @xmath82 such that this balance can be achieved . it is possible that such balancing is a consequence of a spin - fluctuation feedback mechanism @xcite , or emerges directly from the full microscopic spin fluctuation theory . but in the absence of a detailed theory of this effect further progress in the field must await experimental measurement of the effective spin - orbit parameter for the quasiparticles at the fermi surface . + acknowledgements : the authors acknowledge fruitful discussions with prof . y. maeno and prof . i. mazin . this work has been partially supported by the ministry of science and higher education grant no . n202 1878 33 .	in zero magnetic field the superconductor sr@xmath0ruo@xmath1 is believed to have a chiral spin triplet pairing state in which the gap function d - vector is aligned along the crystal @xmath2-axis . using a phenomenological but orbital specific description of the spin dependent electron - electron attraction and a realistic quantitative account of the electronic structure in the normal state we analyze the orientation of the spin triplet cooper pair d - vector in response to an external @xmath2-axis magnetic field . we show that for suitable values of the model parameters a c - axis field of only @xmath3 mt is able to cause a reorientation phase transition of the d - vector from along @xmath2 to the @xmath4 plane , in agreement with recent experiments .
we start by analizing the zero - dimensional ( single variable ) version of the rft langevin equation . the fokker - planck equation associated to eq . ( [ rft ] ) is @xcite : @xmath19 by imposing the detailed balance condition , the associated formal stationary probability distribution is found to be : @xmath20 \label{prft0}\ ] ] where @xmath21 is the effective potential . in figure [ v0rft ] we plot @xmath21 for different values of @xmath5 , and @xmath22 ; for @xmath23 the potential has a minimum at @xmath24 , while for @xmath25 @xmath21 has no maximum or minimum . note , however , that due to the ( non - integrable ) singularity at @xmath26 , the probability eq . ( [ prft0 ] ) is not normalizable , and the only stationary solution is @xmath27 . therefore there is no active phase in this simple 0-dimensional case , and the systems decays towards the absorbing state for any set of paramenter values . let us now study how the single - site effective potential behaves in dimensions larger than zero . in particular , we perform a numerical simulation of eq . ( [ rft ] ) in one dimension . to do so we employ a technique developed by dickman @xcite to deal with numerical simulations of the continuous rft . let us point out that the simulation of this continuous theory with an absorbing state is not a trivial issue , and that , for example , a straightforward discretization of eq . ( [ prft0 ] ) in which eventual negative values of the field ( that may appear due to the discretization ) are fixed to @xmath9 , does not preserve the presence of an absorbing state . dickman s method consists of a discretization of the space , time and also of the field variable , that ensures the presence of an absorbing state ( see @xcite for details ) . in the one - dimensional case the active phase survives to the effect of fluctuations contrarily to what happens in the single - variable case . in figure [ v1rft ] we show the effective potential ( defined as minus the logarithm of the normalized stationary probability distribution ) for different values of @xmath5 . the uppermost curve corresponds to a value of a in the absorbing phase ; the second one to @xmath28 while the two lower ones are in the active phase . note that in all the cases a singularity at the origin of the same type is present , and consequently , for any finite system there is a finite probability for the system in the active phase to go through the potential barrier and decay towards the absorbing state : the active phase is a metastable state . the mean time required for the system to overpass the barrier and collapse to the absorbing state grows exponentially with time , and becomes infinite in the thermodynamic limit . in this way the phase transition appears only in infinitely large systems , and the large system - size limit has to be taken first than the infinite time limit in order to permit the presence of an active phase . in dimensions larger than @xmath29 the same qualitative type of behavior is expected . the fokker - planck equation associated to eq . ( [ mn ] ) in the zero - dimensional case is : @xmath30 by imposing the detailed balance condition , the stationary formal solution is : @xmath31 \label{pmn0}\ ] ] where @xmath21 is the effective potential ; the solution is not normalizable when @xmath32 , and normalizable otherwise . @xmath21 is plotted in figure [ v0mn ] for different values of @xmath5 and @xmath22 ; the two lowermost curves correspond to eq . ( [ pmn0 ] ) in the absorbing phase ( @xmath33 and @xmath34 ) ( where the only stationary solution is @xmath35 ) . the central one ( @xmath36 ) , and the two uppermost curves ( @xmath37 and @xmath38 ) are in the active phase . note that contrarily to the rft the mn exhibits a phase transition even in zero dimensions . observe also that the singularity at the origin in the formal solution eq . ( [ pmn0 ] ) changes its degree as @xmath5 is increased , in contrast with what happens for eq . ( [ prft0 ] ) ; in fact , for @xmath39 ( in the active phase ) the singularity is integrable , and above @xmath40 ( also in the active phase ) the origin becomes _ repelling _ instead of _ this is an essential difference with rft . let us now explore how this property of the single - site potential is modified in higher dimensions . for that , we perform a numerical integration of the stochastic equation defining the model , which presents less technical difficulties than the integration of the rft @xcite . the result are presented in figure [ v1mn ] . qualitatively the potential shape changes in the same way as it does in the 0-dimensional case . above the critical point , there is either an integrable singularity ( uppermost curve ) or a repelling wall ( three other curves ) at the origin . the first case , i.e. , an integrable singularity at the origin occurs in a very tiny region of the parameter space . on the other hand , in the absorbing phase there is a collapse of the probability towards @xmath35 ( non - integrable singularity at the origin ) . therefore , the physics is very different from that of rft : when the system is in the active phase , there is either an integrable singularity at the origin of the potential or a repelling wall . in any case , the situation differs from that in rft , in which there are two locally stable attractors in the active phase : one at the origin ( the stable one ) and one at a different point ( a metastable one ) . that is the reason why a reaction - diffusion system can not be described by an equation as eq . ( [ mn ] ) : in a finite reaction - diffusion system with a non - vanishing particle annihilation rate there is a non - zero probability of reaching the absorbing ( empty ) state for any finite system - size and for any set of parameter values . in systems with mn there is no accessible absorbing state in the active phase , i.e. there is no non - integrable singularity at the origin of the potential ( and consequently no collapse of the probability density to the origin ) , and therefore mn does not capture the physics of reaction - diffusion systems . that fact , does not mean that it is not possible to construct discrete lattice model in the multiplicative noise universality class . in particular , systems exhibiting an unbinding transition from a wall ( as , for example , the problem of local alignment of dna chains @xcite , and wetting transitions @xcite ) belong in this universality class . these models are usually defined in terms of a field variable @xmath41 that flows to @xmath42 in the absorbing phase , and that reach a non - vanishing stationary average value otherwise . in this section we compare the effect of the complex noise appearing in the peliti s field theory eq . ( [ pft ] ) , with the two previously studied cases to get a global picture of the different type of noises that can appear in systems with absorbing states . in order to understand what is the origin of the complex noise in processes as @xmath18 , and to clarify whether a microscopic process like that with a real density field has necessarily to be described by a complex langevin equation , we present a derivation of the peliti s field theory for the annihilation reaction : @xmath18 by employing the exact poisson representation introduced by gardiner and chatuverdi @xcite . note that for this reaction there is no active phase , and the interesting magnitudes are those describing the decay towards the absorbing state . for the sake of simplicity in the notation we present here the 0-dimensional case , extensions to higher dimensions being straightforward . the master equation defining the process is : @xmath43 \label{master}\ ] ] multiplying both sides of eq . ( [ master ] ) by @xmath44 , summing over all @xmath45 from @xmath46 to @xmath47 , and defining the generating function @xmath48 , we get @xmath49 we now introduce the poisson transformation : @xmath50 where @xmath51 is a given function ( see @xcite ) , in terms of which : @xmath52 the poisson transformation has the interesting property that the moments of @xmath53 and @xmath54 can be easily related : @xmath55 ; in particular the first moments are the same for both distributions . the integral over @xmath56 can be taken over different domains of integration ; for the moment let us assume @xmath56 to be a real variable and leave the integration domain undetermined . in terms of @xmath51 , eq . ( [ master2 ] ) reads , @xmath57 \exp ( \alpha ( s-1 ) ) \ ] ] which , integrating by parts , and assuming that the boundary terms give a vanishing contribution to the integral @xcite can be written as : @xmath58 \label{fpp}\ ] ] which is a fokker - planck equation with a negative diffusion coefficient . the langevin equation stochastically equivalent to the previous fokker - planck equation is : @xmath59 where @xmath60 a gaussian noise with amplitude @xmath61 , @xmath14 the complex unit , and @xmath62 has been eliminated by redefining the time as @xmath63 . note that we have arrived to an inconsistency : @xmath64 was assumed to be a real variable and we have arrived to a complex equation ( observe that due to the complex term in eq . ( [ lan ] ) @xmath56 develops an imaginary part even if it is taken to be real at time @xmath65 . in appendix a we present further details on the impossibility of defining a real poissonian representation for the reaction @xmath18 ) . let us now repeat the previous program but performing a complex transformation instead of a real one , i.e. , we take @xmath66 to be @xmath67 , where @xmath68 and @xmath69 are the real and imaginary parts of @xmath56 respectively . this type of transformation leads to a function @xmath54 which is positive and can be identified as a probability distribution @xcite . proceeding in that way we get a new set of langevin equations for the variables @xmath68 and @xmath69 @xmath70 that is equivalent to the original master equation . note that both of the equations in eq . ( [ complex ] ) include the same noise function @xmath60 , and that they could be obtained straightforwardly from eq . ( [ lan ] ) just by writing @xmath71 and separating the real and imaginary parts ) . a typical trajectory of the previous set of equations in the stationary state is shown in figure [ trajectory ] ; it wonders in the complex plane avoiding a region around @xmath46 . even if the stationary solution of the underlying process @xmath18 is a delta function at zero or one particles ( depending on whether the initial condition is even or odd respectively ) , the stationary probability associated to eq.([complex ] ) is not a delta function , but some complicate distribution with @xmath72 and @xmath73 . the value @xmath74 comes from the fact that for initial conditions with @xmath75 even , @xmath76 and , for @xmath75 odd , @xmath77 ; the variable @xmath78 is the average of the two previous possibilities . on the other hand , the expectation value of the imaginary part is zero as expected given the relation among moments of @xmath54 and @xmath53 . it is interesting to note that the effective potential associated to the stationary distribution is a non - differentiable one ; in figure ( [ corte ] ) we show a one - dimensional cut of the stationary potential for different @xmath68 with @xmath79 as computed in a simulation of eq . ( [ complex ] ) . note that contrarily to the cases of the rft and multiplicative noise equation , now the dynamics is not frozen even if the system has relaxed to the absorbing state . we now explore the possibility of finding a real - variable langevin equation describing this class of systems @xcite . as the second equation in eq . ( [ complex ] ) is linear in @xmath69 it is possible to integrate it analytically ; doing so and substituting the result in the first one , we get a closed equation for @xmath68 , that reads : @xmath80 with @xmath81 which is a non - markovian equation ( see appendix b ) . the stationary potential associated to eq.([nonmark ] ) can not be calculated analytically ; the numerical solution is shown in figure [ histoproyect ] . first we observe that it is non - differentiable at @xmath82 ; also we point out that @xmath68 is not absorbing in general ( except for the pathological and unphysical case @xmath83 ) , in other words : due to the presence of the non - markovian terms , proportional to @xmath84 , the system can cross from positive values to negative ones . the role of the complex variable in eq . ( [ lan ] ) is played , after @xmath69 has been integrated out , by the non - markovian terms in eq . ( [ nonmark ] ) , and in both cases the absorbing state of the microscopic associated process is not described by a frozen dynamics in the langevin representation , but by a non - trivial dynamics ( complex or non - markovian ) which statistical properties reproduce those of the reaction - diffusion model . therefore , the nature of the absorbing state in this case is essentially different from those of the previously studied cases . for the sake of completeness let us just mention briefly that a numerical study of the one - dimensional peliti s field theory in terms of a complex langevin equation has been recently published @xcite . the measured magnitudes are in very good agreement with the theoretical predictions coming from renormalization - group and other type of analysis @xcite , confirming that a complex representation captures the physics of microscopic systems as , for example , the procces @xmath18 . we have analyzed different langevin equations associated to systems with absorbing states . systems described by the reggeon field theory langevin equation exhibit a non - integrable singularity at the origin of the single - site potential , that corresponds to a true absorbing state , i.e. , there is an accumulation of probability density at the origin , while the active state is a metastable one for finite system - sizes . systems with multiplicative noise instead change the degree of the singularity at the origin as the control parameter is changed : while in the absorbing phase there is a collapse of the probability density towards the origin , in the active phase there is either an integrable singularity at the origin or it becomes repelling , in which case the the probability to be nearby the origin becomes extremely small , and there is no accessible absorbing state . that is the reason why is not possible to find reaction - diffusion systems ( in which for finite size systems there is always a finite probability of reaching the absorbing state ) in the multiplicative noise universality class . we have also analized some aspects of the annihilation process @xmath85 , which is described by a complex noise langevin equation or alternatively by a real non - markovian langevin equation . this type of langevin equation shows a behavior quite different from that of reggeon field theory and multiplicative noise ; in particular , even if the system is in the absorbing state , there is not a collapse of the probability density to a delta function , and the dynamics is non - trivial . systems with complex noise can alternatively be described by real non - markovian equations . let us consider the pair of reactions @xmath18 and @xmath86 , the first occurring with a rate @xmath87 and the second with @xmath88 . the associated master equation is @xmath89 \nonumber \\ & + & k_1[(n-1 ) p(n-1)- n p(n ) ] . \label{master22}\end{aligned}\ ] ] performing a real poissonian transformation we get : @xmath90 \label{fp2}\ ] ] which is equivalent to the langevin equation @xmath91 interpreted in the ito sense . note that the factor multiplying the noise is positive in the interval @xmath920,k_1/k_2[$ ] , and vanishes at the limits of the previous interval . the formal stationary solution of eq . ( [ fp2 ] ) is : @xmath93 with @xmath94 . considering the poisson representation as defined in @xmath95 $ ] , it is a matter of simple algebra to verify that the boundary terms appearing in the processes of getting eq . ( [ fp2 ] ) from eq . ( [ master22 ] ) give a vanishing contribution . at the same time , trajectories of eq . ( [ lan2 ] ) with initial condition in @xmath95 $ ] do not leave that interval . on the other hand , if the domain of integration was extended over those limits , eq . ( [ lan2 ] ) would develop an imaginary part and the procedure would not be self - consistent . therefore the transformation is well defined only in the real interval @xmath95 $ ] . from a renormalization group point of view the noise - term proportional to @xmath87 in eq . ( [ lan2 ] ) can be argued to be irrelevant rendering the system in the rft universality class . we can now take the limit @xmath96 to see what happens in the peliti s field theory case : the interval in which the poisson representation is defined shrinks down to a single point ; @xmath97 . in the strict limit @xmath98 , a meaningful real poisson representation can not be performed , and a complex representation is required . let us point out as a final remark that it is somehow surprising that the standard renormalization group analysis of the peliti s field theory , based on a path integral representation of eq . ( [ lan ] ) ( or equivalently of eq.([fpp ] ) ) , in which @xmath56 is treated as a real variable , give the right exponents and properties @xcite . we will investigate that apparent paradox in a future work . as a last attempt to write down a one - variable langevin equation with a structure simpler than eq . ( [ nonmark ] ) , and inspired by the rotational quasi - symmetry of the stationary distribution solution ( see figure [ trajectory ] ) , we perform a change of variables to polar coordinates @xmath99 and @xmath100 defined by : @xmath101 , @xmath102 . after changing variables ( for which ito calculus is required @xcite ) , we get : @xmath103 observe that the first one is a deterministic equation , while the second one is stochastic . it is easy to verify that this system does not admit a potential solution ( which is consistent with the stationary potential being non - differentiable ) . this new set of equations permits to derive analytically some of the properties of the stationary probability distribution ( as , for example , the presence of a maximum at @xmath104 ) , but it does not simplify the elimination of one of the variables in favor of the other one to construct a simple one - variable langevin equation . it is a pleasure to acknowledge yuhai tu , geoffrey grinstein , ron dickman , pedro garrido and andrea gabrielli for very useful discussions and comments . i am grateful to the anonymous referee whose comments and criticisms helped me to improve this paper . this work was supported by the european community through grant erbfmbict960925 . h. takayasu and a. yu . tretyakov , phys . lett . * 68 * , 3060 ( 1992 ) ; p. grassberger , f. krause , and t. von der twer , j. phys . * a 17 * , l105 ( 1984 ) . p. grassberger , j. phys . * a 22 * , l1103 ( 1989 ) ; m.h . kim and h. park , phys . * 73 * , 2579 ( 1994 ) ; n. menyhard , j. phys . * a 27 * , 6139 ( 1994 ) ; j. cardy and u.c . tuber , phys . lett . * 77 * , 4780 ( 1996 ) . an equivalent , and probably less confusing , way to write down the langevin equation for the rft consists of omiting the square - root of @xmath75 term in it and substituting @xmath16 by @xmath105 in the noise correlator . when a fock - space formalism is employed to derive eq . ( [ pft ] ) from the master equation defining the process @xmath106 , the @xmath2 are found to be related to _ coherent states_. alternatively , eq . ( [ pft ] ) can be obtained from the master equation by performing a poissonian transformation ( see section i c ) ; in this language the @xmath75 field in the conjugate field of the number of particles . it has been proved that both formalisms , i.e. the fock - space and the poissonian transformation , are rigorously equivalent and that the results obtained using any of them can be exactly translated into the other ( see m. droz and a. mckane , * a 27 * , l467 ( 1994 ) ) .	we present a comparison of three different types of langevin equation exhibiting absorbing states : the langevin equation defining the reggeon field theory , one with multiplicative noise , and a third type in which the noise is complex . each one is found to describe a different underlying physical mechanism ; in particular , the nature of the different absorbing states depends on the type of noise considered . by studying the stationary single - site effective potential , we analyze the impossibility of finding a reaction - diffusion model in the multiplicative noise universality class . we also discuss some theoretical questions related to the nature of complex noise , as for example , whether it is necessary or not to consider a complex equation in order to describe processes as the annihilation reaction , @xmath0 . pacs : = 9.5 in @twocolumnfalse different systems and models appearing in physics as well as in other fields can exhibit absorbing states . an absorbing configuration is one in which a system can get trapped , from which it can not escape @xcite . therefore an absorbing configuration is a fluctuation - free microscopic state . some examples of systems exhibiting absorbing states , among many others , are : chemical reaction - diffusion models of catalysis @xcite , models for the spreading of epidemics or forest fires @xcite , directed percolation @xcite , the contact procces @xcite , models of branching and annihilating random walks @xcite , damage spreading @xcite , and even self - organized systems @xcite . as some control parameter is changed , many of these systems experience a phase transition from an absorbing phase , i.e. , a phase in which the absorbing state is the only stationary state @xcite , to an active phase , characterized by a non - vanishing value of the order parameter . at the critical point , these systems exhibit universal features . it was conjectured some time ago by janssen and grassberger @xcite that all the different systems and models with an unique absorbing state , a single - component order parameter , and no extra symmetry or conservation law belong in the same universality class as directed percolation ( dp ) , which is considered the canonical representative of that vast class of models . in a field theoretical description this universality class is represented by the reggeon field theory ( rft ) , @xcite which in terms of a langevin equation reads @xcite : @xmath1 where @xmath2 is a density field at position @xmath3 and time @xmath4 , @xmath5 and @xmath6 are control parameters , and @xmath7 is a gaussian noise which only non - vanishing correlations are : @xmath8 . the equation is interpreted in the ito sense @xcite . the noise term in eq.([rft ] ) is proportional to the square root of the field , therefore in the absorbing state @xmath9 the dynamics is completely frozen : both the deterministic and the stochastic terms are equal to zero . other higher - order terms could be added to the deterministic part of eq . ( [ rft ] ) but they can be easily argued to be irrelevant in a renormalization group sense . the previous conjecture has been confirmed in a large number of computer simulations and series expansion analysis , and dp - universality class has proven to be extremely robust against the modification of many details in the microscopic models . the conjecture of universality has been extended for multicomponent systems @xcite , as well as for systems with an infinite number of absorbing states @xcite . nevertheless , not all the systems with absorbing states belong in the universality class of dp . some other classes different of dp have been identified , all of them showing some essential physical differences with dp . two relevant examples for what follows are : 1 . particle systems in which evolution occurs only at the interfaces separating occupied ( active ) from empty ( absorbing ) regions belong to the _ compact directed percolation _ ( cdp ) universality class @xcite . examples of this universality class are the one - dimensional diffusion - limited reactions of the type @xmath10 and @xmath11 ( in dimensions larger than one these models are not expected to be in the same universality class as cdp ) . the _ pseudoparticles _ @xmath12 can be thought of as the _ kinks _ separating active from inactive regions , where the dynamics occurs . a field theoretical description for such class of systems was proposed by peliti in @xcite ; its equivalent langevin equation reads : @xmath13 where @xmath2 is a field , @xmath14 is the imaginary unit , and @xmath15 is a gaussian noise with some amplitude @xmath16 . note that @xmath9 is an absorbing configuration . hereafter we refer to eq . ( [ pft ] ) as peliti s field theory . it is important to point out that the field @xmath2 is not the density of @xmath12 particles , but a more abstract field @xcite which expectation value coincides with that of the real density field , and which higher order moments can be also related to higher order moments of the density field . all this will become clearer in a forthcoming section where the explicit derivation of eq . ( [ pft ] ) is performed . + before concluding this epigraph let us point out that there is a class of systems which present an active as well as an absorbing phase with a phase transition separating both of them , and which associated noise should also present a complex structure : this is the so called _ parity - conserving universality class _ @xcite . due to the extra conservation law it is clear by now that these systems are not in the rft universality class . 2 . a new universality class characterized by a noise different from that of the previously described classes has recently been elucidated ( see @xcite and references therein ) : the _ multiplicative noise _ ( mn ) universality class . while in the langevin equation for the rft the noise amplitude is proportional to the square root of the field at each point , in the mn universality class the noise amplitude is proportional to the field itself , i.e. @xmath17 with @xmath2 being a density field at position @xmath3 and time @xmath4 , and @xmath7 a gaussian white noise . obviously this new type of noise is also compatible with the presence of an absorbing state at @xmath9 at which the dynamics is completely suppressed . while there are many microscopic reaction - diffusion models belonging in the dp universality class , and is also easy to identify reaction - diffusion models in the peliti s field theory , no microscopic reaction - diffusion model in the mn universality class has been identified so far . the possibility of constructing a reaction - diffusion model that exhibits the rather striking properties of mn @xcite has been explored in a recent paper by howard and tuber @xcite . they concluded that given the apparent impossibility of finding such type of model in the mn class , the physical meaning of that universality class is unclear . motivated by the previous work @xcite we have further investigate this issue . in what follows we present a comparison of the different noise terms appearing in langevin equations for reggeon field theory , the multiplicative noise , and peliti s field theory , to identify physical differences among them . a simple and intuitive justification of the fact that no reaction - diffusion system can be found in the mn universality class is given . alternatively , we enumerate some other discrete , microscopic models belonging to that class . we also discuss some curious properties of systems with complex noise , and analyze whether a real langevin equation can be written for systems like the annihilation reaction @xmath18 .
recently , a hidden charm resonance named @xmath0 was observed by belle collaboration in the analysis of the @xmath5 process @xcite . the mass and natural width of this resonance are measured to be @xmath6 and @xmath7 , respectively . the product of its two - photon decay width and branching fraction to @xmath8 is @xmath9 for @xmath10 , or @xmath11 for @xmath12 . in literature , the structure of @xmath0 has been proposed to be @xmath13 teraquark state with @xmath14 @xcite , @xmath15 molecular state @xcite and @xmath16-wave charmonium state @xmath17 @xcite . and concerning the quantum numbers of the final states @xmath18 , @xmath0 can also have quantum numbers @xmath19 . in ref . @xcite , it was shown that @xmath0 can not be a @xmath20 exotic @xmath15 molecular state . in this paper , we will accept it as a bound state of @xmath15 to study its strong and electromagnetic decays in the effective lagrangian approach in case of @xmath21 and @xmath2 . since the mass of @xmath0 is about @xmath22mev below the threshold of @xmath3 ( @xmath23mev and @xmath24mev @xcite ) , it is reasonable to regard @xmath0 as a bound state of @xmath25 . and because the quantum numbers of @xmath26 and @xmath27 are @xmath28 and @xmath29 respectively , to form a bound state with quantum numbers @xmath30 or @xmath2 , the coupling between @xmath0 and its constituents should be @xmath31wave . to determine the effective coupling constant between @xmath0 and it constituents @xmath15 , as in our previous work ( for example ref . @xcite ) , we resort to the compositeness condition @xmath32(@xmath33 as the wave function renormalization constant of @xmath0 ) which was early used by nuclear physicists @xcite and is being widely used by particle physicists(see the references in @xcite ) . recently , this method has been applied to study the properties of some `` exotic '' hadrons @xcite and some conclusions were yielded comparing with data . for other interactions , we write down the general effective lagrangian and determine the coupling constants with help of data , theoretical calculation , @xmath34 relation or the vector meson dominance ( vmd ) . as in our previous work @xcite , we introduce a correlation function including a scale parameter @xmath35 to illustrate the distribution of the constituents in the bound state @xmath0 . the parameter @xmath35 is varied to find the physical region where the data can be understood . in the physical region of @xmath35 , the partial widthes for strong and electromagnetic decays are yielded . this paper is organized as the following : in section [ sec : tf ] we will provide the theoretical framework used in this paper . we will present the analytic forms for the radiative and strong decay matrix elements and partial widths of @xmath0 in section [ sec : decay ] . and , the last section is our numerical results and discussions . in this section , we will propose the theoretical framework for the calculation of the strong and electromagnetic decays of @xmath0 . as was mentioned above , we regard @xmath0 as a @xmath36 bound state . and concerning the experimental status , we accept the quantum numbers of @xmath0 as @xmath37 and @xmath2 . for scalar case , one can write the free lagrangian of @xmath0 as @xmath38 with @xmath39 as the mass of @xmath0 . the propagator of @xmath0 can be easily written as @xmath40 which satisfies @xmath41 while for tensor resonance we have the free lagrangian as @xcite @xmath42 where the symmetric tensor @xmath43 denotes the @xmath14 field for @xmath0 and @xmath44 the propagator for @xmath45 is obtained as @xmath46 with respect to the discussions given in first section , one can write the effective lagrangian describing the interaction between @xmath0 and @xmath47 as @xmath48\nonumber\\ & & \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\ ; \times \phi_x((x_1-x_2)^2)\delta(x-\omega_v x_1 - \omega_sx_2 ) \ ; , \label{effelcomp}\end{aligned}\ ] ] where @xmath49 is for scalar resonance case while @xmath50 is for tensor resonance case . @xmath51 and @xmath52 are the effective coupling constants for the interaction between @xmath0 and @xmath3 in scalar and tensor resonance cases , respectively . @xmath53 and @xmath54 are mass ratios which are defined as @xmath55 @xmath56 is a correlation function which illustrates the distribution of the constituents in the bound state . fourier transform of the correlation function reads @xmath57 to write down lagrangian ( [ effelcomp ] ) , for simplicity , we have defined the tensor @xmath58 as a function of the constituents with the explicit form @xmath59 the coupling constants @xmath51 and @xmath52 can be determined with help of the compositeness condition @xmath60 with @xmath61 as the wave function renormalization constant of @xmath0 which is defined as the residual of @xmath0 propagator , i.e. , @xmath62 where @xmath63 for scalar case while @xmath64 for tensor case . for scalar resonance @xmath0 , @xmath65 is its mass operator . but for tensor resonance @xmath0 , @xmath66 relates to its mass operator via relation @xmath67 where @xmath68 denotes terms do not contribute to the mass renormalization of @xmath0 . the mass operator of @xmath0 is illustrated by fig . [ fig : massoperator ] . . ] concerning the feynman diagram depicted fig . [ fig : massoperator ] one can calculate the mass operator explicitly . to get the numerical result of the coupling constant @xmath69 , an explicit form of @xmath70 is necessary . throughout this paper , we take the gaussian form @xmath71 where the size parameter @xmath35 parametrizes the distribution of the constituents inside the molecule . in the following calculation , we will find the physical value of @xmath35 by comparing our calculation of the product of @xmath0 to two - photon partial width and the branching fraction to @xmath72 . it should be noted that choice ( [ gaussian ] ) is not unique . in principle any choice of @xmath70 , as long as it renders the integral convergent sufficiently fast in the ultraviolet region , is reasonable . in this sense , @xmath70 can be regarded as a regulator which makes the ultraviolet divergent integral well defined . with these discussions , we can calculate the effective coupling constant @xmath69 numerically . in the typical nonperturbative region @xmath73gev , using the central value of @xmath0 mass , our numerical result is found to be @xmath74 in fig . [ fig : gxlambda ] we plot the @xmath35 dependence of the coupling constants . one can see that both coupling constants decrease against @xmath35 . this can be understood from the momentum integral of the mass operator . for scalar @xmath0 , the loop integral is quadratically divergent so the derivative of the mass operator which proportional to the inverse of @xmath75 increases against @xmath35 which means the coupling constant @xmath51 decreases against @xmath76 . similar argument can be given for @xmath52 . dependence of the coupling constant @xmath69 . ] the effective lagrangian for the study of strong and electromagnetic decays of @xmath0 consists of two parts : the electromagnetic part @xmath77 and the strong part @xmath78 . the electromagnetic interaction lagrangian @xmath77 includes five parts : @xmath79 from the gauge of the nonlocal and derivative coupling of eq . ( [ effelcomp ] ) , @xmath80 from the gauge of the kinetic term of the charged constituents @xmath26 and @xmath27 , the electromagnetic interaction lagrangian @xmath81 including @xmath26 and @xmath82 , @xmath83 for electromagnetic interaction including @xmath84 and @xmath82 and @xmath85 for electromagnetic interaction including @xmath84 and @xmath26 . one can write @xmath79 by substituting @xmath58 with @xmath86 in eq . ( [ effelcomp ] ) with @xmath87 where the wilson s line @xmath88 is defined as @xmath89 in our following calculation , the nonlocal vertex with one - photon is necessary . the nonlocal vertex with one - photon comes from two sources : one is from covariant derivative and another one is from the expansion of the wilson s line . one can easily derive the feynman rule for the nonlocal vertex with one - photon which comes from the covariant derivative . but to derive the feynman rule for photon from wilson s line , one may use the path - independent prescription suggested in @xcite . the electromagnetic vertex @xmath80 from the gauge of the kinetic terms of the charged constituents can be easily written as @xmath90 \ ; .\end{aligned}\ ] ] one can generally write the effective lagrangian @xmath91 for electromagnetic interaction including @xmath26 and @xmath82 as @xmath92 f_{\mu\nu } \ ; . \label{effesem}\end{aligned}\ ] ] where @xmath93 is the gauged field strength tensor for @xmath94 with definition @xmath95 . and similarly , the general effective lagrangian @xmath83 and @xmath85 can be written as @xmath96f_{\alpha\beta } \ ; , \nonumber\\ { \cal l}_{\rm em}^{as } & = & - i eg_{d_{s1}d_{s0}^{\ast}\gamma}\epsilon_{\mu\nu\alpha\beta } [ d_{s0}^{\ast \ , + } \tilde{d}_{s1 ; \,\mu\nu}^{- } - d_{s0}^{\ast \ , - } \tilde{d}_{s1 ; \,\mu\nu}^{+ } ] f_{\alpha \beta } \ ; .\label{effeaem}\end{aligned}\ ] ] similar as the definition of @xmath93 , we have defined the gauged field strength tensor for @xmath97 with definition @xmath98 . the relevant coupling constants can be determined phenomenologically . confined by the experimental status , one can not fix @xmath99 from data , so we turn to the theoretical calculations ( for example ref . @xcite and references therein ) . from literature , one can see that the minimal result of the theoretical calculation of @xmath100 decay width is @xmath101kev . from this decay width , we get @xmath102gev@xmath103 . the coupling constants @xmath104 and @xmath105 can be determined by using the hqet and branching ratio for relevant processes . first , consider the decay of @xmath106 , the effective lagrangian can be written as @xmath107f_{\mu\nu } \ ; , \end{aligned}\ ] ] where @xmath108 and @xmath109 . from this lagrangian , one can express the decay width as @xmath110 the numerical result of the decay width has been evaluated by several groups . from the references given in ref . @xcite , we see all the results are larger than @xmath111kev . so that we have @xmath112gev@xmath103 . the coupling constant @xmath113 relates to @xmath114 via hqet as @xmath115 so that we have @xmath116 . the coupling constant @xmath117 can be determined by using the relevant branching ratio given in pdg @xcite . from ( [ effeaem ] ) we have @xmath118 using the central value of the branching ratio we have @xmath119 which leads to @xmath120gev@xmath103 . in addition to the lagrangian ( [ effelcomp ] ) , the strong part @xmath78 involves @xmath121-type lagrangian describing the interaction of three vector mesons , the @xmath122-type lagrangian describing the interaction of one scalar meson with two vector mesons , @xmath123-type lagrangian describing the interaction of two scalar mesons with one vector meson , @xmath124-type lagrangian for the interaction of axial - vector with two vector meosns and @xmath125-type lagrangian for axial - vector - scalar - vector meson interaction , i.e. , @xmath126\nonumber\\ & & + ig_{_{\phi d_s^\ast d_s^\ast}}[d_{s ; \,\mu}^{\ast \ , -}(d_{s ; \ , \nu}^{\ast \ , + } \partial^{^{^{\hspace{-0.2cm}\leftrightarrow}}}_\mu \phi_\nu ) + d_{s ; \ , \mu}^{\ast \ , + } ( \phi_\nu\partial^{^{^{\hspace{-0.2cm}\leftrightarrow}}}_\mu d_{s ; \ , \nu}^{\ast \ , - } ) + \phi_\mu ( d_{s ; \ , \nu}^{\ast \ , -}\partial^{^{^{\hspace{-0.2cm}\leftrightarrow}}}_\mu d_{s ; \ , \nu}^{\ast \ , + } ) ] \ ; , \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \cal l}_{\rm str}^{svv } & = & g_{_{\psi d_{s0}^{\ast}d_s^{\ast}}}[d_{s0}^{\ast \ , -}d_{s ; \,\mu\nu}^{\ast \ , + } - d_{s0}^{\ast \ , + } d_{s ; \,\mu\nu}^{\ast \ , -}]\psi_{\mu\nu } + g_{_{\phi d_{s0}^{\ast}d_s^{\ast}}}[d_{s0}^{\ast \ , -}d_{s ; \,\mu\nu}^{\ast \ , + } - d_{s0}^{\ast \ , + } d_{s ; \,\mu\nu}^{\ast \ , -}]\phi_{\mu\nu } \ ; , \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \cal l}_{\rm str}^{ssv } & = & - ig_{_{\psi d_{s0}^{\ast } d_{s0}^{\ast}}}\psi_\mu(d_{s0}^{\ast \ , -}\partial^{^{^{\hspace{-0.2cm}\leftrightarrow}}}_\mu d_{s0}^{\ast \ , + } ) - ig_{_{\phi d_{s0}^{\ast } d_{s0}^{\ast}}}\phi_\mu(d_{s0}^{\ast \ , -}\partial^{^{^{\hspace{-0.2cm}\leftrightarrow}}}_\mu d_{s0}^{\ast \ , + } ) \ ; , \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \cal l}_{\rm str}^{avv } & = & - g_{\psi d_s^\ast d_{s1}}\epsilon_{\mu\nu\alpha\beta } [ d_{s1 ; \,\mu}^- d_{s ; \ , \nu}^{\ast \ , + } - d_{s1 ; \,\mu}^+ d_{s ; \ , \nu}^{\ast \ , -}]\psi_{\alpha\beta } \nonumber\\ & & - g_{\phi d_s^\ast d_{s1}}\epsilon_{\mu\nu\alpha\beta } [ d_{s1 ; \,\mu}^- d_{s ; \ , \nu}^{\ast \ , + } - d_{s1 ; \,\mu}^+ d_{s ; \ , \nu}^{\ast \ , -}]\phi_{\alpha\beta } \ ; , \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \cal l}_{\rm str}^{asv } & = & - i g_{\psi d_{s0}^\ast d_{s1}}\epsilon_{\mu\nu\lambda\sigma}[d_{s0}^{\ast \ , -}\psi_{\mu\nu}d_{s1 ; \,\lambda\sigma}^+ - d_{s0}^{\ast \ , + } \psi_{\mu\nu}d_{s1 ; \,\lambda\sigma}^-]\nonumber\\ & & - i g_{\phi d_{s0}^\ast d_{s1}}\epsilon_{\mu\nu\lambda\sigma}[d_{s0}^{\ast \ , -}\phi_{\mu\nu}d_{s1 ; \,\lambda\sigma}^+ - d_{s0}^{\ast \ , + } \phi_{\mu\nu}d_{s1 ; \,\lambda\sigma}^- ] \ ; . \label{effelast}\end{aligned}\ ] ] because of our less knowledge , we can not determine these coupling constants from data . here , we resort to the vector meson dominance ( vmd ) model @xcite . in the vmd model , the virtual photon in the process @xmath127 is coupled to vector mesons @xmath128 and @xmath129 , which are then coupled to @xmath130 . for zero momentum transfer , one has relation @xmath131 where @xmath132 is the photon - vector - meson mixing amplitude @xmath133 which can be determined from @xmath134 decay width , i.e. , @xmath135 where we did not include electron mass since it is much smaller than vector meson mass . for @xmath128 meson , using @xmath136mev @xcite we have @xmath137mev@xmath138 , while @xmath139mev@xmath140 when @xmath141kev @xcite is applied . concerning that the virtual photon sees the charge of charm quark in @xmath142 meson through @xmath143 coupling and the charge of anti - strange quark in @xmath144 meson through @xmath145 coupling , we have relations @xmath146 from these relations we have @xmath147 and @xmath148 . to determine coupling constants @xmath149 , we make extension to the vmd model used above , i.e. , substituting eq . ( [ vmdrelation ] ) with @xmath150 where @xmath151 and @xmath152 denote the relevant charmed - strange mesons . similarly , eqs . ( [ psiphivmd ] ) should also be extended to @xmath153 from which we yield the relevant coupling constants as @xmath154 to fix the magnitude of coupling constant @xmath155 , we resort to the @xmath34 relation as was used in ref . @xcite from which we have relations @xmath156 to fix the relative signs for the relevant effective lagrangian , one can use hhchpt including @xmath26 and @xmath84 mesons @xcite . but even with this consideration , the relative signs of @xmath85 and @xmath157 to the other terms can not be determined . we leave this as an ambiguity and discuss different cases in the following calculation . in summary , our framework of the interaction lagrangian is @xmath158 \ , . \label{lagint}\end{aligned}\ ] ] up to now , we have fixed all the coupling constants that are necessary for our following calculation of the electromagnetic and strong decays of @xmath0 . in this section , we will present the general forms of the matrix elements and partial widths for the electromagnetic and strong decays of @xmath0 and the feynman diagrams included in our calculation . the four kinds of diagrams depicted in fig . [ fig : em ] and their corresponding crossing ones should be taken into account to study @xmath159 decay . diagrams @xmath160 and @xmath161 are from the final state interaction due to the exchange of @xmath27 , @xmath84 and @xmath26 , diagram @xmath162 arises from the gauge of the nonlocal and derivative coupling between @xmath0 and its constituents @xmath47 but diagram @xmath163 is from the lagrangian ( [ effesem ] ) . ( cross diagrams should be included ) . ] for @xmath0 with quantum numbers @xmath30 , concerning the @xmath164 gauge invariance and the transverseness of the photon polarization vector , one can write down the matrix element for the decay of @xmath165 as @xmath166\epsilon_\alpha(q_1)\epsilon_\beta(q_2 ) \;. \label{matrixems}\end{aligned}\ ] ] while for tensor meson @xmath0 with quantum numbers @xmath167 , its polarization vector satisfies @xmath168 and @xmath169 , so that the matrix element for electromagnetic decay can be written as @xcite @xmath170\frac{q_\mu q_\nu}{q^2 } \label{matrixemt}\\ & & + f^{(2)}_{t \to 2\gamma}\big[(g_{\mu\alpha } - \frac{q_\mu q _ { \alpha}}{q^2})(g_{\nu\beta } - \frac{q_\nu q_{\beta}}{q^2 } ) + ( g_{\mu\beta } - \frac{q_\mu q_{\beta}}{q^2})(g_{\nu\alpha } - \frac{q_\nu q_{\alpha}}{q^2})\big]\big\}\epsilon_{\mu\nu}(p)\epsilon_\alpha(q_1)\epsilon_\beta(q_2 ) \ ; . \nonumber\end{aligned}\ ] ] where @xmath171 . from eqs . ( [ matrixems],[matrixemt ] ) we express the decay width for @xmath0 as @xmath172 where the subindices `` s '' and `` t '' denote the scalar and tensor resonance @xmath0 , respectively . to get the last equation , we have applied the sum of the polarization vector for tensor meson @xcite @xmath173 where @xmath174 . we should take into account the feynman diagrams illustrated in fig . [ fig : strong ] in the study of the strong decay of @xmath175 . furthermore , in addition to these diagrams , their crossing ones should also be included . ( crossing diagrams should be included ) . ] compared to the electromagnetic case , the expression for the matrix element of strong decay is more complicated because the constraint from the gauge invariance is released . when @xmath0 is regarded as a scalar resonance , the matrix element for the strong decay of @xmath176 can be written as @xmath177\epsilon_\alpha(q_1)\epsilon_\beta(q_2 ) \ ; . \label{matrixstrongs}\end{aligned}\ ] ] one can show that , when the gauge invariance is imposed , @xmath178 so expression ( [ matrixstrongs ] ) becomes ( [ matrixems ] ) . similarly , without the constraint from the gauge invariance , in the tensor case , one can write the matrix element for the strong decay of @xmath179 as @xmath180\epsilon_{\mu\nu}(p)\epsilon_\alpha(q_1)\epsilon_\beta(q_2 ) \ ; .\end{aligned}\ ] ] one can prove that when the final vector mesons are both massless particles and the gauge invariance is imposed the following relations can be reduced @xmath181 so expression ( [ matrixemt ] ) for electromagnetic decay matrix element can be yielded . with the help of ( [ sumtensorvect ] ) one can get the analytic forms for the strong decay as @xmath182 -2g_{_{x_s \to v_1v_2}}f_{_{x_s \to v_1v_2}}\big[1-\omega^2\big ] + f_{_{x_s \to v_1v_2}}^2\big[\omega - \frac{1}{\omega}\big]^2\big\ } \ ; , \nonumber\\ \gamma_{\rm t}(x \to j/\psi\phi ) & = & \frac{1}{80\pi m_x^3}\lambda^{1/2}(m_x^2,m_\psi^2,m_\phi^2)\sum_{i\geq j=1}^{5}\big\{c_{ij}f_{x_t \to v_1v_2}^{(i)}f_{x_t \to v_1v_2}^{(j)}\big\ } \ ; , \label{expressionwidthts}\end{aligned}\ ] ] where @xmath183 . @xmath184 is the k@xmath185llen function and @xmath186 are functions of the relevant masses of initial and final states which will be given in appendix . with these discussions , the numerical calculation can be performed via standard loop derivation . since the magnitude of @xmath35 is unknown , we vary its magnitude from @xmath187gev to @xmath188gev to find its physical region where the data can be understood . in our estimate , we use the cental value of the total width , i.e. , @xmath189mev . and , because it is difficult to determine the relative signs between @xmath85 and @xmath190 and other terms , we will consider two cases when we do our numerical calculation , i.e. , the last two terms of eq . ( [ lagint ] ) give positive and negative contributions to the total lagrangian . our results are summarized in tables . [ table : scalarp ] and [ table : scalarm ] . from the numerical results , one can see that the possibility that @xmath0 is a molecular state of @xmath25 can not be ruled out in our model . in the case that @xmath0 has quantum numbers @xmath30 , the physical region of @xmath35 is smaller than the tensor resonance case which means the size of scalar @xmath0 is bigger than the tensor one . we would like to point out that , because we used the minimal values of the theoretical calculation of coupling constants @xmath99 and @xmath191 , our final results about the partial widths can be regarded as lower limit . this is an ambiguity of the present calculation . in fact , the best way to determine these coupling constants is from data , but because of the precision of the data , we can not along this way . when the magnitudes of coupling constants @xmath99 and @xmath191 are improved , the theoretical results of the product of the two - photon decay width and branch fraction to @xmath72 should be larger than the present conclusion . in this case , compared to the tensor @xmath0 , the typical region of @xmath35 for scalar resonance can be reduced to an unphysically small region so one can first rule out the possibility of a scalar molecule . another ambiguity in our calculation of the product of the two - photon decay width and branch fraction to @xmath72 is from the total width of @xmath0 . here we apply the central value , i.e. , @xmath192mev . when a larger total width is applied , the physical region of @xmath35 can be enlarged . but this does not effect the partial widths for strong and electromagnetic decays we predict in the corresponding region of @xmath35 . finally , we conclude that , with the present data and in the framework our model , @xmath0 can be interpreted as @xmath193 molecule . .[table : scalarp ] our numerical results in case of positive sign of eq . ( [ lagint ] ) . [ cols="<,<,<,<,<",options="header " , ] in this appendix , i will present the coefficients @xmath186 in formula ( [ expressionwidthts ] ) . @xmath194 \ ; , \nonumber\\ c_{12 } & = & \frac{-1}{12m^4m_1 ^ 2m_2 ^ 2}\big[\lambda\big(5\lambda(m^2+m_1 ^ 2+m_2 ^ 2 ) + 24 m^2 m_1 ^ 2m_2 ^ 2\big)\big ] \ ; , \nonumber\\ c_{13 } & = & \frac{1}{12m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^2\big((m^2-m_2 ^ 2)^2-m_1 ^ 4\big)\big ] \ ; , \nonumber\\ c_{14 } & = & \frac{-1}{12m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^2\big((m^2-m_1 ^ 2)^2-m_2 ^ 4\big)\big ] \ ; , \nonumber\\ c_{15 } & = & \frac{1}{24m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^3(m^2 - m_1 ^ 2 - m_2 ^ 2)\big ] \ ; , \nonumber\\ c_{22 } & = & \frac{1}{24m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^2 + 44m^2(m_1 ^ 2+m_2 ^ 2)\lambda + 528 m_1 ^ 2m_2 ^ 2 m^4 \big ] \ ; , \nonumber\\ c_{23 } & = & \frac{-1}{12m^4m_1 ^ 2m_2 ^ 2}\big[\lambda(5\lambda + 44 m^2m_1 ^ 2)(m^2-m_1 ^ 2 + m_2 ^ 2)\big ] \ ; , \nonumber\\ c_{24 } & = & \frac{1}{12m^4m_1 ^ 2m_2 ^ 2}\big[\lambda(5\lambda + 44 m^2m_2 ^ 2)(m^2-m_2 ^ 2 + m_1 ^ 2)\big ] \ ; , \nonumber\\ c_{25 } & = & \frac{-1}{24m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^2\big(m^4 - ( m_1 ^ 2 - m_2 ^ 2)^2\big)\big ] \ ; , \nonumber\\ c_{33 } & = & \frac{1}{24m^4m_1 ^ 2m_2 ^ 2}\big[\lambda^2(5\lambda + 44 m^2m_1 ^ 2)\big ] \ ; , \nonumber\\ c_{34 } & = & \frac{-1}{12m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^2\big(m^4-(m_1 ^ 2 - m_2 ^ 2)^2\big)\big ] \ ; , \nonumber\\ c_{35 } & = & \frac{1}{24m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^3(m^2-m_2 ^ 2 + m_1 ^ 2)\big ] \ ; , \nonumber\\ c_{44 } & = & \frac{1}{24m^4m_1 ^ 2m_2 ^ 2}\big[\lambda^2(5\lambda + 44 m^2m_2 ^ 2)\big ] \ ; , \nonumber\\ c_{45 } & = & \frac{-1}{24m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^3(m^2-m_1 ^ 2 + m_2 ^ 2)\big ] \ ; , \nonumber\\ c_{55 } & = & \frac{1}{96m^4m_1 ^ 2m_2 ^ 2}\big[5\lambda^4\big ] \ ; , \end{aligned}\ ] ] where @xmath195 is the k@xmath185llen function and @xmath196 . i would like to thank prof . yu - bing dong from ihep in beijing and prof . m.harada in nagoya university for their valuable discussions and comments . this work is supported in part by the national science foundation of china ( nnsfc ) under grant no . 10905060 and grant - in - aid for scientific research on innovative areas ( no . 2104 ) `` quest on new hadrons with variety of flavors '' from mext . r. m. albuquerque , j. m. dias and m. nielsen , arxiv:1001.3092 [ hep - ph ] . c. amsler _ et al . _ [ particle data group ] , phys . b * 667 * , 1 ( 2008 ) . a. faessler , t. gutsche , v. e. lyubovitskij and y. l. ma , phys . d * 76 * , 014005 ( 2007 ) [ arxiv:0705.0254 [ hep - ph ] ] . s. weinberg , phys . * 130 * , 776 ( 1963 ) ; a. salam , nuovo cim . * 25 * , 224 ( 1962 ) ; y. b. dong , a. faessler , t. gutsche and v. e. lyubovitskij , phys . d * 77 * , 094013 ( 2008 ) [ arxiv:0802.3610 [ hep - ph ] ] . f. giacosa , t. gutsche and v. e. lyubovitskij , phys . d * 77 * , 034007 ( 2008 ) [ arxiv:0710.3403 [ hep - ph ] ] . a. faessler , t. gutsche , v. e. lyubovitskij and y. l. ma , arxiv:0801.2232 [ hep - ph ] . y. l. ma , j. phys . g * 36 * , 055004 ( 2009 ) [ arxiv:0808.3764 [ hep - ph ] ] . s. bellucci , j. gasser and m. e. sainio , nucl . b * 423 * , 80 ( 1994 ) [ erratum - ibid . b * 431 * , 413 ( 1994 ) ] [ arxiv : hep - ph/9401206 ] . j. terning , phys . d * 44 * , 887 ( 1991 ) . s. mandelstam , annals phys . * 19 * , 25 ( 1962 ) . z. w. lin and c. m. ko , phys . c * 62 * , 034903 ( 2000 ) [ arxiv : nucl - th/9912046 ] . r. casalbuoni , a. deandrea , n. di bartolomeo , r. gatto , f. feruglio and g. nardulli , phys . rept . * 281 * , 145 ( 1997 ) [ arxiv : hep - ph/9605342 ] . a. v. anisovich , v. v. anisovich and v. a. nikonov , eur . phys . j. a * 12 * , 103 ( 2001 ) [ arxiv : hep - ph/0108186 ] . a. v. anisovich , v. v. anisovich , m. a. matveev and v. a. nikonov , phys . nucl . * 66 * , 914 ( 2003 ) [ yad . fiz . * 66 * , 946 ( 2003 ) ] [ arxiv : hep - ph/0204330 ] . g. lopez castro and j. h. munoz , phys . rev . d * 55 * , 5581 ( 1997 ) [ arxiv : hep - ph/9702238 ] .	the strong and electromagnetic decays of @xmath0 with quantum numbers @xmath1 and @xmath2 have been studied by using the effective lagrangian approach . the coupling constant between @xmath0 and @xmath3 is determined with the help of the compositeness condition which means that @xmath0 is a bound state of @xmath3 . other coupling constants applied in the calculation are determined phenomenologically . our numerical results show that , using the present data within the present model , the possibility that @xmath0 is a @xmath3 molecule can not be ruled out . # 1@xmath4#1 # 1@xmath4#1
this work is supported by the engineering and physical sciences research council ( uk ) and the deutsche forschungsgemeinschaft in the frame of sfb 787 .	we image the micro - electroluminescence ( el ) spectra of self - assembled inas quantum dots ( qds ) embedded in the intrinsic region of a gaas _ p - i - n _ diode and demonstrate optical detection of resonant carrier injection into a single qd . resonant tunneling of electrons and holes into the qds at bias voltages below the flat - band condition leads to sharp el lines characteristic of individual qds , accompanied by a spatial fragmentation of the surface el emission into small and discrete light - emitting areas , each with its own spectral fingerprint and stark shift . we explain this behavior in terms of coulomb interaction effects and the selective excitation of a small number of qds within the ensemble due to preferential resonant tunneling paths for carriers . self - assembled inas quantum dots ( qds ) are an important model system for investigating the fundamental physics of quantum - confined electrons @xcite . the selective light emission from a small number of qds can be achieved , for example , by lithographically defined small - area diodes @xcite or apertures @xcite , or by the incorporation of qds into microcavities @xcite and nanowires @xcite . such studies have provided information about the electronic properties of the dots and form the basis for novel applications , e.g. optically @xcite and electrically @xcite driven sources of entangled photon pairs for quantum information processing . of particular interest is the possibility of generating sharp el emission lines from individual qds by voltage controlled electrical injection of carriers . despite many works on resonant tunneling injection of carriers in unipolar qd devices @xcite , the simultaneous resonant injection of both electrons and holes required for el emission from an individual qd has received less attention @xcite and is relevant to topical research on electrically driven single qd photon emitters @xcite . here we use micro - electroluminescence ( @xmath0el ) spectroscopy and imaging to investigate the optical emission from a single layer of self - assembled inas quantum dots embedded in the intrinsic region of a _ p - i - n _ light emitting diode . by gradually decreasing the applied bias below the ` flat band ' threshold voltage at which the bias balances the built - in potential in the @xmath1-region , we follow the evolution of the el spectra and the corresponding spatial form of the el emission . we show that resonant tunneling of electrons and holes into the qds at bias voltages below the flat band condition leads to sharp el lines which are characteristic of individual qds , accompanied by a spatial fragmentation of the diode emission into small and discrete light - emitting areas , each with its own spectral fingerprint and stark shift . we explain this behavior in terms of the selective excitation of a small number of qds within the ensemble due to the presence of preferential resonant tunneling paths for carriers . we also discuss the effect of qd charging and coulomb interactions on the resonant excitation of the single qd el emission . this demonstration of bias - controlled excitation of el from an individual qd within an ensemble of dots could be developed further for use in optoelectronic or quantum information applications . our _ p - i - n _ diodes were grown by molecular beam epitaxy on a @xmath2 gaas substrate , which forms the bottom electrical contact of the diode . the layer structure in order of growth on the substrate is as follows : @xmath3 nm and @xmath4 nm p - doped gaas layer with @xmath5@xmath6 and @xmath7@xmath6 , respectively ; a @xmath8 nm intrinsic gaas spacer layer ; a @xmath9 monolayer ( ml ) of inas , which gives rise to a wetting layer ( wl ) and qds with a density of about @xmath10@xmath11 . the qd layer is covered by @xmath12 nm of intrinsic gaas followed by @xmath4 nm n - doped gaas ( @xmath13@xmath6 ) and a @xmath14 nm gaas top layer with @xmath15@xmath6 . the diodes are defined by wet - chemical etching and a ring - shaped gold electrode forms the top electrical contact and provides optical access . here we focus on large area devices with @xmath16 m diameter , containing @xmath17 qds . a schematic band diagram for a bias @xmath18 below the flat band condition , i.e. for @xmath19v , is shown in the inset of fig . 1a . the @xmath0el spectra were recorded at @xmath20k with a spectral resolution of @xmath21ev and a focal spot diameter of about @xmath22 m . the spatial maps of the @xmath0el were recorded by scanning the focusing mirror along the mesa . . inset : schematic diagram of the resonant injection of electrons and holes into a qd . ( b ) colorscale image of the normalized el intensity as a function of photon energy and bias . the dashed lines represent the conditions @xmath23 , @xmath24mev and @xmath25mev . ( c ) and ( d ) @xmath0el spectra showing the fragmentation of the continuous el spectrum into a sharp emission line.,title="fig : " ] figure 1(a ) shows the el spectra for a range of applied biases @xmath18 . as @xmath18 is decreased below the flat band condition , the el spectrum narrows and evolves into single sharp el lines . to track the evolution of the el spectrum with @xmath18 , we show in fig . 1(b ) a color - scale plot of the normalized el intensity versus @xmath18 and the photon energy , @xmath26 . this reveals clearly the narrowing of the qd emission with decreasing bias and the emergence of two distinct el features , labeled a and b and indicated in fig . 1b by dashed lines . a and b both shift to lower energies with decreasing bias with peak energies that run parallel to the line @xmath23 . the energy difference between the excitation energy @xmath27 and the peak energy of each feature is @xmath28mev and @xmath29mev , respectively . this indicates that electron - hole pairs electrically injected into different qds relax approximatively the same energy prior to the radiative recombination in the ground state . we now focus on the @xmath0el spectra in the energy interval @xmath30ev at low applied bias . figure 1c and 1d show the @xmath0el spectrum at a given position on the mesa at different @xmath18 . with decreasing @xmath18 the spectrum evolves from a broad continuous emission into a fragmented spectrum . at @xmath31v the emission consists of a single sharp line only , with a full width at half maximum of @xmath32ev . this linewidth is not determined by the resolution of the spectrometer and is similar to the typical linewidths reported for the low temperature photoluminescence emission of individual inas qds @xcite . this method of el excitation allows the study of such sharp emission lines over several orders of magnitudes in intensity and also reveals a characteristic exponential acoustic phonon broadening and a weak but sharp phonon replica peak , which are reported elsewhere @xcite . the fragmentation of the el spectrum into sharp lines is accompanied by a spatial fragmentation of the el emission . to demonstrate this effect , we recorded @xmath0el spectra at each position of a square grid covering roughly @xmath33 of the diode surface . figure 2 shows spatial maps of the maximum intensity of each spectrum as a function of position for a series of bias voltages . for each image , a scale factor relates the maximum of the colorscale to the maximum at @xmath34v . at @xmath34v the emission is essentially homogeneous across the scan @xcite . at a slightly lower bias , @xmath35v , the diode emission starts to break up into a series of bright spots dominating over a homogeneous background emission . at @xmath36v , the background intensity weakens and several bright spots emerge at distinct positions with a uniform spot size determined by the diameter of the focal spot of the collecting lens . at this voltage the el spectra fragment into sharp lines . at @xmath37v , the number of bright spots decreases and the relative intensities of the individual spots change compared to the image at @xmath36v . the spectra now consist of individual emission lines with no background el . at @xmath38v , the maximum intensities are much lower and the number of visible emission spots is reduced to four . at @xmath31v , only one bright spot is visible and the corresponding spectrum , shown in fig . 1d , consists of a single sharp emission line with an intensity similar to the maximum at @xmath37v . we note that no other el lines could be observed at this bias , suggesting that this emission center originates from a single qd . el intensity for a range of applied biases . for a given bias the el spectrum is recorded for a grid of positions . the images show the maximum intensity of each spectrum as a function of the position on the diode . the number at the bottom right of each image denotes the factor by which the el intensity is scaled compared to the first image.,title="fig : " ] el intensity maps at different photon energies , given by arrows of the same color in ( b ) . the symbols p1-p7 mark the position of the maximum el intensity at this energy . features with intensities below the threshold of 10% of the maximum in a scan are omitted for clarity . ( b ) el spectra recorded for @xmath37v at the positions p1-p7 given in ( a ) . the el spectra are offset for clarity.,title="fig : " ] the el spectra corresponding to the spatial el maps are presented in fig . 3 for @xmath37v . figure 3a shows maps of the normalized el intensity as a function of position at specific photon energies . the individual scans are distinguished by different colors and the corresponding energies are indicated by arrows in the @xmath0el spectra of fig . 3b . the bright spots in fig . 3a and the corresponding el spectra of fig . 3b are labeled as p1-p7 . each spot has approximately the same size and circular shape as the one that gives rise to the sharp el emission in fig . 1d and has a unique spectral fingerprint consisting of a small number of sharp lines . in turn , each emission line originates from exactly one bright spot in the spatial scans @xcite . these observations are consistent with the selective excitation of a small number of dots by carriers tunneling in the diode area of the bright spots . figure 4a shows the bias dependence of the peak intensity of the emission line l1 of fig . 1d at the energy @xmath39ev . the el intensity exhibits sharp peaks at bias voltages of @xmath40v and @xmath41v . these resonances are a clear manifestation of the voltage - tunable resonant tunneling excitation of a single qd detected by optical means . the resonant injection of carriers from the doped contact layers into discrete excited states of the qd is followed by energy relaxation of the carriers into the ground state and radiative recombination . the two bias resonances are separated in energy by @xmath42mev and the differences from the ground state are @xmath43mev and @xmath44mev , both typical values for the quantum confinement energies of self - assembled inas qds . at large biases ( @xmath45v ) the strong growth of the el intensity arises from an increasing contribution of the emission from the qd ensemble , which is excited through carrier injection and redistribution into the extended states of the wetting layer and of the gaas matrix as the flat band condition is approached @xcite . we find resonances on many other el lines in this spectral range , though the number of observed resonances and their amplitudes can differ , as shown for the lines l4 and l5 in fig . 4a . the emission energy of individual el lines depends on the applied bias , as can be seen in fig . 4b , where the energy shift @xmath46 relative to the value at @xmath47v is plotted for several lines l1-l5 . we attribute this bias dependence to the quantum - confined stark effect in the qd @xcite . of particular interest is that the energies of some of the lines remain constant over certain bias ranges . for example , the l1 line does not shift between @xmath48v and @xmath49v , suggesting that the electric field in the intrinsic region remains constant . we propose that resonant tunneling gives rise to an increase in the average charge density in the qd layer , thus screening the local electric field , an effect analogous to the charge build - up effect reported previously for resonant tunneling quantum well diodes @xcite . also note that , although the stark shifts for various el lines differ from each other , the rates of shift with bias are very similar . this indicates that the different bias dependences arise from mesoscopic variations of the potential landscape rather than from differences in the electronic properties of the qds . our data demonstrate that at low temperatures the qd el below the flat band condition is excited by resonant tunneling injection of carriers . the electrons and holes in the doped @xmath50- and @xmath51-type layers adjacent to the @xmath1-region of the diode have an energy spread given by the respective fermi energies ( @xmath52mev ) . since both the electron and hole states have to be aligned with the fermi seas to resonantly excite el , one would expect considerably narrower bias resonances than the ones reported here . on the other hand , the coulomb interaction of a charged qd with the fermi seas of the contacts and with its nearest neighbor qds can provide additional tunneling pathways , thus accounting for the relatively large widths ( @xmath53mv ) of the resonances in fig . 4a . we note that this broadening is significantly larger than the width of individual qd el lines ( @xmath32ev ) . though the number of active qds is constrained by the conditions of resonant tunneling , this constraint is not sufficient to explain the pronounced spectral and spatial fragmentation of the el emission revealed in our study . existing theoretical models @xcite do not predict such an effect either . the high density and uniform distribution of qds ensures that even at the lowest bias several dots could be active . hence we conclude that some qds are coupled more strongly to the reservoirs than others , which is supported by the different number of observable resonances for different qds shown in fig . 4a . since the emission energies of the investigated qds are very similar , this variation of the coupling is likely to be due to local variations of the tunnel distance and barrier height . in previous studies , @xmath0el maps of the emission from the ridge - waveguide regions of ingan quantum well based leds have revealed spatial inhomogeneities due to non - uniform carrier injection caused by crystal degradation @xcite . in our diodes , the el spectra are stable with time , but preferential tunneling paths may arise from mesoscopic fluctuations of the @xmath50- and @xmath51-doped interfaces due to randomly placed dopant atoms in or close to the intrinsic region , crystals defects or strain - related potential minima associated with the qds themselves . such variations would not only explain the spatial and spectral fragmentation of the el spectra , but could also account for some differences in the bias dependence of the stark shifts among various qds . in summary , we have demonstrated how the homogeneous broad band emission of a large quantum dot ensemble fragments into spatially strongly inhomogeneous sharp emission lines from individual quantum dots . each el line exhibits a distinct resonance behavior as a function of the applied bias and a unique stark shift . these effects can be explained in terms of the selective excitation of a small number of qds within the ensemble due to the presence of preferential resonant tunneling paths for carriers . our results provide direct evidence for the resonant and voltage tunable electrical injection of carriers into individual qds and are relevant for future implementation of such structures into electrically controlled single photon leds . in particular , the resonant tunneling excitation of a single dot could reduce quantum decoherence due to interactions with carriers occupying adjacent dots or higher energy continuum states .
tangle functors have been central in the construction of quantum invariants of links and three - manifolds since reshetikhin and turaev s pioneering work @xcite . the starting data is a quasitriangular hopf algebra . this is a hopf algebra @xmath13 along with an invertible element @xmath14 of a completion of @xmath15 that satisfies three equations . if @xmath16 , let @xmath17 , @xmath18 and @xmath19 . the equations are @xmath20 where @xmath21 and @xmath22 are the comultiplication and the flipped comultiplication in @xmath13 , and @xmath23 and @xmath24 an element @xmath25 satisfying the above properties is called a * universal @xmath14-matrix. these conditions imply that @xmath14 satisfies the quantum yang - baxter equation , @xmath26 and if @xmath27 denotes multiplication by @xmath14 followed by flipping the coordinates , then @xmath27 satisfies the standard braid relation , @xmath28 hence quasitriangular hopf algebras give rise to representations of the braid group . tangle functors are the extension of this representation to a representation of the category of framed colored tangles on the category of representations of the quasitriangular hopf algebra . the most important examples of quasitriangular hopf algebras are the drinfeld - jimbo deformations of the universal enveloping algebra of a semisimple lie algebra where the deformation parameter is a complex number that is not a root of unity @xcite . if there is a universal @xmath14-matrix , then the representations of the algebra commute under tensor product , that is if @xmath29 and @xmath30 are two representations of @xmath13 then @xmath31 is equivalent to @xmath32 . roots of unity need to be avoided because the representations of the algebra do not commute under tensor product . these noncommuting representations are sometimes called cyclic * representations . the center of the deformed algebra has a subalgebra @xmath33 of finite index which is a hopf algebra . in fact , @xmath33 is the coordinate ring of the poisson dual of the lie group underlying the original lie algebra . the cyclic representations are associated with points of a branched cover of the dual lie group @xcite . the cyclic representations are used to construct * quantum hyperbolic invariants * of three manifolds . their restriction to the positive part of the quantum group play a role in kashaev s invariant@xcite , the quantum hyperbolic invariants of baseilhac and benedetti @xcite , and implicitly in the representation theory of the kauffman bracket skein algebra @xcite . [ uq]let @xmath34 be a complex number . the algebra @xmath3 is generated by @xmath35 , @xmath36 and @xmath37 with relations , @xmath38 the counit , antipode and comultiplication extend as , @xmath39 @xmath40 @xmath41 this is the version of @xmath3 appearing in @xcite , and @xcite . as @xmath3 is not quasitriangular at roots of unity , something must be done in order to get a tangle functor . in @xcite , they work with a quotient of @xmath3 where @xmath42 and @xmath43 . in this quotient , @xmath44 and @xmath45 . this algebra is quasitriangular , and the @xmath14 matrix is , @xmath46!}q^{\frac{l(l-1)}{2}}e^l\otimes f^l,\ ] ] where @xmath47 . ohtsuki realized that the requirement that @xmath44 was superfluous @xcite , and the same @xmath14-matrix defines a tangle functor based on the larger version of the quantum group . in this paper we find one more extension of the cases where a tangle functor can be constructed . here we assume that @xmath0 is a primitive @xmath1th root of unity where @xmath2 is odd and greater than or equal to @xmath48 . we need @xmath49 and @xmath50 to define a tangle functor . we can not work on the level of the algebra as even when we enforce the relations @xmath49 and @xmath50 in the algebra , there is no universal @xmath14-matrix . hence we work at the level of the representations . we call representations @xmath51 that satisfy @xmath11 and @xmath10 , * semicyclic. first we prove that in these representations a solution to the quantum yang - baxter equation exists and is the image of the @xmath14-matrix in equation [ arr ] . next we prove that for @xmath12-tangles the tangle functor returns the same value as if we colored the tangle with the standard irreducible @xmath2-dimensional representation of @xmath3 . hitoshi murakami tells us that jun murakami already knew this . this does not mean that the functor contains no new information . the functor applied to @xmath52-tangles carries new information beyond the information in the functor coming from the standard @xmath2-dimensional representation . we would like to thank thomas kerler who shared with us his insight into a truly elegant proof that the tangle functor based on semicyclic representations returns the same answer as with the standard @xmath2-dimensional representation . given a @xmath12-tangle the functor returns a universal polynomial in @xmath53 with coefficients in @xmath54 , so that the exponential sum of @xmath36 is equal to the exponential sum of @xmath37 in every monomial appearing with nonzero coefficient . any such monomial can be written as a linear combination of polynomials in @xmath35 and the quantum casimir operator . the actions of these operators agree in the standard @xmath2-dimensional representation and the semicyclic representations . among other things , this means that it is impossible to get quantum hyperbolic invariants from a tangle functor that uses the standard @xmath14-matrix for @xmath3 . throughout this paper @xmath55 where @xmath56 is an odd counting number . the quantum integer @xmath57 , denoted @xmath58 $ ] is defined as , @xmath59=\frac{q^l - q^{-l}}{q - q^{-1}}.\ ] ] the quantum factorial is defined recursively by @xmath60!=1 $ ] , and @xmath61!=[n][n-1]!$ ] . the quantum binomial coefficients are defined by @xmath62=\frac{[n]!}{[k]![n - k]!}.\ ] ] the quantum binomial theorem states that if @xmath63 then , @xmath64 a^kb^{n - k}.\ ] ] the weyl algebra @xmath65 is a hopf algebra generated by @xmath36 , @xmath66 and @xmath6 with relation @xmath67 , antipode , @xmath68 , @xmath69 , counit given by @xmath70 , @xmath71 , and comultiplication @xmath72 and @xmath73 . let @xmath74 be the finite dimensional vector space over the complex numbers with basis @xmath75 where @xmath76 . choose a nonzero complex number @xmath77 . let @xmath78 denote @xmath79 matrices with complex coefficients , identified with @xmath80 via the choice of basis . define a representation of @xmath65 by , @xmath81 by @xmath82 , and @xmath83 when @xmath84 and @xmath85 . cyclic representations of the weyl algebra have been studied by kashaev @xcite , baseilhac and benedetti @xcite and bonahon et al @xcite . let @xmath3 denote the unreduced quantum group as in definition [ uq ] . we remind the reader though , that we are only considering the cases where @xmath86 where @xmath56 is an odd counting number . the following equations and proposition will be useful in showing that the @xmath14-matrix , defined in section [ sec : rmatrix ] , satisfies the quantum yang - baxter equation . by induction @xmath87f^{l-1}\frac{q^{1-l}k - q^{l-1}k^{-1}}{q - q^{-1}}.\ ] ] also @xmath88 where @xmath89 . [ prop : deltah ] @xmath90 the fact that @xmath72 forces @xmath91 . using induction , it can be shown that @xmath92 then , @xmath93 the cyclic representations of @xmath65 defined in section [ weyl ] can be extended to @xmath2 representations of @xmath3 as follows . let @xmath74 be the vector space over the complex numbers with basis @xmath94 where @xmath95 . fix @xmath96 and @xmath97 . let @xmath98 , and @xmath99 when @xmath84 and @xmath100 . notice that @xmath101 and @xmath102 where @xmath103 is the @xmath104 identity matrix . we find a representation @xmath105 of @xmath37 such that @xmath106 . under this condition , the equation @xmath107 leads to the following solution for @xmath108 . denote by @xmath109 the equivalence class of @xmath110 which we treat as the remainder on division by @xmath2 . define @xmath111v_{\overline{i+k-1 } } & : \overline{i+k } \ne \overline{0 } \\ & \\ \sum_{j=0}^{\overline{k-1 } } - \frac{[2(k - j)+n-1]}{a}v_{\overline{i+k-1 } } & : \overline{i+k } = \overline{0 } \end{array } \right.\ ] ] equivalently , @xmath108 can be expressed using the sum to product formula , @xmath112 [ n - k ] v_{\overline{i+k-1 } } & : \overline{i+k } \ne \overline{0 } \\ & \\ \frac{[k ] [ n - k]}{a}v_{\overline{i+k-1 } } & : \overline{i+k } = \overline{0 } \end{array } \right.\ ] ] the definition of these @xmath2 representations leads to the following relations . @xmath113 in fact , all @xmath2 of the representations are isomorphic via conjugation by @xmath114 . @xmath115 these representations are a subclass of those studied in @xcite and are called semicyclic representations. since there is only one semicyclic representation up to isomorphism , we will work with whichever is most convenient . one of the main goals of this paper is to compare the semicyclic and standard irreducible representations of @xmath3 . let @xmath116 denote the standard @xmath2-dimensional irreducible representation of @xmath3 . fix a basis @xmath117 of a vector space @xmath74 , then @xmath118 , and @xmath119 act on this basis as follows . @xmath120 for @xmath121 and @xmath122 , @xmath123[n - i ] v_{i-1}$ ] , @xmath124 . the standard representation , @xmath116 , is similar to the semicyclic representation @xmath125 . in fact , the only difference is @xmath122 while @xmath126 . in the case of the standard irreducible representation @xmath116 , the @xmath14-matrix @xmath127!}q^{\frac{l(l-1)}{2}}e^l\otimes f^l,\ ] ] satisfies @xmath128 where @xmath22 is the flipped comultiplication . this equation also holds for subrepresentations of tensor powers of the two dimensional irreducible representation . this formula first appeared in kirby and melvin @xcite , only they were working in the quotient of @xmath3 by @xmath129 , and @xmath130 . this formula for the @xmath14-matrix also appears in the work of ohtsuki where he notices that the condition @xmath130 is superfluous . it is also used in the unfolded version of the quantum group , where you add a generator @xmath13 with @xmath131 . in the following proposition we will show that the image of @xmath14 in the representations @xmath105 conjugates the image of comulitplication to the image of flipped comultiplication . [ prop : rdeltar ] for all @xmath132 , @xmath133 and @xmath134 , satisfies @xmath135 in the following proof , we suppress the notation @xmath105 and use @xmath136 and @xmath14 to mean their image under the representation @xmath105 . since @xmath50 the proof that @xmath137 and @xmath138 is the same as the proof appearing in @xcite . all that is left to show is @xmath139 . to simplify notation , write @xmath140 to prove the theorem we show that the equation @xmath141 holds for our choice of @xmath14 . using equation [ comult ] we get , @xmath142 using the commutation relations for @xmath143 from equation [ eq : qh ] , move it to the front of the second term and factor it out to get , @xmath144 cancel @xmath143 , distribute and then collect in powers of @xmath36 , to get , @xmath145 there is exactly one term with @xmath146 appearing in it , @xmath147 all other terms will cancel as in the proof in @xcite so we only need to show this term is zero . under all representations @xmath105 , @xmath37 has rank @xmath148 , so @xmath149 has rank @xmath150 . specifically @xmath151 spans the cokernel of @xmath149 , and its kernel is spanned by all other @xmath152 . applying @xmath153 to @xmath151 under @xmath105 gives @xmath154 now we can make a standard inductive computation of the @xmath14-matrix , noting that @xmath155 . going back to equation [ eqn ] , commute the @xmath36 past @xmath156 in the second term , and @xmath6 past @xmath157 in the first , and renumber the first term so that @xmath158 become @xmath159 . we get , @xmath160f^{m-1}\frac{q^{1-m}k - q^{m-1}k^{-1}}{q - q^{-1}})=0.\ ] ] collecting in powers of @xmath36 we get the recursive formula , @xmath161}q^{m-1}c_{m-1}.\ ] ] setting @xmath162 we get the standard formula for the @xmath14-matrix given above . recall , a universal @xmath14-matrix satisfies @xmath163 , and @xmath164 . the first of these two equations holds in semicyclic representations because it is true when @xmath165 , and @xmath50 suffices . this is because the only powers of @xmath36 and @xmath37 in the formula that exceed @xmath148 are in fact powers of @xmath37 , for instance see @xcite for the proof . the second equation does not hold in the semicyclic representations . to see why , apply proposition [ prop : deltah ] to the left hand side @xmath166 . @xmath167 notice that the highest power of @xmath36 or @xmath37 that appears is @xmath148 . since the equation is true when @xmath129 in the algebra it means that the coefficients of @xmath168 all monomials where the powers of @xmath36 and @xmath37 are less than or equal to @xmath148 agree with the answer above . for the right hand side , we have @xmath169 commuting @xmath170 to the front yields , @xmath171 now cancel the exponentiated @xmath172 from both sides . we know that all the terms where @xmath173 cancel with the left hand side so the remainder is , @xmath174 the matrices @xmath175 where @xmath176 ranges from @xmath2 to @xmath177 are linearly independent . hence each of the parts of the sum where @xmath178 needs to be zero . letting @xmath179 we get , @xmath180 consider @xmath181 . the values of this vector under each term of the sum are linearly independent , so the sum is nonzero and the relation @xmath164 does not hold under the semicyclic representations . however , the @xmath14-matrix still satisfies the yang - baxter equation , as long as we are evaluating in the representations @xmath105 . let @xmath182 be the flip @xmath183 . let @xmath184 denote its extension @xmath185 . if @xmath186 then @xmath187 suppose @xmath188 is any of the semicyclic representations of @xmath3 . then @xmath189 the proof follows @xcite . it really only depends on the fact that if we evaluate in @xmath190 , where @xmath191 is an arbitrary representation , and @xmath192 is a semicyclic representation where the kernel of @xmath37 contains @xmath75 , then @xmath193 and @xmath194 when in addition @xmath195 is semicyclic . to simplify notation we suppress the representation from the formulas . in this section , we define a tangle functor for @xmath12-tangles that have been colored with the extended semicyclic representations defined in section [ sec : extending ] . let @xmath198 be a basis for an @xmath2-dimensional vector space @xmath74 over @xmath199 . for @xmath200 and @xmath201 define the cup , cap , and crossing operators as follows . [ cols= " < , < " , ] remark * : in the last step , the exponent of @xmath0 , @xmath202 is replaced with @xmath203 . this is due to the linear relations @xmath204 and @xmath205 coming from the caps . this will be the case for every knot . later we will show that this invariant is the same as kashaev s invariant so we do not find a closed form for the final solution . given a @xmath12-tangle @xmath66 we denote the invariant of @xmath66 coming from the semicyclic representations with @xmath206 by @xmath207 coming from the irreducible semicyclic representations is shown to recover kashaev s invariant . specifically , we show that @xmath207 evaluated at a @xmath208 root of unity is equivalent to kashaev s invariant evaluated at a @xmath208 root of unity for any knot . a word @xmath209 is said to be * balanced * if @xmath210 . in the prior section , @xmath207 was calculated for the figure eight knot . that calculation shows that @xmath207 for any knot will be a sum of balanced words in @xmath36 and @xmath37 with coefficient functions of @xmath0 to a power . the proof will be by induction on the length of a word @xmath219 where the total degree of @xmath36 and @xmath37 are equal . for the base case , consider the word @xmath220 of length 2 . for a fixed integer @xmath2 where @xmath221 , the center of @xmath3 is generated by the standard quadratic casimir defined by @xmath241 . if the casimir is evaluated at the standard irreducible representation , @xmath116 , of @xmath3 then it will clearly not involve any @xmath240 since @xmath242 . the prior proposition shows that the casimir will not involve any @xmath240 when evaluated at the semicyclic irreducible representations discussed in this paper . [ prop : casimir ] let @xmath209 be a balanced word in @xmath36 and @xmath37 , then @xmath219 can be written as a product of the terms @xmath243_q)$ ] where @xmath244_q = \frac{q^rk - q^{-r}k^{-1}}{(q - q^{-1})^2}$ ] and @xmath245 . by induction and using the formula for the casimir , it can be shown that @xmath249_{q^{-1}})$ ] and @xmath250_{q})$ ] . with this in hand , the same induction argument from proposition [ prop : noa ] will show that any balanced word can be written in terms of the casimir and @xmath251_q$ ] . in this proof , we choose to use @xmath125 when calculating @xmath207 . kashaev s invariant is calculated in the same way as @xmath207 except evaluated at the standard irreducible representation @xmath116 . by definition , we have @xmath253 . therefore , the cup and cap operators are identical under either representation . also , @xmath254 acts the same under either representation . by proposition [ prop : casimir ] , we can complete the proof by checking that @xmath255 or @xmath256 evaluates the same under both representations . notice that @xmath257 and the only difference between @xmath258 and @xmath259 is their action on the basis vector @xmath260 . thus we only need to check the action of @xmath256 on @xmath260 under each representation . but , under both representations @xmath261 . 0000 baseilhac , stephane _ quantum coadjoint action and the @xmath262-symbols of @xmath263._interactions between hyperbolic geometry , quantum topology and number theory , 103143 , contemp . , * 541 * , amer . math . soc . , providence , ri , 2011 . baseilhac , stephane ; benedetti , riccardo _ quantum hyperbolic geometry _ , algebr . geom . topol.7 ( 2007 ) , 845917 . bonahon , francis ; liu , xiaobo _ representations of the quantum teichmller space and invariants of surface diffeomorphisms _ , geom . * 11 * ( 2007 ) , 889937 . fok , v. v. ; chekhov , l. o. _ quantum teichmller spaces _ , ( russian ) teoret . mat . * 120 * ( 1999 ) , no . 3 , 511528 ; translation in theoret . and math . phys . * 120 * ( 1999 ) , no . 3 , 12451259 de concini , c. ; procesi , c. _ quantum groups _ d - modules , representation theory , and quantum groups ( venice , 1992 ) , 31140 , lecture notes in math . , * 1565 * , springer , berlin , 1993 . geer , nathan ; patureau - mirand , bertrand _ g - links invariants , markov traces and the semicyclic @xmath264-modules _ , j. knot theory ramifications * 22 * ( 2013 ) , no . 11 , 1350063 , 28 pp . kassel , christian _ quantum groups _ graduate texts in mathematics , * 155 * springer - verlag , new york , 1995 . xii+531 pp . isbn : 0 - 387 - 94370 - 6 . kashaev , r. m. _ quantization of teichmller spaces and the quantum dilogarithm _ , lett . phys . * 43 * ( 1998 ) , no . 2 , 105115 . kashaev , r. m. _ quantum dilogarithm as a 6j - symbol _ , modern phys . a * 9 * ( 1994 ) , no . 40 , 37573768 . kirby , robion ; melvin , paul _ the 3-manifold invariants of witten and reshetikhin - turaev for sl(2,c ) _ , invent . math . * 105 * ( 1991 ) , no . 3 , 473545 . ohtsuki , tomotada quantum invariants . a study of knots , 3-manifolds , and their sets . series on knots and everything , 29 . world scientific publishing co. , inc . , river edge , nj , 2002 . xiv+489 pp . isbn : 981 - 02 - 4675 - 7 . reshetikhin , n. _ quasitriangularity of quantum groups at roots of @xmath150 _ comm . * 170 * ( 1995 ) , no . 1 , 7999 . kashaev , rinat ; reshetikhin , nicolai _ braiding for quantum @xmath265at roots of unity _ noncommutative geometry and representation theory in mathematical physics , 183197 , contemp . math . , * 391 * , amer . soc . , providence , ri , 2005 . reshetikhin , n. yu . ; turaev , v. g._ribbon graphs and their invariants derived from quantum groups _ comm . phys.127 ( 1990 ) , no . 1 , 126 . reshetikhin , n. ; turaev , v. g. _ invariants of 3-manifolds via link polynomials and quantum groups _ invent . math . * 103 * ( 1991 ) , no . 3 , 547597 . turaev , v. g. _ quantum invariants of knots and 3-manifolds _ , de gruyter studies in mathematics , 18 . walter de gruyter & co. , berlin , 1994 . isbn : 3 - 11 - 013704 - 6	let @xmath0 be a @xmath1th root of unity where @xmath2 is odd . let @xmath3 denote the quantum group with large center corresponding to the lie algebra @xmath4 with generators @xmath5 , and @xmath6 . a semicyclic representation of @xmath3 is an @xmath2-dimensional irreducible representation @xmath7 , so that @xmath8 with @xmath9 , @xmath10 and @xmath11 . we construct a tangle functor for framed homogeneous tangles colored with semicyclic representations , and prove that for @xmath12-tangles coming from knots , the invariant defined by the tangle functor coincides with kashaev s invariant .
one of the burning questions in science today is the understanding of dark matter . the quantification of the distribution of dark matter in our universe , at different scales , is of major interest in cosmology roberts75 , rubin2001,salucci2000,deblok2003 , hayashi2007 . at scales of individual galaxies , the estimation of the density of the gravitational mass of luminous as well as dark matter content of these systems , is the relevant version of this exercise . readily available data on galactic images , i.e. photometric observations from galaxies , can in principle , be astronomically modelled to quantify the gravitational mass density of the luminous matter in the galaxy , gallazi2009 , bell2001 ; such luminous matter is however , only a minor fraction of the total that is responsible for the gravitational field of the galaxy since the major fraction of the galactic gravitational mass is contributed to by dark matter . thus , the gravitational mass density of luminous matter , along with constraints on the gravitational mass density of the dark matter content , if available , can help the learning of the total gravitational mass density . however , the learning of this physical density is difficult in light of the sparse and noisy measurements that are available , coercing the practitioner to resort to undertaking simplifying model assumptions . in this paper , we present a new way of quantifying the relative support to such a model assumption in two independent data sets , by comparing the posterior probabilities of models given the two data sets . model selection is a very common exercise faced by practitioners of different disciplines , and substantial literature exists in this field kassraftery , ibf_2001,chipman_2001,ghoshsamanta_2001,barabari_2004,tony , cassella_2009 . in this context , some advantages of bayesian approaches , over frequentist methods has been reported ibf_2004,robert_2001 . much has been discussed in the literature to deal with the computational challenge of bayes factors ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * to name a few ) . at the same time , methods have been advanced as possible resolutions when faced with the challenge of improper priors on the system variables aitkin , ibf , tony . however , the computation of posterior , intrinsic or fractional bayes factors persist as a challenge , especially in the context of discrete , non - parametric and multimodal inference on a high - dimensional state space linkbarker_2006 . this paper demonstrates a novel methodology for quantifying support in two ( or more ) independent data sets from the same galaxy , for the hypothesis that the system state space admits a certain symmetry - namely , isotropy . assuming isotropy implies that the probability density function in this high - dimensional state space of the system , depends only on the magnitude of the state space vector , and does not depend on the inclination of this vector to a chosen direction . the null is nested within the alternative in this problem . at the same time , one of the data sets is small and the other large . also , the dimensionalities of the model parameter vectors sought under the different hypotheses , are also different . indeed in such a case , testing with multiple hypotheses can be performed , such that one hypothesis suggests that the state space that one of the data sets is sampled from , admits this symmetry while the other hypothesis suggests the same for the other data set ( considering the case of 2 available data sets ) . however , in the two cases , little prior information are available on the system parameter vectors . the priors on the model parameters are in fact non - informative ( uniform ) , marked by unknown multiplicative constants . then , as discussed in ibf_1996a , kassraftery , this results in the ratio of the predictive densities of the data sets , under the models , being defined only up to the ratio of these unknown multiplicative constants . thus , the bayes factor is rendered arbitrary . the computation of the bayes factors is then , in principle possible with posterior bayes factors aitkin , intrinsic bayes factors ibf , ibf_1996a or with fractional bayes factors tony . however in this real - life problem that we discuss here , the implementation of bayes factors gets difficult given that the relative support to an undertaken assumption in two distinct data sets is sought by comparing the posterior probability of models that are characterised by different numbers of parameters . secondly , the more complex model in which the simpler model ( the null ) is nested in this situation , is intractable . the null or the simpler model assumes the model parameter space to bear a simple symmetry , thus rendering posterior computation possible . on the contrary , the alternative ( or the more complex model ) does not constrain the geometry of the model parameter space in any way . thus , the simpler model is nested within the more complex model . however , under the alternative hypothesis , i.e. in lieu of such simplifying assumptions , it is not possible to compute the posterior probability of the model parameter vectors . this situation is in principle caused by the complexity of the system , compounded by the paucity of measurements . in the test presented here , learning can be performed - relatively easily under the null - and thereafter , support in the data for the assumption about the symmetry of the model parameter space is quantified . lastly , it is acutely challenging in this high - dimensional non - parametric situation , to achieve intrinsic priors ibf with imaginary training data sets . such training data can be generated from the posterior predictive distribution under the null perez_berger , and is subsequently used to train the prior under the more complex model , to implement it in the computation of the marginal density of the real data . however , the computational difficulty involved in the training of the intrinsic prior even under the null , in this high - dimensional setup , is discouragingly daunting . the implementation of real training data is not possible either , given that any method implemented to learn the gravitational mass density - particularly the similarly motivated bayesian methodology discussed below - might be misled by undersampled data that is likely to result if subsamples are taken from the already small observed data sets that typify this application . it was presented in the last paragraph that the generated training data can not be of practical use in real - life applications of the kind we consider here . however , the implementation of real training data , for the purposes of achieving intrinsic priors , is not possible either . such real training data is obtained as a sub - sample of the available data set . given that for the current application , the data sets are typically small to begin with , estimation of the unknown model parameter vectors using a still smaller ( sub)sample of measurements might be risky in terms of convergence of undertaken inference methods . the difficulties with the computation of bayes factors that we have delineated above , motivates the need for a test that allows for computation of the comparative support in an available data set for one hypothesis to another , and is operational even with non - informative priors , without needing to invoke any other than the available data set , in the computation of the posterior probability distribution of the hypothesis given the data . the test is also motivated to work irrespective of the dimensionality of parameter spaces . motivated by this framework , this paper introduces a new nonparametric test of hypothesis that tests for the existence of global symmetries of the phase space that the available data are sampled from . the test works in parameter space , in the context of non - parametric bayesian inference when for the two or more cases of differently sized samples , little and/or differential prior information are available . this new test involves partitioning the space of the model parameter vectors such that one of the partitions - the subspace @xmath1 - contains those parameters for which the posterior probability density given the data , exceeds the maximal posterior density that can be achieved under the null . outside @xmath1 lie the parameter vectors for which the posterior probability falls below this maximal posterior under the null . the model parameter vectors that lie in @xmath1 are those that underlie the support in the data against the null . thus , the probability of the null given a data set is the complement of the posterior probability density integrated over the subspace @xmath1 ( instead of over the whole of the parameter space ) . such an integral can also be viewed as the probability that a model parameter vector is in the subspace @xmath1 . in this treatment , the probability of a null given the data that is computationally simple to achieve , in the context of bayesian nonparametric inference in high - dimensional state spaces . the paper is organised as follows . section [ sec : application ] discusses the application , in the context of which the new test is introduced . the formulation of a phase space @xmath0 as an isotropic scalar - valued function of the velocity and location vectors of a galactic particle , is discussed in section [ sec : isotropy ] . the details of the bayesian estimation of the unknown parameters of the galaxy , is discussed in section [ sec : chassis ] . the estimated functions of a real galaxy are presented in section [ sec : bayesianlearning ] . the null hypotheses that we test are motivated in section [ sec : testing ] . in section [ sec : priors ] , we discuss the availability of priors on the unknown functions in the relevant literature and how this affects the testing . thereafter , shortcomings of the bayes factor computations in the context of this application are delineated in section [ sec : shortcomings ] . in section [ sec : outline ] , the outline of the new test is presented ; in particular , section [ sec : implementation ] is devoted to a detailed discussion about the implementation of this new methodology . the test is illustrated on simulated data in section [ sec : simulated ] and on real data in section [ sec : results ] . the paper is concluded with a discourse on the implications of the results , in section [ sec : discussions ] . as discussed above , it is difficult to learn the gravitational mass density @xmath2 of dark+luminous matter in galaxies , where @xmath3 , @xmath4 . @xmath2 is positive definite , implying that in any infinitesimally small volume @xmath5 inside the galaxy , gravitational mass is non - zero ( and positive ) . other physically motivated constraints on @xmath2 will be discussed in the next section in the context of learning this function , given the data at hand . while photometric measurements are more readily available , direct detection of dark matter has hitherto been impossible , implying that measurements that can allow for quantification of the gravitational mass density of dark matter , are not achievable . instead , there are effects of the total ( dark+luminous ) gravitational mass that can be measured , though astronomical measurements that bear signature of such effects are hard to achieve in `` early - type '' galaxies , the observed image of which is typically elliptical in shape . of some such astronomical measurements , noisy and partially missing velocities of individual galactic particles have been implemented to learn @xmath2 cote2001,genzel , chakrabarty_somak . from this , when the astronomically modelled gravitational mass density of luminous matter , @xmath6 , is subtracted , the density of the gravitational mass of the dark matter content of early - type galaxies can be learnt . in this paradigm of learning @xmath2 , the data is referred to as partially missing since the noisy measurements of only one component , namely , the component along the line - of - sight that joins the observer to the particle - of the three - dimensional velocity vector of galactic particles , are typically available . we view these measurables as sampled from the phase space @xmath7 of the system , where @xmath7 is the space of all the states that the system can achieve . for a galaxy , @xmath7 is the space of the spatial vectors and velocity vectors of all galactic particles . on other occasions , the dispersion of the line - of - sight component of the velocity vector comprises the measurement . in either case , such kinematic data is expected to track the gravitational field of the galaxy in which the sampled particles play . we ask the question that if the available data include velocity measurements of particles in distinct samples that live in mutually disjoint volumes of @xmath7 , what are the implications for the gravitational mass density estimated using such data sets ? one major implication is that the estimate of @xmath2 obtained using one data set will be in general , different from that obtained using another data set . of course , for the same system , distinct @xmath8 are impossible - in fact , such a fallacious result , if achieved , can be explained as due to the fact that the used data sets have been drawn from mutually insulated volumes of the galactic phase space , the @xmath0s of which do not concur . such is possible , if the galactic phase space is characterised by disjoint volumes , motions in which do not communicate with each other . this is a viable scenario in non - linear dynamics even for systems with moderate complexity ; it is possible that the two types of galactic particles , the data of which are measured , live in distinct basins of attraction that characterise the galactic phase space thomsonstewart . in particular , we configure the question of unequal estimates of gravitational mass density functions , using the available data sets , to the context of the real galaxy ngc 3379 . for this system , multiple data sets are measured for two distinct types of galactic particles pns , bergond . the 3-d spatial location vector of a particle resident in a galaxy is written as @xmath9 , where only the coordinates of the image of the particle - @xmath10 and @xmath11 - can be measured . also , the 3-d velocity vector of a particle is @xmath12 , with only the component @xmath13 being a measurable , i.e. we can only measure the speed with which the particle is approaching us ( the observer ) or receding from us . the galactic phase space @xmath7 is the space of @xmath14 and @xmath15 of all galactic particles . thus , @xmath16 . if we are convinced that motion tracks the gravitational field due to a given gravitational mass density function , then we can write down @xmath17 $ ] , where @xmath18 is a function of @xmath2 . in @xcite , the effort has been to learn @xmath2 from the data using non - parametric modelling . the aim in these work is to make inference on the gravitational mass density @xmath2 given observations of only 1 component @xmath13 out of 3 components of the velocity vector of particles resident in the system , where the dependence of @xmath13 on @xmath2 is intuitively motivated . in order to render the problem tractable , we need to undertake some simplifying model assumptions . it is useful in such situations to assume something simple about symmetries of the system state space . the simplest such assumption considers phase space @xmath7 to be isotropic , i.e. for the phase space probability density function @xmath19 , to be isotropic in @xmath15 and @xmath20 . thus , the phase space @xmath0 is @xmath21 . in general , the @xmath0 of @xmath7 has a time dependence , but here we assume the system to have attained stationarity , i.e. for the phase space @xmath0 to be independent of time . thus , one model assumption that we undertake is that of stationarity of the phase space @xmath0 . such is typically undertaken in the modelling of galaxies since deviation from stationarity can not be checked in the available data since the data sets offer snapshots at a ( nearly ) unique time point in the evolution of the system , the time scale of which @xmath22 years , bt . there are reports of observational signatures of non - stationarity ; + in ngc 3379 these could be considered to include the x - ray image of ngc 3379 3379_xray . all in all , at the level of this paper , we can not say that deviation from equilibrium is ruled out . thus we state that it is under this model assumption of stationarity , that we interpret all results . the phase space @xmath0 is then a scalar - valued , function of 2 vectors , @xmath15 and @xmath14 . now , if a general real - valued function @xmath23 of two vectors @xmath24 , is an isotropic function , then @xmath25 , for any orthogonal transformation matrix @xmath26 truesdell , wang . we recall from the theory of scalar valued functions of two vectors , that if @xmath23 is an isotropic function , its set of invariants ( with respect to @xmath26 ) is @xmath27 . then , the isotropic function of two vectors , @xmath28 , admits the representation @xmath29 truesdell , shi . in general , @xmath30 is the sum of a function of @xmath31 with a function of @xmath32 and a function of @xmath33 . if @xmath34 and @xmath35 are orthogonal vectors , @xmath33=0 . then , it follows that 1 . under the assumption that the phase space @xmath0 is an isotropic scalar valued function of 2 vectors , @xmath36 and @xmath37 , it admits the representation @xmath38 , 2 . if we can write a general phase space @xmath0 as @xmath38 , it implies that it is an isotropic scalar valued function of @xmath36 and @xmath39 , where @xmath40 is a sum of a function of @xmath41 with a function of @xmath42 . ( the inner product @xmath43=0),i.e . @xmath44 with @xmath45 and @xmath46 identified as some real - valued functions . one physical interpretation of the above representation is that the phase space @xmath0 is isotropic if and only if it can be written as a function of the value @xmath40 of the particle energy , where energy of a particle with location @xmath47 and velocity @xmath48 is the function @xmath49 , i.e. @xmath50 n.b . the symbol @xmath51 is to be distinguished from expectation . now total energy is the sum of potential and kinetic energies . thus , to allow for the physical interpretation of @xmath40 as the value of the particle energy , we choose @xmath52 to be the gravitational potential energy , ( written as a function of @xmath53 ) and the kinetic energy @xmath54 . it merits mention that @xmath2 is a known function of @xmath55 as given by a standard equation of physics - the poisson equation : @xmath56 where @xmath57 is the laplacian operator and @xmath58 is a known constant . however , in lieu of information about the nature of the phase space density , and given at least the moderate level of complexity that is envisaged to manifest in phase spaces of galaxies , there is no justification for opting for an isotropic , over an anisotropic prescription for galactic phase spaces . there is a strong logistical motivation for doing so however - the assumption of an isotropic @xmath59 renders the calculations involved in the learning of @xmath2 relatively easier . the details of this learning , given the data and the assumed prescription of isotropy for @xmath59 , are discussed in section [ sec : chassis ] . in principle , learning @xmath2 after relaxing the assumption of isotropy , under a particular restrictive prescription for including anisotropy , can still be possible with a much harder computational procedure ( chakrabarty and saha , under preparation ) . however , it is impossible to learn @xmath2 in the general anisotropic cases , given the typically partially missing nature of the kinematic data , under an unrestricted model for the anisotropy of phase space . thus , the estimate of @xmath2 is sensitive to the prescription chosen for the phase space symmetry - isotropic or not , and if anisotropic , to the value of the `` anisotropy parameter '' . this problem refers to the lack of identifiability among solutions for @xmath2 , caused by lack of knowledge about the state of isotropy of galactic phase spaces . and is referred to as the mass - anisotropy degeneracy in the astrophysical literature @xcite . the problem has been a pressing concern of @xcite when they advance the possibility that anisotropy might be introduced into @xmath59 of galaxy ngc 3379 due to evolutionary reasons , which @xcite had ignored . disagree that this could explain the identification of the low dark matter content in ngc 3379 by @xcite given that the existence of anisotropy was included in the data analysis employed in the earlier work . they stress that `` anisotropy parameter '' recovered from their kinematic data and that obtained from the simulations presented in @xcite , are similar in nature . . here , the anisotropy parameter is a parametrisation of the deviation from an isotropic phase space density , defined as @xmath60 bt , where @xmath61 is the variance in the @xmath62-th component of @xmath14 , @xmath63 . however , the suggestion advanced in this paper is more general . firstly , the learning of @xmath2 is nonparametric , so an explicit parametrisation of anisotropy is not of relevance . importantly , if the test of hypothesis that we develop here , suggests less support in one data set over another , to the model assumption of an isotropic phase space @xmath0 , it would imply that the two data sets have been sampled from distinct phase space density functions that will then be interpreted as fundamentally different . such unequal @xmath0s can not in general be reconciled with each other , merely by adjusting one parameter , the physical interpretation of which is the anisotropy parameter as given in astrophysical literature . in general , the phase space @xmath0 will contain parameters @xmath64 , i.e. the phase space @xmath0 is @xmath65 . given data @xmath66 of the observable phase space coordinates of a sample of @xmath67 particles , the likelihood function @xmath68 can be written using the phase space @xmath0 @xmath69 since by definition , @xmath70 , where the data vector of the @xmath62-th particle is @xmath71 and the spatial and velocity vectors of this particle are @xmath72 and @xmath73 respectively , @xmath74 . however , there are two points of worry at this stage . firstly , we want to learn @xmath2 but it would appear that the likelihood proposed above does not include this function in its definition . generally speaking , in such situations , we would need to embed the sought unknown function ( @xmath2 ) in the definition of the phase space @xmath0 by suggesting that @xmath75 , where @xmath76 . this calls for identification of a function @xmath77 using domain knowledge . the likelihood would then in general be written as @xmath78 where the data are considered to be @xmath79 . in our case , we have already identified that the admissible representation of an isotropic phase space @xmath0 , is as a function of particle energy , ( section [ sec : isotropy ] ) , which is itself the sum of potential and kinetic energies . we have defined @xmath80 , ( see equations [ eqn : eng1 ] , equation [ eqn : eng2 ] ) , where @xmath55 depends on the sought unknown function @xmath2 via poisson equation ( see equation [ eqn : poisson ] ) . thus , a function that has @xmath81 embedded in it , is the energy function @xmath82 that we now re - write as @xmath83 , so that @xmath84 where @xmath85 denotes the euclidian norm of a vector . to summarise , the dependence of @xmath86 on @xmath2 is ensured by suggesting that the phase space @xmath0 depends on value of energy function , @xmath40 . given this , we could suggest that the phase space density is @xmath87 . however , here we assume that the phase space is isotropic and encode this assumption by writing the phase space @xmath0 as a function of energy alone , with no dependence on any other variables ( see section 3.1 ) . in fact , we assume that phase space @xmath0 is @xmath38 , i.e. @xmath88 ( by equation [ eqn : engg ] ) . now , there are no universally accepted parametric forms of the gravitational mass density of all matter in galaxies , known of in the astronomical literature . this urges a fully non - parametric model for @xmath89 . following @xcite , we adopt a fully discretised model such that the range of @xmath90 is binned , with @xmath91 defining the @xmath92-th @xmath15-bin of width @xmath93 . we define that when @xmath90 is in the @xmath92-th @xmath15-bin , @xmath94 , @xmath95 and define the vector @xmath96 . thus , @xmath97 is the discretised form of the unknown function @xmath98 that we seek to learn from the data . in light of this , we view the energy function as @xmath99 . in fact , there is again no guideline in the literature as to the nature of dependence of the phase space @xmath0 on energy , motivating a non - parametric model for @xmath38 too . the relevant range of energy values is discretised such that @xmath100 defines the @xmath101-th @xmath51-bin of width @xmath102 , inside which , @xmath103 , @xmath104 . thus , we set @xmath103 for @xmath40 in the @xmath101-th @xmath51-bin , @xmath104 . we define the vector @xmath105 . indeed @xmath106 is unknown and we attempt to learn it from the data . our second worry is with the suggestion for the likelihood in equation [ eqn : likeli ] . we recall that the data @xmath66 is partially missing as it does not comprise measurements of all 6 phase space coordinates ; in fact @xmath107 . thus , in addition to the line - of - sight speed @xmath13 , the measurables include the spatial locations @xmath108 of the particle on the image of the galaxy . let @xmath109 . the data is then sampled from the @xmath0 @xmath110 of the subspace @xmath111 , so that likelihood should be defined in terms of @xmath110 . in other words , the likelihood function can not contain the unobservable variables @xmath112 . therefore , the definition of likelihood in equation [ eqn : init_likeli ] has to be modified and written in terms of the projected phase space @xmath0 @xmath113 , where this projected @xmath0 results from the projection of @xmath114 onto the space of the observables , @xmath115 . assuming the observed vectors @xmath116 , @xmath117 , to be @xmath118 , the likelihood function is then defined as @xmath119 where the projected phase space @xmath0 for the @xmath62-th data point , @xmath113 , results from projecting the phase space @xmath0 @xmath38 onto the subspace @xmath115 , as in @xmath120 where this projected phase space @xmath0 is normalised at each iteration by a constant given by integrating @xmath121 over all possible values of the observables @xmath122 . it merits mention that @xmath51 inside the integral is the energy function , not to be confused with expectation . in fact , @xmath123 in the second step of equation [ eqn : elaboration ] , elaboration of energy as sum of the potential and kinetic energies invokes the idea that the potential @xmath45 is determined via poisson equation ( equation [ eqn : poisson ] ) , given the gravitational mass density function @xmath124 . the projection from @xmath7 onto @xmath115 in equation [ eqn : projection ] is performed by decomposing the integrals on the rhs as a sum of the contributions to the integrals , from individual @xmath51-bins . the mapping from the space @xmath125 to the range of @xmath40 in the @xmath101-th @xmath51-bin , is identified for any @xmath62 [ see][for details]withsimon , @xmath126 . summing over all @xmath101 s then provides @xmath113 . owing to lack of consensus in the literature on the nature of the gravitational mass densities in galaxies and their phase space @xmath0s , we choose to impose uniform priors for @xmath127 and @xmath128 , @xmath129 , @xmath130 . these are respectively , @xmath131 $ ] and @xmath132 $ ] , . here @xmath133 are experimentally chosen constants and the prior on the phase space density is uniform in [ 0,1 ] since @xmath134 is normalised to be 1 for the most bound orbit , i.e. for the maximum value of @xmath40 . the priors are used along with the likelihood function ( defined above in equation [ eqn : likeli ] ) in bayes rule to write the posterior probability density + @xmath135 above , the posterior density is referred to as @xmath136 to distinguish it from the posterior probability @xmath137 that we actually sample from , after convolving @xmath138 with the distribution of error in the measurements of @xmath13 , @xmath139 . the measurement errors in @xmath10 and @xmath11 are stated by astronomers to be small enough to be ignored . inputs from astronomers involved in the observed data set are considered when modelling the error distribution . typically , the error distribution is considered gaussian with zero mean and variance that is suggested by the relevant observational astronomer(s ) . thus , + @xmath140 + then @xmath106 and @xmath97 are learnt by sampling from this posterior , using adaptive random - walk metropolis - hastings haario . some of the approximations that underlie sampling from this posterior are now discussed , followed by a brief discussion of the inference . this is the approach used by @xcite and + @xcite who assumed an isotropic phase space @xmath0 . now , the undertaken model assumption of phase space isotropy also allows for the identification of @xmath141 , with the spherical radius @xmath142 , where @xmath143 , the radial location of the @xmath62-th particle from the system centre . in other words , the assumption of an isotropic phase space is inclusive of a spherical spatial geometry . then identifying @xmath45 with the gravitational potential energy , the connection between @xmath144 and @xmath145 ( poisson equation ) is recalled as @xmath146 in this geometry . then the @xmath147-th @xmath15-bin discussed above is synonymous to the @xmath147-th radial bin , @xmath129 . the unknown functions abide by the physically motivated constraints that @xmath148 , @xmath149 and @xmath150 . the last constraint is intuitively motivated as valid in a gravitationally bound spherical system that we model the galaxy to be . we view the galaxy as being built by stacking spherical layers on top of each other . then owing to gravity being a force that attracts mass towards the centre , the compactness in the packaging of mass is higher near the centre than away from the centre . in other words , @xmath145 increases as @xmath142 decreases , in general bt . following the trend of the various phase space @xmath0s discussed in astronomical literature bt , @xmath114 is also treated as a monotonically increasing function of energy @xmath51 in this implementation of the methodology . in the adopted discrete model , a simultaneous learning of discretised versions of two univariate functions @xmath38 and @xmath145 is attempted , i.e. the target is to learn the vector @xmath151 and the vector @xmath152 as proxies for the unknown functions . given that the gravitational mass density is non - negative and monotonically non - increasing function of @xmath142 , @xmath153 . again , given that @xmath149 , @xmath154 . in our implementation , to ensure monotonic decline in the gravitational mass density function , it is the difference @xmath155 between the gravitational mass densities in the @xmath147-th and @xmath156-th radial bins that is proposed , as @xmath157 , where @xmath158 is the folded normal distribution folded with mean @xmath159 and variance @xmath160 , @xmath161 , @xmath162 . the motivation behind this choice of the proposal density is that over the support @xmath163 , it is a relatively easy density to sample from , while satisfying the requirement that in general , @xmath164 when @xmath165 . here , the current difference between the gravitational mass density values in the @xmath147 and @xmath156-th radial bins is @xmath166 . the variance @xmath167 is the empirical variance of @xmath168 , computed using values of this difference variable from step number @xmath169 to @xmath170 , where @xmath171 is the current step number and @xmath169 is chosen experimentally to be post - burnin haario , @xmath172 . here @xmath173 and @xmath174 are defined as 0 . then as @xmath147 is varied from @xmath175 to 1 , the proposed @xmath147-th component of the unknown gravitational mass density vector is @xmath176 . @xmath128 is updated similarly , while following an imposed constraint that @xmath114 is monotonically non - decreasing with energy @xmath51 , @xmath130 . however , unlike in the case of the updating of the gravitational mass density value , a monotonically non - decreasing phase space @xmath0 with energy is not motivated by any physically justifiable constraints , though astronomical literature often suggests phase space density functions that typically increase with energy bt . thus , we realise that when the data comprises even partially missing information on the state space coordinates of a system , this methodology allows for the formulation of the likelihood of the parameters describing the sought system function , in terms of the projection of the @xmath0 of the system state space onto the space of the observables , as long as the unknown function can be embedded within the definition of this @xmath0 . such an approach is useful in contexts similar to what we work with here , for applications other than the current one . a notable feature of this methodology is that even when the data are missing , i.e. sampled from a sub - space of the system state space , it is possible to write the likelihood function as a product of the values of the sub - space density , at each data point , where the density of this sub - space is obtained by integrating the unobserved variables out of the state space @xmath0 . ngc 3379 is one of the few elliptical galaxies , for which kinematic information is available for individual members of two different populations of galactic particles - referred to as planetary nebulae ( pne ) and globular clusters ( gcs ) - over an extensive radial range spanning the outer parts of the galaxy . the data used in the work include measurements of @xmath10 , @xmath11 and @xmath13 of 164 pne reported by pns and of 29 gcs by bergond . we refer to the pne data set as @xmath177 and the gc data set as @xmath178 , with respective sample sizes of @xmath179=164 and @xmath180=29 . the learning of the model parameter vectors @xmath106 and@xmath97 is performed in a high - dimensional state space ( dimensionality = @xmath181 ) . the traces of the posterior given either data are presented in figure [ fig : hist_real ] . the marginal posterior densities given either data are shown for one component of the learnt @xmath97 vector , namely @xmath182 ( figure [ fig : marg_real ] ) . the marginal densities are noted to be markedly multimodal . in figure [ fig : fig1 ] , we present the vectors @xmath183 and @xmath184 , learnt from using the @xmath92-th data @xmath185 , @xmath186 . this estimation is performed using the aforementioned bayesian nonparametric methodology , under the model assumption of an isotropic phase space @xmath0 , i.e. a phase space density that is expressed as @xmath38 and approximated in the discretised model of chassis as the vector @xmath106 , ( the @xmath101-th component of which is the value of the phase space density in the @xmath101-th energy - bin ) and the vector @xmath97 , ( the @xmath147-th component of which is the value of the phase space density in the @xmath147-th radial - bin ) . the learnt 95@xmath187 hpds are represented as error bars and the modal values are shown as open circles . . table displaying seeds that individual chains run with data @xmath177 and @xmath178 are started with . the initial choice of the gravitational mass density function is one that is sometimes used in astrophysical literature , @xmath188 . the starting phase space density is chosen to be either of the form @xmath189 or @xmath190 . here the parameters @xmath191 . the chains run with pne data @xmath177 are assigned names characterised by the prefix `` pne - run '' , while runs performed with the gc data @xmath178 are labelled with prefix `` gc - run '' . [ cols="^,^,^,^,^,^",options="header " , ] comparing the computed @xmath192 and @xmath193 across the chains implies that the assumption of isotropy is more likely to be invalid for the phase space from which the pne data @xmath177 are drawn than from which the gc data @xmath178 are drawn . basically , support in real data @xmath177 for the assumption of isotropy is distinct from that in @xmath178 . this implies that the @xmath194 , where phase space @xmath0 that @xmath177 is sampled from is @xmath195 and @xmath196 . however , both data sets carry information on the phase space coordinates @xmath197 of the same galaxy , i.e. both data sets are sampled from @xmath0s that describe the phase space structure of all or some volume inside the same galactic phase space @xmath7 . thus , @xmath198 where @xmath199 is phase space @xmath0 defined in volume @xmath200 and @xmath201 is phase space @xmath0 defined in volume @xmath202 . in terms of the phase space structure of this real galaxy ngc 3379 , we can then conclude that the phase space of the system is marked by at least two distinct volumes , motions in which do not communicate with each other , leading to distinct orbital distributions being set up in these two volumes , which in turn manifests into distinct @xmath0s for these subspaces of the galactic phase space . the pne and gc samples are drawn from such distinct @xmath0s . of course , such an interpretation would hold up if we can rule out extraneous reasons that might be invoked to explain the differential support in @xmath177 and @xmath178 to the assumption of phase space isotropy . such extraneous factors are systematically dealt with in section [ sec : discussions ] . it merits mention that our result that @xmath203 also suggests that the phase space density that the observed gcs in this galaxy live in , is nearly isotropic . the motivating idea is that if different data sets are sampled from different volumes of the galactic phase space , where there is no communication amongst these volumes , the gravitational mass density estimates obtained using these data sets will be different . none of these individual estimates will however tell us of the gravitational mass density function of the whole galaxy , in general . the worrying implication of this is that interpreting one of these estimates of @xmath2 as the galactic estimate , can be completely erroneous . the estimate achieved given one data set will then reveal no more about the entire galaxy than the very isolated phase space volume from which these data on particle motions are sampled . ideally speaking , the gravitational mass density of the galaxy , at any @xmath204 can be is approximated as + @xmath205 where @xmath206 is the estimate obtained using the @xmath92-th data set , @xmath207 , where there are @xmath208 available data sets that have been drawn from @xmath208 mutually insulated volumes or sub - spaces @xmath209 within the galactic phase space , namely from @xmath210 . the more the number of such data sets available , i.e. bigger is @xmath208 , the better is the approximation suggested above . however , in this framework we are always running the risk of misidentifying the estimate of a property of a subspace of the galaxy as that of the galactic property . here , the @xmath92-th data set @xmath211 where @xmath212 is the @xmath0 that describes sub - space @xmath213 . then the statement that @xmath214 and @xmath215 are `` mutually insulated '' implies that motions in @xmath216 do not cross over into @xmath215 and vice versa , resulting in @xmath217 , @xmath218 . thus , @xmath219 are not identically distributed . hence collating all data sets @xmath220 , as a single input into any method that attempts learning the galactic gravitational mass density by invoking phase space densities , will not work , if the method demands the data to be @xmath79 . ( all methods of learning @xmath2 are underlined by the need to invoke the phase space @xmath0 ) . still , if we are in possession of knowledge of the correlation between the @xmath0s @xmath212 and @xmath221 , we could use the collated data set in such a method to learn the galactic mass density . however , as there is absolutely no measured , simulated or theoretical information available on the correlation structure between mutually insulated sub - volumes inside galaxies , such a modelling strategy is untenable . indeed , the development of any such learning strategy , can imply all @xmath208 available data sets jointly , but will be highly sensitive to the non - linear dynamical modelling of the phase space of the particular galaxy in question . at this stage we recall that for the majority of galaxies , @xmath208=1 and @xmath208 is at most 2 for a few galaxies ; examples of these systems include ngc 3379 pns , bergond , cena ( woodley @xmath222 chakrabarty , in preparation ) . thus , we can put a lower bound on the total gravitational mass content of the galaxy , as we now present for ngc 3379 . ngc 3379 is advanced as a dark matter rich galaxy , with the gravitational mass inside a radius of about 20 kpc to be at least as high as about 4 to 10@xmath22310@xmath224m@xmath225 , where m@xmath225 denotes the mass of the sun and the astronomical unit of length , kiloparsec , is abbreviated as kpc . in the above test , a high support in the gc sample towards an isotropic phase space @xmath0 , along with a moderate support in the sampled pne for the same assumption , indicate that the two samples are drawn from two distinct phase space density functions . the expectation that the implementation of the pne and gc data sets will lead to concurring gravitational mass density estimates is foreshadowed by the assumption that both data sets are sampled from the same - namely , the galactic - phase space density @xmath226 . the apparent motivation behind this assumption is that since both samples live in the galactic phase space @xmath7 , they are expected to be sampled from the same galactic phase space density , at the galactic gravitational potential . however , such does not necessarily follow if - for example - @xmath59 is a non - analytic function : @xmath227 then , if the gc data @xmath228 are sampled from the density @xmath229 and pne data @xmath230 , where @xmath231 , @xmath232 , then the statement that both the observed samples are drawn from equal phase space densities is erroneous . if this assumption is erroneous , i.e. if observed data @xmath177 and @xmath178 are sampled from unequal phase space @xmath0s , 1 . firstly it implies that the phase portrait of the galaxy ngc 3379 is split into at least two sub - spaces such that the observed gcs live in a sub - space @xmath233 and the observed pne live in a distinct sub - space @xmath234 , @xmath232 . 2 . secondly that the phase space densities that describe @xmath235 and @xmath236 are unequal . qualitatively we understand that if the galactic phase space @xmath237 is split into isolated volumes , such that the motions in these volumes do not mix and are therefore distinctly distributed in general , the phase space densities of these volumes would be unequal . this is synonymous to saying that @xmath7 is marked by at least two distinct basins of attraction and the two observed samples reside in such distinct basins . thus , a split @xmath7 will readily explain lack of consistency in the estimate of support in the two data sets for the null that the phase space density function that these two data sets are drawn from are isotropic . however , the fundamental question is really about the inverse of this statement . does differential support in @xmath177 and @xmath178 to isotropic pase space @xmath0s necessarily imply a split @xmath7 ? indeed it does , since differential support in @xmath177 and @xmath228 towards isotropy of the @xmath0 that describes the respective native phase space @xmath238 and @xmath239 , implies that the distribution of the phase space vector @xmath240 in the sub - spaces @xmath241 and @xmath242 are distinct . this in turn implies that the phase portrait of this galaxy manifests at least two volumes , the motions in which are isolated from each other . such separation of the motions is possible if these distinct sub - spaces that the two datasets reside in , are separated by separatrices thomsonstewart into isolated volumes , motions across which do not mix . as for a physical reason for the phase space distribution of the pne to be less isotropic than the gc population , we can only speculate at the level of this paper . the gc population being an older component of a galaxy ( than the pne population ) , might have had longer time to equilibrate towards an isotropic distribution . also , pne being end states of stars , the pne population is likely to reside in a flatter component inside the galaxy , than the gc population . galactic phase spaces can be split given that a galaxy is expectedly a complex system , built of multiple components with independent evolutionary histories and distinct dynamical timescales . as an example , at least in the neighbourhood of the sun , the phase space structure of the milky way is highly multi - modal and the ensuing dynamics is highly non - linear , marked by significant chaoticity . the standard causes for the splitting of @xmath7 include the development of basins of attraction leading to attractors , generated in a multistable galactic gravitational potential . basins of attraction could also be triggered around chaotic attractors , which in turn could be due to resonance interaction with external perturbers or due to merging events in the evolutionary history of the galaxy . one worry that astronomers have expressed in the literature about the galaxy ngc 3379 is that the spatial geometry of the system is triaxial and not spherical . for us , the relevant question to ask is if there is support in the data for the consideration of the gravitational mass density to depend on the spherical radius @xmath142 alone . in our work , the fact that the methodology assumes the gravitational mass density to be dependent on @xmath243:=@xmath244 is a manifestation of the broader assumption of phase space isotropy ( see section [ sec : isotropy ] ) . thus , our test for phase space isotropy includes testing for the assumption that the gravitational mass density of the galaxy ngc 3379 bears a dependence on the spatial coordinates @xmath245 , only via @xmath142 , @xmath246 . in other words , we have already tested for the assumption that the gravitational mass density depends on the components of the spatial vector , via the spherical radius @xmath142 . in contrast to the pne and gc samples observed in this galaxy , if two observed sets of galactic particles can be inferred to have been drawn from the same phase space density , we will expect consistency in the gravitational matter density that is recovered by using such data sets in a mass determination formalism . at the end of the discussion presented above , we will naturally want to know what the true gravitational mass density function of ngc 3379 . however , if distinct density estimates are available at a given radius , from @xmath208 independent data sets as @xmath247 , then the lower limit on the galactic gravitational mass density at @xmath142 is @xmath248 . the above results and arguments suggest that it is inherently risky to refer to the gravitational mass density recovered using an observed particle sample - and the gravitational potential computed therefrom - as the gravitational potential of the galaxy . we have demonstrated this with the example of ngc 3379 and shown that inconsistencies in @xmath145 learnt using distinct observed samples , can not be attributed to any other factor except that these observed samples are drawn from distinct and insular sub - spaces within the galactic phase space , and/or the lack of time - independence in the gravitational mass density or phase space density function . i am thankful to dr . john aston for his kind comments . i acknowledge the support of a warwick centre for analytical sciences fellowship .	in lieu of direct detection of dark matter , estimation of the distribution of the gravitational mass in distant galaxies is of crucial importance in astrophysics . typically , such estimation is performed using small samples of noisy , partially missing measurements - only some of the three components of the velocity and location vectors of individual particles that live in the galaxy are measurable . such limitations of the available data in turn demands that simplifying model assumptions be undertaken . thus , assuming that the phase space of a galaxy manifests simple symmetries - such as isotropy - allows for the learning of the density of the gravitational mass in galaxies . this is equivalent to assuming that the phase space @xmath0 from which the velocity and location vectors of galactic particles are sampled from , is an isotropic function of these vectors . we present a new non - parametric test of hypothesis that tests for relative support in two or more measured data sets of disparate sizes , for the undertaken model assumption that a given set of galactic particle data is sampled from an isotropic phase space @xmath0s . this test is designed to work in the context of bayesian non - parametric , multimodal inference in a high - dimensional state space . in fact , the different models that are being compared are characterised by differential dimensionalities of the model parameter vectors . in addition , there is little prior information available about the unknown parameters , suggesting uninformative priors on the parameters in the different models . the problem of model parameter vectors of distinct dimensionalities and the difficulties of computing bayes factors in this context , are circumvented in this test . the test works by identifying the subspace ( of the system phase space ) that is populated by those model parameter vectors , the posterior probability density of which exceed the maximal posterior density achieved under the null . the complement of the probability of the null given a data set is then the integral of the posterior probability density of the model parameters over this sub - space . this integral is the probability that the model parameter lives in this subspace ; in implementational terms this probability is approximated as the fraction of the model parameters that live in this subspace . we illustrate applications of this test with two independent particle data sets in simulated as well as in a real galactic system . the dynamical implications of the results of application to the real galaxy is indicated to be the residence of the observed particle samples in disjoint volumes of the galactic phase space . this result is used to suggest the serious risk borne in attempts at learning of gravitational mass density of galaxies , using particle data . ,
the bd+404124 region consists of a few tens of very young stars associated with the probable herbig ae / be stars bd+404124 (= v1685 cyg ) , lkh@xmath0 224 (= v1686 cyg ) and the young multiple system lkh@xmath0 225 (= v1318 cyg ) , all of which have infrared excesses ( strom et al . 1972a ; hillenbrand et al . 1995 ) . together with ngc 6910 , ngc 6914 , the bd+413731 region and ic 1318 , it is part of the giant star forming region 2 cyg , at a distance of about 1 kpc ( shevchenko et al . the stars in the bd+404124 region are significantly younger ( ages less than 1 million years ) than those in the surrounding ob associations , with the low- and the high - mass stars having formed nearly simultaneously , leading some authors to suggest that star formation in this association might have been induced by the propagation of an external shock wave into the cloud core ( shevchenko et al . 1991 ; hillenbrand et al . the iras and afgl surveys showed a powerful infrared source within the bd+404124 group , but the positional uncertainty did not allow to assign the flux to one of the individual objects in the region . more recent submm continuum maps of the region ( aspin et al . 1994 ; di francesco et al . 1997 ; henning et al . 1998 ) clearly peak on lkh@xmath0 225 , suggesting that it is also dominant in the far - infrared and may be identified with iras 20187 + 4111 ( afgl 2557 ) . aspin et al . ( 1994 ) have shown lkh@xmath0 225 to be a triple system oriented north - south . the most northern and southern components , separated by 5 , appear stellar and are photometrically variable at optical and near - infrared wavelengths . the middle component exhibits strong [ sii ] emission , suggesting it is a nebulous knot of shock - excited material , not unlike many herbig - haro objects ( magakyan & movseyan 1997 ) . whereas bd+404124 dominates the association in the optical , the southern component of lkh@xmath0 225 is brightest in the mid - infrared ( aspin et al . a k - band spectrum of lkh@xmath0 225-south by the same authors shows strong ro - vibrational emission lines of molecular hydrogen . aspin et al . ( 1994 ) estimate the total luminosity of this object to be @xmath2 1600 l@xmath6 , which places it in the luminosity range of the herbig ae / be stars . palla et al . ( 1994 ) obtained near infrared , co , c@xmath7o , cs , and h@xmath1o maser observations of the group . their high resolution vla data show that a h@xmath1o maser source is clearly associated with lkh@xmath0 225-south . moreover , a density concentration in the molecular cloud ( as evidenced by cs j=54 emission ) and a co outflow are both associated with lkh@xmath0 225 . in their model , lkh@xmath0 225 is at the center of a dense molecular core of mass @xmath2 280 m@xmath6 , while bd+404124 lies near the periphery . continuum radio emission has been detected from bd+404124 and a source 42 to the east of bd+404124 ( skinner et al . 1993 ) . the infrared emission - line spectrum of young stellar objects ( ysos ) such as those in the bd+404124 group is dominated by the interaction of the central object with the remnants of the cloud from which it formed . if a young stellar object produces intense uv radiation , either due to accretion or because of a high temperature of the central star , an hii region will develop , producing a rich ionic emission line spectrum . at the edge of the hii region , a so - called photodissociation or photon - dominated region ( pdr ) will be created , in which neutral gas is heated by photoelectrically ejected electrons from grain surfaces . cooling of this neutral gas occurs mainly through emission in atomic fine - structure and molecular lines , allowing us to probe the physical conditions in the pdr through the study of its infrared emission - line spectrum . if the yso has a strong , often collimated , outflow , it will cause a shock wave as the outflow penetrates in the surrounding molecular cloud , heating the gas . in case the shock has a velocity smaller than @xmath2 40 km s@xmath5 , it will not dissociate the molecular material and mainly cool through infrared molecular transitions . because the physical conditions from the pre - shock gas to the post - shock gas change gradually within such a non - dissociative shock , they are often called c(ontinuous)-shocks . for higher shock velocities , the molecules will be dissociated and cooling in the shock occurs mainly through atomic and ionic emission lines . in such a shock , the physical conditions from the pre- to the post - shock gas will change within one mean free path and therefore they are usually termed j(ump)-shocks . in the post - shock gas of a j - shock , molecules will re - form and cool down the gas further . therefore a dissociative shock will produce an infrared spectrum consisting of a combination of ionic and molecular emission lines . in this paper we present _ infrared space observatory _ ( iso ; kessler et al . 1996 ) spectroscopic data obtained at the positions of bd+404124 , lkh@xmath0 224 and lkh@xmath0 225 . some of the h@xmath1 lines included in this data - set were already analyzed in a previous _ letter _ ( wesselius et al . 1996 ) . here we will re - analyze an extended set of h@xmath1 data using the latest iso calibrations , in combination with new data on infrared fine - structure lines , gas - phase molecular absorption bands and solid - state emission features in the bd+404124 group . we will show that the infrared emission lines arise from the combination of a hii region and a pdr at the position of bd+404124 , from a non - dissociative shock at the position of lkh@xmath0 224 , and from the combination of a non - dissociative and a dissociative shock at the position of lkh@xmath0 225 . we will also discuss the unusual absorption - line spectrum seen in the line of sight towards lkh@xmath0 225 and briefly discuss its implications for the physical and chemical evolution of hot cores . we obtained iso short wavelength ( 2.445 @xmath8 m ) spectrometer ( sws ; de graauw et al . 1996a ) and long wavelength ( 43197 @xmath8 m ) spectrometer ( lws ; clegg et al . 1996 ) grating scans at the positions of bd+404124 , lkh@xmath0 224 and lkh@xmath0 225 . for all three positions , both sws scans covering the entire sws wavelength range ( `` aot 01 '' ) and much deeper scans covering small wavelength regions around particularly useful diagnostic lines ( `` aot 02 '' ) were obtained . for lws only scans covering the entire wavelength region were obtained . a full log of the observations is given in table 1 . lcccc object & aot & rev . & date & jd@xmath92450000 + bd+404124 & sws 02 & 142 & 07/04/1996 & 181.037 + & sws 02 & 159 & 24/04/1996 & 197.635 + & sws 01 & 355 & 05/11/1996 & 393.283 + & lws 01 & 768 & 23/12/1997 & 805.525 + lkh@xmath0 224 & sws 02 & 142 & 07/04/1996 & 180.960 + & sws 01 & 858 & 22/03/1998 & 894.920 + lkh@xmath0 225 & sws 02 & 142 & 07/04/1996 & 180.996 + & sws 02 & 355 & 05/11/1996 & 393.306 + & sws 01 & 858 & 22/03/1998 & 894.889 + & lws 01 & 142 & 07/04/1996 & 181.017 + data were reduced in a standard fashion using calibration files corresponding to iso off - line processing software ( olp ) version 7.0 , after which they were corrected for remaining fringing and glitches . to increase the s / n in the final spectra , the detectors were aligned and statistical outliers were removed , after which the spectra were rebinned to a lower spectral resolution . figure 1 shows the resulting iso spectra . plots of all detected lines , rebinned to a resolution @xmath10 of 2000 with an oversampling factor of four ( sws ) , or averaged across scans ( lws ) , are presented in figs . line fluxes for detected lines and upper limits ( total flux for line with peak flux 3@xmath11 ) for the most significant undetected lines are listed in table 2 . for bd+404124 and lkh@xmath0 224 , all detected lines appear at their rest wavelength within the accuracy of the iso spectrometers . for lkh@xmath0 225 , all lines in the sws spectra are systematically shifted by @xmath2 @xmath9100 km s@xmath5 . this could either be a reflection of the real spatial velocity of lkh@xmath0 225 or be caused by a slight offset between our pointing , based on the optical position of the source , and the position of the infrared source . for all three sources , the lines are unresolved , indicating that the velocity dispersion is smaller than a few hundred km s@xmath5 . for each complete spectral scan , the sws actually makes twelve different grating scans , each covering a small wavelength region ( `` sws band '' ) , and with its own optical path . they are joined to form one single spectrum ( fig . 1 ) . because of the variation of the diffraction limit of the telescope with wavelength , different sws bands use apertures of different sizes . for a source that is not point - like , one may therefore see a discontinuity in flux at the wavelengths where such a change in aperture occurs . this effect can indeed be seen in some of our spectra , indicating the presence of extended far - infrared emission throughout the region . since both grating spectrometers on board iso use apertures that are fairly large compared to the separation of sources in most star forming regions , some caution is appropriate in interpreting such measurements . we created a plot with the positions of the sws apertures , overlaid on a k@xmath12-band image of the region ( hillenbrand et al . 1995 ) , shown in fig . as can be seen from this figure , only one strong near - infrared source is included in the bd+404124 and lkh@xmath0 224 measurements , whereas the lkh@xmath0 225 measurement only includes the question - mark shaped infrared nebula formed by the point - like objects lkh@xmath0 225-north and lkh@xmath0 225-south and the bridging nebula lkh@xmath0 225-middle . iso - lws has a beam that is larger ( 6183 fwhm ) than the separation of the objects studied here . therefore the lws measurements of bd+404124 and lkh@xmath0 225 partly cover the same region . the lws spectrum of lkh@xmath0 224 shown in fig . 1 is a weighted average of these spectra . accuracies of the absolute flux calibration in the sws spectra range from 7% in the short - wavelength ( @xmath13 4.10 @xmath8 m ) part to @xmath2 30% in the long wavelength ( @xmath14 29 @xmath8 m ) part ( leech et al . the lws absolute flux calibration is expected to be accurate at the 7% level ( trams et al . 1997 ) . note that the applied flux calibration is based on the assumption that we are looking at a point source . for an extended source , the diffraction losses will be underestimated and in particular at the long - wavelength part of the lws , these corrections will exceed the quoted uncertainty . a spectral energy distribution ( sed ) for bd+404124 was constructed by combining uv , optical , infrared , submm and radio data of bd+404124 from literature ( wesselius et al . 1982 ; strom et al . 1972a ; terranegra et al . 1994 ; hillenbrand et al . 1992 , 1995 ; cohen 1972 ; lorenzetti et al . 1983 ; di francesco et al . 1997 ; corcoran & ray 1998 ; skinner et al . 1993 ; bertout & thum 1982 ) with our new iso spectra ( fig . 6a ) . as can be seen from this plot , the optical to submm photometry forms a smooth curve . at radio wavelengths the slope of the energy distribution appears significantly flatter . we interpret this sed as the sum of three components : continuum emission from the central star in the uv and optical , thermal emission from dust heated by the central star in the infrared to submm and free - free emission from the stellar wind or an hii region at cm wavelengths . the sws spectrum of bd+404124 shows a rising continuum component starting around 13 @xmath8 m . this component is not present in the infrared photometry by cohen ( 1972 ) . this could be due to the smaller aperture size or the chopping throw used for background subtraction by cohen . we interpret this second dust component , only visible in the iso data , as diffuse emission throughout the star forming region . there is general consensus in the literature that the spectral type of bd+404124 is about b2 ve+sh ( merrill et al . 1932 ; herbig 1960 ; strom et al . 1972b ; swings 1981 ; finkenzeller 1985 ; hillenbrand et al . 1995 ) , corresponding to an effective temperature @xmath15 = 22,000 k and surface gravity @xmath16 = 4.0 ( schmidt - kaler 1982 ) . with this spectral type and the optical photometry by strom et al . ( 1972a ) we arrive at a value of @xmath17 towards the optical star . this value agrees well with the @xmath18 derived from the diffuse interstellar band at 5849 ( oudmaijer et al . 1997 ) . with this @xmath18 value and assuming a normal interstellar extinction law ( i.e. @xmath19 = @xmath20 = 3.1 ) , we corrected the data for extinction . a kurucz ( 1991 ) stellar atmosphere model with @xmath15 and @xmath16 corresponding to the star s spectral type ( assuming solar abundance ) was fitted to the extinction - corrected uv to optical sed . the fit is also shown in fig . 6a . as can be seen from this figure , the sed fit is not perfect in the uv . possibly a uv excess due to a strong stellar wind is present in bd+404124 . the fitted kurucz model was used to compute an apparent stellar luminosity @xmath21 for bd+404124 , which was converted to an absolute stellar luminosity using the bd+404124 distance estimate of 1 kpc by shevchenko et al . infrared luminosities were computed by fitting a spline to the infrared to mm data and subtracting the kurucz model from this fit . the derived values for the optical and infrared luminosity of bd+404124 are @xmath22 and @xmath23 l@xmath6 , respectively . these luminosities place the star close to the zero - age main sequence position of a 10 m@xmath6 star in the hertzsprung - russell diagram ( hrd ) . we followed the same procedure to construct a sed for lkh@xmath0 224 , combining optical , infrared and submm photometry from literature ( strom et al . 1972a ; shevchenko , personal communication ; hillenbrand et al . 1995 ; cohen 1972 ; di francesco et al . 1997 ) with our new iso spectra . it is shown in fig . 6b . since lkh@xmath0 224 is strongly variable in the optical , the data used in the sed were selected to be obtained near maximum system brightness , when available . the optical to infrared photometry of lkh@xmath0 224 forms a smooth curve in the sed , which we interpret as due to continuum emission from the star in the optical , and thermal emission from dust heated by the central star in the infrared . there is no agreement in the literature on the spectral type of lkh@xmath0 224 . wenzel ( 1980 ) classified the star spectroscopically as b8e , whereas hillenbrand et al . ( 1995 ) obtained a spectral type of b5 ve . corcoran & ray ( 1997 ) derived a spectral type of a0 for lkh@xmath0 224 , based upon the relative strength of the hei @xmath244471 and mgii @xmath244481 lines . in the optical , lkh@xmath0 224 shows large - amplitude photometric variations ( @xmath14 3 ) due to variable amounts of circumstellar extinction ( wenzel 1980 ; shevchenko et al . 1991 , 1993 ) . in the group of herbig ae / be stars , these large - amplitude variations are only observed when the spectral type is a0 or later ( van den ancker et al . therefore we favour a later spectral type than b for lkh@xmath0 224 . the detection of the caii h and k lines in absorption ( magakyan & movseyan 1997 ) also seems to favour a later spectral type than mid - b . recent optical spectroscopy of lkh@xmath0 224 by de winter ( personal communication ) shows strong , variable , p - cygni type emission in h@xmath0 . the higher members of the balmer series ( up to h15 ) are present and in absorption , together with the caii h and k lines . numerous shell lines are present in the spectrum as well , which could hinder spectral classification solely based upon red spectra . from the available optical spectroscopy , we classify lkh@xmath0 224 as a7e+sh , corresponding to @xmath15 = 7850 k. we estimate the uncertainty in this classification to be around 2 subclasses , or 400 k. with the new spectral type of a7e+sh and the optical photometry by strom et al . ( 1972a ) , we derive an @xmath18 at maximum system brightness for lkh@xmath0 224 of 096 . again fitting a kurucz model to the extinction - corrected photometry , we derive values for the optical and infrared luminosity of lkh@xmath0 224 of @xmath25 and @xmath26 l@xmath6 , respectively . the good fit of the sed to the kurucz model with @xmath15 = 7850 k shows that the photometry is consistent with our adopted spectral type of a7e for lkh@xmath0 224 . comparing the derived parameters of lkh@xmath0 224 with the pre - main sequence evolutionary tracks by palla & stahler ( 1993 ) , we see that it is located in the hrd at the location of a @xmath27 yr old 3.5 m@xmath6 star . this brings the age of the star in good agreement with those of the low - mass members of the bd+404124 group ( hillenbrand et al . 1995 ) . again the same procedure as above was followed to construct a sed for lkh@xmath0 225 ( fig . 6c ) , combining optical to radio measurements from literature ( shevchenko , personal communication ; hillenbrand et al . 1995 ; cohen 1972 ; weaver & jones 1992 ; di francesco et al . 1998 ; henning et al . 1998 ; cohen et al . 1982 ) with our iso data . these data refer to the sum of all components in this object . since lkh@xmath0 225 shows strong variability in the optical and infrared , only data obtained near maximum brightness was used in the sed . like in bd+404124 , the sed of lkh@xmath0 225 consists of a smooth continuum ranging from the optical to the submm . in view of the knowledge of the nature of lkh@xmath0 225 from literature , this sed must be interpreted as the sum of the three components of lkh@xmath0 225 in the optical to submm , each consisting of a hot component with circumstellar dust emission . the plotted iras fluxes of iras 20187 + 4111 ( weaver & jones 1992 ) are smaller than those in the lkh@xmath0 225 lws spectrum . most likely , this is due to the background subtraction applied on the iras data . this means that a cool component is present as well throughout the region , covering an area that is larger than the size of the iras detectors . in view of submm maps of the region ( aspin et al . 1994 ; henning et al . 1998 ) , this seems plausible . as can be seen in fig . 6c , below 8 @xmath8 m the flux levels in the sws spectrum are considerably smaller than those from ground - based photometry at maximum brightness . at longer wavelengths they agree . by extrapolating our iso sws spectrum , we deduce that the total @xmath28 band magnitude of lkh@xmath0 225 was about 83 at the time of our observation . this is within the range of values present in the literature . above 10 @xmath8 m , the sws flux levels are in agreement with ground - based measurements , suggesting that the large - amplitude variability of lkh@xmath0 225 is limited to the optical to near - infrared spectral range . the spectral type of lkh@xmath0 225 is very uncertain . hillenbrand et al . ( 1995 ) give a spectroscopic classification of mid a fe for both lkh@xmath0 225-north and lkh@xmath0 225-south , whereas wenzel ( 1980 ) derives a spectral type of g ke for the sum of both components . aspin et al . ( 1994 ) suggested the southern component to be of intermediate - mass , based on its high bolometric luminosity . unlike for bd+404124 and lkh@xmath0 224 , this makes the direct derivation of @xmath18 very uncertain . adopting @xmath18 = 497 , corresponding to @xmath29 derived from the silicate feature in the next section , results in an optical slope in the energy distribution that is steeper than the rayleigh - jeans tail of a black body this could be because the extinction at the time of the optical observations was smaller than during the iso observations . therefore we used a rather uncertain value of @xmath18 = 236 , based on the observed @xmath30 colour of lkh@xmath0 225 and the intrinsic colours of a f0 star , to correct the photometry of lkh@xmath0 225 for extinction . to obtain a rough measure for the total luminosity of lkh@xmath0 225 , we fitted a kurucz model with @xmath15 = 7200 k , corresponding to a spectral type of f0 , to the extinction - corrected optical photometry . this results in values of @xmath31 and @xmath32 l@xmath6 for the total optical and infrared luminosities of lkh@xmath0 225 . we conclude that at least one of the stars in the lkh@xmath0 225 system must be a massive ( 57 m@xmath6 ) pre - main sequence object in order to explain these luminosities . the iso full grating scans of bd+404124 consist of a relatively smooth continuum , with a number of strong emission lines superimposed . the familiar unidentified infrared ( uir ) emission features at 3.3 , 6.2 , 7.6 , 7.8 , 8.6 and 11.3 @xmath8 m , often attributed to polycyclic aromatic hydrocarbons ( pahs ) , are prominently present . possibly the 12.7 @xmath8 m feature also observed in other sources showing strong pah emission ( beintema et al . 1996 ) is present as well . no 9.7 @xmath8 m amorphous silicate feature , either in emission or in absorption , is visible in the bd+404124 sws spectrum . the slight curvature in the ground - based 813 @xmath8 m bd+404124 spectrum obtained by rodgers & wooden ( 1997 ) , which they interpreted as a silicate feature in emission , is absent in our sws data . the continuum in bd+404124 seems to consist of two distinct components , one from 2.413 @xmath8 m and the other from 13200 @xmath8 m . in the sws range , it is well fit by the superposition of two blackbodies of 320 and 100 k. for the lws spectrum a significant excess at the long - wavelength part remains after the black - body fit . probably a range in temperatures is present starting from 100 k down to much lower temperatures . the jumps in the spectrum at the positions corresponding to a change in aperture suggest that this component is caused by diffuse emission in the star forming region . remarkably , the line flux measured in the 3.3 @xmath8 m pah feature of bd+404124 ( @xmath33 w m@xmath34 ) is much larger than the @xmath35 w m@xmath34 upper limit obtained by brooke et al . pah features also appear absent in the 813 @xmath8 m hifogs spectrum shown by rodgers & wooden ( 1997 ) . this could either be an effect of the different aperture size used ( @xmath36 for sws versus a 14 circular aperture by brooke et al . ) in case the pah emission is spatially extended , or be due to time - variability of the pah emission . if the first explanation is correct , the pah emission should come from a region with a diameter larger than 32 and hence be easily resolvable with ground - based imaging . the iso spectrum of lkh@xmath0 224 consists of one single smooth continuum , with a few emission lines superimposed . no solid - state emission or absorption is apparent in the sws spectrum . again the slight curvature seen in the hifogs 813 @xmath8 m spectrum of lkh@xmath0 224 ( rodgers & wooden 1997 ) is absent in our sws data . in view of the fact that most herbig ae stars show strong silicate emission in the 10 @xmath8 m range ( e.g. van den ancker 1999 ) , the absence of the silicate feature is quite remarkable . it means that the dust emission must come from a region which is completely optically thick up to infrared wavelengths and yet only produce 3 magnitudes of extinction towards the central star in the optical . it is obvious that such a region can not be spherically symmetric , but must be highly flattened or have a hole through which we can seen the central star relatively unobscured . a dusty circumstellar disk might be the prime candidate for such a region . the absence of silicate emission in lkh@xmath0 224 might then be explained by either an intrinsically higher optical depth in its circumstellar disk than those surrounding other herbig ae stars , a view supported by its relative youth , or be caused by a nearly edge - on orientation of the circumstellar disk . the infrared continuum of lkh@xmath0 224 can not be fit by a single blackbody . rather a sum of blackbodies with a range of temperatures starting at several hundreds kelvins and continuing to temperatures of a few tens of k are required to fit the spectrum adequately . again the jumps in the spectrum at the positions corresponding to a change in aperture are present , suggesting that at the longer wavelengths we are looking at diffuse emission in the star forming region . however , the slope of the continuum in lkh@xmath0 224 is significantly different from that of the cool component in bd+404124 . this could either be due to a difference in the maximum temperature or to a difference in temperature gradient of the diffuse component between the two positions . the sws spectrum of lkh@xmath0 225 looks very different . again we have a strong , smooth continuum with many emission lines , but now also absorption lines due to gaseous molecular material as well as strong absorption features due to solid - state material are present . the familiar 9.7 @xmath8 m absorption feature due to the si o stretching mode in amorphous silicates is strong . apart from a 20% difference in the absolute flux levels , the silicate feature looks very similar to what was found by iras lrs ( olnon et al . 1986 ) , suggesting that it is constant in time . the integrated optical depth ( @xmath37 ) of the 9.7 @xmath8 m feature in the iso sws spectrum is 196 cm@xmath5 . the 18 @xmath8 m feature due to the bending mode of si o is also present in absorption with an integrated optical depth of 11 cm@xmath5 . h bending mode of water ice at 3.0 is present in absorption as well ( 156 cm@xmath5 ) . using intrinsic band strengths from literature ( tielens & allamandola 1987 ; gerakines et al . 1995 ) , we have converted the integrated optical depths of the 9.7 @xmath8 m silicate and 3.0 @xmath8 m h@xmath1o features to column densities . the results are a column of @xmath39 molecules for the silicates and @xmath40 molecules for h@xmath1o ice . the 2:1 ratio of these two species is within the range of what is found in other lines of sight ( whittet et al . 1996 and references therein ) . curiously , co@xmath1 ice , which invariably accompanies h@xmath1o ice in all other lines of sight studied by iso ( whittet et al . 1996 ; boogert et al . 1999 ; gerakines et al . 1999 ) is absent in lkh@xmath0 225 . the derived upper limit on the solid co@xmath1/h@xmath1o ratio is 0.04 , much less than the canonical value of 0.15 . no aperture jumps are visible in the sws spectrum of lkh@xmath0 225 . the bulk of the continuum flux must therefore come from an area that is small compared to the smallest sws aperture ( @xmath36 ) . the discontinuity between the sws and lws spectra shows that although lkh@xmath0 225 itself is probably the strongest far - infrared source in the region , the total far - infrared flux of the group is dominated by the diffuse component . like in lkh@xmath0 224 , the sws spectrum of lkh@xmath0 225 can not be fit well with a single blackbody . again a sum of blackbodies with a range of temperatures starting at several hundreds and continuing to temperatures of a few tens of k are required to fit the spectrum adequately . pahs appear absent in our lkh@xmath0 225 sws spectrum . in view of the claim by deutsch et al . ( 1994 ) that the southern component of lkh@xmath0 225 is the strongest uir emitter in the region , this is remarkable . one possible explanation for this apparent discrepancy could be the large degree of variability ( @xmath14 3 in @xmath28 ; allen 1973 ) of lkh@xmath0 225 . in this case the component responsible for the pah emission should have been much fainter than the other component at the time of observation . however , aspin et al . ( 1994 ) showed the southern component to be dominant longward of 3 @xmath8 m . comparison of our lkh@xmath0 225 spectrum with infrared photometry by previous authors shows that the source was not near minimum brightness at the time of the sws observation , making the explanation that the southern component of lkh@xmath0 225 was faint improbable . the most likely explanation for the apparent discrepancy between our non - detection and the reported uir emission ( deutsch et al . 1994 ) might therefore be that the uir emission within the southern component of lkh@xmath0 225 is variable in time . note that because the gas cooling timescale is much longer than the pah cooling timescale , if pah emission has been present within recent years , gas emission from a remnant pdr might be expected in the vicinity of lkh@xmath0 225 . since extinction in the continuum surrounding the 9.7 @xmath8 m feature is small compared to the extinction within this feature , the extinction @xmath41 at wavelength @xmath24 across a non - saturated 9.7 @xmath8 m feature can simply be obtained from the relation @xmath42 . using an average interstellar extinction law which includes the silicate feature ( fluks et al . 1994 ) , we can then convert these values of @xmath41 to a visual extinction , resulting in a value of @xmath43 toward lkh@xmath0 225 . this value is significantly smaller than the @xmath44 obtained from the c@xmath7o column density towards lkh@xmath0 225 ( palla et al . 1995 ) . although smaller , our estimate of @xmath45 is still compatible with the value of @xmath46 derived from the ratio of the 10 s(1 ) and q(3 ) lines of h@xmath1 ( aspin et al . in the sws spectrum of lkh@xmath0 225 several absorption bands are present due to a large column of gas - phase co , co@xmath1 and h@xmath1o in the line of sight . they are shown in fig . molecular absorption bands are absent in the bd+404124 and lkh@xmath0 224 spectra . we compared the observed gas - phase absorption lines to synthetic gas - phase absorption bands computed using molecular constants from the hitran 96 database ( rothmann et al . first the relative population of the different levels within one molecule were computed for an excitation temperature @xmath47 . for the relative abundances of the isotopes , values from the hitran database were adopted . a voigt profile was taken for the lines , with a doppler parameter @xmath48 = @xmath49 for the gaussian component of the voigt profile and the lifetime of the considered level times @xmath50 for the lorentz component . the optical depth @xmath51 as a function of wavelength was then computed by evaluating the sum of all integrated absorption coefficients , distributed on the voigt profile . synthetic absorption spectra were obtained by computing @xmath52 . this procedure for computing absorption spectra is nearly identical to that followed by helmich et al . ( 1996 ) and dartois et al . ( 1998 ) . for comparison with the observations , the spectra were convolved with the iso sws instrumental profile . for the aot 02 observations , the instrumental profile is nearly gaussian with a wavelength - dependent resolving power @xmath53 ranging between 1400 and 3000 ( leech et al . 1997 ) . for our aot 01 observations , the resolution is a factor four lower . best fits of the synthetic absorption spectra to the observed co , co@xmath1 and h@xmath1o bands are also shown in fig . because of the relatively low signal to noise in the spectra , the fits for the different absorption bands were not done independently , but @xmath47 and @xmath48 were required to be the same for all species . this procedure is only valid if the different molecular species are co - spatial . the results of similar fitting procedures in other lines of sight ( e.g. van dishoeck et al . 1996 ; dartois et al . 1998 ; boogert 1999 ) suggest that this is the case . satisfactory fits to the data could be obtained for excitation temperatures of a few 100 k and a doppler parameter @xmath48 between 3 and 10 km s@xmath5 . the derived column of warm co@xmath1 is several times 10@xmath54 cm@xmath34 . for h@xmath1o it is about 10@xmath56 cm@xmath34 , whereas the column of gas - phase co is in the range 10@xmath5610@xmath57 cm@xmath34 . the derived values of @xmath48 are much higher than those expected from purely thermal broadening ( @xmath58 km s@xmath5 , with @xmath59 the molecular weight in amu ; spitzer 1978 ) . this must be caused by macroscopic motion within the column of warm gas in our line of sight toward lkh@xmath0 225 . the fact that our deduced values of @xmath48 are within the range of those found toward hot cloud cores ( helmich et al . 1996 ; van dishoeck & helmich 1996 ; dartois et al . 1998 ) suggests that it is due to turbulence within the cloud rather than a dispersion in large - scale motion such as infall or rotation . the iso sws spectrum of bd+404124 contains several emission lines of the pfund and brackett series of hi . no hi lines were detected in lkh@xmath0 224 and lkh@xmath0 225 . the lines fluxes for bd+404124 were corrected for extinction using the value of @xmath45 = 30 derived in section 3 . we combined these data with ground - based measurements of emission lines in the brackett and paschen series ( harvey 1984 ; hamann & persson 1992 ; rodgers & wooden 1992 ) for bd+404124 to create plots of the hi line flux ratio as a function of upper level quantum number @xmath60 ( fig . the fact that the values for the brackett series lie on a continuous curve shows that they are not influenced much by the different beam sizes used , indicating that they must originate in a compact region . in fig . 8 we also show predicted values for menzel & baker ( 1938 ) case b recombination for a range of temperatures from 1000 to 30,000 k and densities from 10@xmath61 to 10@xmath62 @xmath4 , taken from storey & hummer ( 1995 ) . the fit of the case b values to the data is poor , indicating that at least the lower lines in the pfund and brackett series are not optically thin , as assumed in case b recombination theory , but have @xmath63 @xmath64 1 . an error in our estimate of the extinction towards the line emitting region can not reconcile our observations with the case b values . the poor fit of the case b model predictions to the observed hi line fluxes means that the lines do not originate in the low - density hii region responsible for the [ oiii ] emission ( section 8) . in the limit that all lines are optically thick , the line flux ratios of a static slab of gas in local thermodynamic equilibrium ( lte ) will be dominated by the wings of the line profile . such a slab of gas will produce line fluxes @xmath65 @xmath66 @xmath67 @xmath68 , with @xmath69 the planck function for an electron temperature @xmath70 , and @xmath71 the oscillator strength for the transition ( drake & ulrich 1980 ) . in fig . 8 we also show these values for a range of electron temperatures between 1000 and 100,000 k. again this simple model can not reproduce the observed line flux ratios in bd+404124 , especially for pf@xmath0/br@xmath0 and br@xmath72/br@xmath0 . the most likely explanation for the poor fit is that the lines are more broadened than due to the stark wings of the line profile , probably due to macroscopic motion of the gas . the poor fit of both the case b values and the optically thick static slab model shows that we are not looking at an optically thick static slab of gas , but at an optically thick ionized gas with a significant velocity dispersion . most likely it originates either in an ionized wind , or in an ionized gaseous disk surrounding bd+404124 . without information on the kinematics of the gas , the distinction between these two cases is hard to make . to qualitatively show that such a model can indeed reproduce the observations , we also show in fig . 8 the best fit of a simple lte model for a totally ionized wind following the treatment of nisini et al . this best fit model has a rather high mass loss rate ( @xmath2 10@xmath73 m@xmath6 yr@xmath5 ) , arising from a compact region ( @xmath2 12 r@xmath74 ) in radius . however , the assumption of lte will cause this simple model to significantly overestimate the mass loss rate ( bouret & catala 1998 ) , whereas the assumption that the wind is completely ionized will cause an underestimate of the size of the line emitting region . therefore the values given here should be approached as lower , respectively upper limits . also , a model for an ionized gaseous disk surrounding bd+404124 may give equally good fits to the data . if the lower lines in the pfund and brackett series are optically thick , the region in which these lines originate should also produce optically thick free - free emission at radio wavelengths . in the simple wind model employed above , the br@xmath0 line flux is related to the 6 cm continuum flux by the relation @xmath75 ( mjy ) = @xmath76 @xmath77 ( w m@xmath34 ) @xmath78 , with @xmath79 the wind velocity in units of 100 km s@xmath5 ( simon et al . assuming @xmath80 , this relation yields 0.39 mjy for bd+404124 , in excellent agreement with the observed 6 cm flux of 0.38 mjy ( skinner et al . although highly simplified , this analysis does show that the region responsible for the hi line emission can also produce sufficient amounts of radio continuum emission to account for the observed fluxes and that our interpretation of the radio component in the sed as free - free emission ( section 3 ) is correct . at all three observed positions in the bd+404124 region we detected emission from pure rotational lines of molecular hydrogen . they are listed in table 1 and are shown in figs . 24 . for bd+404124 and lkh@xmath0 224 , 00 s(0 ) to s(5 ) were detected . for lkh@xmath0 225 pure rotational lines up to s(10 ) as well as several ro - vibrational lines were detected . it is clear that in lkh@xmath0 225 the 00 s(6 ) line escaped detection because of its unfortunate location in the h@xmath1o gas phase @xmath81 absorption band ( fig . . the line flux of the 00 s(5 ) line may also be slightly ( @xmath2 10% ) underestimated because of this . we have detected significantly more lines than the preliminary results reported in the a&a iso first results issue ( wesselius et al . 1996 ) , both because of improvements in the sws data reduction since 1996 and because additional data were obtained . therefore a new analysis of these h@xmath1 lines is in place . from the h@xmath1 line fluxes @xmath82 it is possible to calculate the apparent column densities of molecular hydrogen in the upper @xmath83 levels , averaged over the sws beam , @xmath84 , using @xmath85 , with @xmath24 the wavelength , @xmath86 planck s constant and @xmath87 the speed of light . the transition probabilities @xmath88 were taken from turner et al . line fluxes were corrected for extinction using @xmath89 using the average interstellar extinction law by fluks et al . ( 1994 ) . for both bd+404124 and lkh@xmath0 224 we used @xmath45 = 30 , valid for the optical stars . for lkh@xmath0 225 we used the @xmath45 = 154 derived in section 4 . a useful representation of the h@xmath1 data is to plot the log of @xmath90 , the apparent column density in a given energy level divided by its statistical weight , versus the energy of the upper level . the statistical weight @xmath91 is the combination of the rotational and nuclear spin components . it is @xmath92 for para - h@xmath1 ( odd @xmath83 ) and @xmath92 times the ortho / para ratio for ortho - h@xmath1 ( even @xmath83 ) . we have assumed the high temperature equilibrium relative abundances of 3:1 for the ortho and para forms of h@xmath1 ( burton et al . the resulting excitation diagrams are shown in fig . 9 . for lkh@xmath0 225 we also included the measurements of the h@xmath1 ro - vibrational lines by aspin et al . for the lines in common , their values differ somewhat with ours , although the line ratios agree . this is probably an effect of their smaller beam size and indicates that in our measurements the beam filling factor is smaller than one . in fig . 9 the values of aspin et al . were scaled to our measurements . in case the population of the energy levels is thermal , all apparent column densities should lie on a nearly straight line in the excitation diagram . the slope of this line is inversely proportional to the excitation temperature , while the intercept is a measure of the total column density of warm gas . as can be seen from fig . 9 , this is indeed the case in all three stars for the pure rotational lines with an upper level energy up to 5000 k. the fact that the points for ortho and para h@xmath1 lie on the same line proves that our assumption on their relative abundances is correct . we have used the formula for the h@xmath1 column density for a boltzmann distribution given by parmar et al . ( 1991 ) to fit our data points in the low - lying pure rotational levels , using the rotational constants given by dabrowski ( 1984 ) , varying the excitation temperature and column density . the results are shown as the dashed lines in fig . van den ancker et al . ( 1998b and in preparation ) have employed predictions of h@xmath1 emission from pdr , j - shock and c - shock models ( burton et al . 1992 ; hollenbach & mckee 1989 ; kaufman & neufeld 1996 ) , to determine the excitation temperature from the low - lying pure rotational levels as a function of density @xmath60 and either incident fuv flux @xmath93 ( in units of the average interstellar fuv field g@xmath94 = @xmath95 erg cm@xmath34 s@xmath5 sr@xmath5 ; habing 1968 ) or shock velocity @xmath96 in an identical way as was done for the observations presented here . they arrived at the conclusion that the pdr and j - shock models allow a fairly small ( 200540 k ) range of excitation temperatures , whereas for c - shocks this range is much larger ( 1001500 k ) . in the model predictions for shocks , @xmath97 does not depend much on density , whereas for pdrs it does not depend much on @xmath93 . once the mechanism of the h@xmath1 emission is established , it can therefore be used to constrain @xmath96 or @xmath60 in a straightforward way . the values for @xmath97 measured for lkh@xmath0 224 and lkh@xmath0 225 fall outside the range of @xmath97 values predicted by either pdr or j - shock models . they are similar to the values found in other regions containing shocks ( van den ancker et al . if the observed pure rotational h@xmath1 emission is indeed due to a planar c - shock , its shock velocity should be around 20 km s@xmath5 . in section 4 we saw that our bd+404124 sws full grating scan contains strong emission in the pah bands . this indicates that a pdr is present at that location . if we assume that the observed h@xmath1 emission is entirely due to this pdr , the observed @xmath97 of 470 k points to a rather high ( @xmath98 ) density within the pdr . for lkh@xmath0 225 , ro - vibrational as well as pure rotational emission from h@xmath1 was detected ( fig . both the higher pure rotational lines and the ro - vibrational lines significantly deviate from the straight line defined by the lower pure rotational lines in the excitation diagram and do not form a smooth line by themselves , as expected in the case of multiple thermal components . such a situation can either occur because of fluorescence of h@xmath1 in a strong uv field ( black & van dishoeck 1987 ; draine & bertoldi 1996 ) , be due to x - ray or euv heating ( wolfire & knigl 1991 ; tin et al . 1997 ) , or be due to the formation energy of re - formed h@xmath1 in post - shock gas ( shull & hollenbach 1978 ) . although the h@xmath1 spectra resulting from the possible formation mechanisms do differ in the relative prominence of transitions with @xmath99 ( black & van dishoeck 1987 ) , the distinction between these situations is rather difficult to make with the detected lines in lkh@xmath0 225 . by summing all the apparent column densities for the ro - vibrational and higher pure rotational lines we obtain an estimate for the observed total column of non - thermal h@xmath1 in lkh@xmath0 225 of @xmath100 cm@xmath34 , corresponding to @xmath101 m@xmath6 or 0.3 earth masses within the sws beam . this value of @xmath102(h@xmath103 ) is within the range predicted by uv fluorescence models . however , the absence of pah emission in lkh@xmath0 225 at the time of the iso observations suggests that there was no strong uv radiation field at that time . therefore we consider the possibility that the ro - vibrational h@xmath1 emission originates in a shock more likely . interestingly , in lkh@xmath0 225 we have detected two pairs of lines arising from the same energy level : the 10 q(1 ) & 10 o(3 ) and the 10 q(3 ) & 10 o(5 ) lines . for these pairs of lines @xmath84 must be identical . therefore they can be used to derive a differential extinction between the two wavelengths of the lines using @xmath104 . using the interstellar extinction curve by fluks et al . ( 1994 ) , we can then convert the differential extinction to a value of @xmath45 . the results of this procedure are values for @xmath45 of @xmath105 and @xmath106 for the 10 q(1 ) & 10 o(3 ) and 10 q(3 ) & 10 o(5 ) line pairs , respectively . these extinction values are somewhat larger than the @xmath107 derived from the silicate feature in section 4 . but they agree well with @xmath44 obtained from the c@xmath7o column density towards lkh@xmath0 225 ( palla et al . if the slope of the infrared extinction law is approximately interstellar , the extinction towards the region producing the low - lying pure rotational h@xmath1 lines in lkh@xmath0 225 must be smaller than this value though ; such a large extinction would cause the extinction - corrected 00 s(3 ) line to deviate significantly from the thermal distribution line in the excitation diagram , since it is located within the 9.7 @xmath8 m silicate feature in the extinction curve . we conclude that in lkh@xmath0 225 either there must be two distinct sources of h@xmath1 emission , a 710 k thermal component suffering little extinction presumably a c - shock and a heavily extincted , deeply embedded component possibly a j - shock , or the slope of the infrared extinction law towards this region must be much steeper than interstellar . at both the positions of bd+404124 and lkh@xmath0 225 , several emission lines due to the @xmath108 = 00 transitions of gas - phase co were detected in the long - wavelength part of the lws spectra ( table 2 ) . similar to what was done for the h@xmath1 emission in the previous section , we constructed co excitation diagrams , using molecular data from kirby - docken & liu ( 1978 ) . they are shown in fig . the temperature and column of co resulting from the boltzmann fit to these excitation diagrams are 500 k and @xmath109 cm@xmath34 and 300 k and @xmath110 cm@xmath34 for bd+404124 and lkh@xmath0 225 , respectively . we estimate the errors in these fit parameters to be around 50 k in temperature and 50% in co column . the observed co lines have critical densities of around 10@xmath111 @xmath4 . therefore the observation of these lines also implies densities of at least this order of magnitude in the originating region . the co excitation temperature of 500 k found at the position of bd+404124 is comparable to that found from the h@xmath1 lines . the co / h@xmath1 mass fraction of 0.01 is compatible for what one would expect for the warm gas in a pdr . however , assuming that densities of around 10@xmath111 @xmath4 would exist in the entire bd+404124 pdr would be implausible . we therefore conclude that if these co lines and the h@xmath1 both originate in the large - scale environment of bd+404124 , the pdr must have a clumpy structure ( e.g. burton et al . 1990 ) . the co emission arising from the neighbourhood of lkh@xmath0 225 is remarkably different from that seen in h@xmath1 : here the excitation temperature of the co is much lower than that found in h@xmath1 . the co / h@xmath1 mass fraction of 0.03 is also much higher than that found at the position of bd+404124 . it seems likely that another component than the c - shock responsible for the h@xmath1 emission needs to be invoked to explain the low co excitation temperature . again this region must have densities of around 10@xmath111 @xmath4 . a possible disk or extended envelope around lkh@xmath0 225 could have both the temperature and densities required to explain the observed co spectrum . in order to interpret our observations as arising in either a pdr , c - shock or j - shock , we compare our results with theoretical models of such regions ( tielens & hollenbach 1985 ; hollenbach & mckee 1989 ; kaufman & neufeld 1996 ) . important constraints come from the observed fine structure lines . the mere presence of certain lines helps us to identify the mechanism responsible for the observed emission . in contrast to both c- and j - shocks , pdrs do not produce significant quantities of [ si ] emission ( tielens & hollenbach 1985 ) . c - shocks only contain trace fractions of ions and hence can not explain the [ feii ] , [ siii ] or [ cii ] emission . 0.17 cm lccccccccccc object & & & & & + & @xmath60 [ @xmath4 ] & @xmath93 [ g@xmath94 ] & @xmath114 [ 10@xmath115 sr ] & & @xmath60 [ @xmath4 ] & @xmath96 [ km s@xmath5 ] & @xmath114 [ 10@xmath115 sr ] & & @xmath60 [ @xmath4 ] & @xmath96 [ km s@xmath5 ] & @xmath114 [ 10@xmath115 sr ] + bd+404124 & 10@xmath117/10@xmath111 & 10@xmath117/10@xmath118 & 0.55/0.05 & & & & & & & & + lkh@xmath0 224 & & & & & @xmath2 10@xmath118 & 20 & @xmath2 0.10 & & & & + lkh@xmath0 225 & & & & & @xmath2 10@xmath118 & 20 & @xmath2 0.20 & & 10@xmath11710@xmath118 & 60 & 0.12 + in bd+404124 the h@xmath1 lines and the presence of pah emission strongly suggest the presence of a pdr . no [ si ] was detected , so no shock is required to explain the emission from this region . the [ feii ] and [ siii ] emission observed at the position of bd+404124 may also originate in a pdr . the [ oiii ] emission lines detected by sws can not be produced in a pdr . the ratio of the [ oiii ] 52 and 88 @xmath8 m lines is sensitive to the electron density @xmath119 in the line forming region . it hardly depends on the temperature . we computed the expected ratio for a range of temperatures and densities , in the approximation that we can treat [ oiii ] as a five level ion following the treatment by aller ( 1984 ) , using collisional strengths from aggarwal et al . the results are shown in fig . we conclude that the [ oiii ] lines observed by lws are formed in a hii region with @xmath119 = 0.62.2 @xmath120 @xmath121 @xmath4 . in view of the 22,000 k effective temperature of bd+404124 , it is surprising that [ oiii ] emission is detected , since simple photo - ionization models predict that for stars with @xmath15 @xmath13 32,000 k most oxygen should remain neutral . clearly another ionization model must be present here . possibly the region also responsible for the hi lines could collisionally ionize the oxygen and we are observing the outer , low - density part of an ionized wind . apart from the pdr , a hii region surrounding bd+404124 may also contribute significantly to the observed [ feii ] and [ siii ] emission . therefore we have used the photo - ionization code cloudy ( version 90.04 ; ferland 1996 ) to generate model predictions for line strengths for an hii region surrounding bd+404124 . a kurucz ( 1991 ) model with parameters appropriate for bd+404124 was taken as the input spectrum , and a spherical geometry and a constant electron density of 2 @xmath120 @xmath121 @xmath4 throughout the hii region were assumed . according to this model , 30% of the observed [ siii ] 34.82 @xmath8 m flux and 40% of the 25.99 @xmath8 m [ feii ] flux will originate in the hii region . from comparing the remainder of the fluxes of the atomic fine structure lines to the pdr models of tielens & hollenbach ( 1985 ) , we derive that the pdr should be fairly dense ( @xmath2 10@xmath117 @xmath4 ) and have an incident far - uv field of about 10@xmath117 g@xmath94 . the extent of this pdr should be around 180 square arcseconds , or about half of the sws beam . in view of the observed h@xmath1 spectrum ( section 7 ) , a c - shock appears the most likely candidate for explaining the observed [ si ] emission in lkh@xmath0 224 . a c - shock with a shock velocity of around 20 km s@xmath5 can explain both the h@xmath1 and [ si ] emission . however , the density and extent of the region are poorly constrained : the emission may either arise in a dense ( @xmath2 10@xmath111 @xmath4 ) region of only a few square arcseconds , or arise in a region of about 10@xmath6110@xmath117 @xmath4 with a high beam filling factor . since it is observed in the largest sws aperture , the observed [ siii ] emission could be due to contamination from lkh@xmath0 225 ( see fig . 5 ) . for lkh@xmath0 225 the situation appears more complicated . the detection of [ si ] clearly shows that a shock must be present in the region . a c - shock can explain the observed thermal h@xmath1 component , with similar parameters as for lkh@xmath0 224 , but can only explain part of the [ si ] emission . the [ feii ] and [ siii ] emission could either come from a deeply embedded hii region , a deeply embedded pdr , or a j - shock . in view of the necessity to have an additional source of [ si ] emission , the most likely candidate might be a j - shock . a j - shock model with @xmath96 @xmath2 60 km s@xmath5 and @xmath60 @xmath2 10@xmath11710@xmath118 @xmath4 can indeed reproduce the observed [ si ] and [ siii ] intensities . however , to reproduce the observed [ feii ] 26.0 @xmath8 m strength , iron would have to be about eight times more depleted than assumed in the tielens & hollenbach pdr models . the observed [ cii ] 157.7 @xmath8 m emission at the position of lkh@xmath0 225 can not be explained by shocks . this could very well be due to contamination by the bd+404124 pdr or the galactic [ cii ] background , however . fit parameters for the pdr , j - shock and c - shock models used to explain the h@xmath1 , co and fine structure emission lines are listed in table 3 . the summed model predictions for the line fluxes are listed in table 2 . in the previous sections we have seen that in a small ob association like the bd+404124 group , different phenomena contribute to at first glance similar line spectra . the emission at the position of bd+404124 is well reproduced by the combination of a hii region around bd+404124 with a pdr . in section 7 we derived values of the far - uv radiation field for its pdr from the observed emission lines . it is useful to compare those values to those expected from the central star to see whether that is a sufficient source of radiation , or another exciting source needs to be invoked . from the kurucz ( 1991 ) model with @xmath15 = 22,000 k and @xmath122 , fitted to the extinction - corrected uv to optical sed of bd+404124 ( section 3 ) , we compute that bd+404124 has a total far - uv ( 613.6 ev ) luminosity of @xmath123 l@xmath6 . at a location @xmath124 au from the central star , this field will have diluted to the value of @xmath125 g@xmath94 obtained from the infrared lines . at the bd+404124 distance of 1 kpc , this corresponds to an angular separation of 5 , consistent with the constraints imposed by the sws aperture . we conclude that bd+404124 can indeed be responsible for the incident uv flux upon the pdr observed in that vicinity . compared to the far - uv luminosity of bd+404124 , the other two sources only show little uv emission ( 3 and 6 l@xmath6 for lkh@xmath0 224 and 225 , respectively ) , explaining why we do nt observe a pdr at those positions . lccccccccccccc object & & & & & & & ref . + & co & co@xmath1 & h@xmath1o & & co & co@xmath1 & h@xmath1o & & co & co@xmath1 & h@xmath1o & & + lkh@xmath0 225 & 10@xmath5610@xmath57 & 15 @xmath120 10@xmath54 & 0.53 @xmath120 10@xmath56 & & @xmath13 6 @xmath120 10@xmath54 & @xmath13 3 @xmath120 10@xmath126 & 7.8 @xmath120 10@xmath54 & & @xmath14 16 & @xmath14 3 & @xmath2 13 & & ( 1 ) + t tau & 10@xmath710@xmath56 & @xmath13 1 @xmath120 10@xmath54 & @xmath13 1 @xmath120 10@xmath7 & & @xmath13 5 @xmath120 10@xmath54 & 1.4 @xmath120 10@xmath54 & 5.4 @xmath120 10@xmath54 & & @xmath14 20 & @xmath13 0.7 & @xmath13 1.9 & & ( 2 ) + afgl 4176 & 2 @xmath120 10@xmath56 & 5 @xmath120 10@xmath127 & 2 @xmath120 10@xmath7 & & @xmath13 5 @xmath120 10@xmath126 & 1.2 @xmath120 10@xmath54 & 9 @xmath120 10@xmath54 & & @xmath14 400 & @xmath2 0.04 & @xmath2 2.2 & & ( 3,4,5,6 ) + w33a & @xmath128 & & @xmath129 & & @xmath130 & @xmath131 & @xmath132 & & @xmath2 40 & & @xmath2 0.04 & & ( 3,7,8,9 ) + afgl 2136 & 1.0 @xmath120 10@xmath56 & 1 @xmath120 10@xmath126 & 2 @xmath120 10@xmath7 & & 1.8 @xmath120 10@xmath54 & 6.1 @xmath120 10@xmath54 & 5.0 @xmath120 10@xmath7 & & @xmath2 50 & @xmath2 0.02 & @xmath2 0.4 & & ( 3,4,5,6 ) + rafgl 7009s & 6.1 @xmath120 10@xmath7 & 1.0 @xmath120 10@xmath54 & @xmath133 2 @xmath120 10@xmath7 & & 1.8 @xmath120 10@xmath7 & 2.5 @xmath120 10@xmath7 & 1.1 @xmath120 10@xmath56 & & @xmath2 3.4 & @xmath2 0.04 & @xmath133 0.18 & & ( 10 ) + elias 29 & @xmath134 & & @xmath135 & & @xmath136 & @xmath137 & @xmath138 & & @xmath14 55 & & @xmath14 0.3 & & ( 9 ) + afgl 2591 & 1.1 @xmath120 10@xmath56 & 1 @xmath120 10@xmath126 & 2 @xmath120 10@xmath7 & & @xmath13 4 @xmath120 10@xmath126 & 2.7 @xmath120 10@xmath54 & 1.7 @xmath120 10@xmath7 & & @xmath14 270 & @xmath2 0.04 & @xmath2 1.1 & & ( 3,4,5,6 ) + ngc 7538 irs9 & 1.4 @xmath120 10@xmath56 & 8.0 @xmath120 10@xmath127 & @xmath13 3 @xmath120 10@xmath54 & & 1.0 @xmath120 10@xmath7 & 1.2 @xmath120 10@xmath7 & 8.0 @xmath120 10@xmath7 & & @xmath2 14 & @xmath2 0.01 & @xmath13 0.04 & & ( 3,4,5,6 ) + references : ( 1 ) this paper ; ( 2 ) van den ancker et al . ( 1999 ) ; ( 2 ) mitchell et al . ( 1990 ) ; ( 3 ) van dishoeck et al . ( 1996 ) ; ( 4 ) de graauw et al . ( 1996b ) ; ( 5 ) van dishoeck & helmich ( 1996 ) ; ( 6 ) chiar et al . ( 1998 ) ; ( 7 ) gerakines et al . ( 1999 ) ; ( 8) boogert ( 1999 ) ; ( 9 ) dartois et al . ( 1998 ) . for lkh@xmath0 224 we showed that the infrared emission - line spectrum could be explained by a single non - dissociative shock . a c - shock with very similar parameters may also be present at the position of lkh@xmath0 225 . comparing our iso sws aperture positions ( fig . 5 ) with the image of the bipolar outflow centered on lkh@xmath0 225 ( palla et al . 1995 ) , we note that this outflow covers both lkh@xmath0 224 and lkh@xmath0 225 . one possible identification for the c - shock in both lkh@xmath0 224 and lkh@xmath0 225 might therefore be the shock created as the outflow originating from lkh@xmath0 225 drives into the surrounding molecular cloud . apart from the large - scale c - shock , we have seen that probably a j - shock is present in lkh@xmath0 225 as well . this can be naturally identified with the herbig - haro knot lkh@xmath0 225-m observed in [ sii ] in the optical ( magakyan & movseyan 1997 ) . it remains unclear whether the ro - vibrational h@xmath1 lines observed in this object are due to h@xmath1 fluorescence by ly@xmath0 photons or to h@xmath1 re - formation in the post - shock gas . spectroscopy of lines with higher @xmath108 is required to clarify this . in the line of sight toward lkh@xmath0 225 , a large column consisting of warm gas - phase co , co@xmath1 and h@xmath1o and solid water ice and silicates was detected . the column of gas - phase water was found to be about 10 times higher that that of water ice . no co@xmath1 ice was detected toward lkh@xmath0 225 , demonstrating that the bulk of the co@xmath1 is in the gas - phase . in table 4 we compare the derived gas / solid ratio for lkh@xmath0 225 with those found in the lines of sight towards other young stellar objects . both the h@xmath1o and co@xmath1 gas / solid ratios are much higher than those found in other ysos . for co@xmath1 , the lower limit to the gas / solid is even a factor 100 higher than that found in the source with the previously reported highest ratio ! at this point we can only speculate as to why this is the case . one possible explanation could be that the radiation field of bd+404124 is heating the outer layers of the lkh@xmath0 225 envelope , evaporating the co@xmath1 ice . if this is the case , bd+404124 should be located closer to the observer than lkh@xmath0 225 , suggesting that lkh@xmath0 225 is located at the far side of the `` blister '' in which the bd+404124 group is located . another explanation could be that the lkh@xmath0 225 core itself is unusually warm . since the abundance of gas - phase water is sensitive to the temperature ( charnley 1997 ) , this would also explain the high abundance of water vapour in the line of sight toward lkh@xmath0 225 . because lkh@xmath0 225 is less massive than the luminous objects in which the previous gas / solid ratios have been determined , this high temperature might reflect a more evolved nature of the lkh@xmath0 225 core . future research , such as a temperature determination from gas - phase co measurements , should be able to distinguish between these possibilities and show whether we have discovered the chemically most evolved hot core known to date , or are looking through a line of sight particularly contaminated by a nearby ob star . aggarwal , k.m . , baluja , k.l . , tully , j.a . 1982 , mnras 201 , 923 allen , d.a . 1973 , mnras 161 , 145 aller , l.h . 1984 , _ `` physics of thermal gaseous nebulae '' _ , dordrecht , reidel aspin , c. , sandell , g. , weintraub , d.a . 1994 , a&a 282 , l25 beintema , d.a . , van den ancker , m.e . , molster , f.j . 1996 , a&a 315 , l369 bertout , c. , thum , c. 1982 , a&a 107 , 368 black , f.h . , van dishoeck . e.f . 1987 , apj 322 , 412 boogert , a.c.a . 1999 , phd thesis , groningen university boogert , a.c.a . , ehrenfreund , p. , gerakines , p. et al . 1999 , a&a , in press bouret , j.c . , catala , c. 1998 , a&a 340 , 163 brooke , t.y . , tokunaga , a.t . , strom , s.e . 1993 , aj 106 , 656 burton , m.g . , hollenbach , d.j . , tielens , a.g.g.m . 1990 , apj 365 , 620 burton , m.g . , hollenbach , d.j . , tielens , a.g.g.m . 1992 , apj 399 , 563 charnley , s.b . 1997 , apj 481 , 396 chiar , j.e . , gerakines , p.a . , whittet , d.c.b . 1998 , apj 498 , 716 cohen , m. 1972 , apj 173 , l61 cohen , m. , bieging , j.h . , schwartz , p.r . 1982 , apj 253 , 707 corcoran , m. , ray , t.p . 1996 , a&a 321 , 189 corcoran , m. , ray , t.p . 1998 , a&a 331 , 147 dabrowski , i. 1984 , canadian j. phys . 62 , 1639 dartois , e. , dhendecourt , l. , boulanger , f. , jourdain de muizon , m. , breitfellner , m. , puget , j.l . , habing , h.j . 1998 , a&a 331 , 651 de graauw , th . , haser , l.n . , beintema , d.a . 1996a , a&a 315 , l49 de graauw , th . , whittet , d.c.b . , gerakines , p.a . 1996b , a&a 315 , l345 deutsch , l.k . , iyengar , m.a . , hora , j.l . , hoffmann , w.f . , dayal , a. , butner , h.m . , fazio , g.g . 1994 , aas 185 , 8416 di francesco , j. , evans , n.j . , harvey , p.m. , mundy , l.g . , guilloteau , s. , chandler , c.j . 1997 , apj 482 , 433 draine , b.t . , bertoldi , f. 1996 , apj 468 , 269 drake , s.a . , ulrich , r.k . 1980 , apjs 42 , 351 ferland , g.j . 1996 , univ . of kentucky physics department internal report finkenzeller , u. 1985 , a&a 151 , 340 fluks , m.a . , plez , b. , th , p.s . , de winter , d. , westerlund , b.e . , steenman , h.c . 1994 , a&as 105 , 311 gerakines , p.a . , schutte , w.a . , greenberg , j.m . , van dishoeck , e.f . 1995 , a&a 296 , 810 gerakines , p.a . , whittet , d.c.b . , ehrenfreund , p. et al . 1999 , apj 522 , 357 habing , h.j . 1968 , bull . netherlands 19 , 421 hamann , f. , persson , s.e . 1992 , apjs 82 , 285 harvey , p.m. 1984 , pasp 96 , 297 helmich , f.p . , van dishoeck , e.f . , black , j.h . 1996 , a&a 315 , l173 henning , t. , burkert , a. , launhardt , r. , leinert , c. , stecklum , b. 1998 , a&a 336 , 565 herbig , g.h . 1960 , apjs 4 , 337 hillenbrand , l.a . , strom , s.e . , vrba , f.j . , keene , j. 1992 , apj 397 , 613 hillenbrand , l.a . , meyer , m.r . , strom , s.e . , skrutskie , m.f . 1995 , aj 109 , 280 hollenbach , d.j . , mckee , c.f . 1989 , apj 342 , 306 kaufman , m.j . , neufeld , d.a . 1996 , apj 456 , 611 kessler , m.f . , steinz , j.a . , anderegg , m.e . 1996 , a&a 315 , l27 kirby - docken , k. , liu , b. 1978 , apjs 36 , 359 kurucz , r.l . 1991 , in _ `` stellar atmospheres beyond classical models '' _ ( eds . davis philip , a.r . upgren , k.a . janes ) , l. davis press , schenectady , new york , p. 441 leech , k. et al . 1997 , `` sws instrument data users manual '' , issue 3.1 , sai/95 - 221/dc lorenzetti , d. , saraceno , p. , strafella , f. 1983 , apj 264 , 554 magakyan , t.y . , movsesyan , t.a . 1997 , pisma astron . 23 , 764 ( astron . 23 , 666 ) menzel , d.h . , baker , j.g . 1938 , apj 88 , 52 merrill , p.w . , humason , m.l . , burwell , c.g . 1932 , apj 76 , 156 mitchell , g.f . , maillard , j.p . , allen , m. , beer , r. , belcourt , k. 1990 , apj 363 , 554 nisini , b. , milillo , a. , saraceno , p. , vitali , f. 1997 , a&a 302 , 169 olnon , f.m . , raimond , e. & iras science team 1986 , a&as 65 , 607 oudmaijer , r.d . , busfield , g. , drew , j.e . 1997 , mnras 291 , 797 palla , f. , stahler , s.w . 1993 , apj 418 , 414 palla , f. , testi , l. , hunter , t.r . , taylor , g.b . , prusti , t. , felli , m. , natta , a. , stanga , r.m . 1995 , a&a 293 , 521 parmar , p.s . , lacy , j.h . , achtermann , j.m . 1991 , apj 372 , l25 rodgers , b. , wooden , d.h . 1997 , aas 191 , 4709 rothmann , l.s . , gamache , r.r . , tipping , r.h . 1996 , j. quant . transfer 48 , 469 schmidt - kaler , th . 1982 , in `` landolt brnstein catalogue '' , vol vi/2b shevchenko , v.s . , ibragimov , m.a . , chernysheva , t.l . 1991 , astron . 68 , 466 ( sva 35 , 229 ) shevchenko , v.s . , grankin , k.n . , ibragimov , m.a . , melnikov , s.y . , yakubov , s.d . 1993 , ap&ss 202 , 121 shull , j.m . , hollenbach , d.j . 1978 , apj 220 , 525 simon , m. , felli , m. , cassar , l. , fischer , j. , massi , m. 1983 , apj 266 , 623 skinner , s.l . , brown , a. , stewart , r.t . 1993 , apjs 87 , 217 spitzer , l. 1978 , `` physical processes in the interstellar medium '' , wiley interscience , new york storey , p.j . , hummer , d.g . 1995 , mnras 272 , 41 strom , k.m . , strom , s.e . , breger , m. , brooke , a.l . , yost , j. , grasdalen , g. , carrasco , l. 1972a , apj 173 , l65 strom , s.e . , strom , k.m . , yost , j. , carrasco , l. , grasdalen , g. 1972b , apj 173 , 353 swings , j.p . 1981 , a&as 43 , 331 terranegra , l. , chavarra , c. , diaz , s. , gonzalez - patino , d. 1994 , a&as 104 , 557 tielens , a.g.g.m . , allamandola , l.j . 1987 , in _ `` interstellar processes '' _ , eds . hollenbach & h.a . thronson jr . ( dordrecht : reidel ) , p. 397 tielens , a.g.g.m . , hollenbach , d.j . 1985 , apj 291 , 722 tin , s. , lepp , s. , gredel , r. , dalgarno , a. 1997 , apj 481 , 282 trams , n. et al . 1997 , `` iso - lws instrument data users manual '' , issue 5.0 , sai/95 - 219/dc turner , j. , kirby - docken , k. , dalgarno , a. 1977 , apjs 35 , 281 van den ancker , m.e . , de winter , d. , tjin a djie , h.r.e . 1998a , a&a 330 , 145 van den ancker , m.e . , wesselius , p.r . , tielens , a.g.g.m . , waters , l.b.f.m . 1998b , in _ `` iso s view on stellar evolution '' _ , eds . waters et al . , ap&ss 255 , 69 van den ancker , m.e . 1999 , phd thesis , university of amsterdam van den ancker , m.e . , wesselius , p.r . , tielens , a.g.g.m . , van dishoeck , e.f . , spinoglio , l. 1999 , a&a 348 , 877 van dishoeck , e.f . , helmich , f.p . 1996 , a&a 315 , l177 van dishoeck , e.f . , helmich , f.p . , de graauw , th . 1996 , a&a 315 , l349 weaver , w.b . , jones , g. 1992 , apjs 78 , 239 wenzel , w. 1980 , mitt . ver . sterne 8 , 182 wesselius , p.r . , van den ancker , m.e . , young , e.t . 1996 , a&a 315 , l197 wesselius , p.r . , van duinen , r.j . , de jonge , a.r.w . , aalders , j.w.g . , luinge , w. , wildeman , k.j . 1982 , a&as 49 , 427 whittet , d.c.b . , schutte , w.a . , tielens , a.g.g.m . 1996 , a&a 315 , l357 wolfire , m.g . , knigl , a. 1991 , apj 383 , 205	we present the results of iso sws and lws grating scans towards the three brightest members of the bd+404124 group in the infrared : bd+404124 ( b2ve ) , lkh@xmath0 224 ( a7e ) and the embedded source lkh@xmath0 225 . emission from the pure rotational lines of h@xmath1 , from ro - vibrational transitions of co , from pahs , from hi recombination lines and from the infrared fine structure lines of [ feii ] , [ siii ] , [ si ] , [ oi ] , [ oiii ] and [ cii ] was detected . these emission lines arise in the combination of a low - density ( @xmath2 10@xmath3 @xmath4 ) hii region with a clumpy pdr in the case of bd+404124 . the lower transitions of the infrared hi lines observed in bd+404124 are optically thick ; most likely they arise in either a dense wind or a circumstellar disk . this same region is also responsible for the optical hi lines and the radio continuum emission . in the lines of sight towards lkh@xmath0 224 and lkh@xmath0 225 , the observed emission lines arise in a non - dissociative shock produced by a slow ( @xmath2 20 km s@xmath5 ) outflow arising from lkh@xmath0 225 . toward lkh@xmath0 225 we also observe a dissociative shock , presumably located closer to the outflow source than the non - dissociative shock . in the line of sight towards lkh@xmath0 225 we observed absorption features due to solid water ice and amorphous silicates , and due to gas - phase h@xmath1o , co and co@xmath1 . no solid co@xmath1 was detected towards lkh@xmath0 225 , making this the first line of sight where the bulk of the co@xmath1 is in the gas - phase .
superfluid pairing plays an important role in the study of nuclear structure . in infinite systems the pairing gap @xmath9 vanishes at a temperature @xmath10 , which is the critical temperature of the superfluid - normal ( sn ) phase transition . in deformed nuclei , the coriolis force , which breaks the cooper pairs , increases with the total angular momentum up to a certain critical angular momentum @xmath11 , where all cooper pairs are broken and the nucleus undergoes the sn phase transition , as suggested by the mottelson - valatin effect @xcite the combination of temperature and angular momentum effects causes the pairing reentrance phenomenon , which was predicted 50 years ago by kammuri @xcite . the physical interpretation of this phenomenon was given by moretto @xcite as follows . in spherical nuclei at @xmath12 0 , the total angular momentum @xmath13 is made up by the nucleons from the broken pairs , which occupy the single - particle levels around the fermi surface and block them against the scattered pairs . as @xmath13 increases , more levels around the fermi surface are blocked , which make the pairing correlations decrease until a sufficiently large total angular momentum @xmath14 , where the pairing gap @xmath9 completely vanishes in infinite systems . the increase of @xmath4 spreads the quasiparticles , which occupy the levels near the fermi surface , farther away from it . some levels become partially unoccupied , which are available for scattered pairs . as the result , when @xmath4 reaches a critical value @xmath15 , the pairing correlations are energetically favored , and the pairing gap reappears . as @xmath4 goes higher , the increase of a large number of quasiparticles quenches and eventually breaks down the pairing gap at @xmath16 @xmath17 . it is now well known that , because of thermal fluctuations in finite nuclei , the paring gap does not vanish at @xmath18 and/or @xmath11 , but moronically decreases with increasing @xmath4 and/or @xmath13 . the pairing reentrance in finite nuclei is also modified so that the pairing gap remains finite after its first reappearance at finite @xmath4 and @xmath13 . it has been shown within the phonon damping model ( pdm ) @xcite that the presence of a thermal pairing gap at @xmath19 makes the width of the giant dipole resonance ( gdr ) remain nearly unchanged at low temperatures ( @xmath20 1 mev ) in heavy nuclei . this lecture will show the most recent experimental facts that support this mechanism . it also demonstrates that the recently observed enhancement of the nuclear level densities @xcite is a clear evidence of pairing reentrance in @xmath1pd @xcite . in the description of the gdr width in hot rotating nuclei , the pdm has been extended to include the effect of angular momentum @xcite . it will be shown that the predictions of the pdm agree well with the latest experimental data for the gdr in highly excited @xmath2mo . finally , the gdr widths and energies predicted by the pdm and experimentally extracted are also used to calculate the shear viscosity of finite hot nuclei . the pairing pairing hamiltonian for a spherical system rotating about the symmetry @xmath21 axis , where the total angular momentum @xmath13 is completely determined by its @xmath21-projection @xmath22 alone , has the form @xmath23 , where @xmath24 are the creation ( annihilation ) operators of a particle ( neutron or proton ) in the @xmath25-th state , whereas @xmath26 , @xmath27 and @xmath28 are the single - particle energies , chemical potential and rotation frequency for an axially deformed system , respectively . this hamiltonian was used to derive the ftbcs1 equations , which consist of a set of ftbcs ( finite - temperature bcs ) -based equations , corrected by the effects of quasiparticle - number fluctuations . the ftbcs1 equation for the pairing gap is written as a sum of the level - independent term @xmath9 and level - dependent term @xmath29 as @xmath30 where @xmath31 , @xmath32 , @xmath33/2 $ ] , @xmath34/2 $ ] , @xmath35 , @xmath36^{-1}$ ] . with the quasiparticle - number fluctuations @xmath37 . the equations for the particle number and total angular momentum are @xmath38 $ ] , @xmath39 , respectively , with @xmath40 being the angular momentum projection quantum number . by solving these ftbcs1 equations , one can easily calculate the total energy @xmath41 and entropy @xmath42 of the system , from which the level density @xmath43 is calculated by using the invert laplace transformation of the grand partition function @xcite . the total level density @xmath44 of a system with the total energy @xmath41 is a sum of @xmath13-dependent level densities ( densities of states ) , namely @xmath45 . the @xmath13-dependent level density @xmath46 is calculated by differentiating @xmath43 , that is @xmath47 . recently , a series of measurements has been conducted at the bhabha atomic research center for the reaction @xmath48c + @xmath49nb @xmath50 @xmath51ag@xmath52 @xmath50 @xmath1pd@xmath52 @xmath53 p at the projectile energy @xmath54 40 - 45 mev @xcite . in these experiments the exclusive proton spectra in coincidence with a @xmath55-ray multiplicity detector array reveal broad structures at higher multiplicities and energies . the obtained spectra including these broad structures can be fitted by using the statistical model only after multiplying the phenomenological level densities for @xmath1pd by an enhancement function , which depends on excitation energy and angular momentum . it has been suggested in ref . @xcite that this enhancement might come from the pairing reentrance . the level - weighted pairing gaps @xmath56 ( with @xmath57 being the sum of all levels ) obtained within the conventional ftbcs show no pairing reentrance . the ftbcs neutron and proton gaps shown in panels ( a ) and ( b ) of fig . [ prolate ] decrease with increasing @xmath58 ( or @xmath4 ) and collapses at some @xmath59 , whose value decreases as @xmath13 increases . at the same time , the ftbcs1 gaps do not vanish at @xmath60 0 . at @xmath61 20@xmath62 they decrease monotonically as @xmath58 increases and remain finite even at @xmath60 15 mev . the pairing reentrance shows up at @xmath63 20@xmath62 for neutrons and at @xmath63 30@xmath62 for protons , where the gaps first increase with @xmath58 then decrease as @xmath58 increases further but do not vanish even at high @xmath58 . since this pairing reentrance takes place at rather high values of @xmath13 , where the pairing gaps already vanish within the ftbcs , its origin resides in the quasiparticle number fluctuations @xmath64 in the ftbcs1 gap equation . our preliminary calculations in ref . @xcite show that the corresponding level densities obtained within the ftbcs are smooth for all values of @xmath58 and @xmath13 , whereas those predicted by the ftbcs1 have a local enhancement at 2 @xmath65 4.5 mev for all values of @xmath13 under consideration . for @xmath66 0.276 , the ftbcs1 predictions agree fairly well with the empirical level densities obtained at @xmath67 30@xmath62 [ see figs . [ prolate ] ( c ) [ prolate ] ( f ) ] . for @xmath68 , a better agreement between theoretical predictions and empirical level densities is seen at @xmath69 30@xmath62 ( not shown ) . the fact that theoretical and empirical level densities agree well for @xmath67 30@xmath62 at @xmath66 0.276 , and for @xmath69 30@xmath62 at @xmath68 is a clear indication of a transition from a prolate shape at @xmath67 30@xmath62 to an oblate shape at higher @xmath13 . the pdm hamiltonian includes the independent single - particle ( quasiparticle ) field , gdr phonon field , and the coupling between them @xcite . the single - particle energies @xmath26 are obtained within the woods - saxon potentials . the gdr width @xmath70 consists of the quantal width , @xmath71 , and thermal width , @xmath72 , as @xmath73 . including thermal pairing , the quantal and thermal widths are given as @xmath74^{2}(1-n_{p}-n_{h } ) \delta[e_{\rm gdr}-e_{p}-e_{h}]~$ ] , and @xmath75^{2}(n_{s'}-n_{s } ) \delta[e_{\rm gdr}-e_{s}+e_{s'}]~,$ ] where @xmath76 , @xmath77 ( @xmath78 ) , @xmath79 are quasiparticle occupations numbers , which are well approximated with the fermi - dirac distribution for independent quasiparticles , @xmath80^{-1}$ ] . the parameter @xmath81 is chosen so that @xmath82 at @xmath12 0 is equal to gdr s width at @xmath12 0 , whereas the parameter @xmath83 is chosen so that , with varying @xmath4 , the gdr energy @xmath84 does not change significantly . the latter is found as the solution of the equation @xmath85 , where @xmath86 is the energy of the gdr phonon before the coupling between the phonon and single - particle mean fields is switched on , and @xmath87 is the polarization operator owing to this coupling ( its expression is given in ref . the gdr strength function is found as @xmath88/\{(\omega - e_{gdr})^2+[\gamma_q(\omega ) + \gamma_t(\omega)]^{2}\}~.$ ] the representation @xmath89 $ ] is used for the @xmath90-function with @xmath91 0.5 mev in numerical calculations . the gdr widths predicted by the pdm , thermal shape fluctuation model ( tsfm ) , and the phenomenological model called critical temperature included fluctuation model ( ctfm ) are shown in fig . [ width ] in comparison with the experimental systematics for @xmath92tc @xcite , @xmath93tl @xcite , and @xmath94pb @xcite . the pdm results for open shell nuclei include the effect of non - vanishing thermal pairing gap owing to thermal fluctuations in finite nuclei . the pdm is the only up - to - date semi - microscopic model that is able to describe well the experimental data in the entire temperature region including @xmath20 1 mev as well as high @xmath4 , where the gdr width seems to saturate . to describe the non - collective rotation of a spherical nucleus , the @xmath21-projection @xmath22 of the total angular momentum @xmath13 is added into the pdm hamiltonian as @xmath95 , where @xmath55 is the rotation frequency @xcite . the latter and the chemical potential are defined , in the absence of pairing , from the equation @xmath96 and @xmath97 , where @xmath98 is the particle number and @xmath99 are the single - particle occupation numbers , @xmath100 $ ] , and @xmath101 . the prediction of the pdm for the gdr line shape agrees remarkably well with the latest experimental date for @xmath2mo at high @xmath4 and @xmath13 @xcite , as shown in fig . for the application of hydrodynamics to nuclear system , the quantum mechanical uncertainty principle requires a finite viscosity for any thermal fluid . kovtun , son and starinets ( kss ) @xcite conjectured that the specific viscosity , which is the ratio @xmath5 of shear viscosity @xmath3 to the entropy volume density @xmath102 , is bounded at the lower end for all fluids by the universal value , which is called the kss bound ( or unit ) , @xmath103 . according to collective theories , one of the fundamental explanations for the giant resonance damping is the friction term ( viscosity ) of the neutron and proton fluids . by using the green - kubo s relation , ref . @xcite expresses the shear viscosity @xmath104 at finite @xmath4 in terms of the gdr s energies and widths at zero and finite @xmath4 as @xmath105^{2}}{e_{gdr}(t)^{2}+[\gamma(t)/2]^{2}}~. \label{eta1}\ ] ] the predictions for the shear viscosity @xmath3 and the specific viscosity @xmath5 by several theoretical models for @xmath106sn and @xmath94pb are plotted as functions of @xmath4 in fig . [ eta&ratio ] in comparison with the empirical results . the latter are extracted from the experimental systematics for gdr in tin and lead regions @xcite making use of eq . ( [ eta1 ] ) . it is seen from fig . [ eta&ratio ] that the predictions by the pdm agree best with the empirical results . the ratio @xmath5 decreases sharply with increasing @xmath4 up to @xmath107 1.5 mev , starting from which the decrease gradually slows down to reach ( 2 - 3 ) kss units at @xmath12 5 mev . the fermi - liquid drop model ( fldm ) has a similar trend up to @xmath107 2 - 3 mev , but at higher @xmath4 ( @xmath108 3 mev for @xmath106sn or 2 mev for @xmath94pb ) both @xmath3 and @xmath5 with @xmath4 sharply increase . at @xmath12 5 mev the fldm model predicts the ratio @xmath5 within ( 3.7 - 6.5 ) kss units , that is roughly 1.5 2 times larger than the pdm predictions . the tsfm in two versions as adiabatic model ( am ) and phenomenological tsfm ( ptsfm ) show a similar trend for @xmath3 and @xmath5 , provided @xmath109 in the ptsfm calculations is reduced to 0.72@xmath110 . however both versions of tsfm overestimate @xmath3 at @xmath111 1.5 mev . based on these results and on a model - independent estimation , it is concluded that the pdm predicts 1.3 @xmath112 4 kss units for medium and heavy nuclei at @xmath12 5 mev . this is about ( 3 - 5 ) times smaller ( and of much less uncertainty ) than the value between ( 4 - 19 ) kss units predicted by the fldm for heavy nuclei @xcite , where the same lower value @xmath1130.6@xmath110 was used . this lecture has confirmed that thermal pairing included in the phonon damping model ( pdm ) is indeed responsible for the nearly constant width of the giant dipole resonance ( gdr ) at low temperature @xmath0 1 mev . it also demonstrates that the enhancement , which has been observed in the recent experimentally extracted nuclear level densities in @xmath1pd at low excitation energy and various angular momenta , is the first experimental evidence of the pairing reentrance in finite ( hot rotating ) nuclei . the extension of the pdm to finite angular momentum offers predictions in excellent agreement with the latest experimental data of gdr in the compound nucleus @xmath2mo . the specific shear viscosity @xmath5 in several medium and heavy nuclei is calculated from the gdr widths and energies at zero and finite temperatures and found to decrease with increasing @xmath4 to reach @xmath6 at @xmath7 5 mev , which is almost the same value obtained for quark - gluon - plasma at @xmath8 170 mev . the calculations were carried out on the riken integrated cluster of clusters ( ricc ) system . thanks are due to m. ciemala for permission to use the experimental gdr line shapes in @xmath2mo @xcite . 9 b.r . mottelson and j.g . valatin , phys . rev . lett . * 5 * , 511 ( 1960 ) . t. kammuri , prog . . phys . * 31 * , 595 ( 1964 ) . l. g. moretto , phys . b * 35 * , 379 ( 1971 ) ; l.g . moretto , nucl . a * 185 * , 145 ( 1972 ) . n. dinh dang and a. arima , phys . lett * 80 * , 4145 ( 1998 ) ; n. dinh dang and a. arima , phys . c * 68 * , 044303 ( 2003 ) . et al . _ , a * 707 * , 343 ( 2002 ) ; a. mitra _ et al . _ , a * 765 * , 277 ( 2006 ) ; a. mitra _ et al . _ , j. phys . g * 36 * , 095103 ( 2009 ) ; a. mitra _ et al . _ , epj web of converences * 2 * , 04004 ( 2010 ) . v. datar , a. mitra , and d. chakrbarty , _ unpublished_. n. dinh dang , n. quang hung , b.k . agrawal , v.m . datar , a. mitra , and d.r . chakrabarty , _ in preparation_. n. dinh dang , phys . c * 85 * , 064323 ( 2012 ) . n. quang hung and n. dinh dang , phys . c * 84 * , 054324 ( 2011 ) . b. dey , d. mondal , d. pandit , s. mukhopadhyay , s. pal , s. bhattacharya , a. de , k. banerjee , n. dinh dang , n. quang hung , and s.r . banerjee , phys . b * 731 * , 92 ( 2014 ) . n. dinh dang and n. quang hung , phys . c * 86 * , 044333 ( 2012 ) . a. schiller and m. thoennessen , atomic data and nuclear data tables * 93 * , 548 ( 2007 ) . m. ciemala , ph.d . thesis , niewodniczanski institute of nuclear physics pan , krakow , report no . 2062/pl ( 2013 ) ( http://www.ifj.edu.pl/publ/reports/2013/2062.pdf ) p.k . kovtun , d.t . son , and a.o . starinets , phys . * 94 * , 111601 ( 2005 ) . n. dinh dang , phys . c * 84 * , 034309 ( 2011 ) . n. auerbach and s. shlomo , phys . . lett . * 103 * , 172501 ( 2009 ) .	recent results are reported showing the effects of thermal pairing in highly excited nuclei . it is demonstrated that thermal pairing included in the phonon damping model ( pdm ) is responsible for the nearly constant width of the giant dipole resonance ( gdr ) at low temperature @xmath0 1 mev . it is also shown that the enhancement observed in the recent experimentally extracted nuclear level densities in @xmath1pd at low excitation energy and various angular momenta is the first experimental evidence of the pairing reentrance in finite ( hot rotating ) nuclei . in the study of gdr in highly excited nuclei , the pdm has been extended to include finite angular momentum . the results of calculations within the pdm are found in excellent agreement with the latest experimental data of gdr in the compound nucleus @xmath2mo . finally , an exact expression is derived to calculate the shear viscosity @xmath3 as a function of @xmath4 in finite nuclei directly from the gdr width and energy at zero and finite @xmath4 . based on this result , the values @xmath5 of specific shear viscosity in several medium and heavy nuclei were calculated and found to decrease with increasing @xmath4 to reach @xmath6 at @xmath7 5 mev , that is almost the same value obtained for quark - gluon - plasma at @xmath8 170 mev .
the two asymmetric b - factories at slac and kek have provided a striking confirmation of the ckm paradigm @xcite . existing experimental information from the indirect cp violation parameter , @xmath0 for the @xmath1 , semileptonic @xmath2 and @xmath3 mixing along with lattice calculations predict that in the sm , @xmath4 @xcite . this is in very good agreement with the belle and babar result @xcite : + _ cp(b^0k^0 ) = 2=.726.037 [ psik ] this leads to the conclusion that the ckm phase of the standard model ( sm ) is the dominant contributor to @xmath5 . that , of course , also means that cp - odd phase(s ) due to beyond the standard model ( bsm ) sources may well cause only small deviations from the sm in b - physics . + actually , there are several reasons to think that bsm phase(s ) may cause only small deviations in b - physics . in this regard , sm itself teaches a very important lesson . we know now that the ckm phase is 0(1 ) ( actually , the cp violation parameter @xmath6 is 0(.3 ) @xcite ) . the cp effects that it causes on different observables though is quite different . in k - decays , the cp asymmetrics are @xmath7 . in charm physics , also there are good reasons to expect small observable effects . in top physics , the ckm phase causes completely negligible effects @xcite . thus only in b - decays , the large asymmetries ( often 0(1 ) ) are caused by the ckm phase . so even if the bsm phase(s ) are 0(1 ) it is unlikely that again in b - physics they will cause large effects just as the sm does . situation with regard to bsm cp - odd phase(s ) ( @xmath9 ) is somewhat reminiscent of the neutrino mass ( @xmath8 ) @xcite . there was no good reason for @xmath8 to be zero ; similarly , there are none for @xmath9 to be zero either . in the case of @xmath10 , there were the solar @xmath11 results that were suggestive for a very long time ; similarly , in the case of @xmath9 , the fact that in the sm , baryogenesis is difficult to accomodate serves as the beacon . + it took decades to show @xmath8 is not zero : @xmath12 had to be lowered from @xmath13 around 1983 down to @xmath14 before @xmath15 was established via neutrino oscillations . we can hope for better luck with @xmath9 but there is no good reason to be too optimistic ; therefore , we should not rely on luck but rather we should seriously prepare for this possibility . + to recapitulate , just as the sm - ckm phase is 0(1 ) , but it caused only @xmath16 cp symmetries in k- decays , an 0(1 ) bsm - cp - odd phase may well cause only very small asymmetries in b - physics . to search for such small effects : + 1 ) we need lots and lots of clean b s ( _ i.e. _ @xmath17 ) or more ) + 2 ) intensive study of @xmath18 mesons ( in addition to b s ) becomes very important as comparison between the two types of b - mesons will teach us how to improve quantitative estimates of flavor symmetry breaking effects . + 3 ) we also need clean predictions from theory ( wherein item 2 should help ) . improved searches for bsm - cp - odd phase(s ) can be subdivided into the following main categories : + + a ) indirect searches with theory input + b ) indirect searches without theory input + c ) direct searches . + among the four parameters of the ckm matrix , @xmath19 and @xmath6 , @xmath20 , @xmath21 @xcite are known quite precisely ; @xmath22 and @xmath6 still need to be determined accurately . efforts have been underway for many years to determine these parameters . the angles @xmath23 , of the unitarity triangle ( ut ) can be determined once one knows the 4-ckm parameters . a well studied strategy for determining these from experimental data requires knowledge of hadronic matrix elements . efforts to calculate several of the relevant matrix elements on the lattice , with increasing accuracy , have been underway for past many years . a central role is played by the following four inputs @xcite : * @xmath24 from the lattice with @xmath0 from experiment * @xmath25 from the lattice with @xmath26 from experiment * @xmath27 from the lattice with @xmath28 from experiment * @xmath29 from experiment , along with input from phenomenology especially heavy quark symmetry as well as the lattice . as mentioned above , for the past few years , these inputs have led to the important constraint : @xmath30 which is found to be in very good agreement with direct experimental determination , eq . [ psik ] . despite severe limitations ( e.g. the so - called quenched approximation ) these lattice inputs provided valuable help so that with b - factory measurements one arrives at the very important conclusion that in @xmath31 the ckm - phase is the dominant contributor ; any new physics ( np ) contribution is unlikely to be greater than about 15% . what sort of progress can we expect from the lattice in the next several years in these ( indirect ) determination of the ut ? to answer this it is useful to look back and compare where we were to where we are now . perhaps this gives us an indication of the pace of progress of the past several years . lattice calculations of matrix elements around 1995 @xcite yielded ( amongst other things ) @xmath32 , whereas the corresponding error decreased to around @xmath33 around 2001 @xcite . in addition to @xmath34 , such calculations also now constrain @xmath35 with an error of around @xmath36 @xcite . there are three important developments that should help lattice calculations in the near future : + 1 . exact chiral symmetry can be maintained on the lattice . this is especially important for light quark physics . 2 . relatively inexpensive methods for simulations with dynamical quarks ( esp . using improved staggered fermions @xcite ) have become available . this should help overcome limitations of the quenched approximation . 3 . about a factor of 20 increase in computing power is now being used compared to a few years ago . as a specific example one can see that the error on @xmath24 with the 1st use of dynamical domain wall fermions @xcite now seems to be reduced by about a factor of two @xcite . in the next few years or so errors on lattice determination of ckm parameters should decrease appreciably , perhaps by a factor of 3 . so the error in @xmath37 ; @xmath38 etc . while this increase in accuracy is very welcome , and will be very useful , there are good reasons to believe , experiment will move ahead of theory in direct determinations of unitarity angles in the next 5 years . ( at present , experiment is already ahead of theory for @xmath39 ) . one of the most exciting developments of recent years in b - physics is that methods have been developed so that all three angles of the ut can be determined cleanly with very small theory errors . this is very important as it can open up several ways to test the sm - ckm paradigm of cp violation ; in particular , the possibility of searching for small deviations . let us very briefly recapitulate the methods in question : + * time dependent cp asymmetry ( tdcpa ) measurements in @xmath40 type of final states should give the angle @xmath34 very precisely with an estimated irreducible theory error ( ite ) of @xmath41 @xcite . * direct cp ( dircp ) studies in @xmath42 gives @xmath43 very + cleanly @xcite . * tdcpa measurements in @xmath44 gives @xmath45 _ and _ also @xmath34 very cleanly @xcite . * in addition , tdcpa measurements in @xmath46 type modes also gives @xmath43 very cleanly @xcite . * determination of the rate for the cp violating decay @xmath47 is a very clean way to measure the wolfenstein parameter @xmath6 , which is indeed the cp - odd phase in the ckm matrix @xcite . + it is important to note that the ite for each of these methods is expected to be @xmath48 , in fact perhaps even @xmath49 . * finally let us briefly mention that , tdcpa studies of @xmath50 or @xmath51 or @xmath52 gives @xmath53 @xcite . however , in this case , isospin conservation needs be used and that requires , _ assuming _ that electro - weak penguins ( ewp ) make negligible contribution . this introduces some model dependence and may cause an error of order a few degrees , _ i.e. _ for @xmath53 extraction the ite may well end up being o(a few % ) . however , given that there are three types of final states each of which allows a determination of @xmath53 , it is quite likely that further studies of these methods will lead to a reduction of the common source of error originating from isopsin violation due to the ewp . + it is extremely important that we make use of these opportunities afforded to us by as many of these very clean redundant measurements as possible . in order to exploit these methods to their fullest potential and get the angles with errors of order ite will , for sure , require a super - b factory(sbf ) @xcite . + this in itself constitutes a strong enough reason for a sbf , as it represents a great opportunity to precisely nail down the important parameters of the ckm paradigm . below we briefly discuss why the precision extraction of @xmath43 seems so promising . for definiteness , let us recall the basic features of the ads method @xcite . in this interference is sought between two amplitudes of roughly similar size _ i.e. _ @xmath54 and @xmath55 where the @xmath56 and @xmath57 decay to common final states such as the simple two body ones like @xmath58 , @xmath59 , @xmath60 , @xmath61 or they may also be multibody modes e.g. the dalitz decay @xmath62 , @xmath63 etc . it is easy to see that the interference is between a colored allowed b decay followed by doubly cabibbo suppressed d decay and a color - suppressed b decay followed by cabibbo allowed d decay and consequently then interference tends to be maximal and should lead to large asymmetries . for a given ( common ) final state of @xmath56 and @xmath57 the amplitude involves three unknowns : the color suppressed br(@xmath55 ) , which is not directly accessible to experiment @xcite , the strong phase @xmath64 and the weak phase @xmath43 . corresponding to each such final state ( fs ) there are two observables : the rate for @xmath65 decay and for the @xmath66 decay . thus , if you stick to just one common fs of @xmath56 , @xmath57 , you do not have enough information to solve for @xmath43 . if you next consider two common fs of @xmath56 and @xmath57 then you have one additional unknown ( a strong phase ) making a total of 4-unknowns with also 4-observables . so with two final states the system becomes soluble , i.e. we can then use the experimental data to solve not only the value of @xmath43 but also the strong phases and the suppressed br for @xmath55 . with n common fs of @xmath56 and @xmath57 , you will have 2n observables and n + 2 unknowns . we need @xmath67 i.e. @xmath68 . the crucial point , though , is that there are a very large number of possible common modes of @xmath56 and @xmath57 which can all be used to improve the determination of @xmath43 . let us briefly mention some of the relevant common modes of @xmath56 and @xmath57 : * the cp - eigenstate modes , originally discussed by glw @xcite : @xmath69 [ @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 ; @xmath75 , .... * cp - non - eigenstates ( cpnes ) , discussed by gls @xcite : @xmath76 , @xmath77 ... these are singly cabibbo suppressed modes . * cpnes modes originally discussed by ads @xcite : @xmath78[@xmath79 , @xmath80 , @xmath81 .... ] * there are also many multibody modes , such as the dalitz @xmath56 decays : @xmath82 @xcite or @xmath62 @xcite etc ; and also modes such as @xmath83 , @xmath84 , or indeed @xmath85 @xcite . furthermore , multibody modes such as @xmath86 or @xmath87 @xcite can also be used . [ un_det ] and fig . [ over_det ] show how combining different strategies helps a great deal . in the fig we show @xmath88 versus @xmath43 . as indicated above when you consider an individual final state of @xmath56 and @xmath57 then of course there are 3 unknowns ( the strong phase , the weak phase ( @xmath43 ) and the unmeasureable " br ) and only two observables ( the rate for @xmath65 and the rate for @xmath66 ) . so in the figure , for a fixed value of @xmath43 , we search for the minimum of the @xmath88 by letting the strong phase and the unmeasureable " br take any value they want . fig . [ un_det ] and [ over_det ] show situtation with regard to under determined and over determined cases respectively . the upper horizontal line corresponds roughly to the low luminosity i.e. comparable to the current b - factories@xcite whereas the lower horizontal curve is relevant for a super b - factory . in fig . [ un_det ] in blue is shown the case when only the input from ( glw ) cpes modes of @xmath56 is used ; note all the cpes modes are included here . you see that the resolution on @xmath43 then is very poor . in particular , this method is rather ineffective in giving a lower bound ; its upper bound is better . in contrast , a single ads mode ( @xmath58 ) is very effective in so far as lower bound is concerned , but it does not yield an effective upper bound ( red ) . note that in these two cases one has only two observables and 3 unknowns . in purple is shown the situation when these two methods are combined . then at least at high luminosity there is significant improvement in attaining a tight upper bound ; lower bound obtained by ads alone seems largely unaffected . shown in green is another under determined case consisting of the use of a single ads mode , though it includes @xmath89 as well @xmath90 ; this again dramatically improves the lower bound . from an examination of these curves it is easy to see that combining information from different methods and modes improves the determination significantly @xcite . next we briefly discuss some over determined cases ( fig . [ over_det ] ) . in purple all the cpes modes of @xmath56 are combined with just one doubly cabibbo - suppressed ( cpnes ) mode . here there are 4 observables for the 4 unknowns and one gets a reasonable solution at least especially for the high luminosity case . the black curve is different from the purple one in only one respect ; the black one also includes the @xmath91 from @xmath92 where subsequently the @xmath91 gives rise to a @xmath56 . comparison of the black one with the purple shows considerable improvement by including the @xmath91 . in this case the number of observables ( 8) exceeds the number of unknowns ( 6 ) . actually , the @xmath91 can decay to @xmath56 via two modes : @xmath93 or @xmath94 . bondar and gershon @xcitehave made a very nice observation that the strong phase for the @xmath43 emission is opposite to that of the @xmath95 emission . inclusion ( blue curves ) of both types of emission increases the number of observables to 12 with no increase in number of unknowns . so this improves the resolving power for @xmath43 even more . the orange curves show the outcome when a lot more input is included ; not only @xmath96 , @xmath97 , @xmath56 , @xmath91 but also dalitz and multibody decays of @xmath56 are included . but the gains now are very modest ; thus once the number of observables exceeds the number of unknowns by a few ( say o(3 ) ) further increase in input only has a minimal impact . let us briefly recall that another important way to get these angles is by studying time - dependent cp ( tdcp ) ( or mixing - induced cp ( mixcp ) ) violation via @xmath98 . once again , all the common decay modes of @xmath56 and @xmath57 can be used just as in the case of direct cp studies involving @xmath99 decays . therefore , needless to say input from charm factory @xcite also becomes desirable for mixcp studies of @xmath98 as it is for direct cp using @xmath99 . it is important to stress that this method gives not only the combinations of the angles ( 2@xmath100 ) but also in addition this is another way to get @xmath34 cleanly @xcite . in fact whether one uses @xmath99 with dircp or @xmath101 with tdcp these methods are very clean with ( as indicated above ) the ite of @xmath102 . however , the tdcp studies for getting @xmath43 ( with the use of @xmath34 as determined from @xmath103 ) is less efficient than with the use of dircp involving @xmath99 . once we go to luminosities @xmath104 , though , the two methods for @xmath43 should become competitive . note that this method for getting @xmath34 is significantly less efficient than from the @xmath103 studies @xcite . 0.1 in ._projections for direct determination of ut . _ [ cols="<,^,^,^,^",options="header " , ] the new millennium marks the spectacular success of b - factories leading to a milestone in our understanding of cp - violation ; in particular , for the first time ckm paradigm of cp violation is quantitatively confirmed . direct measurement of @xmath105 by the b - factories agrees remarkably well with the theoretical expectation from the sm to about 10% . furthermore , first relatively crude _ direct _ determination of the other two angles ( @xmath53 & @xmath43 ) also are consistent with theoretical expectations . while these findings are good news for the sm , at the same ime , they imply that most likely the effect of bsm cp - odd phase on b - physics is likely to be a small perturbation . thus discovery of new bsm - cp - odd source(s ) of cp violation in b - physics is likely to require very large , clean , data samples and extremely clean predictions from theory . for the search of such small deviations approximate null tests of the sm gain new prominence . also important for this purpose is the drive to directly determine all three angles of the ut with highest precision possible , _ i.e. _ with errors roughly around the errors allowed by theory . it should be clear that to accomplish this important goal would require a super - b factory . specifically regarding penguin - dominated hadronic fs , that have been much in the recent news , the current data does not show any convincing signal for deviation from the sm ; however , it is a very important and sensitive test for new physics and its of vital importance to reduce the experimental errors to o(5% ) ; for this purpose too a sbf may well be needed . outside of b - physics , k - unitarity triangle , neutron electric dipole moment and top quark dipole moment are also very important _ approximate null tests _ of the sm that should be pursued vigorously . i want to thank the organizers ( and , in particular , mario greco ) for their kind invitation . useful discussions with david atwood , tom browder , chun - khiang chua , tim gershon , masashi hazumi , jim smith and especially hai - yang cheng are gratefully acknowledged . this research was supported in part by doe contract no . de - fg02 - 04er41291 .	recent highlights in cp violation phenomena are reviewed . b - factory results imply that cp - violation phase in the ckm matrix is the dominant contributor to the observed cp violation in k and b - physics . deviations from the predictions of the ckm - paradigm due to beyond the standard model cp - odd phase are likely to be a small perturbation . therefore , large data sample of clean b s will be needed . precise determination of the unitarity triangle , along with time dependent cp in penguin dominated hadronic and radiative modes are discussed . _ null tests _ in b , k and top - physics and separate determination of the k - unitarity triangle are also emphasized . = 11.6pt
* optical clock operation * the strontium lattice clock was operated under conditions very similar to those of a previous campaign @xcite . a new interrogation laser system @xcite improved the estimated optimum fractional frequency instability of the sr clock , expressed as the allan deviation @xmath5 , to below @xmath6/@xmath7 with averaging time @xmath8 @xcite . for best accuracy of the lattice clock , we reduced the duty cycle of the clock interrogation and thus reduced the power dissipation in the experimental apparatus that leads to thermal gradients . these gradients limit our knowledge of the ac- and dc - stark line shift due to blackbody radiation ( bbr ) and thereby cause the largest uncertainty contribution to our lattice clock . with this reduced duty cycle , our clock still achieved an estimated instability of about @xmath9/@xmath7 during the measurement campaign . the clock uncertainty has been evaluated similarly to @xcite . the uncertainty budget is summarized in tab . [ tab : unc ] . the systematic uncertainty @xmath10 is thus an order of magnitude smaller than ptb s cs fountain clock s uncertainty of @xmath11 @xcite . further details are given in sec . [ sec : meth ] . .corrections applied to the measured sr lattice clock frequency and their uncertainties during the frequency measurement given in fractional units . for a detailed description of the individual contributions the reader is referred to @xcite and sec . [ sec : meth ] . [ cols= " < , > , < , > , < " , ] from october 6 through 15 , 2014 , the lattice clock was operated together with our primary cs clock to measure the si frequency of the sr clock transition , which we will discuss first . the frequencies of the cs clock and the sr lattice clock have been compared via femtosecond - frequency combs ( fs in fig . [ fig : network ] ) . a continuously operated , high - performance hydrogen maser ( vremya - ch vch-1003 m ) used as a flywheel is connected to both clocks , e.g. for the generation of a time - scale steered by either the cs fountain clock ( ts(cs ) , @xcite ) or the lattice clock ( ts(sr ) ) . during the measurement campaign , the lattice clock was operated over a total up - time of @xmath12 s ( shown in fig . [ fig : stab ] ( b ) ) , together with the almost continuously running fountain clock ( availability @xmath13 ) . from this simultaneous comparison of the cs and the sr clock with the maser , the lattice clock s frequency can be calculated in the si unit hertz as realized by the cs fountain clock . for white frequency noise , which is the dominating noise type in atomic frequency standards , the statistical clock uncertainty for a given averaging time @xmath8 is equal to the allan deviation @xmath14 @xcite . thus , the fountain clock s instability @xmath15/@xmath7 clearly dominates the statistical uncertainty of the measurement given by @xmath16 . however , using only the joint up - times of both clocks does not make any use of the information available from the maser : its frequency is more stable than the fountain s for periods of up to @xmath17 s and can be measured accurately even during short up - times of the lattice clock . therefore , it is reasonable to take the maser frequency value measured with the sr clock as representative for longer intervals to improve @xmath18 . in general , due to the maser instability there is a potential difference between the maser frequency averaged over time @xmath19 and an extended time @xmath20 . the lack of knowledge about this potential difference is expressed as an additional uncertainty @xmath21 , which , given the knowledge of the maser flywheel s properties , can be calculated . * data analysis * the uncertainty @xmath21 arises from the extrapolation of the mean flywheel frequency measured with our sr clock over time @xmath19 to the longer time @xmath20 . with the help of parseval s theorem , @xmath21 can be expressed by the flywheel s spectrum of frequency fluctuations @xmath22 and a weighting function @xmath23 that describes the respective measurement intervals ( see sec . [ sec : meth ] ) . the maser spectrum @xmath22 is extracted from different comparisons : the fast fluctuations and , thus , the stability at short averaging times @xmath8 are obtained from a direct comparison with the sr clock . the allan deviation of a long continuous data set is presented in fig . [ fig : stab ] ( a ) ( triangles ) . mid- and low - frequency information about the maser s frequency fluctuations are obtained from preceding maser maser comparisons ( diamonds ) and the measurements against the fountain clocks ( dots , circles ) , respectively . from these data , we derive a model for @xmath22 that includes flicker frequency , white frequency , and white phase noise contributions . a linear maser drift was omitted here , because the analysis was designed to be drift - insensitive ( sec . [ sec : meth ] ) . of the relevant oscillators . solid black line fountain clock csf2 , dashed black line sr lattice clock , data points measured maser stability ( triangles : vs. lattice clock , diamonds : maser comparisons , empty dots : against fountain clocks , filled dots : ditto without linear drift ) , solid ( dashed ) green line noise model for the maser with ( without ) linear drift removed . the inset again shows the stabilities of the maser ( green ) and fountain clock ( black ) . the red curve shows the additional uncertainty @xmath21 when the maser is used as a flywheel and data is extrapolated from an interval @xmath12 s to @xmath20 . dashed vertical lines indicate the direct and the optimum extrapolation measurement times . + * b * : frequency deviation between the nominal 100 mhz maser output and the sr lattice clock averaged over 10 s assuming the sr clock transition frequency equal to the recommendation of the cipm for the secondary representation of the second @xcite , in total 267000 s. + * c * : weighting function used to derive the calibration uncertainty of the hydrogen maser s frequency with respect to the sr lattice clock for an interval of @xmath24 s. + * d * : time error of a simulated time - scale ts(sr ) ( red solid line ) with respect to an ideal reference . it is shown starting with the first corrected interval . the shaded areas show the 1-@xmath5 time uncertainty range of ts(sr ) ( red ) and ts(cs ) ( gray ) including statistical and systematic contributions . + ] the given sr up - times @xmath19 ( see fig . [ fig : stab ] ( b ) ) and the length and distribution of the extended time @xmath20 , which we are free to choose , determine the weighting function @xmath23 ( see fig . [ fig : stab ] ( c ) for the exemplary case of @xmath25 ) and , thus , together with @xmath22 the corresponding extrapolation uncertainty @xmath21 ( inset of fig . [ fig : stab ] ( a ) ) . due to the highly reliable cs fountain it was not necessary to consider the influence of its down - times on @xmath23 . for a flywheel - assisted sr absolute frequency measurement we optimize @xmath20 such that the overall statistical measurement uncertainty @xmath26 is minimized . as shown in the inset of fig . [ fig : stab ] ( a ) , the extrapolation uncertainty reaches the statistical uncertainty of the primary clock csf2 at about @xmath24 s ; further extension would degrade the combined statistical uncertainty @xmath18 . we can therefore enlarge the data set from 267000 s to about @xmath24 s. in consequence , the fractional systematic uncertainty of the primary fountain clock of @xmath11 becomes the largest contribution to the overall uncertainty of this frequency measurement of our strontium lattice clock . the result for the sr clock transition frequency is 429228004229872.97(16 ) hz , which is in very good agreement with previous high - accuracy measurements @xcite and provides a metrologically important confirmation that is necessary to build confidence in view of a re - definition of the si - unit second . the excellent agreement between absolute frequency measurements of strontium lattice clocks in numerous institutes is shown in fig . [ fig : freq ] ( a ) . [ fig : freq ] ( b ) points out that the systematic uncertainties of present - day sr clocks worldwide range from similar to to well below the cs systematics of the absolute frequency measurement with the lowest overall uncertainty ever achieved @xcite . @xmath27 5s5p @xmath28 transition in @xmath29sr . the values have been obtained from various references @xcite . ptb 14 has been obtained in this work . the vertical line indicates the frequency recommended by the cipm in 2013 for the secondary representation of the second by sr lattice clocks @xcite , the dashed lines show the assigned uncertainty . + * b * : listing of the systematic uncertainty @xmath30 of the best sr clocks worldwide @xcite . the gray line indicates the cs systematic uncertainty of the absolute frequency measurement with the smallest overall uncertainty so far @xcite . ] * realizing an optical time - scale * the approach discussed here can be used not only to reduce the statistical uncertainty of frequency measurements , but also to establish a continuous time - scale from a frequency standard with an availability of considerably less than 100 % . this situation is not particular to optical clocks @xcite , but nowadays prevalent when time - scales are established in metrology institutes or contributing to utc @xcite . in the following , we consider the combination of the sr lattice clock and the continuously running maser flywheel to constitute a time - scale based on an optical clock . the main difference to the absolute frequency measurement is that now we are not free to chose the extended time interval @xmath20 since it is determined by the time - scale s origin and the duration for which the time - scale has to be provided . for convenience , we will discuss an optical time - scale that covers the 12 days of our flywheel - assisted absolute frequency measurement campaign . the predictability of a time - scale is quantified by its uncertainty : the inset in fig . [ fig : stab ] ( a ) shows the frequency uncertainty @xmath31 of an ensemble of masers all having the same instability like our maser whose frequencies are extrapolated from @xmath19 to a period @xmath20 ( red line ) . the curve can be related to an accumulated time - scale uncertainty via @xmath32 when realizing a time - scale by correcting such maser s frequency by interrupted measurements with our optical clock . to yield the overall optical time - scale uncertainty , this contribution has to be added in quadrature to the accumulated time uncertainty of the sr clock s scale unit , which is at least an order of magnitude smaller . we choose a numerical approach in order to demonstrate the potential performance of a real - time steered time - scale ts(sr ) during the measurement interval of interest since an experimental characterization would suffer from the instability of available reference time - scales . ts(sr ) is compared to the uncertainty of a theoretical time - scale ts(cs ) based on a cs reference with 100 % up - time ( assuming the parameters of ptb s fountain clock csf2 ) . in order to provide time whenever needed we choose a real - time implementation even though it bears tougher constraints , which can result in a degraded instability compared to a post - processing approach . in general , the steering algorithm can be optimized with respect to a stable scale unit and a good long - term performance . both design goals coincide for 100 % clock up - time but they diverge the longer the clock down - times are that must be bridged with a less stable flywheel . for ts(sr ) we focus on a low statistical uncertainty for long averaging times to reach the clock s systematics as quickly as possible . to estimate the error of ts(sr ) , we numerically generate frequency data traces on a 10 s grid for the maser flywheel and the sr clock using the time intervals of fig . [ fig : stab ] ( c ) . while the clock trace accounts for the statistical noise @xmath33 , the maser trace reflects the noise and drift characteristics of our maser model ( see sec . [ sec : meth ] ) . this data is used to simulate a time - scale ts(sr ) ( see fig . [ fig : stab ] ( d ) , starting with the first corrected period ) . at the beginning of the measurement interval ( @xmath34 corresponds to the modified julian date ( mjd ) @xmath35 ) , the time error @xmath36 between the time - scale and an ideal one is set to zero . the steering algorithm works as follows : the average maser frequency is calculated regularly using the frequency information available up to that point from the sr clock . this regular period is chosen to one hour . the average @xmath37 is used as estimate for the maser frequency @xmath38 over the full elapsed time @xmath39 . the corresponding time error is applied to the phase stepper in fig . [ fig : network ] to correct the time - scale from the maser as @xmath40 . to attain a complete picture , one also has to account for the sr clock s systematics @xmath10 . this is achieved by setting off the fractional clock frequency from its ideal value by @xmath41 , which results in a minor time error of about 15 ps over the 12 days . we observe a very stable behaviour of ts(sr ) ( fig . [ fig : stab ] ( d ) ) ; it is rather smooth and no excursions larger than 150 ps are observed . the initial transient response at day 2 is an artefact of the initialization of the real - time time - scale and could be reduced considerably in post - processing . between day 6 and 9 , when no data from the lattice clock is available , ts(sr ) starts to deviate according to the free - running maser instability , but once there is new optical data available ( @xmath42 days ) the time offset does not increase further and is even reduced due to a improved estimate of the past maser average frequency . in addition , fig . [ fig : stab ] ( d ) shows uncertainty bands of independent time - scale realizations of ts(sr ) and ts(cs ) as shaded areas including statistical and systematic contributions . in case of ts(sr ) , the statistical uncertainty connected to the interrupted optical clock operation @xmath43 is calculated by the method described above , applying a weighting function that reflects the elapsed time and up - time of the optical clock . in this case , @xmath43 is not insensitive to a linear drift of the flywheel for all times and thus an uncertainty contribution considering this fact is added in quadrature . however , in a real operation scenario the drift is usually known , e.g. from long - term comparisons against utc , and can easily be corrected . for the considered 12-days period , the uncertainty of ts(sr ) is below 200 ps at all times . while the overall uncertainty of ts(cs ) is dominated by the reference clock s systematics , that of ts(sr ) is dominated by @xmath43 . thus , with increasing optical clock availability the uncertainty of ts(sr ) would be further reduced . yet , even with the optical clock operation periods at hand , which can be considerably increased for the realization of a time - scale , the uncertainty of ts(sr ) is clearly below that of the traditional time - scale at all relevant times . moreover , the uncertainty of ts(cs ) is increasing much faster . the formalism presented here is applicable to arbitrary flywheel - assisted clock measurement scenarios . therefore , it allows assessing the scaling of the time errors for longer measurements . for example , when hypothetically providing ts(sr ) for a longer interval by repeating the measurement campaign illustrated in fig . [ fig : stab ] ( c ) in a similar way , both campaigns can be considered statistically independent and , thus , the overall time error essentially increases with the square - root of the combined campaigns total duration @xmath44 and correspondingly the statistical extrapolation uncertainty of the average frequency scales with @xmath45 . this paper demonstrates that optical frequency standards and in particular strontium lattice clocks have reached a maturity that they now can be used in combination with a high - quality commercially available flywheel oscillator to generate a time - scale that shows beyond state - of - the - art performance . this is essential for actual optical clock applications and a redefinition of the si - unit second . the measurement presented in fig . [ fig : stab ] ( b ) did not explicitly aim for maximum up - time and can therefore be regarded a lower limit of capability of our apparatus . yet , fig . [ fig : stab ] ( d ) shows that a time error of below 200 ps was achieved for a measurement interval of 12 days . continuation of the campaign would have led to a time error of about 300 ps over 30 days , the typical comparison duration to establish utc . this is a remarkably low value in view of the deviations in the ns - range between the best time - scales reported in the monthly circular - t @xcite , which , however , includes an additional time link error of about @xmath46 ns . concerning the flywheel oscillator , several choices are available besides the h - maser used here to achieve lower scale unit distortions for certain gap durations . cryogenic sapphire microwave oscillators @xcite , cryogenic optical reference resonators @xcite , and lasers stabilized by the method of spectral hole burning @xcite show excellent mid- and long - term stability and could serve this purpose . even an optical clock that is optimized for reliability instead of accuracy can be envisaged . ultimately , an ensemble of flywheels is conceivable providing lowest instability for all occurring gap durations . today s satellite - based comparison techniques can readily provide time - scale comparisons with uncertainties below 1 ns @xcite . thus , having averaged for a year or more , the well - established satellite network would allow the intercontinental comparison of flywheel - assisted optical clocks in the low @xmath47 regime . even further , esa s future space mission atomic clock ensemble in space ( aces ) will provide enhanced frequency comparison capabilities over intercontinental distances @xcite , such that flywheel - supported optical clock comparisons will no longer be limited by the satellite link between them . thus , intercontinental clock comparisons with uncertainty in the @xmath48 regime after averaging over a month is within reach . the current optical clocks can be developed towards mostly autonomous systems that will increase the average availabilities @xcite . for example , a clock availability of 70 % would immediately improve the time - scale uncertainty further to below 50 ps over 30 days assuming our system and a worst - case - scenario of a single 7 h interruption per day , which corresponds to a frequency uncertainty of @xmath49 . * sr lattice clock operation * to interrogate the 698 nm @xmath50 clock transition of @xmath29sr , an atomic beam from an oven is zeeman - slowed and loaded into a magneto - optical trap ( mot ) operated on the strong 461 nm @xmath51 transition . after less than 300 ms loading , the atoms are transferred into a second - stage mot operated on the 689 nm intercombination line , @xmath52 , that allows for laser cooling of the atoms to a few micro - kelvin . during the last cooling phase , the atoms are loaded into a nearly horizontally oriented 1d - optical lattice at the stark - shift cancellation wavelength of strontium near 813 nm @xcite . in the lattice , the atoms are spin - polarized to either the @xmath53 or @xmath54 level of the ground state . the polarization of the sample is purified by a short @xmath55 pulse on the clock transition in a homogeneous magnetic bias field and subsequent removal of remaining ground state atoms . after removing high - energy atoms from the lattice by temporarily reducing the lattice depth @xcite , the clock transition is interrogated in a bias field of about 25 @xmath56 t . the excitation probability is estimated from the atoms in ground and excited state after the interrogation . to derive an error signal free from linear zeeman shift the laser is alternatingly tuned to points of maximum slope of the @xmath57 transitions . a single sequence lasts about 1 s , or , if a dead time is introduced to reduce heat load and thermal gradients , one interrogation every 2 s is performed . different from our previous set - up @xcite , the 429 thz clock laser system interrogating the @xmath29sr @xmath58 @xmath59 clock transition was replaced such that stabilization to the cryogenic silicon cavity @xcite is no longer required @xcite . the new clock laser system uses a 48 cm long reference resonator made from ultra - low expansion glass @xcite . it enables regular coherent interrogation of 640 ms duration , which leads to a resonance linewidth of about 1.4 hz full - width at half - maximum . * sr lattice clock uncertainty contributions * _ bbr shift : _ the stark shift caused by bbr was determined experimentally @xcite , and validated theoretically @xcite as well as experimentally @xcite . recently , the dynamic correction factor @xmath60 , reflecting the difference in frequency shift between bbr and a static electric field with equal rms amplitude , was refined @xcite . in our experiment , the uncertainty due to bbr is limited by the uncertainty of the representative temperature @xmath61 . temperature differences of 1.3 k at maximum measured between the warmest ( @xmath62 ) and coldest ( @xmath63 ) point of the apparatus lead under the assumption of a uniform probability distribution of @xmath61 to the most probable temperature of @xmath64 k and an uncertainty of @xmath65 of the fractional frequency correction of @xmath66 . in addition , radiation from the hot oven reaches the atoms and we estimate a corresponding correction of @xmath67 @xcite . _ zeeman shifts : _ the linear zeeman shift , as well as the vector light shift , are removed by stabilizing the interrogation laser to the mean frequency of the @xmath68 zeeman components @xcite . moreover , this scheme provides continuous monitoring of the total magnetic field experienced by the atoms and thus enables a characterization and correction of the second - order zeeman shift . we find a shift of @xmath69 . the uncertainty is governed by the uncertainty of the magnetic field derived from the line splitting . the uncertainty of the correction coefficient and the vector light shift can be neglected . _ collisions : _ cold collisions and collisions with the background gas may cause frequency shifts . we measured the cold collision shift by variation of the atom density in the lattice and found a shift of @xmath70 . from the lattice lifetime of 7 s , background gas collisions lead to a shift of @xmath71 based on @xcite . _ line pulling : _ line pulling is induced by overlapping transitions caused by atoms populating different levels or by polarization imperfections of the excitation laser such that not only @xmath55- but also @xmath5-transitions are driven simultaneously . investigation of the preparation sequence and the laser polarization as seen by the atoms allows us to assign a negligible correction with small uncertainty to the line pulling effect of @xmath72 . _ lattice light shifts : _ by varying the lattice depth between the usual spectroscopy setting of 72 @xmath73 and 110 @xmath73 , with the recoil energy of the lattice @xmath73 , we find a stark shift cancellation frequency of 368554471(3 ) mhz for a mutually parallel lattice polarization and magnetic field orientation . this leads to a fractional uncertainty of the shift of @xmath74 . besides the determination of the stark shift cancellation frequency under the given experimental conditions of lattice polarization and orientation , possible shifts due to multi - photon and higher - order multi - pole excitations have to be accounted for @xcite . using the coefficients of @xcite for e2/m1 transitions and two - photon induced light shifts ( hyperpolarisability ) @xcite , these higher - order effects cause shifts of @xmath75 and @xmath76 , respectively . _ tunnelling : _ resonant tunnelling in the lattice causes energy bands with finite widths that can be populated non - uniformly , which in turn causes a tunnelling - induced frequency shift . bandwidths and corresponding shifts can be reduced by a tilt or acceleration of the lattice such that tunnelling is no longer resonant @xcite . as we prepare the atoms almost exclusively in the vibrational ground state in longitudinal direction and operate in a deep lattice , even a small lattice tilt of @xmath77 present in our set - up is sufficient to reduce the uncertainty due to tunnelling to @xmath78 . _ probe light shift : _ the highly advanced interrogation laser @xcite enables long interrogation times of 640 ms . therefore , the necessary intensity to drive a @xmath55-pulse and thus the light shift caused by the clock laser are very small . with the coefficient published in @xcite , we find less than @xmath79 of fractional shift , which we use as uncertainty . _ optical path length : _ changes of the optical path length between atoms and the clock laser cause a first - order doppler shift and therefore errors in the frequency measurement . although continuous drifts due to , e.g. , slow temperature changes may not be overly harmful @xcite , clock - cycle - synchronous vibrations or rf heating of the aom may . lattice clocks offer the unique opportunity to stabilize the optical path length from the interrogation laser to the atom position determined by the mirror that forms the standing wave of the 1d lattice . evaluation of the lock signal @xcite reveals an upper limit of path length - induced shifts of @xmath79 . _ servo error : _ in practice , all reference cavities used to build narrow bandwidth lasers exhibit a frequency drift due to ageing of the resonator materials and residual temperature variations . any drift of the clock laser frequency causes a servo error , since frequency corrections are applied discontinuously . we reduce this effect by continuously sweeping the laser frequency with a rate that is updated from the atomic signal . then , only non - linearities in the drift cause a servo error . as our new reference cavity shows an exceptionally linear drift , the fractional uncertainty is below @xmath48 . _ dc stark shift : _ static electric fields in the interrogation region , for instance produced by patch charges on isolating surfaces , can cause substantial frequency shifts @xcite . we have evaluated possible offset fields from the analysis of the observed stark shift induced by intentionally applied electric fields . with an upper limit for the offset field of about 16(1 ) v / m , we expect a dc stark shift of about @xmath80 , which we take as uncertainty @xcite . * theory of data extrapolation * for the analysis we use the normalized frequency fluctuations @xmath81 defined as @xmath82 with the frequency @xmath83 and the nominal frequency @xmath84 . these fractional quantities allow to simply compare fluctuations between oscillators of different average frequencies with frequency ratio @xmath85 . @xmath86 to express the uncertainty introduced by the extrapolation , we follow an approach similar to the one used to relate allan deviations and the oscillator frequency noise spectrum ( see e.g. @xcite and references therein ) . averaging the flywheel frequency @xmath81 over one set of time intervals @xmath87 with total duration @xmath88 can be written as @xmath89 with the weighting function @xmath90 : @xmath91 a similar approach is used for atomic clocks to describe the atoms response to frequency fluctuations of the interrogation oscillator @xcite . replacing the weighting function in eq . [ eq : udev ] with the combined one @xmath92 yields the difference @xmath93 of the mean flywheel frequencies averaged over the two sets of time intervals @xmath94 and @xmath95 . the uncertainty @xmath21 due to extrapolating the mean flywheel frequency from @xmath94 to @xmath95 is given by the standard deviation of this difference @xmath96 : @xmath97 parseval s theorem relates this variance to the single - sided power spectral density @xmath22 of the flywheel oscillator s frequency fluctuations as : @xmath98 with @xmath99 being the fourier transform of the weighting function . note that the intervals @xmath94 and @xmath95 can deliberately be chosen symmetric about their common center - of - gravity . in this case , a pure linear drift of @xmath81 will lead to a coincidence of @xmath100 and @xmath101 . thus , @xmath96 and @xmath21 will not be affected by the linear drift of the flywheel oscillator . * model of maser spectrum * based on the observed maser frequency fluctuations after removing a linear frequency drift of @xmath102day , the fractional instability of the maser @xmath103 was modelled by three noise contributions that have been added in quadrature : a contribution of @xmath104 from white phase noise , a white frequency noise part of @xmath105 and a flicker floor of @xmath106 @xcite . these values were converted to corresponding spectral power densities of frequency fluctuations @xcite that were used in the calculation of @xmath21 using eq . [ eq : udev2 ] . we gratefully acknowledge andreas bauch for helpful discussions . this work was supported by the centre of quantum engineering and space - time reseach ( quest ) , the german research foundation ( dfg ) through the rtg 1729 fundamentals and applications of ultra - cold matter and the crc 1128 geo - q relativistic geodesy and gravimetry with quantum sensors , and the project international time - scales with optical clocks ( itoc ) , which is part of the european metrology research programme ( emrp ) . the emrp is jointly funded by the emrp participating countries within euramet and the european union . 60ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://www.bipm.org/en/measurement-units/new-si/ [ `` , '' ] ( ) link:\doibase 10.1103/physrevlett.104.070802 [ * * , ( ) ] link:\doibase 10.1038/ncomms3109 [ * * , ( ) ] link:\doibase 10.1103/physreva.87.023806 [ * * , ( ) ] link:\doibase 10.1126/science.1240420 [ * * , ( ) ] link:\doibase 10.1038/nature12941 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.113.210802 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2015.5 [ * * , ( ) ] link:\doibase 10.1016/j.crhy.2015.03.012 [ * * , ( ) ] , link:\doibase 10.1038/nphys2834 [ * * , ( ) ] , in http://www.eftf.org/previousmeetings.php[__ ] ( , , ) pp . link:\doibase 10.1088/1367 - 2630/16/7/073023 [ * * , ( ) ] link:\doibase 10.1063/1.3682002 [ * * , ( ) ] link:\doibase 10.1088/0026 - 1394/49/3/180 [ * * , ( ) ] link:\doibase 10.1088/0026 - 1394/50/2/133 [ * * , ( ) ] link:\doibase 10.1364/ol.40.002112 [ * * , ( ) ] @noop `` , '' ( ) , link:\doibase 10.1088/0026 - 1394/47/1/008 [ * * , ( ) ] link:\doibase 10.1088/0026 - 1394/49/1/012 [ * * , ( ) ] link:\doibase 10.1088/0026 - 1394/47/1/004 [ * * , ( ) ] cipm , http://www.bipm.org/utils/en/pdf/cipm2012-en.pdf [ `` , '' ] ( ) link:\doibase 10.1103/physrevlett.98.083002 [ * * , ( ) ] link:\doibase 10.1140/epjd / e2007 - 00330 - 3 [ * * , ( ) ] link:\doibase 10.1088/0026 - 1394/45/5/008 [ * * , ( ) ] link:\doibase 10.1364/ol.34.000692 [ * * , ( ) ] \doibase doi:10.1088/0026 - 1394/48/5/022 [ * * , ( ) ] link:\doibase 10.1143/apex.5.022701 [ * * , ( ) ] link:\doibase 10.1364/oe.20.022034 [ * * , ( ) ] link:\doibase 10.7567/apex.7.012401 [ * * , ( ) ] link:\doibase 10.1364/ol.39.004072 [ * * , ( ) ] link:\doibase 10.1088/0256 - 307x/32/9/090601 [ * * ( ) , 10.1088/0256 - 307x/32/9/090601 ] @noop `` , '' ( ) @noop `` , '' ( ) link:\doibase 10.1038/ncomms7896 [ * * , ( ) ] @noop `` , '' ( ) link:\doibase 10.1088/0026 - 1394/44/1/014 [ * * , ( ) ] in http://www.eftf.org/previousmeetings.php[__ ] ( ) pp . [ `` , '' ] link:\doibase 10.1063/1.4709479 [ * * , ( ) ] link:\doibase 10.1364/ol.23.001031 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2012.217 [ * * , ( ) ] link:\doibase 10.1364/ol.39.005102 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2011.215 [ * * , ( ) ] link:\doibase 10.1088/0026 - 1394/47/5/005 [ * * , ( ) ] link:\doibase 10.1140/epjst / e2009 - 01041 - 7 [ * * , ( ) ] link:\doibase 10.1088/0957 - 0233/23/5/055201 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.173005 [ * * , ( ) ] link:\doibase 10.1109/tim.2013.2242597 [ * * , ( ) ] link:\doibase 10.1109/tim.2010.2088470 [ * * , ( ) ] link:\doibase 10.1103/physreva.87.012509 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.210801 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.180802 [ * * , ( ) ] link:\doibase 10.1103/physreva.72.033409 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.040404 [ * * , ( ) ] http://www.opticsinfobase.org/abstract.cfm?uri=ol-32-13-1812 [ * * , ( ) ] link:\doibase 10.1007/s00340 - 012 - 4952 - 6 [ * * , ( ) ] link:\doibase 10.1109/tuffc.2007.337 [ * * , ( ) ] in http://tycho.usno.navy.mil/ptti/1987/vol%2019_13.pdf[__ ] ( , , ) pp . in link:\doibase 10.1109/freq.1999.840736 [ _ _ ] ( ) pp . in http://www.vremya-ch.com/english/materials/index.html[__ ]	optical clocks are not only powerful tools for prime fundamental research , but are also deemed for the re - definition of the si base unit second as they surpass the performance of caesium atomic clocks in both accuracy and stability by more than an order of magnitude . however , an important obstacle in this transition has so far been the limited reliability of the optical clocks that made a continuous realization of a time - scale impractical . in this paper , we demonstrate how this dilemma can be resolved and that a time - scale based on an optical clock can be established that is superior to one based on even the best caesium fountain clocks . the paper also gives further proof of the international consistency of strontium lattice clocks on the @xmath0 accuracy level , which is another prerequisite for a change in the definition of the second . the international system of units ( si ) is the universal base for all measurements and thus constitutes the backbone of natural sciences and engineering . the definitions themselves and their realizations have been adapted over the years to profit from state - of - the - art research results . important modifications are at hand for the electrical units and the mass , which will be defined through agreed - upon values of fundamental constants @xcite . due to their impressive progress @xcite , optical clocks surpass caesium fountain clocks , which currently realize the si - second with lowest uncertainty , in terms of stability and accuracy of the frequency realizations of the unperturbed atomic transition . this has caused discussion about the need for a re - definition of this unit @xcite . a first step has been the acknowledgement of suitable optical transitions in neutral atoms and ions as secondary representations of the si unit `` second '' by the commit international des poids et mesures ( cipm ) @xcite . given the outstanding role of the si , it is of paramount importance that adaptations must be prepared carefully . hence , changes will be acceptable only if @xcite * an obvious candidate is identified , * the transition is smooth , and * the new approach is practical . while the first point stimulates ongoing research , the second point has already been addressed satisfactorily for strontium optical lattice clocks @xcite by measurements of the strontium clock transition frequency with today s best possible accuracy ; the results in this paper confirm the previous measurement results . the latter prerequisite , however , has not been achieved so far for the candidate systems as existing optical clocks have generally been considered less reliable and thus less suitable for the actual implementation of a practical time - scale , which continuously accumulates the atomic seconds . is used to correct the classical time - scale ts(cs ) generated from the maser utilizing a phase stepper ( @xmath1 ) . an equivalent scheme is applicable when referencing the time - scale ts(sr ) to an optical frequency standard . for that purpose , the clock laser light is down - converted to the microwave regime using a femtosecond frequency comb ( fs ) before comparing against the flywheel . moreover , both maser offsets can be analysed to yield the sr clock frequency in si units . ] time - scales provide us with coordinates for the position of events in time , much like a coordinate system does for the positioning in space . in particular , it needs to be realized without interruption to provide a continuous coordinate or to measure time intervals and to synchronize distant events . today , the universal coordinated time - scale utc is post - processed from a weighted monthly average of the time kept by some 400 microwave atomic clocks around the world that are intercompared via satellite links . to obtain time whenever needed , the time labs generate local time - scales in real - time typically by steering the output frequency of a continuously running flywheel oscillator ( fig . [ fig : network ] ) , e.g. a hydrogen maser , on a monthly period to stay synchronized with the utc clock ensemble @xcite . some time labs synchronize their flywheel oscillators on an even shorter period to match the frequency of a local primary cs clock @xcite . the origin of the time error of such real - time time - scales with respect to an ideal time - scale is twofold : first , the accuracy of the local scale unit is limited by imperfections of the atomic reference(s ) . even for the best caesium fountain clocks this error can integrate to about 1 ns after one month . second , missing information from utc or down - times of the local atomic reference clock introduce a time error depending on the instability of the flywheel oscillator . due to the averaging , utc is more stable than most local time - scales . thus , in practice the quality of a local time - scale is typically assessed by comparing to utc , which , however , is not an ideal reference . due to its construction , a deviation between the rate of utc and the second of the local fountains can be as large as 0.1 ns / day , which accumulates to a time error of about 3 ns at the end of the month . optical clocks are expected to be capable of reducing the time error of a likewise realized local time - scale by more than one order of magnitude ( time @xmath2 ps after a month ) provided the time error arising from their down - times is limited to an acceptable level and thus significantly improve the time - scale s predictability . together with improved time link technologies @xcite , a network of optical atomic clocks will allow the generation of a much more stable utc and thus labelling events in time all over the world more precisely . this would be beneficial for global navigation systems , astro- and fundamental physics . in this article , we demonstrate how ptb s strontium lattice clock is now able to maintain a local time - scale with a time error of less than 200 ps compared to an ideal reference over about 12 days , or @xmath3 in fractional units . over this period , the optical clock is allowed to operate with long interruptions ( overall clock availability : @xmath4 ) , and down - times are bridged by a hydrogen maser flywheel . the distortion of the scale unit due to the use of the flywheel while the clock is offline causes the dominant contribution to the achieved residual time error . yet , this shows for the first time that already today optical frequency standards with limited availability can actually serve as atomic references to support a local time - scale over extended periods , yielding a long - term performance better than of the ones referenced to the best current microwave clocks even if they are operated free of interruptions . further improvements of the time - scale based on an optical clock are at hand by ever higher availability of the optical clock . moreover , the maser - flywheel - assisted sr data together with data from one of our cs fountains acquired in parallel enabled us to measure the sr clock transition frequency with minimum uncertainty . so far this could only be achieved by operating the clocks for long times and/or by averaging over several of the best fountain clocks @xcite .
the thesis of j .- p . serre , at the beginning of the fifties , introduced the idea of using the cohomology of loop spaces in topology . topologists have quickly studied the cohomology of iterated loop spaces @xmath4 which are spaces of maps @xmath5 on spheres @xmath6 . the computation of the cohomology of @xmath4 would make possible the determination of homotopy classes of these maps @xmath5 but , for the moment , this computation is well understood only when @xmath7 is the @xmath0-fold suspension of a space @xmath8 . besides , we have a comprehensive knowledge of the homotopy classes of finite cell complexes in a few particular cases only ( which do not include spheres ) for which the homotopy vanishes in dimension @xmath9 . our works @xcite give an algebraic device to model iterated loop spaces @xmath4 on all spaces @xmath7 and for every value of @xmath0 . the goal of this paper is to explain how to make explicit a structured complex computing the cohomology of a profinite completion of the iterated loop spaces @xmath10 from this model , by putting together results of @xcite and of another article @xcite . the obtained complex is defined purely algebraically , in terms of the internal structure of certain operads , and inherits nice operadic structures from which we hope to get information of new nature on the topology of @xmath10 . the @xmath2-operads give the core structures of our construction . for our purpose , we consider @xmath2-operads in the category of differential graded modules ( dg - modules for short ) and we use the noun @xmath2-algebra to refer to any category of algebras associated to an @xmath2-operad in that ambient category . in that context , the @xmath2-operads refer to the homotopy type of operads which are weakly equivalent to the chain operad of boardman - vogt little @xmath0-cubes . most models of @xmath2-operads come in nested sequences @xmath11 such that @xmath12 is an @xmath13 operad , an operad weakly equivalent to the operad @xmath14 of associative and commutative algebras . the @xmath15-operads , which form the initial terms of nested sequences of the form ( * ) , are weakly equivalent to the operad @xmath16 of associative non - commutative algebras . intuitively , an @xmath15-algebra structure is interpreted as a structure which is homotopy associative in the strong sense but non - commutative , while an @xmath17-algebra is fully homotopy ( associative and ) commutative . the @xmath2-algebra structures , which are intermediate between @xmath15 and @xmath17 , are interpreted as structures which are homotopy associative in the strong sense and homotopy commutative up to some degree . a commutative algebra naturally inherits an @xmath2-algebra structure , for all @xmath18 . throughout this paper , a space means a simplicial set and we associate to a simplicial set @xmath7 the standard normalized complex @xmath19 with coefficients in the ground ring @xmath20 of our base category of dg - modules . as we deal with pointed spaces , we use the reduced complex @xmath21 in our constructions rather than the whole complex @xmath19 . to handle structures without unit ( like these reduced complexes @xmath22 ) , we also remove @xmath23-ary operations from our operads . the cochain algebras @xmath22 inherit an @xmath17-algebra structure by @xcite , for which we have an explicit construction ( see @xcite ) giving an action of certain @xmath17-operads equipped with a filtration of the form ( * ) on @xmath22 . in that context , we immediately see that the @xmath17-algebra @xmath22 inherits an @xmath2-algebra structure for all @xmath0 . now a natural homology theory @xmath24 is associated to any category of algebras over an @xmath2-operad . for a cochain algebra @xmath25 , the theorems of @xcite , put together , imply that the homology @xmath26 determines the cohomology of a profinite completion of the @xmath0-fold loop space @xmath4 . but we observe in @xcite that @xmath2-operads are in a sense koszul and we prove , using a good @xmath2-operad @xmath27 , that a cofibrant model of @xmath2-operad is yielded by an operadic cobar construction @xmath28 , where @xmath29 is an operadic desuspension of the dual cooperad of @xmath27 in dg - modules . this result implies that the homology @xmath24 of any @xmath27-algebra can be computed by a nice structured complex of the form @xmath30 , where @xmath31 is the connected @xmath32-cofree coalgebra on @xmath33 and @xmath34 is a twisting coderivation determined by the @xmath2-algebra structure of @xmath33 ( this construction is reviewed in [ koszuldualityreview ] ) . the goal of this paper is to determine the differential of this complex @xmath35 for the cochain algebra of a sphere @xmath36 , where @xmath37 . in @xcite , we observe that the action of @xmath28 on @xmath38 , by way of the morphism @xmath39 , is determined by a composite of : 1 . [ corestructures : embedding ] the image under the cobar construction @xmath40 of a morphism @xmath41 dual in dg - modules to the embedding @xmath42 ; 2 . [ corestructures : twist ] the operadic suspension of an augmentation morphism @xmath43 which , by operadic bar duality , is equivalent to a morphism @xmath44 on the suspension of the operad @xmath45 associated to the standard category of @xmath3-algebras , the classical version of homotopy lie algebra structure . the existence of a morphism @xmath44 reflects the existence of a restriction functor from @xmath46-algebras to @xmath3-algebras and has been used in the literature in the case @xmath47 , in conjunction with the deligne conjecture , to determine the homotopy type of certain deformation complexes ( see for instance @xcite ) . in @xcite , we prove that these morphisms @xmath48 are unique up to a right homotopy in the category of operads , and hence , are characteristic of the structure of an @xmath46-operad . we give a conceptual definition of the complex @xmath49 in terms of the embedding @xmath42 , of the morphism @xmath44 and of the composition structure of the operad @xmath27 . we will obtain a result of a particularly simple form also because the reduced normalized complex @xmath38 is a free @xmath20-module of rank one . we give all details in [ spherekoszulduality ] . for this introduction , we are simply going to state the final outcome of our construction : the application of our result to iterated loop space cohomology as it arises from @xcite . we dualize the construction to ease the formulation of our result . we form , for any @xmath50 , the free complete @xmath27-algebra @xmath51 on one generator @xmath52 of degree @xmath53 . the elements of this algebra @xmath54 are just formal power series @xmath55 where @xmath56 is an @xmath57-ary operation of the operad @xmath58 . our result makes appear a composition product @xmath59 , borrowed to @xcite , defined termwise by sums of operadic substitution operations @xmath60 running over all entries of @xmath61 . we observe that the existence of a morphism @xmath62 is formally equivalent to the existence of an element @xmath63 , of degree @xmath64 , satisfying the relation @xmath65 , where @xmath66 refers to the canonical differential of @xmath54 . we embed this element @xmath67 into @xmath51 , for any @xmath50 . we have then : we consider the twisted complex @xmath68 defined by adding the map @xmath69 to the natural differential of the free complete @xmath27-algebra @xmath51 . when the ground ring is @xmath70 , we have an identity @xmath71 between the continuous cohomology of the profinite completion of @xmath10 and the homology of the dual complex of @xmath68 in @xmath70-modules ( we take the continuous dual with respect to the topology of power series ) . the profinite completion of the theorem is defined by the limit of a diagram @xmath72 where @xmath73 ranges over homotopy finite approximations of @xmath74 ( see @xcite ) . this formulation is slightly abusive because the profinite completion of @xmath10 would be the limit of a diagram of homotopy finite approximations of the space @xmath10 itself . nevertheless results of @xcite imply that these limits coincide . the theorem has an @xmath75-version where the profinite completion of @xmath10 is replaced by a @xmath76-profinite approximation , the limit of a diagram @xmath72 where @xmath73 now ranges over homotopy @xmath76-finite approximations of @xmath74 ( see again @xcite ) . the topological results of @xcite are stated in the @xmath75-setting , but the arguments work same over @xmath70 . the above theorem is not intended for computational purposes and we should not undertake any computation with it the iterated bar complexes of @xcite , which are equipped with nice filtrations , seem better suited for such an undertaking . the actual goal of this article , with this theorem , is to make explicit the connection between the cohomology of iterated loop spaces of spheres and the homotopical structures of @xmath2-operads . in subsequent work , we plan to apply the theorem to gain qualitative information on the structure of iterated loop space cohomology by using connections , explained in @xcite , between @xmath2-operads and quantum field theory . in [ koszuldualityreview ] , we briefly review the definition of the homology @xmath77 associated to an operad @xmath78 , the applications of the bar duality of operads to the definition of complexes @xmath79 computing @xmath77 , and we explain how to apply this construction to @xmath2-operads with the result of @xcite . in [ freecompletealgebras ] , we explain the definition of the free complete algebras of the form @xmath51 and their duality connection with cofree coalgebras @xmath80 on a dg - module @xmath81 of rank one . in [ spherekoszulduality ] , we study the applications of operadic bar duality to the cochain algebra of spheres @xmath38 and we determine the form of the complex @xmath49 from which we obtain the outcome of our theorem . in [ koszuldualityreview ] and [ spherekoszulduality ] , we deal with a particular @xmath2-operad , already used in the proof of @xcite , which arises from a certain filtration of an @xmath17-operad introduced by barratt - eccles in @xcite . in [ homotopyinvariance ] , we explain that the homotopy type of the complex @xmath82 which occurs in our main theorem does not depend on the choice of the @xmath2-operad @xmath27 and of the element @xmath83 associated to a morphism @xmath84 . to begin with , we review the background of our constructions . we are only going to give a brief survey of this background since our main purpose is to explain our conventions . we refer to @xcite for a more detailed account with the same overall conventions as in the present article . to begin with , we explicitly specify which base category of dg - modules is considered throughout the article . the ground ring , fixed once and for all , is denoted by @xmath20 . for us a dg - module refers to a @xmath70-graded @xmath20-module @xmath81 equipped with an internal differential @xmath85 that decreases the degree by one , and we take this category of dg - modules as base category @xmath86 . the category of dg - modules is equipped with its standard tensor product @xmath87 together with the symmetry isomorphism @xmath88 involving a sign . this sign is determined by the standard sign convention of differential graded algebra , which we follow . the notation @xmath89 is used to represent any sign arising from an application of this convention . the ground ring @xmath20 , viewed as a dg - module of rank @xmath90 concentrated in degree @xmath23 , defines the unit object for the tensor product of dg - modules . the morphism sets of any category @xmath91 are denoted by @xmath92 . the morphisms of dg - modules are the degree and differential preserving morphisms of @xmath20-modules @xmath93 . the category of dg - modules comes also equipped with internal hom - objects @xmath94 characterized by the adjunction relation @xmath95 with respect to the tensor product @xmath87 . recall briefly that a homogeneous element of @xmath94 is just a morphism of @xmath20-modules @xmath93 raising degrees by @xmath96 . the differential of a homomorphism @xmath97 is defined by the ( graded ) commutator @xmath98 , where we consider the internal differentials of @xmath81 and @xmath99 . the elements of the dg - hom @xmath94 are called homomorphisms to be distinguished from the actual morphisms of dg - modules , the elements of the morphism set @xmath100 . the dual of a dg - module @xmath81 is the dg - module such that @xmath101 , where we again view the ground ring @xmath20 as a dg - module of rank @xmath90 concentrated in degree @xmath23 . we always consider operads ( and cooperads ) in the category of dg - modules . we adopt the notation @xmath102 for the whole category of operads in dg - modules . we use the notation @xmath103 for the initial object of the category of operads which reduces to the ground ring @xmath104 in arity @xmath90 and is trivial otherwise . in general , we tacitely assume that an operad @xmath78 satisfies @xmath105 ( we say that @xmath78 is non - unitary ) . in some cases , we may assume that an operad satisfies @xmath106 too . we say in this situation that @xmath78 is a connected operad . cooperads @xmath107 are always assumed to satisfy the connectedness assumptions @xmath108 and @xmath109 in order to avoid difficulties with infinite sums in coproducts . we use the notation @xmath110 ( respectively @xmath111 ) for the category of non - unitary ( respectively connected ) operads in dg - modules . we call @xmath112-object the structure , underlying an operad , formed by a collection @xmath113 where @xmath114 is a dg - modules equipped with an action of the symmetric group on @xmath57 letters @xmath115 . a connected operad inherits a canonical augmentation @xmath116 which is the identity in arity @xmath117 and is trivial otherwise . we adopt the notation @xmath118 for the augmentation ideal of any operad @xmath78 equipped with an augmentation @xmath116 . we have in the connected case @xmath119 and @xmath120 for all @xmath121 . any connected cooperad @xmath107 has a coaugmentation @xmath122 and a coaugmentation coideal @xmath123 defined like the augmentation and the augmentation ideal of a connected operad . the suspension of a dg - module @xmath81 is the dg - module @xmath124 defined by the tensor product @xmath125\otimes c$ ] where , for all @xmath126 , we adopt the notation @xmath127 $ ] for the free graded @xmath20-module of rank @xmath90 concentrated in degree @xmath128 . the operadic suspension of an operad @xmath78 , is an operad @xmath129 characterized by the commutation relation @xmath130 at the level of free algebras . basically , the operad @xmath129 is defined arity - wise by the tensor products @xmath131\otimes\operatorname{\mathsf{p}}(n)^{\pm}$ ] where the notation @xmath89 refers to a twist of the natural @xmath132-action on @xmath133 by the signature of permutations . we have an operadic suspension operation defined in the same way on cooperads . note that we can apply the suspension of dg - modules arity - wise to any operad @xmath78 ( or cooperad ) to produce a @xmath112-object @xmath134 such that @xmath135\otimes\operatorname{\mathsf{p}}(r)$ ] , but this suspended @xmath112-object does not inherit an operad structure . the category of algebras in dg - modules associated to an operad @xmath78 is denoted by @xmath136 . the free @xmath78-algebra associated to a dg - module @xmath81 is denoted by @xmath137 . recall simply that this free @xmath78-algebra is defined by the dg - module of generalized symmetric tensors @xmath138 , where the @xmath115-quotient identifies tensor permutations with the action of permutations on @xmath139 . note that we can remove the @xmath23 term of this expansion since we assume @xmath105 . we use the notation @xmath140 where @xmath141 , @xmath142 , to represent the element of @xmath137 defined by the tensor @xmath143 . we may also use the graphical representation @xmath144\ar@{.}[r ] & \cdots\ar@{.}[r]\ar@{}[d]\|{\displaystyle{\cdots } } & x_r\ar[dl ] \\ \ar@{.}[r ] & * + < 8pt>[f]{p}\ar[d]\ar@{.}[r ] & \\ \ar@{.}[r ] & 0\ar@{.}[r ] & } } \ ] ] for these elements @xmath145 . in the case of a cooperad @xmath107 , we use the notation @xmath80 for the connected cofree @xmath107-coalgebra such that @xmath146 . in this definition , we take the same generalized symmetric tensor expansion as a free algebra over an operad for the underlying functor of a cofree coalgebra over a cooperad and therefore , we can safely adopt the functional and graphical representation of elements of free algebras of operads to denote the elements of @xmath80 . note that @xmath80 does not agree with the standard cofree connected coalgebras of the literature in usual cases ( like cocommutative coalgebras ) precisely because we take coinvariants instead of invariants , but we generally apply the definition of @xmath80 to cooperads @xmath107 equipped with a free @xmath112-module structure so that the result is the same if we replace coinvariants by invariants in the definition of @xmath80 . the coaugmentation morphism of a cooperad yields an embedding @xmath147 which identifies the dg - module @xmath81 with a summand of @xmath80 . similarly , if @xmath78 is an augmented operad , then we have a split embedding @xmath148 , yielded by the operad unit of @xmath78 . the morphism @xmath149 induced by the augmentation of @xmath78 gives a canonical retraction of @xmath148 so that @xmath81 is canonically identified with a summand of @xmath137 . the category of dg - modules @xmath86 is equipped with its standard model structure in which the weak - equivalences are the morphisms which induce an isomorphism in homology , the fibrations are the degreewise surjections ( see @xcite ) . the category of non - unitary operads @xmath110 inherits a full model structure from dg - modules so that a morphism @xmath150 is a weak - equivalence ( respectively , a fibration ) if @xmath151 forms a weak - equivalence ( respectively , fibration ) of dg - modules @xmath152 in each arity @xmath153 . the cofibrations are characterized by the right - lifting - property with respect to acyclic fibrations . the category of @xmath112-objects in dg - modules also inherits a full model structure with the same definition for the class of weak - equivalences ( respectively , fibrations ) . any operad cofibration with a cofibrant operad as domain defines a cofibration of @xmath112-objects , but the converse implication does not hold . therefore we say that an operad @xmath78 is @xmath112-cofibrant when its unit morphism @xmath154 defines a cofibration in the category of @xmath112-objects . the category of algebras over a @xmath112-cofibrant operad @xmath78 inherits a semi - model structure in the sense that the axioms of model categories are satisfied for @xmath78-algebras only when we restrict the applications of the right lifting property of ( acyclic ) cofibations and of the factorization axiom to morphisms with a cofibrant domain . this is enough for the usual constructions of homotopical algebra . naturally the weak - equivalences ( respectively , fibrations ) of @xmath78-algebras are the morphisms of @xmath78-algebras @xmath155 which define a weak - equivalence ( respectively , a fibration ) in the category of dg - modules . in certain constructions , a homomorphism @xmath156 of degree @xmath157 is added to the internal differential of a dg - module @xmath81 to produce a new dg - module , with the same underlying graded @xmath20-module as @xmath81 , but the map @xmath158 as differential . since we have @xmath159 , the relation of differential @xmath160 amounts to the equation @xmath161 in @xmath162 . in this situation , we say that @xmath34 is a twisting homomorphism and we just use the pair @xmath163 to refer the new dg - module defined by the addition of @xmath34 to the internal differential of @xmath81 . when we apply this construction to an algebra @xmath33 over an operad @xmath78 , we assume that the twisting homomorphism satisfies the derivation relation @xmath164 to ensure that the twisted dg - module @xmath165 inherits a @xmath78-algebra structure from @xmath33 . we then say that @xmath166 is a twisting derivation . we call quasi - free @xmath78-algebra the twisted @xmath78-algebras @xmath165 associated to a free @xmath78-algebra @xmath167 . we have a one - one correspondence between derivations on free @xmath78-algebras @xmath168 and homomorphisms @xmath169 . we use the notation @xmath170 for the derivation associated to @xmath171 . in one direction , we have @xmath172 . in the other direction , the derivation @xmath173 is determined from @xmath169 by the derivation formula @xmath174 , for all @xmath175 . for a free @xmath78-algebra @xmath137 , a derivation @xmath173 satisfies the equation of twisting homomorphisms if and only if the associated homomorphism @xmath169 satisfies the equation @xmath176 in @xmath177 . we also apply the definition of twisted dg - modules to coalgebras over cooperads . we have a notion of quasi - cofree @xmath107-coalgebra , dual to the notion of quasi - free algebra over an operad , consisting of a twisted dg - module @xmath178 such that @xmath179 is a cofree @xmath107-coalgebra and @xmath180 is a @xmath107-coalgebra coderivation uniquely determined by a homomorphism @xmath181 such that @xmath182 in @xmath183 . in the context of coalgebras , we determine the twisting coderivation @xmath184 associated to a homomorphism @xmath185 by a graphical expression of the form : @xmath186\ar@{.}[r ] & \cdots\ar@{.}[r]\ar@{}[d]\|{\displaystyle{\cdots } } & x_r\ar[dl ] \\ \ar@{.}[r ] & * + < 8pt>[f]{c}\ar[d]\ar@{.}[r ] & \\ \ar@{.}[r ] & 0\ar@{.}[r ] & } } \right\ } = \sum_{i } \pm\left\{\vcenter{\xymatrix@h=6pt@w=3pt@m=2pt@!r=1pt@!c=1pt { x_1\ar[drr]\ar@{.}[r ] & \cdots\ar@{.}[r ] & \alpha(x_i)\ar[d]\ar@{.}[r ] & \cdots\ar@{.}[r ] & x_r\ar[dll ] \\ \ar@{.}[rr ] & & * + < 8pt>[f]{c}\ar@{.}[rr]\ar[d ] & & \\ \ar@{.}[rr ] & & 0\ar@{.}[rr ] & & } } \right\ } \\ + \sum_{\substack{\tau\in\theta_2(r)\\ \rho_{\tau}(c ) } } \pm\left\{\vcenter{\xymatrix@h=6pt@w=4pt@m=2pt@r=8pt@c=4pt { & & \save [ ] .[rrd]!c * + < 4pt>[f-,]\frm{}+<6pt>\frm{\{}+<0pt>\frm{\ } } \restore\ar@{}[]!l-<8pt,0pt>;[d]!l-<8pt,0pt>_{\displaystyle\alpha } x_\ar@{.}[r]\ar[dr ] & \cdots\ar@{.}[r ] & x_\ar[dl ] & & \\ x_\ar@{.}[r]\ar@/_6pt/[drrr ] & \cdots & & + < 6pt>[f]{c_}\ar@{.}[r]\ar@{.}[l]\ar[d ] & & \cdots\ar@{.}[r ] & x_\ar@/^6pt/[dlll ] \\ \ar@{.}[rrr ] & & & * + < 6pt>[f]{c_}\ar[d]\ar@{.}[rrr ] & & & \\ \ar@{.}[rrr ] & & & 0\ar@{.}[rrr ] & & & } } \right\}.\end{gathered}\ ] ] the notation @xmath187 refers to a cooperadic coproduct along a tree @xmath188 of an element @xmath189 and the expressions @xmath190 inside the sum refer to the factors of this cooperadic coproduct ( see ( ? ? ? * proposition 4.1.3 ) for details ) . the notation @xmath191 refers to the category of trees with two vertices . the second sum ranges over all trees @xmath192 and , for each @xmath192 , over all terms of the expansion of @xmath187 . in the first sum , we just consider the restriction of @xmath185 to the summand @xmath193 and we sum up over all inputs of @xmath189 . the purpose of this section is to review applications of operadic bar duality to the definition of homology theories for algebras over an operad . the first reference addressing applications of homotopical algebra to algebras over operads in the setting of unbounded dg - modules over a ring is the article @xcite . the definition of a homology theory with trivial coefficients , in terms of quillen s homotopical algebra , was given earlier in @xcite in the context of @xmath194-graded dg - modules over a field of characteristic @xmath23 . the application of operadic bar duality to homotopical algebra of algebras over operads is explained in that reference in the same characteristic @xmath23 setting . briefly , the authors of @xcite prove that the existence of a cofibrant model given by an operadic cobar construction @xmath195 for an operad @xmath78 gives rise to an explicit cofibrant replacement functor on the category of @xmath78-algebras which can be used for homology computations . in @xcite we explain how to extend the applications of operadic bar duality to the context of unbounded dg - modules over a ring . more specifically , we prove that the functorial cofibrant replacement of algebras over an operad , defined @xcite , works in that framework provided that we restrict ourself to @xmath112-cofibrant operads and to algebras which are cofibrant as dg - modules . we review the definition of these cofibrant replacement functors very briefly before explaining the applications to the homology of algebras over operads in our setting . we explain at each step how the constructions apply to @xmath2-operads . in this section ( and in the next one ) , we assume that @xmath27 is the particular @xmath2-operad , arising from a filtration of the barratt - eccles operad @xcite , used in the proof of the koszul duality result of @xcite . this operad has , among others , the advantage of being free of finite rank over @xmath20 in each arity . thus we can dualize the dg - modules @xmath196 without special care . besides , the operad @xmath27 is connected so that the collection of dual dg - modules @xmath197 inherits a well - defined cooperad structure . to begin with , we say a few words about the operadic cobar construction @xmath195 . the operadic cobar construction is a functor which associates a connected operad @xmath195 to any cooperad @xmath107 . we do not really need the explicit definition of @xmath195 . just recall that @xmath195 is a cofibrant as an operad whenever the cooperad @xmath107 is @xmath112-cofibrant and that any morphism @xmath198 towards an operad @xmath78 is fully determined by a homogeneous morphism @xmath199 , of degree @xmath157 , vanishing on @xmath200 , and satisfying the equation : @xmath201 & \cdots & i_r\ar[dl ] \\ & * + < 6pt>[f]{c}\ar[d ] & \\ & 0 & } } \right\ } = \sum_{\substack{\tau\in\theta_2(i)\\ \rho_{\tau}(\gamma ) } } \pm\lambda_\left\{\vcenter{\xymatrix@h=6pt@w=4pt@m=2pt@r=8pt@c=4pt { & i_\ar[dr ] & \cdots & i_\ar[dl ] & \\ i_\ar[drr ] & \cdots & * + < 6pt>[f]{\theta(c_)}\ar[d ] & \cdots & i_\ar[dll ] \\ & & * + < 6pt>[f]{\theta(c_)}\ar[d ] & & \\ & & 0 & & } } \right\},\ ] ] for any @xmath189 , where @xmath202 refers to the composition operation of the operad @xmath78 and we adopt conventions similar to [ background : twistedobjects ] to represent the expansion of the coproduct of a cooperad element @xmath189 . we refer to @xcite for details on this matter . we adopt the notation @xmath203 for the morphism associated to @xmath199 and we refer to @xmath204 as the twisting cochain associated to @xmath205 . for our @xmath2-operad , we have : [ koszuldualityreview : koszuldualitystatement ] the operad @xmath27 has a cofibrant model of the form @xmath206 , where @xmath29 is the @xmath0-fold operadic desuspension of the dual cooperad of @xmath27 in dg - modules . the proof of this assertion in @xcite is divided in several steps . the first step involves the definition of a morphism @xmath207 towards the operad of commutative algebras @xmath14 . the weak - equivalence @xmath208 is defined in a second step by a lifting process from the morphism @xmath209 . suppose now we have a @xmath112-cofibrant connected operad @xmath78 together with a cofibrant model of the form @xmath210 , where @xmath107 is a @xmath112-cofibrant operad . let @xmath211 be the augmentation of this cofibrant model . to any @xmath78-algebra @xmath33 , we associate a ( connected ) quasi - cofree @xmath107-coalgebra @xmath212 and a quasi - free @xmath78-algebra of the form @xmath213 . the homomorphism @xmath185 which determines the twisting coderivation of @xmath214 maps an element @xmath215 to the evaluation of the operation @xmath216 on @xmath217 . the homomorphism @xmath171 which determines the twisting derivation of @xmath218 is given by the composite @xmath219 where @xmath220 is yielded by the universal @xmath107-coalgebra structure of the quasi - cofree @xmath107-coalgebra @xmath214 . then : [ koszuldualityreview : cofibrantreplacements ] we have a weak - equivalence @xmath221 induced on generators @xmath222 by the augmentation of the cofree @xmath107-coalgebra @xmath223 and @xmath218 defines a cofibrant replacement of @xmath33 in @xmath78-algebras provided that @xmath33 is cofibrant as a dg - module . we refer to @xcite for the version of this statement in the context of @xmath194-graded dg - modules over a field characteristic @xmath23 and to @xcite for the proof in our setting . in @xcite , we check that the cooperad @xmath29 associated to our @xmath2-operad @xmath27 is @xmath112-cofibrant ( like the operad @xmath27 itself ) . hence the result of proposition [ koszuldualityreview : cofibrantreplacements ] holds for algebras over our @xmath2-operad . recall that the action of an operad @xmath78 on an algebra @xmath33 is determined by an operad morphism @xmath224 , where @xmath225 , the endomorphism operad of @xmath33 , is the operad such that @xmath226 . the evaluation of an operation @xmath141 on a tensor @xmath217 is defined by the evaluation of the map @xmath227 on @xmath228 . observe that the homomorphism @xmath229 in the definition of @xmath214 is , by adjunction , equivalent to the homomorphism @xmath230 such that @xmath231 is the composite @xmath232 giving the action of @xmath195 on @xmath33 through @xmath205 . hence the twisting derivation of @xmath214 is fully determined by the restriction of the @xmath78-algebra structure of @xmath33 to @xmath195 , and not by the @xmath78-algebra structure itself . in [ spherekoszulduality ] , we apply this observation to the operad @xmath27 because , in the case @xmath36 , we have a good characterization of the morphism @xmath233 . let @xmath78 be an augmented operad . the indecomposable quotient of @xmath78-algebra @xmath33 is defined as the quotient dg - module of @xmath33 under the image of operations @xmath234 , where @xmath76 ranges over the augmentation ideal of @xmath78 . thus : @xmath235 for a @xmath112-cofibrant augmented operad @xmath78 , we have a homology theory @xmath236 defined by the formula @xmath237 for any @xmath78-algebra @xmath33 , where @xmath218 is a cofibrant replacement of @xmath33 . in the case of a quasi - free @xmath78-algebra , we have an identity @xmath238 . thus , when we apply the definition of the homology @xmath77 to the functorial cofibrant replacement @xmath239 , we obtain the well - known assertion : [ koszuldualityreview : koszulhomology ] provided that the @xmath78-algebra @xmath33 is cofibrant as a dg - module , we have the relation @xmath240 . in the special case @xmath241 , this relation gives the identity : @xmath242 before applying this construction to the cochain algebra of a sphere , we go back to the general study of coalgebras @xmath80 . throughout the paper , we adopt the notation @xmath127 $ ] for a dg - module of rank one concentrated in ( lower ) degree @xmath128 . the reduced normalized cochain complex of a sphere @xmath243 is an instance of dg - module of this form . since upper gradings are changed to opposite lower gradings , we obtain precisely @xmath244 $ ] . the purpose of this section is to study coderivations of cofree coalgebras @xmath80 over a dg - module @xmath245 $ ] with the aim to determine , in the next section , the complex @xmath246 associated to @xmath36 . to make our description more conceptual , the idea , explained in the introduction , is to define the complex in a dual setting of generalized power series algebras rather than in a coalgebra setting . therefore we study the duality process first . from now on , we assume , if not explicitly recalled , that @xmath52 is a variable of degree @xmath53 . in our presentation , we consider the free @xmath78-algebra @xmath247 on a variable @xmath52 , which can be defined as the free @xmath78-algebra @xmath248 on the free dg - module of rank one spanned by @xmath52 . by definition , we have an identity @xmath249 . to define the free complete @xmath78-algebra @xmath250 , we just replace the sum in the expansion of the free @xmath78-algebra by a product ( formed within the category of dg - modules ) : @xmath251 . the free complete @xmath78-algebra is equipped with the topology defined by the nested sequence of dg - submodules @xmath252 . by proposition 1.2.18 ) , the dual in dg - modules of a connected cooperad @xmath107 always forms a connected operad . now , if @xmath107 is equipped with a free @xmath112-structure , then the dual of the connected cofree coalgebra @xmath80 on @xmath245 $ ] can be identified with a complete free algebra @xmath250 over the operad @xmath253 , where @xmath52 is a variable of degree @xmath53 , because we have in this case @xmath254^{\otimes r})_{\sigma_r})^{\vee } \simeq(\operatorname{\mathsf{d}}(r)^{\vee}\otimes\operatorname{\bbbk}[-m]^{\otimes r})^{\sigma_r } \simeq(\operatorname{\mathsf{d}}(r)^{\vee}\otimes\operatorname{\bbbk}[-m]^{\otimes r})_{\sigma_r}$ ] and the duality transforms the sum into a product . in the sequel , we are rather interested in a converse duality operation , from free complete algebras over operads to coalgebras over cooperads . the dual of a connected operad @xmath255 does not inherit a cooperad structure in general , because we need a map @xmath256 to separate the factors of the coproduct of @xmath255 , but this condition is fulfilled when each dg - module @xmath139 is free locally of finite rank over @xmath20 and @xmath257 does form a cooperad in that situation . in the sequel , we apply the duality of dg - modules to the free complete @xmath78-algebra @xmath250 . in this case , the continuous dual with respect to the topology of @xmath250 is more natural to consider and the notation @xmath258 is used to refer to it . thus , we have by definition @xmath259 . besides , we will always assume that @xmath78 is equipped with a free @xmath112-structure when we apply the duality to @xmath250 . in that setting , we obtain : [ freecompletealgebras : dualityisomorphism ] assume again that @xmath52 is a homogeneous variable of degree @xmath53 . if the operad @xmath78 is equipped with a free @xmath112-structure and is locally a free module of finite rank over @xmath20 , then we have an isomorphism @xmath260)$ ] , where @xmath257 is the dual cooperad of @xmath78 . let @xmath261 denote the continuous homomorphism associated to an element @xmath262)$ ] , @xmath189 , where we use the notation @xmath263 for the canonical generator of the @xmath20-module @xmath264 $ ] . this homomorphism can be defined termwise by the relation @xmath265 , for @xmath266 , @xmath267 , and the duality formula @xmath268 for @xmath141 . the relation @xmath260)$ ] is just converse to the relation @xmath269)^{\vee}\simeq\widehat{\operatorname{\mathsf{p}}}(x)$ ] of [ freecompletealgebras : duality ] for @xmath257 . by suspension , we deduce from these duality relations an isomorphism between the suspended dg - module @xmath270 and the dg - module of homomorphisms @xmath271)\rightarrow\operatorname{\bbbk}[m]$ ] . recall that we also have a degree - preserving bijection between homomorphisms @xmath271)\rightarrow\operatorname{\bbbk}[m]$ ] and coderivations @xmath272)\rightarrow\operatorname{\mathsf{d}}(\operatorname{\bbbk}[m])$ ] . in fact this bijection defines an isomorphism of dg - modules when we equip the collection of coderivations @xmath272)\rightarrow\operatorname{\mathsf{d}}(\operatorname{\bbbk}[m])$ ] with the natural dg - module structure coming from dg - module homomorphisms . now , on the side of the free complete @xmath78-algebra @xmath250 , we obtain : [ freecompletealgebras : continuousderivations ] for any free complete @xmath78-algebra @xmath250 on one variable @xmath52 of degree @xmath273 , we have degree - preserving bijections between : 1 . the elements @xmath274 , 2 . the derivations of @xmath78-algebras @xmath275 which are continuous with respect to the topology of @xmath250 . this bijection defines an isomorphism of dg - modules too when we equip the collection of continuous derivations with the natural dg - module structure coming from dg - module homomorphisms . in the sequel , we adopt the notation @xmath276 for the continuous derivation associated to an element @xmath274 . to check the proposition , we essentially observe that @xmath277 is defined termwise by the formal identity @xmath278 extending the formula of [ background : twistedobjects ] . the sum runs over all inputs of the operation @xmath141 . then we use the continuity of the derivation to identify @xmath277 on the whole dg - module @xmath250 . in the converse direction , we simply retrieve the element @xmath279 associated to a derivation @xmath277 by the identity @xmath280 . from the explicit construction of the derivation @xmath276 , we obtain : [ freecompletealgebras : derivationadjunction ] let @xmath274 . let @xmath281)\rightarrow\operatorname{\bbbk}[m]$ ] be the homomorphism associated to @xmath279 by the duality isomorphism @xmath282)^{\vee}$ ] of [ freecompletealgebras : duality ] . the adjoint homomorphism of the derivation @xmath276 corresponds under the duality isomorphism @xmath260)$ ] to the coderivation @xmath283)\rightarrow\operatorname{\mathsf{d}}(\operatorname{\bbbk}[m])$ ] associated to @xmath284 . formal from the expression of the derivation @xmath277 in the proof of proposition [ freecompletealgebras : continuousderivations ] of the coderivation @xmath285 in [ background : twistedobjects ] , and from the expression of the duality isomorphisms . the construction of quasi - free algebras over an operad can be adapted to free complete algebras . in that situation , we consider twisted dg - modules of the form @xmath286 where the twisting morphism @xmath287 is a continuous derivation of @xmath250 associated to some element @xmath274 . as in the context of quasi - free algebras , the derivation relation implies that the equation of twisting homorphisms @xmath288 is fulfilled for @xmath287 if and only if it holds on the generating element @xmath52 of @xmath250 . because of the relation @xmath280 , this equation amounts to the identity @xmath289 in @xmath250 . the purpose of the next paragraphs is to explain that the composition structure considered in the introduction of this article models expressions of the form @xmath290 , where @xmath291 , and can be used to characterize twisting derivations on @xmath250 ( as well as twisting coderivations on the cofree coalgebra @xmath269)$ ] by the duality statement of proposition [ freecompletealgebras : derivationadjunction ] ) . to begin with , we review the definition of the composition structure considered in the introduction for a free complete algebra @xmath250 over any operad @xmath78 . this definition is borrowed to @xcite ( we also refer to @xcite for a variant of this composition structure ) . one first defines the composite of elements of homogeneous order @xmath292 by a sum of operadic substitution operations @xmath293 running over all entries of @xmath294 . this termwise composition operation is continuous , and hence , extends to the completion @xmath295 . for our purpose , we just consider the restriction of the obtained composition operation to the dg - module @xmath296 , in order to obtain the operation : @xmath297 note that this composition operation decreases degrees by @xmath273 , and becomes degree preserving only when we form @xmath273-desuspensions of the dg - module @xmath250 . this composition operation @xmath298 is , after desuspension , an instance of a pre - lie structure in the sense of @xcite , a product satisfying the identity @xmath299 the next lemma gives our motivation to introduce the pre - lie composition operation : [ freecompletealgebras : compositederivations ] for any pair @xmath291 , we have the identity @xmath300 in @xmath250 . this lemma is obvious from the explicit expression of the derivation @xmath277 associated to any @xmath274 in the proof of proposition [ freecompletealgebras : continuousderivations ] . this lemma implies : [ freecompletealgebras : twistingderivations ] the derivation @xmath276 associated to any @xmath274 defines a twisting derivation on @xmath250 if and only if we have the identity @xmath301 in @xmath250 . in the next sections , we use the identity of lemma [ freecompletealgebras : compositederivations ] to express the derivation @xmath277 associated by proposition [ freecompletealgebras : continuousderivations ] to any element @xmath274 . before proceedings to the study of the chain complex @xmath49 , we record some applications of the pre - lie composition structure to the definition of morphisms on the operad of @xmath3-algebras . first of all , recall that this operad is defined by the cobar construction @xmath302 , where @xmath303 is the dual cooperad of the operad of commutative algebras . the operad @xmath304 comes equipped with a weak - equivalence @xmath305 , where @xmath306 is the operad of lie algebras . hence , we have an identity @xmath307 and the homology @xmath308 is generated as an operad by an operation @xmath309 satisfying the identities of a lie bracket . this observation is used in [ homotopyinvariance ] . for the moment , we just want to record the following general proposition : [ freecompletealgebras : linfinitymorphisms ] if the operad @xmath78 is equipped with a free @xmath112-structure , then we have a bijection between : 1 . the operad morphisms @xmath310 ; 2 . and the elements @xmath274 , of degree @xmath64 , and such that the relation @xmath301 holds in @xmath250 , where we still assume that @xmath52 is a variable of degree @xmath53 . the cobar construction commutes with operadic suspensions . therefore we have an identity @xmath311 . since @xmath312 for all @xmath313 , the homomorphism @xmath314 associated to @xmath203 is equivalent to a collection of invariant elements in @xmath315 . we use the assumption about the @xmath112-structure of @xmath78 and the existence of an isomorphism @xmath316 to identify this collection with an element of the form @xmath274 . we easily check that the equation of [ koszuldualityreview : cobarconstruction ] for the twisting cochain @xmath204 amounts to the equation @xmath301 for the element @xmath279 and the proposition follows . the aim of this section is to make explicit the complex @xmath317 computing @xmath318 for the cochain algebra of a sphere @xmath36 . we still consider the particular @xmath2-operads @xmath27 arising from a filtration @xmath319 of the chain barratt - eccles operad @xmath320 . we do not need to review the definition of these operads @xmath27 and of the barratt - eccles operad @xmath320 . we are simply going to explain a representation of the @xmath2-algebra structure of @xmath38 in terms of realizations , for these operads @xmath27 , of intrinsic homotopical structures of @xmath2-operads . this representation will give the form of the complex @xmath317 associated to @xmath36 . we review the definition of the @xmath321-algebra structure of @xmath38 first , before explaining applications of the koszul duality result of @xcite to the cochain algebra @xmath38 . recall that the ( reduced normalized ) cochain complex of a @xmath128-sphere @xmath322 is identified with the free @xmath20-module of rank @xmath90 concentrated in ( lower ) degree @xmath323 : @xmath324.\ ] ] the associated endomorphism operad @xmath325 is identified with the @xmath128-fold operadic desuspension of the commutative operad @xmath14 because we have identities : @xmath326^{\otimes r},\operatorname{\bbbk}[-d ] ) = \operatorname{\bbbk}[rd - d ] = \lambda^{-d}\operatorname{\mathsf{c}}(r).\ ] ] hence , the @xmath17-structure of @xmath243 is determined by an operad morphism @xmath327 . the work @xcite gives an explicit representation of this morphism for the chain barratt - eccles operad . for our purpose , we use that @xmath328 is identified with a composite @xmath329 where @xmath330 denotes the augmentation of the barratt - eccles operad ( to which we apply the operadic suspension functor ) while @xmath331 denotes a new operad morphism @xmath332 ( to which we also apply the operadic suspension ) discovered in @xcite and defined by an explicit formula . in ( * ? * observation 0.1.10 ) , we observe that @xmath332 has restrictions giving operad morphisms @xmath333 , for each @xmath334 , and we prove : [ spherekoszulduality : suspensionkoszuldual ] the weak - equivalences @xmath208 , already considered in theorem [ koszuldualityreview : koszuldualitystatement ] , fit a commutative diagram @xmath335^{\sigma^}\ar@{.>}[d]_{\psi_{n-1}}^{\sim } & b^c(\operatorname{\mathsf{d}}_n)\ar@{.>}[d]_{\psi_n}^{\sim } \\ \operatorname{\mathsf{e}}_{n-1}\ar[r]_{\iota } & \operatorname{\mathsf{e}}_n } } , \ ] ] where @xmath336 is the embedding morphism and @xmath337 is the morphism associated to @xmath333 by functoriality of the construction @xmath338 . we recall , just after theorem [ koszuldualityreview : koszuldualitystatement ] , that the weak - equivalence @xmath339 arises as a lifting @xmath340!r+<4pt,0pt>;[r ] & \operatorname{\mathsf{e}}\ar[d]^{\epsilon } \\ b^c(\operatorname{\mathsf{d}}_m)\ar@{.>}[ur]^{\psi_m}\ar[rr]_{\phi_m } & & \operatorname{\mathsf{c}}}\ ] ] of a certain morphism @xmath341 . the notation @xmath330 refers again to the augmentation of the barratt - eccles operad . by cobar duality , we have an equivalence @xmath342 , where : @xmath343^{\iota^}\ar@{.>}[d]_{\psi_n^{\sharp}}^{\sim } & b^c(\operatorname{\mathsf{e}}_{n-1}^{\vee})\ar@{.>}[d]_{\psi_{n-1}^{\sharp}}^{\sim } \\ \lambda^n\operatorname{\mathsf{e}}_n\ar[r]_{\sigma } & \lambda^{n-1}\operatorname{\mathsf{e}}_{n-1 } } } .\ ] ] the notation @xmath344 now refers to the morphism associated to the embedding @xmath336 by functoriality of the construction @xmath345 . from this result , we deduce : [ spherekoszulduality : spherealgebramodel ] for every @xmath346 , we have a commutative diagram @xmath347_{\iota^}\ar[rr]^{\sim } & & \operatorname{\mathsf{e}}_n\ar[d]^{\sigma}\ar@{^{(}->}[]!r+<4pt,0pt>;[r ] & \operatorname{\mathsf{e}}\ar[d]^{\sigma } \\ \vdots\ar[d]_{\iota^ } & & \vdots\ar[d]^{\sigma } & \vdots\ar[d]^{\sigma } \\ b^c(\lambda^{-n}\operatorname{\mathsf{e}}_{m}^{\vee})\ar[r]_(0.43){\simeq } & \lambda^{m - n}b^c(\lambda^{-m}\operatorname{\mathsf{e}}_{m}^{\vee})\ar[r]_(0.6){\sim}\ar@/_1em/[drr]_{\lambda^{m - n}\phi'_{m } } & \lambda^{m - n}\operatorname{\mathsf{e}}_{m}\ar@{^{(}->}[]!r+<4pt,0pt>;[r ] & \lambda^{m - n}\operatorname{\mathsf{e}}\ar[d]^{\sim } \\ & & & \lambda^{m - n}\operatorname{\mathsf{c}}= \operatorname{\mathit{end}}_{\bar{n}^(s^{n - m } ) } } \ ] ] giving the action of the @xmath2-operad @xmath348 on @xmath38 , where @xmath349 is a morphism homotopic to @xmath341 in the model category of operads . note : one can easily observe that the cobar construction commutes with operadic suspensions . to determine the form of the complex @xmath350 , we essentially use the observation of [ koszuldualityreview : endomorphismoperads ] and the representation of the morphism @xmath351 supplied by this theorem . by ( theorem 5.2.2 ) or ( * ? ? ? * theorem c ) , any @xmath352-algebras @xmath353 and @xmath354 having the same underlying dg - module @xmath33 but a different @xmath352-structure determined by morphisms @xmath355 are connected by a chain of weak - equivalences of @xmath352-algebras when the morphisms @xmath356 are homotopic in the category of operads . hence , in our study of the cochain algebra @xmath38 , any choice of morphism @xmath351 in the homotopy class of the composite @xmath357 will give the right result . for that reason , we can replace the morphism @xmath349 in theorem [ spherekoszulduality : spherealgebramodel ] by @xmath341 , or any homotopy equivalent morphism @xmath358 . besides , the construction of @xcite gives a morphism @xmath341 well - characterized up to homotopy only . let us review the definition of these morphisms @xmath359 before going further in our study . the morphisms @xmath43 are , according to the recollections of [ koszuldualityreview : cobarconstruction ] , determined by homomorphisms of @xmath112-modules @xmath360 satisfying a certain equation . in @xcite , we observe that the definition of @xmath361 amounts to the definition of a collection of elements @xmath362 of degree @xmath363 , or equivalently to an element @xmath67 of degree @xmath64 in the completed @xmath58-algebra @xmath364 , where @xmath52 is again a variable of degree @xmath53 . the element @xmath365 simply represents the term of order @xmath57 of @xmath63 . recall that @xmath366 for all @xmath313 . the equivalence between @xmath365 and @xmath361 is given by the adjunction relation @xmath367 , for any @xmath368 , where @xmath369 is any representative of @xmath365 in @xmath370 . in the relation , we use the isomorphism @xmath371 defined by the norm of the @xmath115-action ( recall that @xmath115 acts freely on @xmath372 ) . by inspection of ( * lemma 1.1.2 ) , we immediately see that the equation of the homomorphism @xmath361 amounts to the equation @xmath373 in term of the element @xmath63 , where @xmath374 refers to the composition operation of [ freecompletealgebras : compositionstructure ] . hence , we have by proposition [ freecompletealgebras : twistingderivations ] : [ spherekoszulduality : twistingderivation ] the element @xmath63 associated to the homomorphism @xmath361 defining @xmath341 determines , for each @xmath50 , a twisting derivation @xmath375 such that @xmath69 , for any @xmath376 . then we observe : [ spherekoszulduality : derivationadjunctionrelation ] the adjoint homomorphism of the derivation @xmath375 corresponds , under the isomorphisms @xmath377)\simeq\lambda^{-n}\operatorname{\mathsf{e}}_n^{\vee}(\operatorname{\bbbk}[m - n ] ) = \operatorname{\mathsf{d}}_n(n^(s^{n - m}))\ ] ] to the twisting coderivation @xmath378 of the complex @xmath379 computing @xmath380 . in [ koszuldualityreview : endomorphismoperads ] , we observe that the homomorphism @xmath381)\rightarrow\operatorname{\bbbk}[m - n]$ ] defining the coderivation of @xmath382 . corresponds , by adjunction , to the twisting cochain @xmath383 } = \operatorname{\mathit{end}}_{n^(s^{n - m})}$ ] determining the action of the operad @xmath28 on @xmath384 . by theorem [ spherekoszulduality : spherealgebramodel ] , this twisting cochain is given by a composite : @xmath385},\ ] ] where @xmath344 is , up to operadic suspension , the dual of the embedding morphism @xmath42 . if we apply the formula of @xmath361 in terms of the element @xmath63 , then we obtain the identity @xmath386 for any @xmath387 , where we go back to the notation used around proposition [ freecompletealgebras : dualityisomorphism ] for the elements of @xmath388)$ ] . thus we immediately see that @xmath185 agrees with the homomorphism associated to @xmath83 in [ freecompletealgebras : dualityisomorphism]-[freecompletealgebras : continuousderivations ] , and the adjunction relation between the coderivation @xmath389 and the derivation @xmath69 associated to @xmath67 follows from the assertion of proposition [ freecompletealgebras : derivationadjunction ] . from this statement , we conclude : [ spherekoszulduality : homologycomplex ] we have an identity @xmath390 , where we consider the twisting derivation @xmath69 of lemma [ freecompletealgebras : continuousderivations ] and the continuous dual of the twisted complete @xmath58-algebra @xmath68 . the theorem of the introduction follows from this result and : [ spherekoszulduality : topologicalinterpretation ] take @xmath391 as ground ring . for any space @xmath7 whose homology @xmath392 is a degreewise finitely generated @xmath70-module , we have an identity @xmath393 where @xmath394 refers to the profinite completion of the @xmath0-fold loop space @xmath4 ( of which we take the continuous cohomology ) . we have explained around theorem [ spherekoszulduality : spherealgebramodel ] that the result of theorem [ spherekoszulduality : homologycomplex ] does not depend on the choice of the morphism @xmath341 in a certain homotopy class because any such choice gives rise to weakly - equivalent models of the @xmath2-algebra underlying @xmath384 . we prove in the next section that a wider homotopy invariance property holds for theorem [ spherekoszulduality : homologycomplex ] . we have a result similar to lemma [ spherekoszulduality : derivationadjunctionrelation ] for the complex @xmath395 computing the natural cohomology theory @xmath396 associated to @xmath2-operads . as a byproduct , we also have a cohomological analogue of the result of theorem [ spherekoszulduality : homologycomplex ] . at the complex level , we obtain an identity of the form @xmath397 where we equip the dg - module @xmath54 with the twisting cochain such that @xmath398 . in the previous section , we have used a filtration layer of the barratt - eccles operad as a preferred @xmath2-operad , but we mention in the introduction of this article that the result of our theorems holds for any choice of @xmath46-operad @xmath58 and for any choice of morphism @xmath43 . the goal of this section is precisely to prove this wide homotopy invariance property . recall that @xmath399 is identified with the operad of associative algebras @xmath16 for @xmath400 , with the @xmath273-gerstenhaber operad @xmath401 for @xmath402 . the associative operad is generated as an operad by an associative product operation @xmath403 , of degree @xmath23 . the @xmath273-gerstenhaber operad is generated by an associative and commutative product operation @xmath404 of degree @xmath23 and by a lie operation @xmath405 of degree @xmath406 satisfying a graded version of the poisson distribution relation . for the moment , we still consider the @xmath46-operad @xmath58 of @xcite , arising from the filtration of the barratt - eccles operad . in this context , the morphisms @xmath43 are characterized by the relation @xmath407 and @xmath408 in homology where we use the isomorphism @xmath409 ( established in @xcite ) , because we have : note that the relation @xmath408 is obvious for @xmath402 since @xmath14 is concentrated in degree @xmath23 . for the element @xmath63 associated to @xmath341 , the relation @xmath407 amounts to the identity @xmath414 = \mu$ ] in @xmath415 , where @xmath416 refers to the component of @xmath67 of order @xmath417 . recall that @xmath418 and the relation @xmath419 implies at order @xmath417 that @xmath416 defines a cycle in @xmath420 ( see ( * ? ? ? * lemma 1.1.2 ) ) . now we plan to extend the result of theorem [ spherekoszulduality : homologycomplex ] to @xmath2-operads @xmath421 which are @xmath112-cofibrant but not necessarily finitely generated in all degree and for each arity . in that context , we consider the operad @xmath304 , associated to the usual category of @xmath422-algebras , defined by the cobar construction @xmath302 , where @xmath303 is the dual cooperad of the operad of commutative algebras . recall that @xmath308 is isomorphic to the operad of lie algebras @xmath306 which is generated , as an operad , by an operation @xmath309 satisfying the identities of a lie bracket ( see [ freecompletealgebras : linfinityoperad ] ) . in the next statement , we use the operadic suspension of this operad @xmath423 for which the homology is generated , as an operad , by a lie bracket operation @xmath309 of degree @xmath417 . since we can not form the dual of @xmath424 when @xmath424 is not free of finite rank in all degree and for each arity , we can not use the relationship between twisting elements @xmath425 and morphisms @xmath426 , but we still have : [ homotopyinvariance : morphismrepresentation ] the definition of a morphism @xmath427 , amounts to the definition of an element @xmath428 , of degree @xmath64 , vanishing at order @xmath90 , and satisfying @xmath429 in @xmath430 . moreover , we have the identity @xmath431 in homology if and only if @xmath432 = \mu$ ] in @xmath433 , where @xmath434 represents the term of order @xmath417 of @xmath435 . the first part of the proposition is established in proposition [ freecompletealgebras : linfinitymorphisms ] . the second assertion of the proposition is an obvious consequence of the relation , established in the proof of proposition [ freecompletealgebras : linfinitymorphisms ] , between the homomorphism @xmath436 which determines @xmath48 and the element @xmath435 . suppose now we have an @xmath46-operad @xmath424 together with a morphism of the form of proposition [ homotopyinvariance : morphismrepresentation ] and let @xmath435 be the associated element in the free complete @xmath424-algebra @xmath430 . we aim at comparing the pair @xmath437 to a pair @xmath438 formed from the barratt - eccles operad . we already know that the operads @xmath58 and @xmath58 are connected by a chain of weak - equivalences : @xmath439{\phi}\pi_m\xrightarrow[\sim]{\psi}\xi_m.\ ] ] we can moreover assume that the morphism @xmath440 and @xmath441 are fibrations . we have then : [ homotopyinvariance : morphismcorrespondence ] we have elements @xmath442 and @xmath63 such that @xmath443 and @xmath444 and satisfying the identity @xmath445 = [ \omega_m(2 ) ] = \mu$ ] in homology , where as usual @xmath446 ( respectively , @xmath416 ) denotes the term of order @xmath417 of @xmath447 ( respectively , @xmath67 ) . the idea is to define the components @xmath448 of the element @xmath442 by induction on @xmath57 . for this , we use that the equation @xmath449 amounts at order @xmath57 to an equation of the form : @xmath450 where we use the standard partial composite notation of the operadic substitution operation occurring in the definition of the composition product @xmath451 . thus , whenever appropriate elements @xmath452 such that @xmath453 are defined for @xmath454 , the existence of @xmath448 amounts to the vanishing of an obstruction in @xmath455 . now this obstruction does vanish just because @xmath441 induces an isomorphism in homology , preserves operadic structures , and carries ( [ obstructionequation ] ) to the same equation for @xmath435 . hence we are done with the definition of @xmath447 . [ homotopyinvariance : complexequivalences ] we have a chain of weak - equivalences @xmath457{\phi}(\widehat{\pi}_m(x),\partial_m)^{\vee } \xrightarrow[\sim]{\psi}(\widehat{\xi}_m(x),\partial_m)^{\vee}\ ] ] between the continuous duals of the chain complexes formed from the elements of lemma [ homotopyinvariance : morphismcorrespondence ] . the commutation of @xmath440 and @xmath441 with operad structures readily implies that @xmath440 and @xmath441 induce dg - module morphisms between the twisted complexes . by the standard spectral sequence argument , we obtain that these dg - module morphisms are weak - equivalences since @xmath440 and @xmath441 induce weak - equivalences at the level of cofree coalgebras . [ homotopyinvariance : statement ] let @xmath458 be any nested collection of @xmath112-cofibrant operads equivalent to the nested sequence of the chain operads of little cubes . suppose we have an operad morphism @xmath459 satisfying @xmath431 in homology and consider the equivalent element @xmath428 such that @xmath429 . in this theorem , we do not have to assume that the operad @xmath424 are finitely generated . in particular , we can apply the theorem to the @xmath2-operads considered in @xcite formed by the semi - algebraic chain complex of the fulton - macpherson operad . b. fresse , _ koszul duality of operads and homology of partition posets _ , _ in _ `` homotopy theory : relations with algebraic geometry , group cohomology , and algebraic @xmath462-theory '' , contemp . math . * 346 * , amer . ( 2004 ) , 115215 . , _ operadic cobar constructions , cylinder objects and homotopy morphisms of algebras over operads _ , _ in _ `` alpine perspectives on algebraic topology ( arolla , 2008 ) '' , contemp . math . * 504 * , amer . math ( 2009 ) , pp . 125189 . v. hinich , v. schechtman , _ on homotopy limit of homotopy algebras _ , _ in _ `` @xmath462-theory , arithmetic and geometry ( moscow , 19841986 ) '' , lecture notes in mathematics * 1289 * , springer - verlag ( 1987 ) , pp .	the goal of this article is to make explicit a structured complex computing the cohomology of a profinite completion of the @xmath0-fold loop space of a sphere of dimension @xmath1 . our construction involves : the free complete algebra in one variable associated to any fixed @xmath2-operad ; an element in this free complete algebra determined by a morphism from the operad of @xmath3-algebras to an operadic suspension of our @xmath2-operad . as such , our complex is defined purely algebraically in terms of characteristic structures of @xmath2-operads . we deduce our main theorem from several results obtained in a previous series of articles namely : a connection between the cohomology of iterated loop spaces and the cohomology of algebras over @xmath2-operads , and a koszul duality result for @xmath2-operads . we use notably that the koszul duality of @xmath2-operads makes appear structure maps of the cochain algebra of spheres .
this paper is devoted to a study of families of equilibrium measures on the real line in the polynomial external fields . we consider these measures as functions of a parameter @xmath0 representing either the total mass , time or temperature . these families are regarded as models of many physical processes , which motivates their intense study in the context of the mathematical physics . but equilibrium problems for the logarithmic potential play an important role also in analysis and approximation theory ; in particular , they provide a general method used in the theory of orthogonal polynomials . in this sense , the subject is essentially `` bilingual '' and has so many ramifications that we opted for writing an extended introduction , instead of a dull enumeration of our results . one of our main goals is a formal investigation of the dependence of the key parameters of the equilibrium measure from the parameter @xmath0 , and in particular , their singularities as functions of @xmath0 . in thermodynamical terms , these singularities are closely related to phase transitions , and we start the introduction with the discussion of this problem in the context of statistical mechanics and thermodynamics . it does not mean that we actually intend to interpret our results in any nonstandard way . we want , on the contrary , to make a short review of a few existing standard interpretations , which could be helpful during `` translations '' in our bilingual area . then we formally introduce basic facts related to the equilibrium measure and mention briefly some applications in analysis and approximation theory , returning at the end to the topic of the coulomb gas and random matrices . in the course of these discussions we describe in general terms the main results of the paper and its structure . determinantal random point processes ( or self - avoiding point fields ) are pervasive in statistical mechanics , probability theory , combinatorics , and many other branches of mathematics @xcite . among the examples of such processes we can mention 2-d fermionic systems or coulomb gases , plancherel measures on partitions , two - dimensional random growth models , random non - intersecting paths , totally - asymetric exclusion processes ( tasep ) , quantum hall models , and random matrix models , to mention a few . many of these models lead to the so - called _ orthogonal polynomial ensembles _ @xcite , an important subclass of determinantal random point processes . the large scale behavior of such ensembles is described in terms of the asymptotics of the underlying family of orthogonal polynomials , and in the last instance , by the related equilibrium measure solving an extremal problem from the potential theory . one of the earliest and probably best known examples of orthogonal polynomial ensembles is the joint distribution of the eigenvalues of a random @xmath1 hermitian matrix drawn from a unitary ensemble ( see @xcite ) . more precisely , we endow the set of @xmath1 hermitian matrices @xmath2 with the joint probability distribution @xmath3 where @xmath4 is a given function with enough increase at @xmath5 to guarantee the convergence of the integral in the definition of the normalizing constant @xmath6 then @xmath7 induces a ( joint ) probability distribution @xmath8 on the eigenvalues @xmath9 of these matrices , with the density @xmath10 and with the corresponding _ partition function _ @xmath11 this result can be traced back to a classical theorem by h. weyl , see @xcite and also @xcite . notice that @xmath7 is invariant under unitary transformations of the hermitian matrices , while measure @xmath8 defines the equilibrium statistical mechanics of a dyson gas of @xmath12 particles with positions @xmath13 on @xmath14 in the potential @xmath15 , where they interact by the repulsive electrostatic potential of the plane ( the logarithmic potential ) . on the other hand , is also the joint probability density function of an @xmath12-point orthogonal polynomial ensemble with weight @xmath16 ( see e.g. @xcite ) . free energy _ of this matrix model is defined as @xmath17 regardless its interpretation , a particularly important problem is to analyze its asymptotic behavior in the thermodynamic limit , i.e. as @xmath18 . a rather straightforward fact , the existence of the limit @xmath19 ( infinite volume free energy ) has been established under very general conditions on @xmath15 , see e.g. @xcite and section [ subsect : equilibrium_analytic ] below . an important property of the infinite volume free energy @xmath20 is its analyticity with respect to the parameters of the problem . the values of the parameters at which the free energy is not analytic are the _ critical points _ ; curves of discontinuity of some derivatives of the free energy connecting the critical points divide the parameter space into different phases of the model . thus , critical points are points of phase transition , and they are going to be a center of our case study . the observation that the distribution can be regarded as the gibbs ensemble on the weyl chamber @xmath21 with hamiltonian @xmath22 allows to foretell the fundamental fact that the value of @xmath20 is given by the solution of a minimization problem for the weighted logarithmic energy . the corresponding minimizer is the _ equilibrium measure _ associated to the problem . ] . this measure , which is a one - dimensional distribution on @xmath14 , is also a model for the limit distribution of the eigenvalues @xmath9 in . indeed , the multidimensional probability distribution @xmath8 is concentrated for large @xmath12 near a single point @xmath23 , which is actually the minimizer for the corresponding discrete energy of an @xmath12-point distribution . in other words , for large @xmath12 the measure @xmath8 is close to @xmath24 . as @xmath18 , the discrete equilibrium measure converges to the continuous one , so that the thermodynamic limits are essentially described by the ( continuous ) equilibrium measures ( see section [ sec : equilibriuminanalysis ] ) . when @xmath15 is real - analytic , the support of such a measure ( or the asymptotic spectrum of the corresponding unitary ensemble ) is comprised of a finite number of disjoint intervals ( or `` cuts '' ) , and the number of these intervals is a fundamental parameter . for instance , the free energy @xmath25 has a full asymptotic expansion in powers of @xmath26 ( `` topological large @xmath12 expansion '' ) if and only if this support is a single interval ; otherwise , oscillatory terms are present ( this was observed in @xcite , and studied systematically in @xcite ) . the particularly interesting phenomena occur precisely in the neighborhood of the values of the parameters at which the number of the connected components of the support of the equilibrium measure changes . any change in the number of cuts is a phase transition in the sense specified above , but there are also phase transitions of other kinds ( not related to a change in the number of cuts ) : see sections [ subsec : classification ] and [ sec : phasetrans ] for details . we specialize our analysis to the polynomial potential . it is convenient to write @xmath15 in the form @xmath27 where @xmath28 is a polynomial of an even degree and positive leading coefficient . this case is of great interest , see e.g. @xcite , to cite a few references . a common situation is when @xmath29 in such a way that @xmath30 although the coefficients of @xmath28 ( `` coupling constants '' of the model ) play a role of the variables in the problem , the parameter @xmath31 stands clearly out , as it was already mentioned above . recall that it can be regarded either as a temperature ( from the point of view of statistical mechanics ) or time ( from the perspective of a dynamical system ) , and will correspond to the total mass of the equilibrium measure in the external field @xmath28 on @xmath14 . one of the goals of this paper is the description of the evolution of the limiting spectrum of the unitary ensemble for as time ( temperature ) @xmath0 grows from zero to infinity , paying special attention to the mechanisms underlying the increase ( `` birth of a cut '' ) and decrease ( `` fusion of two cuts '' or `` closure of a gap '' ) in the number of its connected components . equivalently , we study one - parametric families of equilibrium measures on @xmath14 in a polynomial external field with the total mass @xmath0 of the measure as the parameter . this problem has been addressed in several publications before , here we only mention a few of them . for instance , @xcite studied the fusion of two cuts for the quartic potential @xmath28 , proving the phase transition of the third order for the free energy ( in other words , that @xmath32 ) . this work found continuation in @xcite , where in particular the birth of the cut in the same model was observed ( but not proved rigorously ) . it turns out actually that the third order phase transition of the infinite volume free energy is inherent to all possible transitions existing in this model . the evolution of the support of the equilibrium measure in terms of the parameters of the polynomial potential has been studied also in @xcite , where a connection with some pde s and integrable systems have been exploited . however , despite of this intense activity , the picture is not complete ; even the monographic chapter @xcite contains imprecisions . open questions exist actually for the first non - trivial case of a quartic field , and this paper adds a number of new details to this particular picture ; some of them seem to be significant . for instance , we present a simple characterization of the case when the equilibrium measure has one cut _ for all _ values of the parameter @xmath31 , or when a singularity of type iii ( see the definition in section [ sec : revisited ] ) can occur . we also study in more detail the system of differential equations governing the dynamics of the endpoints of the support of the equilibrium measure . in particular , we analyze the behavior of the infinite volume free energy and of other magnitudes of the system near the singular points , revealing some interesting universal properties . finally , but not less important , we show that all the known and new results related to the outlined problem can be systematically derived from two basic facts in the potential theory . one of them is a representation for the cauchy transform of the equilibrium measure . this representation is a corollary of the fact that on the real line any equilibrium measure in an analytic field is a critical measure , which allows us to apply a variational technique presented in @xcite , and systematically in @xcite . another fact is a buyarov rakhmanov differentiation formula @xcite for the equilibrium measure with respect to its total mass and some of its immediate consequences , which we complement with a unified treatment of the differentiation formulas with respect to any coupling constant , a result which to our knowledge is new . we provide further details in section [ sec : critical ] . in this section we introduce basic notation and mention a number of fundamental facts on the equilibrium measures on the real line , necessary for the rest of the exposition . for more details the reader can consult the original papers @xcite and the monograph @xcite . for a finite borel measure @xmath33 with compact support @xmath34 on the plane we can define its _ logarithmic potential _ @xmath35 and its _ logarithmic energy _ , @xmath36= -\iint \log \|z - x\|\,d\sigma ( x)d\sigma ( z).\ ] ] suppose further that a real - valued function @xmath28 , called the _ external field _ , is defined on @xmath34 . then we introduce the _ total _ ( or `` chemical '' ) _ potential _ , @xmath37 ( defined at least where @xmath28 is ) and the _ total energy _ , @xmath38= i [ \sigma ] + 2\int \varphi(z ) \ , dz,\ ] ] respectively . this definition makes sense for a very wide class of functions @xmath28 , although for the purpose of this paper it is sufficient to consider basically real - analytic , actually polynomial , external fields on the real axis . as usual , we assume also that @xmath39 condition that is automatically satisfied for any real non - constant polynomial of even degree and positive leading coefficient . then for each @xmath31 there exists a unique measure @xmath40 with compact suport @xmath41 , minimizing the total energy @xmath42= \min_{\sigma \in { { \mathcal m}}_t } i_{\varphi } [ \sigma]\ ] ] in the class @xmath43 of positive borel measures @xmath33 compactly supported on @xmath14 and with total mass @xmath0 ( that is , @xmath44 ) . moreover , @xmath45 is completely determined by the equilibrium condition satisfied by the total potential @xmath46 where the _ equilibrium _ or _ extremal constant _ @xmath47 can be written as @xmath48-\int \varphi \ , d\lambda_t \right).\ ] ] when @xmath49 is a compact set on @xmath14 and the external field @xmath28 is @xmath50 the corresponding energy minimizer @xmath51 in the class of all probability measures @xmath52 , denoted by @xmath53 , is known as the _ robin measure _ of @xmath49 , its energy is the _ robin constant _ of @xmath49 , @xmath54,\ ] ] and the logarithmic capacity of @xmath49 is given by @xmath55 . measure @xmath53 is characterized by the the fact that @xmath56 together with the equilibrium condition @xmath57 for a general external field the main technical problem when finding the equilibrium measure is that its support is not known a priori and has to be found from the equilibrium conditions . once the support is determined , the measure itself can be recovered from the equations presented by the equality part in , which is an integral equation with the logarithmic kernel . after differentiation it is reduced to a singular integral equation with a cauchy kernel whose solution ( on reasonable sets ) has an explicit representation . hence , solving the support problem is the key , and it is essentially more difficult . in the one cut case , its endpoints are determined by relatively simple equations , but in general , straightforward solutions are not available . for a polynomial ( actually , real analytic ) field the support @xmath58 is a union of a finite number of intervals . equations may be written for the endpoints of those intervals ( and we better know the number of intervals in advance ) , but these equations are not easy to deal with . for instance , in the case of the polynomial field they may be interpreted as systems of equations on periods of an abelian differential on a hyperelliptic riemann surface ( see e.g. remark [ remark : riemann ] ) , but such systems are usually far from being simple . in this context , our approach is based on the dynamics of the family @xmath58 . as we have mentioned above , one of our goals is to describe the dynamics of the family of supports @xmath58 as the mass @xmath0 ( which is also `` time '' , or according to @xcite , the `` temperature '' ) changes from @xmath59 to @xmath60 . the detailed discussion starts in sections [ sec : critical ] and [ sec : polyn ] below . observe finally that a study of the family @xmath61 with total mass @xmath0 as a parameter in a fixed external field @xmath28 may be reduced by a simple connecting formula to the family of unit equilibrium measures with respect to family of fields @xmath62 . we have @xmath63 in some cases such a reduction is reasonable , but more often it brings more problems than benefits . in particular , we believe that the total mass as a parameter has particular advantages in the problem under consideration . the significant progress in the theory of rational approximations of analytic functions ( pad - type approximants ) and in the related theory of orthogonal polynomials in the 1980 s is due in part to the logarithmic potential method ; see the original papers @xcite and the monographs @xcite , although the list is not complete . a novel important ingredient of the techniques developed during that period was precisely the introduction of the weighted equilibrium measure ( among other types of equilibria ) . for instance , a typical application of the notion of equilibrium measures for orthogonal polynomials is the following fundamental result from @xcite ( see also @xcite ) , presented here in a simplified form : let @xmath28 be a continuous external field , @xmath64 , and let polynomial @xmath65 be defined by @xmath66 this polynomial is also characterized by the extremal property @xmath67 now , if @xmath29 in such a way that @xmath68 , then @xmath69 where @xmath70 is the weak-@xmath71 convergence . from the point of view of the spectral theory , zeros of @xmath72 are eigenvalues of the truncated @xmath1 jacobi matrices associated with the weight @xmath73 , and shows that @xmath74 can be naturally interpreted as the limit spectrum of the infinite jacobi matrix , associated with the scaling @xmath75 . for details related to applications of equilibrium measures in spectral theory of discrete sturm - liouville operators see @xcite . another large circle of applications of ( continuous ) equilibrium measures @xmath74 is related to the discrete analogue of this notion . let @xmath76 denote the set of all point mass measures on @xmath14 of total mass @xmath77 . the corresponding discrete energies are defined by @xmath78:= \sum_{i\neq j } \log\frac{1}{\|\zeta_i-\zeta_j\| } , \quad e_\varphi[\mu]:= e[\mu]+2\int \varphi\ , d\mu\ ] ] ( cf . ) . let @xmath79 be a minimizer ( not necessarily unique ) for @xmath80 $ ] in the class @xmath81 ; points @xmath82 are also known as _ weighted fekete points_. we have @xmath83 \longrightarrow i_\varphi [ \lambda_1 ] \quad \text{as } n\to \infty;\ ] ] for the proof , see e.g. @xcite , where some applications of fekete points in approximation theory are also discussed . furthermore , for a class of external fields @xmath28 , the polynomials @xmath84 satisfy a second order differential equation with polynomial coefficients and constitute an example of the so - called heine - stieltjes polynomials ; see @xcite for further details on the algebraic and analytic properties of these polynomials . as it was shown in @xcite , the limit zero distribution of such polynomials in the classical case studied by stieltjes @xcite is governed again by the ( continuous ) one - parametric family of equilibrium measures . also a related work worth mentioning is @xcite , where deformation with respect to the parameters of the external field were considered . let us return to the random models mentioned in the beginning of this introduction . an important fact , already mentioned there , is that the eigenvalue distribution of the unitary ensemble of random matrices may be equivalently written in terms of the discrete energy . indeed , since we deal only with symmetric functions of a vector @xmath85 , we may identify @xmath86 with the non - ordered collection @xmath87 of real points . this collection uniquely determines measure @xmath88 with @xmath89 by @xmath90 so that we can identify also @xmath91 and write @xmath92 instead of @xmath93 . now , we define @xmath94 where @xmath95 . distribution @xmath96 coincides with @xmath97 in if we identify @xmath98 and use also connection . as observed before , is a canonical gibbs distribution for a dyson gas . in this connection , the classical heine s formula @xmath99 represents the orthogonal polynomial @xmath72 with respect to the weight @xmath100 as the average of polynomials with random real zeros distributed according to @xmath101 . in the last decade the distribution @xmath102 and its characteristics , such as the correlation function , has been intensively investigated in relation to the eigenvalue statistics of the unitary ensembles of hermitian matrices . one of the relevant features of @xmath102 in the asymptotic ( thermodynamic ) regime is that it exhibits non - trivial phase transitions or critical behavior , with universal properties that attract increasing attention . the existence of such phase transitions and their character has important physical interpretations , see e.g. @xcite . the equilibrium measure @xmath61 ( and consequently , the parameters upon which it depends ) is called _ regular _ ( see e.g. @xcite ) if the following conditions hold : ( a ) @xmath61 is absolutely continuous on @xmath14 with respect to the lebesgue measure and @xmath103 does not vanish in the interior of @xmath58 ; ( b ) @xmath103 has a square root vanishing at the endpoints of the connected components of the support @xmath58 ; and ( c ) the inequality in is strict in @xmath104 . otherwise , the equilibrium measure is singular . following the terminology of @xcite , an equality in at a point in @xmath104 is a _ singularity of type i _ ( this is also known as `` birth of the cut '' , see below ) ; zeros of @xmath103 that lie in the interior of @xmath58 are called _ singular points of type ii _ ( corresponding to a closure of a gap ) , while the endpoints of the connected components of @xmath58 where @xmath103 has a higher order vanishing are called _ singular points of type iii_. it is rather easy to see that the infinite volume free energy or the correlation kernels are analytic functions of the parameters of the problem as long as the underlying equilibrium measure is regular . hence , phase transitions ( in the physical sense of break of analyticity of the free energy ) are intimately connected to the singularities of the equilibrium measure ; this is the central idea of this work . our goal is to discuss the mechanisms of these phase transitions and to analyze the behavior of the free energy near its critical values . the asymptotic behavior of a random matrix ensemble near singular points of the equilibrium measure has been studied intensively ( see @xcite , to mention a few ) . a remarkable feature of this behavior is revealed in the so - called double scaling limit , where the parameters near the critical value are coupled with the @xmath12 ( the size of the matrix ) . in this case , with an appropriate scaling of the variables , the asymptotic behavior of the correlation kernel near the critical point depends only on the type of the singularity rather than on the potential itself , which is one of the manifestations of the universality in the random matrix models . it is known that the asymptotic behavior of the correlation kernel in a double scaling limit at the singular points is connected to the hierarchies of the painlev equations . in the general polynomial case , the local behavior at the birth of the cut was described in @xcite ( in terms of orthogonal polynomials with respect to the freud weight @xmath105 on @xmath14 ) , and proved almost simultaneously in @xcite using the riemann - hilbert techniques . for the type ii singularity @xcite , the kernel can be written in terms of the hastings mcleod solution of the painlev ii equation , while for the type iii singularity , the kernel is related to the painlev i transcendent @xcite . the double - scaling asymptotic analysis of this sort is beyond the scope and techniques of this paper . in the study of measures minimizing the total energy functional it is often convenient to consider simultaneously solutions of certain more general but closely related types of equilibrium problems and associated measures , which we call _ critical measures_. we discuss here two kinds of critical measures the local minima and the saddle points of the energy , used to derive an equation for the cauchy transform of equilibrium measure ( see theorem [ lem : mfra09a ] below ) . in this connection we make also a few general remarks on the equations that can be useful for the constructive determination of the equilibrium measures . we fix a smooth external field @xmath106 and @xmath31 . recall that the pair consisting of a positive measure @xmath107 ( the equilibrium measure with total mass @xmath0 ) and and an ( equilibrium ) constant @xmath108 is uniquely defined by the ( equilibrium ) conditions ; measure @xmath109 is also the unique global minimizer of the total energy in the class @xmath110 . under the assumptions above , it is absolutely continuous with respect to the lebesgue measure , so that @xmath111 on @xmath14 . consequently , the problem of the constructive determination of @xmath61 boils down to finding its support @xmath58 , a problem that has been under investigation for a few decades . the main conclusion is that this problem has a simple constructive solution if @xmath58 is a single interval ( see e.g. @xcite or ( * ? ? ? iv ) ) ; when the support is not connected , a satisfactory constructive solution is not available . let us provide some details in this respect . assume that the support is a union of @xmath112 disjoint intervals , @xmath113 , \quad a_1<a_2<\dots < a_{2p}.\ ] ] differentiating the equality part of the equilibrium conditions one obtains the singular integral equation @xmath114 this equation on the unknown function @xmath115 , given a fixed @xmath58 in and a fixed @xmath116 , is a well known object in the theory of singular integral equations . a general solution of this equation may be written in terms of singular integrals ( see ( * ? ? ? * ch . 5 , 4 ) , as well as @xcite ) . the explicit expression for this solution is not so relevant here , but it renders a system of @xmath117 equations on the @xmath118 endpoints @xmath119 of the support @xmath58 ( see equations below ) that are central to our discussion . condition @xmath106 implies that @xmath120 , and in particular , that the density of the equilibrium measure is bounded on @xmath58 . thus , we are interested in the bounded solutions of , and by @xcite , it yields the following result : assume @xmath31 fixed . a bounded solution for exists if and only if @xmath121 furthermore , in this case the bounded solution is unique and is given by the formula @xmath122 as we have mentioned above , this theorem is a corollary of a representation for general solution of @xcite ; an alternative derivation of equations may be found in ( * ? ? ? * ch . 6 ) . a general solution of satisfies @xmath123 where constants @xmath124 are not necessarily the same on different intervals @xmath125 . in order to get we must impose @xmath126 extra conditions , @xmath127 since @xmath61 is uniquely determined by @xmath28 and @xmath128 , these are equations on @xmath129 s . actually , equations and , complemented with the normalization condition @xmath130 , constitute a system of @xmath118 equations on @xmath118 unknowns , the endpoints @xmath129 of the intervals in @xmath58 . using the expression for @xmath61 in they can be explicitly rewritten in terms of @xmath28 . we omit these calculations here , but see below , where these equations will be rewritten in a form more suitable for our needs . the question in what sense these equations determine @xmath131 is not trivial . actually , we still need to verify the inequalities @xmath132 , @xmath133 , along with @xmath134 , @xmath135 . if they are satisfied , the problem is solved . can this solution be regarded as constructive ? we can certainly assert this in the one - cut situation . but we are inclined to give a rather negative answer in the multi - cut case . to the best of our knowledge , the general support problem is actually a whole research program . let us point out that once @xmath58 has been established , it determines the partition of the total mass @xmath0 into its `` portions '' @xmath136 these values are closely related to the so - called t hooft parameters in the gauge theories . we can make additional progress under the assumption that the external field is analytic on the real line . more exactly , we assume that there is a domain @xmath137 containing @xmath14 and a function @xmath138 such that @xmath139 , @xmath140 . [ lem : mfra09a ] with the assumptions and notation introduced above , there exists an analytic function @xmath141 in @xmath137 , real - valued on @xmath14 , such that @xmath142 where @xmath143 is the cauchy transform of @xmath61 . formula can be obtained considering variations of the plane of the form @xmath144 , with @xmath145 , @xmath146 and @xmath147 , which induce variations on measures , and equating to zero the corresponding first variation of the total energy . in this way , the theorem is a direct corollary of a more general theorem on equilibrium measures of @xmath148-curves from @xcite or @xcite , and the fact that the real line is an @xmath148-curve in any real - analytic field . the notion of the @xmath148-property turned out to be crucial in the extension of the results on the real line to the equilibrium in a harmonic external field in @xmath149 , considering the complex plane as a conductor . in the described situation the representation remains valid , as well as many of its corollaries derived in this paper , with some obvious modifications . however , we will not go into details here and refer the reader to @xcite for a more thorough discussion . the proof of theorem [ lem : mfra09a ] actually carries over to a more general situation , e.g. when @xmath150 is meromorphic ( in this case @xmath151 might have poles at the location of poles of @xmath150 ) . in a more physical context , was obtained for instance as the saddle - point approximation to the large @xmath12 limit of the number of planar feynman diagrams of quantum field theory @xcite . the meromorphic @xmath150 has found applications also in the study of the asymptotic distribution of zeros of heine - stieltjes polynomials @xcite or in the analysis of the phase transition in matrix model with logarithmic action @xcite . let us point out briefly some corollaries of theorem [ lem : mfra09a ] . clearly , allows us to recover easily the density of the equilibrium measure @xmath45 on @xmath14 : @xmath152 and its total potential , @xmath153 with an appropriate choice of the lower limit of integration and of the branch of the square root . it follows further ( as it was shown in @xcite ) that the support @xmath58 of @xmath61 consists of a finite number of intervals , whose endpoints are simple real zeros of @xmath154 ; the unknown zeros of @xmath154 become the main parameters determining @xmath61 and its support . in this representation we do not have equations anymore ( they are now `` embedded '' in the representation , which automatically defines a bounded density ) . equations now take the form @xmath155 we could continue rewriting further the conditions on the unknown parameters , taking into account e.g. the explicit form of @xmath150 and identity , but in no case the equivalent equations will be elementary , or at least suitable for a truly constructive solution . one approach that renders an effective procedure for investigation of those equations is to consider the dependence of the system from its parameters . the main parameter is @xmath156 , and it is the central object in this paper . details are presented in subsections [ subsec : dynamical_first ] and [ subsec : dynamical ] below . a `` dynamical '' description of the support @xmath157 was proposed by buyarov and rakhmanov in @xcite ( see also ( * ? ? ? * formula ( 1.8 ) and lemma 1 ) , where the particular one - cut case was analyzed ) . for an external field @xmath158 , @xmath31 and a borel measure @xmath159 let us define two sets , @xmath160 when @xmath161 , we omit the reference to the measure , writing @xmath58 instead of @xmath162 and @xmath163 instead of @xmath164 . then the equilibrium condition can be expressed in the following equivalent form : @xmath165 and @xmath61 is the unique measure in @xmath166 satisfying . in the regular case , @xmath167 otherwise , points from @xmath168 are called in @xcite _ singular points _ of type i , and represent also a `` birth of a cut '' : @xmath169 ( see ( * ? ? ? * theorem 2 ) ) ; moreover , for an analytic and real - valued @xmath28 , @xmath168 is at most denumerable ( thus , @xmath170 ) . the following second main ingredient of our analysis is actually a simplified version of the result in @xcite , but can be derived also from theorem [ lem : mfra09a ] : [ lem : buyarov / rakhmanov:99 ] for the given real - analytic external field @xmath28 on @xmath14 , the family of the equilibrium measures @xmath61 is monotonically increasing , and for every @xmath31 , @xmath171=\rho(s_{t } ) , \quad \frac{d i_{\varphi } [ \lambda_t]}{d t}= 2 c_t.\ ] ] here @xmath53 and @xmath172 are correspondingly the robin measure and the robin constant of the compact @xmath49 , and @xmath47 is the extremal constant defined in . in the terminology of the random matrix ensembles , theorem [ lem : buyarov / rakhmanov:99 ] states that , up to a factor 2 , the first variation of the infinite volume free energy is the equilibrium constant , while the second variation is the robin constant of the support ( limit spectrum ) @xmath58 . formulas have been rediscovered in less generality in several works , see e.g. @xcite . the first formula in should be used in general in a weak sense : for any continuous and compactly supported function @xmath173 on @xmath14 , @xmath174 however , since in our context the equilibrium measures are absolutely continuous with respect to the lebesgue measure on @xmath14 , the differentiation of the measure @xmath61 with respect to @xmath0 in can be understood in the strong sense of the standard differentiation of its density @xmath175 . the detailed proof of theorem [ lem : buyarov / rakhmanov:99 ] ( under much less restrictive assumptions on @xmath28 ) was given in @xcite . here we just point out that for the real - analytic external field formulas can be established directly from theorem [ lem : mfra09a ] . indeed , a variation with respect to @xmath0 of the equilibrium property and of the normalization condition @xmath176 yield that @xmath177 is a probability measure with the logarithmic potential constant on @xmath58 . formula shows that it is absolutely continuous with respect to the lebesgue measure on @xmath14 , and that its density is of the form @xmath178 where @xmath179 is a polynomial . these property characterize the robin measure of @xmath58 , which establishes the first formula in . moreover , the constant value of @xmath180 on @xmath58 must be the robin constant of @xmath58 , so that by differentiating we get that @xmath181 . finally , by , @xmath182 = t c _ { t } + \int \varphi \ , d\lambda_t.\ ] ] from the first two identities in , just established , it follows that @xmath183}{d t } = c _ { t } + t \frac{d c_t}{d t } + \frac{d } { d t } \int \varphi \ , d\lambda_t = c _ { t } + t \rho(s_{t } ) + \int \varphi \ , d\omega_{s_t},\ ] ] and it remains to observe that @xmath184 since @xmath185 for @xmath186 , we can always assume that the external field @xmath28 is normalized by @xmath187 in ( * ? ? ? * theorem 2 ) it was additionally proved that with this normalization , the external field @xmath28 has the form @xmath188 where @xmath189 is the green function of the domain @xmath190 with pole at infinity , and the extremal constant @xmath47 has a limit when @xmath191 : @xmath192 here we bring to consideration another useful technique related to the variation of the external field . let @xmath193 be a one - parametric family of external fields depending on a parameter @xmath194 . without loss of generality , let @xmath195 ; for simplicity , we assume here that @xmath28 is real - analytic in both variables . the following result , extending the previous theorem , holds : [ thm : generalparameter ] let @xmath31 be fixed . assume that @xmath196 is real - analytic in both variables , and @xmath197 uniformly in @xmath198 in a neighborhood of @xmath199 . then @xmath200 where the ( in general ) signed measure @xmath201 is uniquely determined by the following conditions : @xmath202 let @xmath201 be any limit point ( in the weak topology ) of the family @xmath203 ( which exists due to its weak compactness ) . the first property of @xmath201 in is obvious ; the second one is a consequence that measures in are neutral ( their mass is 0 ) . finally , with our analyticity assumptions on @xmath28 it easily follows from ( see also ( * ? ? ? * lemma 5.3 ) ) that @xmath204 in the interior of the support of @xmath205 . consequently , @xmath206 which implies the third equality in . conditions determine @xmath201 uniquely . indeed , the logarithmic potential of the difference of two signed measures @xmath201 satisfying is constant and equal to @xmath59 on @xmath58 . it remains to use that the logarithmic energy is a positive definite functional on neutral signed measures ( see e.g ( * ? ? ? * theorem 1.16 ) ) . clearly , theorem [ thm : generalparameter ] can be extended to more general ( e.g. , smooth ) external fields @xmath28 , but we prefer to keep the settings as simple as possible , and still sufficient for our purposes . as it was mentioned above ( see ) , we can reduce our analysis to unit ( probability ) measures on @xmath14 , regarding the total mass or temperature @xmath0 as a particular parameter of the external fields of the form @xmath207 . consequently , it is convenient to clarify the relation between theorems [ lem : buyarov / rakhmanov:99 ] and [ thm : generalparameter ] . according to , @xmath208 the derivative of @xmath209 at a given @xmath31 satisfies with @xmath210 , and from theorem [ thm : generalparameter ] it follows that @xmath211 is a signed measure @xmath201 characterized by the equilibrium conditions . in this particular case @xmath201 can be computed explicitly : observe that @xmath212 so that applying theorem [ lem : buyarov / rakhmanov:99 ] we get @xmath213 obviously , the right hand side is a neutral measure ( @xmath214 ) satisfying , with the equilibrium constant equal to @xmath215 the reciprocal statement is also valid : formula shows that the logarithmic potential of @xmath216 is constant on @xmath58 , fact that combined with the monotonicity of @xmath61 yields the first formula in . after the brief review of the necessary technical background we have carried out in the previous section , we can start our study of equilibrium measures in a polynomial external field . we define @xmath217 where the coefficients @xmath218 ( the _ coupling constants _ ) are the parameters of the problem . for convenience , we adopt the normalization @xmath219 as well as include the total mass or temperature @xmath220 as an additional coupling constant . in this way , the equilibrium measure @xmath221 is a function of @xmath222 real parameters @xmath218 , @xmath223 . first , we reformulate theorem [ lem : mfra09a ] and its corollaries specifically for the case at hand , that is , when @xmath28 is defined by . the theorem implies that for the associated equilibrium measure @xmath45 with total mass @xmath31 and support @xmath225 , the function @xmath226 is a polynomial in @xmath227 of degree @xmath228 with coefficients depending of @xmath0 . actually , if we denote by @xmath229 the polynomial part of the expansion of @xmath173 at infinity , then the normalization @xmath176 implies that @xmath230 let @xmath231 , @xmath232 , denote the real zeros of @xmath154 of odd multiplicities . observe that @xmath154 can not have complex zeros of odd multiplicity , and each real zero of odd multiplicity of @xmath154 is an end point of @xmath58 , so that expression holds . we define monic polynomials @xmath233 and @xmath234 by @xmath235 ( notice that @xmath236 is a monic polynomial of degree @xmath237 whose zeros have even multiplicities and , therefore , the monic polynomial @xmath234 of degree @xmath238 is correctly defined ) . by , for @xmath239 there should be at least one zero of @xmath234 in each gap @xmath240 $ ] ; this readily yields the bound @xmath241 . naturally , both polynomials @xmath233 and @xmath234 , as well as the endpoints @xmath129 of the support @xmath58 , are functions of the parameter @xmath0 , fact that we usually omit from notation for the sake of brevity . in what follows we understand by @xmath242 and @xmath243 the holomorphic branches of these functions in @xmath244 determined by the condition @xmath245 and @xmath246 for @xmath247 , as well as @xmath248 denotes the boundary value of @xmath173 on @xmath14 from the upper half - plane . by , @xmath249 the factorization @xmath250 plays a fundamental role in the sequel . in particular , the @xmath151-representation of the the equilibrium measure in may be equivalently written in terms of the following @xmath224-representation : @xmath251 in a one - cut case ( @xmath252 ) the equalities and render a system of @xmath222 algebraic equations on the zeros of @xmath233 and @xmath234 , matching the number of unknowns . taking into account additionally that for @xmath252 equation is algebraic in @xmath119 s , this shows that in this ( and basically , only in this ) case all parameters of the equilibrium measure @xmath61 are algebraic functions of the coefficients @xmath253 of the external field . we conclude this section with two trivial but useful identities : @xmath254 and @xmath255 ( see ) . the @xmath224-representation introduced above has several advantages : it allows us to recast the discussion on the evolution of the support @xmath58 and of the phase transitions for the free energy in terms of the zeros of @xmath233 and @xmath234 . this will be carried out in the next two subsections . a more general analysis of the variation of @xmath58 as a function of all coupling constants is done in subsection [ subsec : variations ] . our first task is to rewrite the differentiation formulas of theorem [ lem : buyarov / rakhmanov:99 ] in terms of the zeros of @xmath233 and @xmath234 , which can be regarded as phase coordinates of a dynamical system . recall that if @xmath58 has the form , then its robin measure @xmath256 ( see the definition in section [ subsect : equilibrium_analytic ] ) is @xmath257 while for the complex green function @xmath258 with pole at infinity we have @xmath259 here @xmath260 is a real monic polynomial of degree @xmath126 ; for @xmath252 , @xmath261 , while for @xmath239 it is uniquely determined by conditions @xmath262 it follows in particular that for @xmath263 , @xmath264 , \quad j=1 , \dots , p-1.\ ] ] in order to denote the differentiation with respect to the total mass @xmath0 we will use either @xmath265 or the dot over the function , indistinctly . recall that @xmath129 are the zeros of @xmath233 , and denote by @xmath266 the zeros of @xmath234 . [ thm : dynamical system - t ] if @xmath233 and @xmath234 have no common zeros then @xmath267 if additionally all zeros of @xmath234 are simple , then the second set of equations in simplifies to @xmath268 taking the square root in both sides of and differentiating the resulting formula with respect to @xmath0 ( recall that @xmath28 does not depend on @xmath0 ) , we get @xmath269 by theorem [ lem : buyarov / rakhmanov:99 ] and , @xmath270 and further , with the help of , @xmath271 assertions of the theorem follow by equating residues of the rational functions in the last equation above . a subset of equations from corresponding to @xmath272 has been obtained in several places before , in particular , in the relevant work of bleher and eynard @xcite , although using a different approach . that paper is , also probably , one of the first works where some analytic properties of phase transitions in dyson gases were rigorously studied . it contains several noteworthy results , such as a discussion of the string equations and the double scaling limit of the correlation functions when simultaneously the volume goes to infinity and the parameter @xmath0 approaches its critical value with an appropriate speed . equations are similar in form to a system of odes studied by dubrovin in @xcite for the dynamics of the korteweg - de vries equation in the class of finite - zone or finite - band potentials . curiously , dubrovin s equations govern the evolution of the `` spurious '' poles of the diagonal pad approximants to rational modifications of a markov function ( cauchy transform @xmath273 ) @xcite , and are equivalent to the equations obtained by one of the authors in @xcite in terms of the harmonic measure of the support of @xmath201 . we will see in the sequel that conditions of theorem [ thm : dynamical system - t ] are satisfied for any @xmath0 except for a finite number of values . these values of @xmath0 may be called critical ; they correspond to some of the phase transitions but not to all of them . there are other significant values , also presenting phase transitions , which are not critical in the above mentioned sense . as it follows from theorems [ lem : buyarov / rakhmanov:99 ] and [ thm : generalparameter ] ( see also @xcite and ( * ? ? * theorem 1.3 . ( iii ) ) ) , the endpoints of the support of the equilibrium measure , its density function , and the corresponding equilibrium energy ( or the infinite volume free energy ) are analytic functions of the coefficients of the external field , except for a finite number of values where the analyticity breaks down . these critical points divide the parameter space into different phases of the model and represent phase transitions . some of them ( but not all ) correspond also to the change of topology of the support of the equilibrium measure . recall the classification of the singularities of the equilibrium measure we summarized in subsection [ sec : revisited ] . from the point of view of the evolution in @xmath0 and with the @xmath151- ( or @xmath224- ) representation at hand these singularities are now generically classified as follows . * * singularity of type i * is a bifurcation @xmath274 , representing a birth of the cut , is the event at a critical time @xmath275 when a simple zero of @xmath234 ceases to exist and at its place two new zeros of @xmath233 are born . formally , at @xmath275 the inequality , @xmath276 is no longer strict in @xmath104 , and the equality is attained at some point @xmath277 , where we will have @xmath278 . at this point a bifurcation of the zero @xmath279 occurs . + a significant property of this situation is that the phase transition occurs by saturation of the inequality ; the moment @xmath275 of the bifurcation is not defined by the dynamical system , i.e. it is not its singular point . all the phase parameters @xmath280 may be analytically continued through @xmath275 . the solution of the system for @xmath281 would give us a critical , but not an equilibrium , measure . * * singularity of type ii * is the opposite event @xmath282 , or fusion of two cuts , consisting in the collision and subsequent disappearance of two zeros of @xmath233 ( note that a zero of the complex green function , `` trapped '' in the closing gap , disappears simultaneously ) , and an appearance of a ( double ) zero of @xmath234 , followed by the splitting of this real double zero into two complex simple zeros . the collision of two zeros of @xmath233 is a critical point of the dynamical system . + we note that as a rare event ( event of a higher co - dimension ) it may happen that a number of other cuts were present in the vanishing gap @xmath283 immediately before the collision ; they all disappear at the moment of collision of @xmath284 and @xmath285 . this will be accompanied by an appearance for a moment of a zero of @xmath234 of an even multiplicity higher than two . + thus , a double zero ( or of an even higher multiplicity ) of @xmath234 is present at the moment of collision inside @xmath58 . according to , the density of @xmath61 vanishes at these points with an integer even order . * * singularity of type iii * are the endpoints of @xmath58 where @xmath154 has a multiple zero ; they correspond to the case when @xmath233 and @xmath234 in have a common real zero . additionally , a special situation is created when two complex - conjugate simple zeros of @xmath234 collide on @xmath104 , and either bounce back to the complex plane or continue their evolution as two real simple roots of @xmath234 ( at this moment , two new local extrema of the total potential @xmath286 in @xmath104 are born ) . at these values of the parameters the free energy is still analytic , so this is not a phase transition in the sense we agreed to use in this study , although the colliding zeros of @xmath234 lose analyticity with respect to @xmath0 . in a certain sense , the singularity of type iii is a limiting case of this phenomenon , when it occurs at an endpoint of @xmath58 . the reader should be aware of a certain freedom in our terminology regarding the singular points : we refer to singularities meaning both the value of the parameter @xmath0 at which a bifurcation occurs , and the point on @xmath14 where the actual bifurcation takes place . we hope that the correct meaning in each case is clear from the context and will not lead the reader to confusion . we return to the analysis of the local behavior at the singularities of the system in terms of @xmath287 in section [ sec : phasetrans ] . so far we have been regarding the equilibrium measure @xmath61 and its @xmath151- and @xmath224-representations as functions of the total mass @xmath0 , assuming that @xmath28 is fixed . now we discuss a more general problem : the dependence of @xmath288 and , correspondingly , of @xmath151 , @xmath233 , and @xmath234 from all the coefficients @xmath218 of the external field . it is a remarkable fact that the differentiation formulas with respect to the coupling constants @xmath218 , @xmath289 , have the same form as the differentiation formula with respect to @xmath290 , and that they all can be obtained in a unified way . we start with a simple technical observation : [ existence_h ] given points @xmath231 and the corresponding monic polynomial @xmath291 , there exist polynomials @xmath292 , @xmath293 , @xmath294 , uniquely determined by the following conditions : @xmath295 and @xmath296 clearly , @xmath297 , where @xmath260 is the numerator of @xmath298 introduced in subsection [ subsec : dynamical ] , see . furthermore , for each @xmath299 , means that the coefficients corresponding to powers @xmath300 of the laurent expansion of @xmath301 at infinity vanish , which renders @xmath302 linear equations , additional to @xmath126 linear equations on the @xmath303 coefficients of @xmath292 . the corresponding homogeneous linear equations are obtained by setting @xmath304 along with . in particular , every @xmath292 is of degree at most @xmath305 , and according to , has a zero in each interval @xmath306 , @xmath307 , which yields only the trivial solution for this system . since we are going to write all differentiation formulas in a unified way , we prefer to use here the notation @xmath308 with @xmath220 . [ thm : generaldiff ] let the polynomial external field @xmath28 be given by . let also @xmath309 denote the corresponding equilibrium measure of mass @xmath31 , and let polynomials @xmath233 and @xmath234 be the @xmath224-representation of this equilibrium measure , see . then @xmath310 where polynomials @xmath292 are given in lemma [ existence_h ] . moreover , @xmath311 in consequence , @xmath312 and @xmath313 for @xmath314 , formulas and are just a restatement of theorem [ thm : dynamical system - t ] . furthermore , from theorem [ thm : generalparameter ] it follows that for @xmath315 , @xmath316 where @xmath317 is a signed measure on @xmath58 satisfying @xmath318 it follows from that the multivalued analytic function @xmath319 has a single - valued real part , which is continuous in @xmath149 and satisfies @xmath320 taking into account we conclude that @xmath321 since for @xmath315 , @xmath322 identity of follows for all remaining @xmath302 s . finally , shows that @xmath323 is a homogeneous function of degree 1 of the vector of coupling constants @xmath324 , so that is just euler s theorem for such a function . formula is obtained by replacing and correspondingly in the left and right hand sides of . evaluating at the zeros of @xmath154 we obtain a set of algebraic identities , @xmath325 called _ hodograph equations _ in @xcite . solving them we could find the main parameters of the equilibrium measure and of its support . however , their explicit character is misleading : in the multi - cut case the dependence of the coefficients of @xmath292 from the coupling constants @xmath218 s is highly transcendental , and as the authors of @xcite point out , except for the simplest examples , equations are extremely difficult to solve , `` even by numerical methods '' . [ remark : riemann ] alternatively , following the general methodology put forward in @xcite , we can derive the identities on the endpoints of the connected components of @xmath58 considering the hyperelliptic riemann surface @xmath326 of @xmath327 and its deformations depending on the set of coupling constants @xmath218 , imposing the condition that the partial derivatives with respect to the parameters @xmath218 of the corresponding normalized abelian differentials of this surface are given by a meromorphic differential @xmath328 on @xmath326 . this is equivalent to the set of the so - called _ whitham equations _ @xcite on @xmath326 . actually , polynomials @xmath292 defined in lemma [ existence_h ] , appear in the explicit representation of these normalized abelian differentials of the third ( @xmath314 ) and second kind ( @xmath299 ) on @xmath326 . the key connection with the equilibrium problem is provided by identity , which shows that the differential @xmath329 , with @xmath28 given by , can be extended as this meromorphic differential @xmath328 on @xmath326 . this approach was used in @xcite to obtain in particular an analogue of , and developed further in @xcite . again , a direct consequence of theorem [ thm : generaldiff ] is the possibility to rewrite the differentiation formulas in terms of the zeros @xmath129 of @xmath233 and @xmath266 of @xmath234 : [ thm : odebis ] if under assumptions of theorem [ thm : generaldiff ] , @xmath233 and @xmath234 have no common zeros then for @xmath330 , @xmath331 if additionally all zeros of @xmath234 are simple , then the second set of equations in simplifies to @xmath332 it is a consequence of that @xmath333 observe that @xmath334 analogous formula is valid for @xmath335 . hence , both the left and the right hand sides in are rational functions in @xmath198 , with possible poles only at the zeros of @xmath233 and @xmath234 . the necessary identities are established by comparing the corresponding residues at each pole . for instance , with the assumption that the zeros of @xmath233 and @xmath234 are disjoint , the residue of the left hand side of at @xmath336 is equal to @xmath337 , which yields the first set of equations in ( recall that by construction , all zeros of @xmath233 are simple ) . the analysis of the residues at @xmath338 gives us the remaining identities . the proof shows how the statement can be modified in the case of coincidence of some zeros of @xmath233 and @xmath234 ( in other words , in the case of roots of @xmath154 of degree higher than 2 ) . moreover , the evolution of @xmath154 is such that as long as the right - hand sides in remain bounded , all zeros of @xmath154 , and hence , @xmath154 itself , are in @xmath339 . we can rewrite the equations on @xmath129 s in in a weaker form : @xmath340 which are the _ whitham equations in hydrodynamic form _ ( see ( * ? ? ? * eq . ( 77 ) ) ) . the simplest case to consider is when @xmath341 , so that @xmath58 consists of a single interval @xmath342 $ ] . in this situation , @xmath343 , @xmath344 , and under assumptions that @xmath233 and @xmath234 have disjoint zeros we obtain from that for @xmath345 , @xmath346 this can be rewritten as a system of pde , @xmath347 which is a rescaled form of the continuum limit of the toda lattice ( see @xcite ) . this set of equations for the endpoints of @xmath58 in the one - cut case was derived previously in @xcite . in an analogous fashion , @xmath348 which yields the following differential relations : @xmath349 again , these can be rewritten as dynamical systems , @xmath350 and @xmath351 \displaystyle \partial_3 a_2 + \frac{3(a_2-a_1)(4a_1 ^ 2 + ( a_1+a_2)^2 ) } { 16}\ , \partial_0 a_2 = 0 . \end{array}\ ] ] in this section we take a closer look at the behavior of the equilibrium energy for the external field when these phase transitions occur , considering only the variation of the total mass ( equivalently , temperature or time ) @xmath0 . in order to simplify our analysis , we make some generic assumptions on the three types of phase transitions described in section [ subsec : classification ] . along this section we denote by @xmath275 the critical time at which the phase transition occurs , and use the @xmath151- and @xmath352-representations of the equilibrium measure @xmath61 as introduced in section [ subsec : r - representation ] . * * singularity of type i : * at a time @xmath275 a real zero @xmath279 of @xmath234 ( a double zero of @xmath154 ) splits into two simple zeros @xmath353 , and the interval @xmath354 $ ] becomes part of @xmath58 ( _ birth of a cut _ ) ; we assume that for @xmath0 in a neighborhood of @xmath275 , @xmath279 is a simple zero of @xmath234 ; * * singularity of type ii : * at a time @xmath275 two simple zeros @xmath355 and @xmath356 of @xmath233 ( simple zeros of @xmath154 ) collide ( _ fusion of two cuts _ ) . * * singularity of type iii : * at a time @xmath275 a pair of complex conjugate zeros @xmath279 and @xmath357 of @xmath234 ( double zeros of @xmath154 ) collide with a simple zero @xmath358 of @xmath233 , so that @xmath359 as @xmath360 . additionally , in subsection [ subs : iv ] we analyze the scenario when at a time @xmath275 a pair of complex conjugate zeros @xmath279 and @xmath357 of @xmath234 ( double zeros of @xmath154 ) collide at @xmath361 and either bounce back to the complex plane or become two simple real zeros @xmath362 of @xmath234 ( these real zeros are new local extrema of the total potential @xmath286 on @xmath104 ) . obviously , for a general polynomial potential some of these phase transitions can occur simultaneously : for a given value @xmath275 two or more cuts could merge , while a new cut is open elsewhere , together with a type iii singularity at some endpoints of @xmath363 . still , the basic `` building blocks '' of all phase transitions are precisely the four cases described above , which we proceed to study . it was established in @xcite that in the second case , when at @xmath275 two components of @xmath58 merge into a single cut in such a way that @xmath28 is regular for @xmath364 , the equilibrium energy can be analytically continued through @xmath275 from both sides . for the case of a quartic potential @xmath28 it was shown in @xcite that the energy and its first two derivatives are continuous at @xmath275 , but the third derivative has a finite jump . this is a third order phase transition , observed also in a circular ensemble of random matrices @xcite . with respect to the singularity of type iii , it is mentioned in @xcite that in this case the free energy is expected to have an algebraic singularity at @xmath365 , `` but this problem has not been studied yet in details '' . in this section we extend this result to a general multi - cut case and show that for the three types of singularities the energy and its first two derivatives ( but in general not the third one ) are continuous at the critical value @xmath275 . the character of the discontinuities is also analyzed , and we can summarize our findings as follows : in all cases there is a parameter , that we call @xmath366 , expressing geometrically the `` distance '' to the singularity . in the case of a birth of a new cut , this is the size of this new component of @xmath58 ; for a fusion of two cuts , this is the size of the vanishing gap , while for the singularity of type iii we can take it as the distance between two colliding zeros of @xmath154 . in the three cases the first two derivatives of the equilibrium energy @xmath367 $ ] are continuous , while the third derivative has a discontinuity @xmath368 . we gather in the table below a rough information about the character of the dependence of these parameters from @xmath369 at the corresponding singularities , see subsections [ subs : birth][subs : iii ] for a detailed asymptotics and further discussion . @xmath370 [ cols="^,^,^,^ " , ] @xmath370 clearly , all zeros @xmath129 , @xmath266 of @xmath154 ( and hence , @xmath154 itself ) are continuous functions of @xmath0 . moreover , it follows from that when a singularity occurs , all @xmath129 s and @xmath266 s not involved in the phase transition are smooth functions of @xmath0 , with the degree of smoothness , roughly speaking , 1 more than the smoothness of @xmath366 in the corresponding column of the table above . recall that we assume that at @xmath275 a simple real zero @xmath279 of @xmath234 ( a double zero of @xmath154 ) splits into two simple zeros @xmath353 , and the interval @xmath354 $ ] becomes part of @xmath58 . first of all we want to estimate the size of the new cut as a function of @xmath0 . we use the notation introduced above , indicating explicitly the dependence from the parameter @xmath0 . let us remind the reader in particular that polynomial @xmath260 is the numerator of the derivative of the complex green function , see . our assumptions can be written as follows : for a small @xmath371 , there exist polynomials @xmath372 , @xmath373 and @xmath374 , such that @xmath375,\\ ( x - a_-)(x - a_+)\bm a(x;t ) , & t\in ( t , t+\varepsilon),\\ \end{cases } \\ b(x;t)&=\begin{cases } ( x - b ) \bm b(x;t ) , & t\in ( t-\varepsilon , t],\\ \bm b(x;t ) , & t\in ( t , t+\varepsilon),\\ \end{cases } \\ h(x;t)&=\begin{cases } \bm h(x;t ) , & t\in ( t-\varepsilon , t],\\ ( x-\zeta)\bm h(x;t ) , & t\in ( t , t+\varepsilon),\\ \end{cases}\end{aligned}\ ] ] where @xmath376 , @xmath279 and @xmath377 are real - valued continuous functions of @xmath0 such that @xmath378 we denote this common value by @xmath379 ; this is the place where the new cut is born at @xmath275 . remember that it is imposed only by the saturation of the inequality constraint in outside of @xmath58 . polynomials @xmath372 , @xmath373 and @xmath374 are continuous with respect to the parameter @xmath380 , but represent , generally speaking , different real - analytic functions of @xmath0 for @xmath381 and @xmath281 . let us denote @xmath382 , omitting from the notation when possible the explicit dependence on @xmath0 . from it follows that @xmath383 , \\ \label{locala } \dot{a_\pm } & = \pm \frac{2 h ( a_\pm ) } { ( a_+-a_-)\bm q(a_\pm ) } = \pm \frac{2 ( a_\pm -\zeta ) \bm h ( a_\pm ) } { ( a_+-a_-)\bm q(a_\pm ) } , \quad t\in ( t , t+\varepsilon).\end{aligned}\ ] ] it is convenient to introduce two new variables for @xmath281 : @xmath384 adding both equations in we get @xmath385 which shows in particular that @xmath386 and we conclude that the function defined piecewise as @xmath387 is also @xmath339 in @xmath388 . analogously , subtracting equations in yields that for @xmath389 , @xmath390 or equivalently , @xmath391 let us denote by @xmath358 the largest zero of @xmath372 satisfying @xmath392 . then condition reads as @xmath393 clearly , @xmath394 on the other hand , @xmath395 from and it follows now that if we denote @xmath396 then @xmath397 notice that the solution of the ode @xmath398 , with @xmath399 , satisfies @xmath400 so that implies that @xmath401 this is consistent with the @xmath402 scaling used e.g. in @xcite . we turn next to the asymptotic behavior of the robin constant @xmath403 of the support @xmath58 , which according to theorem [ lem : buyarov / rakhmanov:99 ] is the second derivative of the infinite volume free energy . let us study first the following model situation , from which the general conclusion is readily derived . assume that @xmath404 is a union of a finite number of disjoint real intervals , and @xmath405 $ ] is disjoint with @xmath406 for @xmath407 . we denote @xmath408 , and slightly abuse notation , writing @xmath409 instead of @xmath410 . observe that with this notation , @xmath411 . our goal is to study the behavior of @xmath409 as @xmath412 . as usual , @xmath413 stands for the green s function of @xmath414 with a pole at @xmath415 . the well - known relation holds : @xmath416 which defines the constant @xmath417 . [ lem : localrobin ] there exists a constant @xmath418 ( uniform in @xmath406 and @xmath419 ) such that for @xmath407 , @xmath420 let @xmath201 be the balayage of the unit point mass at @xmath419 onto @xmath406 , so that ( see @xcite ) @xmath421 comparing it with we conclude that @xmath422 indeed , the function in the right hand side is harmonic in @xmath423 , vanishing on @xmath406 , and has the appropriate logarithmic behavior when @xmath424 . recall that we denote by @xmath425 and @xmath426 be the robin measures of @xmath406 and @xmath427 , respectively . for a parameter @xmath428 define @xmath429 observe that since @xmath430 is bounded on @xmath406 , and @xmath431 is bounded away from zero on @xmath406 , @xmath33 is a positive measure for sufficiently small values of @xmath428 . consider the potential @xmath432 for such values of @xmath433 ; we want to find its bounds on @xmath434 . let us start with @xmath435 . we have @xmath436 the first parenthesis in the right hand side is @xmath437 by definition of @xmath425 ; the second one is equal to @xmath438 , where we use . so , we need to estimate the third parenthesis . we have @xmath439 so that @xmath440 using the straightforward bounds @xmath441 we conclude that @xmath442 gathering all estimates in we see that @xmath443 now we look for similar bounds on @xmath427 . observe that @xmath444 by the well - known property of @xmath425 , the first parenthesis is equal to @xmath445 ; by identity , @xmath446 while by the equilibrium condition , @xmath447 . thus , @xmath448 now we set the value of the parameter @xmath433 to guarantee that @xmath449 in other words , we take @xmath450 observe that @xmath451 for @xmath452 , so that we can choose @xmath453 such that @xmath454 defined in is a probability measure for @xmath455 . we are ready to establish the sought bounds on @xmath427 . since @xmath456 is constant on @xmath427 , we have @xmath457 clearly , both terms are harmonic on @xmath458 , so we conclude that @xmath459 where @xmath460 is uniform with respect to small variations of @xmath406 and @xmath419 . it remains to gather , and , and use the well - known property @xmath461 which concludes the proof of the lemma . in order to use this result in the situation of a birth of a cut it is enough to notice that @xmath462 is a differentiable function of @xmath0 at @xmath275 , and so is @xmath463 . taking into account we get @xmath464 in particular , @xmath403 is continuous but non - differentiable at @xmath275 , since it has an infinite jump : @xmath465 while the limit @xmath466 obviously exists and is finite . expression is consistent with the @xmath467 jump found in @xcite for the quartic potential ( the simplest transition from one to two cuts , see section [ sec : quartic ] ) . we turn our attention to the critical transition when two intervals from @xmath58 collide and merge into a single interval . more precisely , we assume that at a time @xmath275 two simple zeros @xmath355 and @xmath356 of @xmath233 ( simple zeros of @xmath154 ) coalesce . again , we start by estimating the size of the gap between two merging cuts as a function of @xmath0 . using the notation introduced above , our assumptions can be written as follows : for a small @xmath371 , there exist polynomials @xmath372 , @xmath373 and @xmath374 , such that @xmath468,\\ \bm h(x;t ) , & t\in ( t , t+\varepsilon),\\ \end{cases}\end{aligned}\ ] ] where @xmath469 , @xmath279 and @xmath377 are continuous functions of @xmath0 such that @xmath470 again , we denote this common value by @xmath379 . notice that @xmath469 , @xmath279 and @xmath377 are real - valued for @xmath471 $ ] , while @xmath472 for @xmath389 . polynomials @xmath372 , @xmath373 and @xmath374 are continuous with respect to the parameter @xmath380 , but represent , generally speaking , different real - analytic functions of @xmath0 for @xmath381 and @xmath281 . we denote @xmath382 , omitting from the notation when possible the explicit dependence on @xmath0 . from it follows that @xmath473 adding the first two equations in and using the notation introduced in we get @xmath474 so that @xmath475 analogously , from , @xmath476 and we conclude that the function defined piecewise as @xmath477 is @xmath339 in @xmath388 . furthermore , by , @xmath478 condition reads as @xmath479 with the change of variables @xmath480 in the integrand we obtain that @xmath481 where @xmath482 so that @xmath483 uniformly for @xmath484 $ ] . thus , @xmath485 analogously , by , @xmath486 with the same change of variables as above , we get @xmath487 where now @xmath488 so that @xmath489 uniformly for @xmath484 $ ] . consequently , @xmath490 combining and we see that @xmath491 using it in we conclude that @xmath492 this formula shows that two cuts can come together only at zeros @xmath379 of @xmath234 for which @xmath493 . furthermore , we have @xmath494 now we turn to the simplified model problem . assume that @xmath263 , @xmath495 where @xmath496 $ ] are pairwise - disjoint intervals , and assume that for certain @xmath497 , @xmath498 where @xmath419 is fixed and @xmath412 . no other endpoint @xmath129 depends on @xmath366 . by , for @xmath499 , we have @xmath500 with @xmath501 and @xmath502 where all @xmath503 $ ] , for @xmath504 ; notice that @xmath505 . zeros @xmath506 do depend on @xmath366 , so we will write @xmath507 when we want to make it explicit . observe also that @xmath508 and @xmath509 , @xmath510 , are the zeros of the derivative of the complex green s function for @xmath511 . [ lemma : merger ] with the notation above , @xmath512 moreover , @xmath513 with @xmath514 consider first the vector - valued function @xmath515 , assigning each value of @xmath366 to the corresponding vector @xmath516 , defined by equations . it is clearly differentiable for any @xmath517 . if we denote @xmath518 then by the implicit function theorem , @xmath519 consider the @xmath520-th row of the matrix in the right - hand side of for @xmath412 . we have @xmath521 since only the numerator of the integrand depends on @xmath506 , it is easy to see that @xmath522 on the other hand , the minor of the matrix in the right - hand side of obtained after eliminating the @xmath520-th row and column is clearly invertible for @xmath523 : it corresponds to the system for @xmath511 , and the endpoints of @xmath511 are free ends where no phase transition occurs . hence , expanding the matrix in along its @xmath520-th row we conclude that it is invertible for small values of @xmath455 , and we can write @xmath524 observe that for @xmath525 , @xmath526 while for @xmath527 , @xmath528 this proves . let us be more precise about the asymptotics of @xmath529 . by , @xmath530 thus , defining @xmath531 and making in the integral above the appropriate change of variables we obtain an expression for @xmath532 : @xmath533 where @xmath534 since @xmath535 , and defining @xmath536 , @xmath537 , we get that @xmath538 so that @xmath539 where @xmath540 we claim that in the asymptotic expression above we can replace @xmath541 by @xmath542 . this is a direct consequence of and the fact that in @xmath543 , only the numerator depends on @xmath366 . this proves the lemma . as in the case of the birth of a cut , we study the asymptotics of the robin constant @xmath544 as @xmath412 . observe that for @xmath545 and @xmath546 , @xmath412 , @xmath547 with the @xmath548 term uniform in @xmath198 . recall that for all @xmath549 , @xmath550 is a monic polynomial of degree @xmath305 , and by , @xmath551 uniformly in the endpoints of @xmath511 . this motivates us to define @xmath552 the existence of this limit will be established next . meanwhile , it is clear that @xmath553 is a polynomial of degree at most @xmath554 ( for @xmath555 , function @xmath543 in is a constant , so that by , @xmath556 in this case ) . assume that @xmath557 ; equations for @xmath558 yield : @xmath559 dividing it through by @xmath560 , using and considering limit when @xmath412 we obtain the following equations on @xmath553 : @xmath561 where @xmath562 is defined in . this renders a system of @xmath305 linear equations with @xmath305 unknowns ( the coefficients of @xmath553 ) , which has a unique solution ( as the consideration of the corresponding homogeneous system clearly shows ) , and in particular , establishes the existence of the limit in . from we have for @xmath563 , @xmath564 so that @xmath565 and the integral is convergent for every sufficiently small @xmath549 . taking into account , and , we conclude that @xmath566 . hence , dividing the identity above through by @xmath560 and using the definition of @xmath553 we get @xmath567 we summarize this in the following lemma : [ lem : robinasympt ] under the assumptions above , @xmath568 where the constant @xmath569 is given by the right hand side of , and the polynomial @xmath553 is uniquely defined by the equations for @xmath557 , or @xmath556 for @xmath570 . now we go back to the phase transition when two cuts merge . using formula we see that @xmath571 observe that this expression involves explicitly the point @xmath379 where two cuts merged ; clearly , this is not the case if we consider the limit of @xmath572 as @xmath573 . we could conclude from here that @xmath572 at @xmath275 has finite but , in general , different values from the left and from the right . in particular , @xmath403 is not differentiable at @xmath275 . [ example : merge2 ] as an illustration , let us consider the case when @xmath58 has only two cuts that merge into a single interval at @xmath275 . using the notation above , this means that @xmath574 , with @xmath575 for @xmath364 , @xmath576 , and @xmath577 . thus , by , @xmath578 since @xmath579 $ ] for @xmath580 , we have @xmath581 . using , @xmath582 so that @xmath583 where we take the values @xmath584 . consider now @xmath364 ; with the notations above , @xmath585 where @xmath586 furthermore , @xmath587 gathering these identities , setting @xmath588 and taking into account the expression of @xmath589 , we get @xmath590 from and we can easily compute the finite jump of @xmath591 at @xmath275 . further simplifications are obtained for the quartic potential , when @xmath592 . equations above boil down now to : @xmath593 ( this expression matches , after due transformations , formula ( 2.41 ) in @xcite ) , and @xmath594 so that @xmath595 in order to compare these formulas with those obtained in @xcite , we must set @xmath596 and @xmath597 , and denote @xmath598 . it yields @xmath599 this value differs slightly from the one in @xcite , probably due to a minor error in formula ( 2.55 ) therein . observe finally that that the jump in or is strictly negative , unless @xmath379 ( place of collision ) coincides with the midpoint of the interval @xmath600 $ ] . for the quartic external field it takes place if and only if it is symmetric , i.e. attains its global minimum at two distinct points . this conclusion is a straightforward consequence of the formulas in . indeed , if @xmath601 and @xmath602 , then the first formula in gives us that @xmath603 since the situation is invariant by translation in @xmath14 , we can conclude that if @xmath379 is the midpoint of the interval @xmath600 $ ] then @xmath28 has two global minima , situated at @xmath604 and a global maximum at @xmath605 . the reciprocal is also immediate . in this section we want to clarify the character of the phase transition in the case of a type iii singularity . to keep things simple , let us restrict our attention to the quartic case ( @xmath606 in ) , but the conclusions readily follow for the general situation . assume that for @xmath607 the type iii singularity occurs , without loss of generality , at the right endpoint of the support . this means that for @xmath380 , @xmath608 , @xmath609 where @xmath610 and @xmath611 , in such a way that both @xmath612 and @xmath613 are small with respect to @xmath614 , as @xmath615 . by , for @xmath616 , @xmath617 for the sake of brevity we denote @xmath618 then @xmath619 by , @xmath620 combining it with , we arrive at @xmath621 observe that by , @xmath622 for @xmath0 in a small neighborhood of @xmath275 , so that we can use @xmath366 as a new variable . dividing and by we get @xmath623 notice that the right hand sides are analytic at @xmath624 and @xmath625 as long as @xmath626 . thus , for any @xmath627 there exists an analytic solution @xmath628 of satisfying the initial conditions @xmath629 the series expansion of this solution at @xmath624 gives @xmath630 replacing it in yields @xmath631 if we denote @xmath632 , we will obtain from here that @xmath633 or @xmath634 we can rewrite it as @xmath635 and finally , @xmath636 the analytic function of @xmath366 in the left hand side of is invertible , and straightforward computations show that @xmath637 on the other hand , by , @xmath638 so that yields @xmath639 formulas and show that both @xmath640 and @xmath641 have the asymptotic order @xmath642 in a neighborhood of the critical value @xmath275 at which @xmath279 impacts @xmath643 and returns to the complex plane , i.e. where the type iii phase transition occurs . this is consistent with the result obtained previously in ( * ? ? ? * lemma 8.1 ) . it is convenient to point out also a certain universal behavior this transition exhibits , expressed in the fact that the leading terms in and depend on the initial value @xmath644 only . in particular , we see that @xmath279 impacts @xmath643 from the complex plane with the asymptotic slope of @xmath645 , both for the incidence and the reflexion angles , as it were for the actual reflection law in optics . we switch now to the analysis of the robin constant @xmath403 ; in the case we are studying , @xmath579 $ ] in a small neighborhood of @xmath275 , so that with @xmath646 , @xmath647 from we obtain that @xmath648 using that @xmath649 as @xmath615 and expressions and , we get that @xmath650 and @xmath651 taking into account we conclude that @xmath652 in other words , in the case of a singularity of type iii , we have again a third order phase transition with an infinite algebraic jump of the third derivative of the free energy at the critical time , with the exponent @xmath653 . we finally turn our attention to the situation created when a pair of complex conjugate zeros @xmath279 and @xmath357 of @xmath234 collide at @xmath361 and become two new simple zeros @xmath654 , @xmath655 of @xmath234 . it was mentioned that all @xmath129 s are analytic through @xmath275 , and thus by , @xmath234 and @xmath154 are also analytic functions of the parameter @xmath0 . using the notation introduced above , our assumptions can be written as follows : for a small @xmath371 , there exist polynomials @xmath372 , @xmath373 and @xmath374 , continuous with respect to the parameter @xmath380 , but represent , generally speaking , different real - analytic functions of @xmath0 for @xmath381 and @xmath281 , such that @xmath656 where @xmath279 and @xmath657 are continuous functions of @xmath0 such that @xmath658 notice that @xmath472 for @xmath364 ; without loss of generality , @xmath611 . a priori , we do not assume that @xmath657 are real - valued for @xmath389 , so the two possibilities are either @xmath362 or @xmath659 . we denote @xmath660 , omitting from the notation when possible the explicit dependence on @xmath0 . from it follows that @xmath661 subtracting / adding both equations in we easily get that @xmath662 it follows in particular that @xmath663 observe that these formulas show that the collision of @xmath279 and @xmath664 on @xmath104 can occur only at a position @xmath379 where @xmath665 moreover , assuming that for @xmath281 the new zeros @xmath666 are complex conjugate , the same formulas apply . this yields the partial conclusion : _ the scenario when the complex zeros of @xmath234 collide at @xmath104 and bounce back to the complex plane can occur only when @xmath379 is either a zero or a pole of @xmath667 . _ this is the situation , for instance , when @xmath379 coincides with one of the endpoints of @xmath58 , and in this case we get a type iii phase transition . thus , let us assume that @xmath668 this is always the case , for instance , if @xmath363 is a single interval . then , @xmath362 for @xmath389 . denoting again @xmath669 , we obtain in the same fashion that @xmath670 finally , we have mentioned that a singularity of type iii is a limit case of the situation analyzed here , when @xmath379 coincides with one of the @xmath129 s . furthermore , under assumption , a collision of a pair of complex conjugate zeros of @xmath234 at @xmath361 is followed by a type i phase transition ( birth of a new cut ) . however , these two phenomena can not occur simultaneously : if @xmath671 , conditions @xmath672 ( see ) are incompatible . in this section we consider in detail a particularly important case of a quartic potential , i.e. when @xmath606 in the representation . observe that this is the first non - trivial situation , since for @xmath673 ( quadratic polynomial ) all calculations are rather straightforward . according to section [ sec : polyn ] , for @xmath31 , @xmath58 has the form with either @xmath252 ( one - cut regime " or one - cut case " ) or @xmath570 ( two - cut case " ) . additionally to the description of all possible scenarios for the evolution of @xmath58 as @xmath0 travels the positive semi axis , we characterize here the quartic potentials @xmath28 for which @xmath58 is connected _ for all values of _ @xmath31 , as well as those for which the singularity of type iii ( higher order vanishing of the density of the equilibrium measure @xmath61 ) or the birth of new local extrema occur . roughly speaking , the evolution of @xmath58 can be described qualitatively as follows : for the quartic potential @xmath28 there exists a two - sided infinite sector on the plane , centered at a global minimum of @xmath28 and symmetric with respect to the horizontal line passing through this minimum and with the slope @xmath674 , such that @xmath58 is a single interval for all values of @xmath31 if and only if the other critical points of @xmath28 lie outside of this sector . otherwise , the positive @xmath0-semi axis splits into two finite subintervals and an infinite ray . the finite subinterval containing @xmath675 ( which may degenerate to a single point @xmath675 ) corresponds to the one - cut situation , and for the neighboring finite interval the support @xmath58 has two connected components . finally , the infinite ray corresponds again to the one - cut case . recall ( see sections [ subsec : classification ] and [ sec : phasetrans ] ) that the transition from one to two cuts occurs always by saturation of the inequality in , while the transition from two to one cut occurs by collision of some zeros of the right hand side of . let us give the rigorous statements . for any quartic real polynomial @xmath28 with positive leading coefficient we define the value @xmath676 as follows : let @xmath677 denote a point where @xmath28 attains its global minimum on @xmath14 ( which can be unique or not ) , and @xmath678 , any other critical point of @xmath28 ( zero of @xmath679 ) . then @xmath680 geometrically , @xmath676 is the square of the slope of the straight line joining @xmath681 and @xmath682 . notice that a real cubic polynomial has either 3 real zeros , or one real and two complex conjugate zeros , so that there is no ambiguity in the definition of @xmath683 . next , we define the critical slope : let @xmath684 denote the only positive root of the equation @xmath685 explicitly , @xmath686{5 \left(3072 \sqrt{6}-3107\right)}>0.\ ] ] alternatively , @xmath687 where @xmath688 is the only real solution of the equation @xmath689 [ thm : characterization ] let the quartic potential @xmath28 be given , and @xmath520 denote the critical value as described above . then , case @xmath690 : : : @xmath58 is a single interval for all values of @xmath31 , no singularities occur . moreover , non - real zeros of @xmath154 in move monotonically away from the real line if and only if @xmath691 . case @xmath692 : : : @xmath58 is a single interval for all values of @xmath31 , but there exists a ( unique ) value of @xmath0 for which a type iii singularity occurs ( and this is the unique phase transition ) . case @xmath693 : : : @xmath58 evolves from one cut to two cuts , and then back to one cut , presenting once the birth of new local extrema , a singularity of type i and a singularity of type ii , in this order . no other singularities occur . case @xmath694 : : : if @xmath28 attains its global minimum at a single point , then @xmath58 evolves from one cut to two cuts , and then back to one cut , presenting once a singularity of type i and a singularity of type ii , in this order . no other singularities occur . + if @xmath28 attains its global minimum at two different points , then @xmath58 evolves from two cuts to one cut , and only a singularity of type ii is present . we obviously consider @xmath31 ; the value @xmath675 is not regarded as a singularity . observe that for @xmath28 with more than one local extrema on @xmath14 there are no type iii phase transitions . we can also easily characterize the quartic external fields @xmath28 for which singularity of type iii occurs ( i.e. such that the zeros of @xmath679 lie on the critical line ) directly in terms of their coefficients . indeed , if @xmath695 , then @xmath696 so , without loss of generality , we may assume that @xmath697 let @xmath698 be a real zero of @xmath679 ; according to theorem [ thm : characterization ] , @xmath61 will have a type iii singularity for a certain value of @xmath0 if and only if the other two zeros of @xmath679 are of the form @xmath699 , @xmath700 , and the value @xmath701 is a root of the polynomial . in particular , @xmath702 comparing and we conclude that @xmath703 eliminating @xmath698 and @xmath704 from we find that @xmath705 the resultant of the polynomials in the left hand side of and is an integer multiple of @xmath706 . since both polynomials share a common root , @xmath520 , the resultant must vanish , and we conclude the following : for an external field @xmath679 with derivative of the form the equilibrium measure @xmath61 develops a type iii singularity if and only if @xmath707 using the substitution we can easily extend this result to the general case : _ for an external field @xmath679 such that @xmath708 the equilibrium measure @xmath61 develops a type iii singularity if and only if _ @xmath709 for instance , direct substitution shows that @xmath710 satisfies this condition . according to , in this case we will have a singularity of type iii for a finite value of @xmath275 , where the density of the equilibrium measure vanishes with the exponent @xmath711 . the value of @xmath712 can be found by the procedure described at the end of this section . theorem [ thm : characterization ] is a consequence of theorems [ thm : mainresult ] and [ thm : mainresult2 ] below , where some additional finer results on the dynamics of the equilibrium measure as a function of @xmath0 are established . since the problem is basically invariant under homotopy , horizontal and vertical shifts in the potential @xmath28 , as well as mirror transformation @xmath713 of the variable , in the rest of this section without loss of generality we assume that @xmath714 with both @xmath715 and @xmath716 in the closed right half plane . we have @xmath717 so that we may suppose that one of the following two generic situations takes place : case 1 : : : @xmath679 has three real roots , and @xmath718 ; case 2 : : : @xmath679 has one real root , at @xmath719 , @xmath720 is in the first quadrant , and @xmath721 . condition in case 1 is equivalent to saying that @xmath28 has on @xmath14 two local minima , at @xmath719 and @xmath722 , and a maximum at @xmath723 , in such a way that @xmath724 ( obviously , any general situation when @xmath679 has three real roots can be reduced to case 1 by an affine change of variables and by adding a constant to @xmath28 ) . since the situation @xmath725 ( or @xmath726 ) is equivalent to an even external field , in what follows we assume that the inequalities are strict : @xmath727 . case 2 means that @xmath28 has only one local extremum on the real line . in this way , the only case excluded from the analysis is @xmath728 , for which @xmath729 and the situation is trivial . for @xmath31 the polynomial @xmath154 in the right hand side of has degree 6 , and the identity takes the form @xmath730 for certain constants @xmath731 . the following technical result will be useful in what follows : [ lemma : equations ] assume that for @xmath154 in , @xmath732 or equivalently , @xmath733 . then if @xmath28 is given by , the following identities hold : @xmath734 this is a straightforward consequence of replacing in and equating the coefficients in both sides . [ remk1 ] it is important to observe that the identities remain valid under a homothetic transformation @xmath735 for @xmath736 . in other words , a linear scaling in space yields a quartic scaling in time ( or temperature ) . let us consider first case 1 ( with strict inequalities ) . the theorem below is the quantitative description of the following evolution : for small values of temperature @xmath0 a single cut is born in a neighborhood of the origin . the other two ( double ) zeros of @xmath154 are real and close to @xmath723 and @xmath722 , moving in opposite directions . at the first critical temperature @xmath737 a bifurcation of type i occurs : the rightmost double zero splits into two simple real zeros , giving birth to a second cut in the spectrum . this configuration is preserved until both cuts merge at a quartic point at a temperature @xmath738 ( phase transition of type ii ) . after that , two complex conjugate double roots of @xmath154 drift away to infinity in the complex plane , so that for the remaining situation we are back in the one - cut case . for the sake of convenience , here we use all introduced notations interchangeably , @xmath739 [ thm : mainresult ] let @xmath28 be an external field given by with @xmath740 . then there exist two critical values @xmath741 such that : * * ( phase 1 ) : * for @xmath742 , @xmath233 and @xmath234 in have the form @xmath743 with @xmath744 , and @xmath579 $ ] ( one - cut case ) . + parameters @xmath129 , @xmath266 are functions of @xmath0 and satisfy the system of differential equations @xmath745 with the initial values @xmath746 for @xmath742 , @xmath747 and function @xmath748 is monotonically decreasing and positive in @xmath749 , with @xmath750 * * ( 1st transition , phase transition of type i ) : * for @xmath737 such that @xmath751 we have @xmath752 , with @xmath753 , which is a singular point of type i ( `` birth of a cut '' ) . * * ( phase 2 ) : * for @xmath754 , @xmath233 and @xmath234 in have the form @xmath755 with @xmath756 , and @xmath579\cup [ a_3,a_4]$ ] ( two - cut case ) . + moreover , these values satisfy the system of differential equations @xmath757 parameters @xmath758 and @xmath759 are continuous at @xmath737 , while @xmath760 the value of @xmath761 is determined by @xmath762 in particular , @xmath763 the critical temperature @xmath764 is determined by the collision condition @xmath765 . * * ( 2nd transition , phase transition of type ii ) : * for @xmath738 , @xmath233 and @xmath234 in have the form @xmath766 where @xmath767 so that @xmath768 , and @xmath769 $ ] . points @xmath770 , @xmath771 and @xmath772 can be found from the first three equations in ( with @xmath773 , @xmath774 and @xmath775 ) , while the value of @xmath764 is obtained from the fourth equation . + since @xmath776 , @xmath777 is a singular point of type ii , and @xmath778 has a root of order 4 at @xmath779 . * * ( phase 3 ) : * for @xmath780 , @xmath233 and @xmath234 in have the form @xmath781 with @xmath782 , @xmath783 , and @xmath769 $ ] ( one - cut case ) . + moreover , these values satisfy the system of differential equations of the form ( setting now @xmath784 ) ; in particular , @xmath785 grows monotonically with @xmath0 , and @xmath786 if @xmath725 , then @xmath787 . the evolution is as described , except for @xmath788 ( phase 1 is missing ) . using and representation we conclude that there exists @xmath453 such that for @xmath789 we are in the one - cut case , and formulas hold , with both @xmath790 and @xmath791 close to @xmath715 and @xmath716 , respectively . in this situation , @xmath792 , so that is a particularization of , and the inequalities are just straightforward consequences of . observe also that implies that @xmath793 furthermore , for any finite initial positions @xmath794 , @xmath795 , with @xmath796 , the solution of the system of differential equations exhibits collision ( of @xmath643 and @xmath790 ) in finite time . indeed , by , for @xmath31 ( and before the collision ) , @xmath797 since @xmath798 , solving the corresponding ode we conclude that @xmath799 analogously , @xmath800 replacing these bounds in we get @xmath801 or @xmath802 thus , a collision will occur by time @xmath275 if @xmath803 since @xmath804 , @xmath805 as @xmath806 , the integral in the left - hand side diverges as @xmath807 , which proves that there will always be a collision in a finite time @xmath712 . observe that function @xmath808 in is well - defined in the whole interval @xmath809 , and that the integrand in is , up to a constant , the analytic continuation of the density of @xmath61 , see . using we conclude that @xmath810 however , at the collision time @xmath712 , @xmath811 which shows that there is a unique time @xmath812 for which @xmath751 . from the positivity of the measure @xmath61 and expression it is easy to conclude that all roots of @xmath154 , which are not endpoints of the support , need to be double . hence , for @xmath813 , the rightmost double root @xmath791 splits into a pair of simple real roots @xmath814 and @xmath772 , giving rise to formula . we apply again theorem [ thm : dynamical system - t ] with @xmath815 , which yields . observe that it is not straightforward to deduce the sign of @xmath816 from @xmath817 taking into account the initial values we see that for a small @xmath453 and @xmath818 , @xmath377 is close to @xmath814 , so that @xmath819 . this implies that immediately after the birth of a cut ( @xmath737 ) , @xmath820 , so that point @xmath790 still moves to the left `` by inertia '' . in that range of time , @xmath643 and @xmath814 ( and in consequence , also @xmath790 and @xmath377 ) are in the collision course , and collision occurs in finite time , when all these four points merge _ simultaneously_. this critical time @xmath821 can be characterized by the appearance of a quadruple root @xmath822 of @xmath778 inside @xmath823 , so that the system is valid . taking into account the monotonicity of @xmath58 , we see that for @xmath780 the quadruple root @xmath822 of @xmath824 splits into two complex - conjugate roots @xmath790 and @xmath825 , and formulas hold . adding the equations for @xmath266 s ( @xmath784 ) we obtain @xmath826 observe that @xmath827 is an equation of an `` east - west opening '' rectangular ( or equilateral ) hyperbola @xmath828 with its vertices at @xmath770 and @xmath772 , and @xmath829 corresponds to the connected component @xmath830 of its complement in @xmath149 containing the segment joining @xmath770 and @xmath772 . figure1 ( 17,46)@xmath831 ( 77,46)@xmath832 ( 3,79)@xmath833 ( 33,69)@xmath834 hence , we conclude from that @xmath835 moreover , the monotonicity of the support @xmath58 ( or equivalently , the fact that @xmath836 ) implies that @xmath837 clearly , for @xmath780 , @xmath838 , so that in this range of @xmath0 , @xmath839 grows monotonically , and we have a one - cut case for all @xmath780 . finally , since by assumptions , @xmath840 for all @xmath841 , using the representation we conclude that @xmath842 regarding the second limit in , observe that from equations , taking @xmath784 , we get @xmath843 an immediate consequence of these identities is that the centers of masses of the zeros of @xmath233 and of the zeros of @xmath234 are always in a collision course . a comparison of the coefficients at @xmath844 and @xmath845 in both sides of yields the system @xmath846 an assumption that @xmath847 ( collision ) in the last two equations in implies that @xmath848 , which is possible only if @xmath725 . this is the symmetric case not considered here . hence , we conclude that in our situation there exist the limits @xmath849 by using this in the first identity in we conclude the proof of . next , we turn to case 2 . [ thm : mainresult2 ] consider the external field given by with @xmath850 , and with @xmath851 in the first quadrant . then : 1 . if @xmath852 where @xmath853 , and @xmath520 is the only positive root of the equation , then for all @xmath31 , polynomials @xmath233 and @xmath234 in have the form @xmath854 with @xmath855 , and @xmath58 consists of a single interval @xmath600 $ ] ( one - cut case ) . + if @xmath856 , then functions @xmath857 , @xmath643 and @xmath858 grow monotonically with @xmath0 from their initial positions @xmath59 , @xmath59 and @xmath859 , respectively , to @xmath60 . otherwise ( i.e. , if @xmath860 ) , there exists a critical value @xmath861 such that @xmath858 is monotonically decreasing for @xmath862 , monotonically increasing for @xmath863 , and @xmath864 . 2 . if @xmath865 then there exist three critical values @xmath866 such that : * for @xmath862 , polynomials @xmath233 and @xmath234 in have the form with @xmath855 ; @xmath858 is monotonically decreasing for @xmath862 . at the critical value @xmath867 , @xmath868 , which is a birth of new local extrema ; without loss of generality , we assume @xmath869 . during this phase , @xmath58 consists of a single interval @xmath600 $ ] . * for @xmath870 , polynomials @xmath233 and @xmath234 in have the form , with @xmath744 , and @xmath579 $ ] ( still one - cut case ) . + at @xmath737 , function @xmath808 defined in satisfies @xmath751 , and @xmath871 is a singular point of type i ( `` birth of a cut '' ) . * for @xmath754 , phase 2 described in theorem [ thm : mainresult ] , takes place . second transition , also described in theorem [ thm : mainresult ] , occurs at @xmath738 by the same mechanism ( collision of four zeros of @xmath154 ) . this situation corresponds to the two - cut case . * finally , for @xmath780 we have phase 3 as described in theorem [ thm : mainresult ] . we are back in the one - cut case . as in the proof of theorem [ thm : mainresult ] , using and representation we conclude that there exists @xmath453 such that for @xmath789 we are in the one - cut case , and formulas hold . using the equivalence we conclude , in particular , that if @xmath872 , that is , if @xmath873 , then @xmath839 grows monotonically with @xmath0 , varying from @xmath874 to @xmath60 as @xmath0 travels @xmath875 . clearly , in this situation we have a one - cut case for all @xmath31 , which establishes the first part of ( a ) . as it was discussed before , for a fixed @xmath233 , the solutions @xmath790 and @xmath825 of exhibit collision in a finite time . the second part of case ( a ) corresponds to the situation when for a @xmath861 before the collision , @xmath876 ( see figure [ fig : hyperbola ] ) . finally , consider the case when @xmath790 hits the real line in a time @xmath867 . clearly , for @xmath877 , @xmath878 , so that @xmath879 , where @xmath880 . assume first that @xmath881 , @xmath882 . at @xmath883 we have that @xmath233 and @xmath234 in have the form @xmath884 with ( without loss of generality ) @xmath885 . as @xmath0 becomes greater than @xmath867 , the only possibility is that @xmath790 splits into two real zeros , so we are left in the situation of phase 1 of case 1 ( see theorem [ thm : mainresult ] ) . from this point the evolution of @xmath233 , @xmath234 and @xmath58 follows exactly the case 1 . the boundary between ( a ) and ( b ) is precisely when @xmath886 that is , when @xmath887 has a zero at @xmath790 of order 5 ; using the expression in we get that for @xmath883 , @xmath888 we can use the identities from lemma [ lemma : equations ] with @xmath773 , @xmath889 and @xmath890 . moreover , by the homogeneity of these identities ( see remark [ remk1 ] ) we may assume that @xmath891 , with @xmath892 : the boundary between case ( a ) and case ( b ) will be given by the ray @xmath893 , with @xmath894 . thus , from we obtain @xmath895 these equations must have a solution with @xmath896 for @xmath31 . solving the first two for @xmath770 and @xmath790 and replacing the solutions in the third equation ( preferably , using a symbolic algebra software ) we obtain @xmath897 which is equivalent to . direct calculations show that the discriminant of the cubic polynomial above is negative , so it has a unique ( positive ) root @xmath898 , which satisfies identities , and two complex conjugate roots . finally , a substitution of the obtained values for @xmath899 in the last equation renders @xmath900 . we conclude this section with a number of remarks that we consider relevant . the results from theorem [ thm : odebis ] allow us to explore the dependence of the support @xmath58 ( and of the corresponding equilibrium measure @xmath61 ) from the rest of the parameters of the problem , in our setting , from the position of the critical points @xmath715 and @xmath716 . for instance , when @xmath58 consists of a single interval ( `` one - cut case '' ) differential relations apply . observe e.g. that in the case when the center of masses of @xmath715 and @xmath716 increases ( hence , @xmath901 in decreases ) , both endpoints of @xmath58 move to the right . in a certain sense , the considerations above show that the most general situation corresponds to case 2 with @xmath715 lying in the complex plane but below the critical ray emanating from the origin and passing through @xmath902 ( or , in the terminology introduced at the beginning of this section , when @xmath903 ) . this and only this situation exhibits all possible transitions occurring in the quartic case . observe that the external field with @xmath850 , and with @xmath851 in the first quadrant is convex if and only if @xmath904 , where @xmath520 is the critical value given by . hence , the persistence of the one - cut case for all @xmath31 under assumptions of convexity ( see e.g. @xcite ) is a consequence of the statement above . [ rem : bleher ] let us return to the analysis of bleher and eynard in @xcite , where they considered the external field of the form @xmath905 this family always exhibits a second transition ( see theorem [ thm : mainresult ] ) or a singular point of type ii for @xmath906 , in such a way that @xmath907 . according to our analysis , for @xmath908 there is also a first transition ( birth of a cut ) at @xmath909 , not described in @xcite , but predicted in @xcite and found numerically in @xcite . expression is an alternative parametrization of the quartic external field ; the range @xmath910 covers all the cases when a two - cut @xmath58 arises , described in theorem [ thm : characterization ] . however , with the purpose of characterizing the existence of phase transitions we find the description given in our theorem [ thm : characterization ] more transparent . the analysis in @xcite was extended in @xcite to the case when @xmath28 is a polynomial of even degree ( greater than @xmath911 ) such that in the critical time @xmath738 the density of the equilibrium measure has a zero of higher order @xmath222 , @xmath912 . we finish this section observing that using the system of equations we can find algebraic equations for the quadruple critical points @xmath913 and @xmath822 and for the corresponding endpoints of the support @xmath914 and @xmath915 , respectively ( so that these values are algebraic functions of @xmath715 and @xmath716 , and hence , of the coefficients of the external field ) . indeed , solving the first two equations in for @xmath358 and @xmath916 , we get @xmath917 and replacing it in the third one yields @xmath918 if @xmath718 , this is equivalent to finding real roots of the polynomial @xmath919 for the values of the parameter @xmath920 $ ] . its discriminant ( easily found with the help of a computer algebra system ) , @xmath921 is positive for @xmath920 $ ] , and the roots of @xmath922 for @xmath923 are @xmath924 and @xmath925 . hence , this polynomial has three positive simple roots in the indicated range of @xmath926 . however , only the middle one guarantees that @xmath927 which , according to , is a necessary condition for having real solutions @xmath358 and @xmath916 in . as observed , for @xmath923 this root is @xmath928 , while for @xmath929 it is @xmath925 . hence , in the case when @xmath718 , the recipe for finding the ( unique ) critical point is as follows : with @xmath930 $ ] , find the root @xmath931 of lying in @xmath932,\ ] ] and take @xmath933 . then the endpoints of the support are obtained by replacing @xmath774 in , and taking @xmath934 ; the value of @xmath764 is finally computed from the fourth equation in . in this case , as we have seen , @xmath777 is a singular point of type ii ( zero of the density of the equilibrium measure ) . let us consider the case when @xmath720 is in the first quadrant , and @xmath721 . denoting @xmath935 and dividing by @xmath936 , we arrive at the polynomial equation @xmath937 whose discriminant , as natural , is given by the left hand side of with @xmath520 replaced by @xmath938 . thus , for @xmath939 , with @xmath520 defined in , polynomial @xmath922 has only one real root , which yields non - real values of @xmath358 and @xmath916 in . in this case , no quadruple critical point appears . if on the contrary , @xmath940 , then @xmath922 has three positive roots ; only two of them give real values of @xmath358 and @xmath916 in . summarizing , in the case when @xmath941 , the recipe for finding two critical points is as follows : find the real roots @xmath931 of , take @xmath942 and replace it in . if @xmath943 , then @xmath944 , @xmath945 and @xmath946 ; the value of @xmath867 is computed from the fourth equation in . if on the contrary , @xmath947 , then @xmath948 , @xmath949 and @xmath950 ; the value of @xmath764 is computed again from the fourth equation in . the first and the second authors have been supported in part by the research projects mtm2011 - 28952-c02 - 01 ( a.m .- f . ) and mtm2011 - 28781 ( r.o . ) from the ministry of science and innovation of spain and the european regional development fund ( erdf ) . additionally , the first author was supported by the excellence grant p09-fqm-4643 and the research group fqm-229 from junta de andaluca . a.m .- f . and e.a.r . also thank aim workshop `` vector equilibrium problems and their applications to random matrix models '' and useful discussions in that research - stimulating environment . we gratefully acknowledge also several constructive remarks from razvan teodorescu ( university of south florida , usa ) , arno kuijlaars ( university of leuven , belgium ) and tamara grava ( sissa , italy ) . a. i. aptekarev and w. van assche . asymptotics of discrete orthogonal polynomials and the continuum limit of the toda lattice . , 34(48):1062710637 , 2001 . symmetries and integrability of difference equations ( tokyo , 2000 ) . 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 . , 52(11):13351425 , 1999 . a. a. gonchar and e. a. rakhmanov . on the convergence of simultaneous pad approximants for systems of functions of markov type . , 157:3148 , 234 , 1981 . number theory , mathematical analysis and their applications . a. a. gonchar and e. a. rakhmanov . equilibrium measure and the distribution of zeros of extremal polynomials . , 125(2):117127 , 1984 . translation from mat . 134(176 ) , no.3(11 ) , 306 - 352 ( 1987 ) . a. a. gonchar and e. a. rakhmanov . equilibrium distributions and degree of rational approximation of analytic functions . , 62(2):305348 , 1987 . translation from mat . 134(176 ) , no.3(11 ) , 306 - 352 ( 1987 ) . a. b. j. kuijlaars and k. t .- generic behavior of the density of states in random matrix theory and equilibrium problems in the presence of real analytic external fields . , 53(6):736785 , 2000 . g. lpez and e. a. rakhmanov . rational approximations , orthogonal polynomials and equilibrium distributions . in _ orthogonal polynomials and their applications ( segovia , 1986 ) _ , volume 1329 of _ lecture notes in math . _ , pages 125157 . springer , berlin , 1988 . a. martnez - finkelshtein and e. a. rakhmanov . on asymptotic behavior of heine - stieltjes and van vleck polynomials . in _ recent trends in orthogonal polynomials and approximation theory _ , volume 507 of _ contemp . _ , pages 209232 . soc . , providence , ri , 2010 . n. i. muskhelishvili . . dover publications inc . , new york , 1992 . translated from the second ( 1946 ) russian edition and with a preface by j. r. m. radok . corrected reprint of the 1953 english translation . e. a. rakhmanov . strong asymptotics for orthogonal polynomials . in _ methods of approximation theory in complex analysis and mathematical physics ( leningrad , 1991 ) _ , volume 1550 of _ lecture notes in math . _ , pages 7197 . springer , berlin , 1993 . h. stahl . orthogonal polynomials of complex - valued measures and the convergence of pad approximants . in _ fourier analysis and approximation theory ( proc . budapest , 1976 ) , vol . volume 19 of _ colloq . jnos bolyai _ , pages 771788 . north - holland , amsterdam , 1978 . ` a. martnez - finkelshtein ( [email protected] ) department of mathematics university of almera , spain , and instituto carlos i de fsica terica y computacional granada university , spain e. a. rakhmanov ( [email protected] ) department of mathematics , university of south florida , usa r. orive ( [email protected] ) department of mathematical analysis , university of la laguna tenerife , canary islands , spain `	the paper is devoted to a study of phase transitions in the hermitian random matrix models with a polynomial potential . in an alternative equivalent language , we study families of equilibrium measures on the real line in a polynomial external field . the total mass of the measure is considered as the main parameter , which may be interpreted also either as temperature or time . our main tools are differentiation formulas with respect to the parameters of the problem , and a representation of the equilibrium potential in terms of a hyperelliptic integral . using this combination we introduce and investigate a dynamical system ( system of ode s ) describing the evolution of families of equilibrium measures . on this basis we are able to systematically derive a number of new results on phase transitions , such as the local behavior of the system at all kinds of phase transitions , as well as to review a number of known ones .
mesoscopic circuits with ferromagnetic elements are generating a growing interest , both for the fundamentally new effects they can exhibit due to the lifting of spin - degeneracy , and for the possibilities of technological advances , e.g. towards nanospintronics@xcite . the description of the interface between ferromagnetic and non - magnetic elements is crucial to understand the behavior of these devices . obviously , one first has to take into account the spin - polarization of the electronic transmission probabilities through the interfaces , a property which generates the widely known magnetoresistance effects@xcite . importantly , the phases of the scattering amplitudes also depend on the spin in general . this so - called spin - dependence of interfacial phase shifts ( sdips ) is less frequently taken into account . however , the sdips has already been found to affect the behavior of many different types of mesoscopic conductors with ferromagnetic leads , like diffusive normal metal islands @xcite , resonant systems @xcite , coulomb blockade systems@xcite , or luttinger liquids@xcite . the sdips can also have numerous types of consequences in superconducting / ferromagnetic ( @xmath0 ) hybrid systems @xcite . it can e.g. produce effective field effects in superconducting electrodes@xcite , or a phase shift of the spatial oscillations of the superconducting correlations induced by the proximity effect in a diffusive @xmath1 electrode@xcite . the sdips has also been predicted to induce triplet correlations in a @xmath2 electrode in contact with several ferromagnetic insulators with non - collinear polarization directions@xcite . recently , mean current measurements through @xmath0 quantum points contacts ( qpcs ) have raised much interest as a possible means to determine the polarization of a @xmath1 material . however , the effects of the sdips in these devices have not raised much attention so far . in most cases , the experimental results were interpreted in terms of a generalization of the blonder - tinkham - klapwick ( btk ) model@xcite to the spin - dependent case @xcite . we would like to emphasize that the description of a @xmath0 interface in terms of a delta - function barrier naturally includes a finite sdips but does not allow one to analyze its effects separately . moreover , the sdips included in the generalized btk model is specific for the delta - function barrier , and for a more realistic interface potential the sdips can be different . this is the reason why the influence of the transmission probabilities and of the sdips have to be investigated separately . the scattering approach seems perfectly adapted for this purpose since it allows to account explicitly for scattering phases@xcite . in a few works , current transport through @xmath0 qpcs was described with a single - channel scattering description@xcite . however , the sdips was not taken into account in these works . in this article , we study how the conductance and the current noise through a single channel @xmath0 qpc are affected by the sdips , using a scattering approach . we would like to emphasize that the current noise in hybrid @xmath0 systems has draw little attention so far@xcite , although it can provide more information on the system considered . we find that the behavior of the qpc depends on its spin - dependent normal state transmission probabilities @xmath3 , but also on the difference @xmath4 between the normal state reflection phases of up spins and down spins against @xmath1 . the device is thus more difficult to characterize than in the spin - degenerate case since there are two more parameters to determine , i.e. @xmath5 and @xmath4 . we show that a finite sdips @xmath6 can strongly modify the behavior of the qpc , for instance by shifting the andreev peaks appearing in the qpc conductance to subgap voltages , as found by zhao et al . for @xmath7 quantum point contacts@xcite , or by producing subgap peaks or dips in the differential fano factor curves . at finite temperatures , we find that the sdips - induced shift of the andreev peaks in the conductance can be difficult to distinguish from a gap reduction . we show that , in this context , a noise measurement can be very useful to characterize better the contact properties . this is particularly clear in cases where the qpc differential fano factor versus bias voltage is qualitatively modified by the sdips while the conductance curve remains similar to the curves obtained without a sdips . this paper is organized as follows : section ii.a introduces expressions of the conductance and noise of a @xmath0 interface in the scattering formalism , section ii.b presents an explicit expression of the scattering matrix of a single channel @xmath0 qpc , section ii.c discusses the conductance and noise of this qpc at zero temperature , and section ii.d discusses the finite temperature case . section iii concludes . we consider a superconducting / ferromagnetic ( @xmath0 ) interface with @xmath2 a conventional bcs conductor with a s - wave symmetry . we assume that the @xmath2 and @xmath1 leads can be considered as ballistic , so that the @xmath0 interface can be described with the scattering approach@xcite and the leads with bogoliubov - de gennes ( bdg ) equations@xcite . we use a gauge transformation to set the origin of the quasiparticles energies to the fermi level of @xmath2 , i.e. @xmath8@xcite . due to the bias voltage @xmath9 applied to the qpc , the fermi level of @xmath1 is @xmath10 , with @xmath11 the absolute value of the electron charge . in the following , we will denote by @xmath12 electron states with spin @xmath13 and by @xmath14 hole states in the @xmath15 spin band , with @xmath16 a spin component collinear to the polarization of @xmath1 . a hole @xmath14 corresponding to an empty electronic state at an energy @xmath17 carries the same energy @xmath18 as an electron @xmath12 occupying a state with energy @xmath18 . these two types of quasiparticles are coupled by andreev reflection processes occurring at the @xmath0 interface . one can thus write @xmath19 for @xmath20 , with @xmath21 and @xmath22 the scattering matrix of the @xmath0 interface for quasiparticles carrying an energy @xmath18 . here , we consider the single channel case , so that @xmath23 [ @xmath24 refers to the annihilation operator associated to the incident [ outgoing ] state for a particle with type @xmath25 and energy @xmath18 of lead @xmath26 . in the following , we use a picture consisting of both positive and negative energy states . the operator @xmath27 associated to the total current flowing through the device at time @xmath28 can be calculated as the sum of the electron and hole currents of the @xmath29 space , which replaces the summation on the spin direction@xcite . more precisely , one has @xmath30 with @xmath31 for @xmath32 . here , @xmath33 is the coordinate along the leads and @xmath34 is the transverse coordinate . the field operator @xmath35 associated with particles of type @xmath36 in lead @xmath1 is defined as @xmath37k_{f}% ( \varepsilon)z}+b_{f,\gamma}(\varepsilon)e^{-i\lambda\lbrack\gamma ] k_{f}(\varepsilon)z})\end{aligned}\ ] ] with @xmath38=-1 $ ] , @xmath39=+1.$ ] we have introduced above different quantities which characterize the conduction channel of the qpc , i.e. the transverse wave function @xmath40 , the wavevector @xmath41 , and the velocity of carriers @xmath42 . the operator conjugated to @xmath35 is denoted @xmath43 . neglecting the energy dependence of @xmath44 ( see e.g. ref . ) , one finds@xmath45\iint\limits_{-\infty , -\infty}^{+\infty,+\infty}d\varepsilon_{1}d\varepsilon_{2}e^{\frac { i(\varepsilon_{1}-\varepsilon_{2})t}{\hbar}}\nonumber\\ & \times\sum\limits_{\substack{\alpha_{1},\alpha_{2}\\m_{1},m_{2}}% } a_{m_{1},\alpha_{1}}^{\dagger}(\varepsilon_{1})\mathcal{a}_{f , m_{1},m_{2}% } ^{\gamma,\alpha_{1},\alpha_{2}}(\varepsilon_{1},\varepsilon_{2}% ) a_{m_{2},\alpha_{2}}(\varepsilon_{2 } ) \label{iinst}%\end{aligned}\ ] ] with@xmath46 in the above eqs . , capital latin indices correspond to the lead @xmath1 or @xmath2 , greek indices correspond to the electron or hole band of the space @xmath29 , and @xmath47 is the identity matrix in the subspace of states of type @xmath36 of lead @xmath1 . in this paper , we study the average current @xmath48 flowing through the interface and the zero - frequency current noise @xmath49 \left [ \hat { i}(0)-\langle i\rangle\right ] \right\rangle $ ] . equations ( [ iinst ] ) and ( [ a ] ) lead to the expressions @xcite@xmath50 and @xmath51 we have introduced above the fermi factors @xmath52\mu_{m}\right ) /k_{b}t]\right ) ^{-1}$ ] with @xmath53 the temperature . in the limit @xmath54 , eqs . ( [ igal ] ) and ( [ sgal ] ) fulfill the fluctuation dissipation theorem , i.e. @xmath55 with @xmath56*. * in the limit @xmath57 , simplified expressions of @xmath58 and @xmath2 can be obtained . in order to account for the two spin bands in the same way , we use a symmetry property stemming from the structure of the bdg equations , i.e. @xmath59 , with @xmath60 , @xmath61 , @xmath62 and @xmath63 ( see derivation in appendix a ) . equations ( [ igal ] ) and ( [ sgal ] ) combined with the unitarity of @xmath22 then lead to @xmath64 with @xmath65 and@xcite @xmath66 \label{sfns}%\end{aligned}\ ] ] with @xmath67 . note that eq . ( [ gfns ] ) is valid only if @xmath68 , with @xmath32 , can be considered as independent from @xmath9 . the integration domain @xmath69 in eq.([sfns ] ) is @xmath70 $ ] for @xmath71 and @xmath72 $ ] for @xmath73 , similarly to the fact that the elements @xmath74 contributing to @xmath75 in eq . ( [ gfns ] ) must be taken at @xmath76 depending on the sign of @xmath9 . before concluding this section , we note that in the multichannel case , the elements @xmath77 would be matrices relating the incoming and outgoing states of the different channels of leads @xmath78 and @xmath26 . in this case , one could generalize straightforwardly eqs . ( [ igal ] ) , ( [ sgal ] ) , ( [ gfns ] ) and ( [ sfns ] ) by applying to their right - hand sides a trace on the channels space and using @xmath79 . in section ii , we focus on the single channel case . we briefly discuss possible extensions of this work to the multichannel case in section iii . we consider a @xmath0 interface which is narrow compared to the coherence length of @xmath2 , so that it can be modeled as a specular @xmath80 interface in series with a ( possibly ) dirty @xmath81 interface , with the length of @xmath82 tending to zero@xcite . we would like to emphasize that the @xmath82 layer is introduced artificially , it is merely a trick to describe the superconducting interface on a scale much shorter that the coherence length . the matrix @xmath22 of the @xmath0 interface can expressed in terms of the scattering matrix @xmath83 of electrons with spins @xmath13 on the @xmath81 interface and of the amplitude @xmath36 of the andreev reflections at the @xmath80 interface . one finds in particular @xcite , @xmath84@xmath85@xmath86 ^{-1}%\]]@xmath87 ^{-1 } \label{s4}%\ ] ] from eqs . ( [ gfns ] ) and ( [ sfns ] ) , at @xmath57 , calculating the current and the noise through the @xmath0 interface only requires one to know the two elements of @xmath22 given above . for @xmath88 , one must use eqs . ( [ igal ] ) and ( [ sgal ] ) , so that the whole @xmath89 matrix is necessary ( the other elements of @xmath89 are given in appendix b ) . there remains to introduce explicit expressions for @xmath36 and @xmath90 . first , the amplitude @xmath36 can be calculated using the ballistic bdg equations to model the @xmath80 interface@xcite , with a step approximation for the gap @xmath91 of @xmath2 . this gives @xmath92 for @xmath93 and @xmath94 for @xmath95 . secondly , in the single channel case , the unitary of @xmath90 leads to @xmath96{ll}% $ \sqrt{r_{\sigma}}\exp\left [ i\varphi_{ff}^{\sigma}\right ] $ & $ \sqrt{t_{\sigma}}\exp\left [ i\varphi_{fn}^{\sigma}\right ] $ \\ $ \sqrt{t_{\sigma}}\exp\left [ i\varphi_{nf}^{\sigma}\right ] $ & $ \sqrt{r_{\sigma}}\exp\left [ i\varphi_{nn}^{\sigma}\right ] $ % \end{tabular } \ \ \ \ \right]\ ] ] with @xmath97 and @xmath98 $ ] . in principle , the scattering phases @xmath99 , with @xmath100 , are spin - dependent because electrons are affected by a spin - dependent scattering potential at the @xmath0 interface . for simplicity , we will assume that @xmath101 , @xmath3 and @xmath99 are independent from @xmath18 and @xmath9 , which ensures the validity of eq . ( [ gfns ] ) ( see ref . ) . note that the expression that we have introduced in this section for @xmath102 implies @xmath103 and @xmath104 , which is not a general property of superconducting hybrid devices ( see e.g. ref . ) . this section discusses the conductance and current noise of the single channel @xmath0 qpc at zero temperature ( @xmath57 ) . for @xmath105 , no quasiparticle propagates in @xmath2 , so that the unitarity of @xmath106 leads to @xmath107 . one can thus calculate @xmath58 and @xmath2 from eqs . ( [ gfns ] ) and ( [ sfns ] ) by using @xmath108 } \label{thesingle}%\end{aligned}\ ] ] for @xmath93 , with @xmath109 $ ] . we note that in the absence of a sdips , i.e. @xmath110 , eq . ( [ thesingle ] ) is in agreement with eq . ( 7 ) of ref . . for @xmath111 , one can calculate @xmath58 from eq . ( [ gfns ] ) by using @xmath112 } \label{wmonocanal}%\end{aligned}\ ] ] for @xmath113 . the expression of @xmath2 is too complicated to be given here . the above eqs . describe a phenomenon analogous to the one predicted for a @xmath7 quantum point contact@xcite . the denominators of eqs . ( [ thesingle ] ) and ( [ wmonocanal ] ) contain interference terms which describe iterative reflection processes between the @xmath80 and @xmath81 interfaces ( andreev bound states ) . a quasiparticle can interfere with itself after two back and forth travels between @xmath2 and @xmath1 , one as an electron @xmath12 and one as a hole @xmath14 . the conductance and noise depend on the phase difference @xmath114 because the @xmath81 scattering matrix for holes is @xmath115 . this picture is in fact valid at any temperatures and voltages . in the general case , the whole @xmath102 matrix is necessary to calculate the current and noise , but all the elements of @xmath102 have the same denominator as the one appearing at the right hand side of eq . ( [ thesingle ] ) and ( [ wmonocanal ] ) for @xmath116 and @xmath113 respectively [ see appendix b ] . in addition , we have checked analytically that , in the general case , @xmath58 and @xmath2 depend on the phases of the @xmath81 scattering matrix through the parameter @xmath4 only ( they are actually periodic functions of @xmath4 with a @xmath117-periodicity ) . calculated for @xmath118 and different values of @xmath119 , as a function of the sdips parameter @xmath120 , at zero temperature ( @xmath57 ) . right panels : reduced zero - bias conductance @xmath121 of the qpc versus @xmath122 , for @xmath123 ( upper right panel ) , @xmath124 ( bottom right panel ) and different values of @xmath4 , i.e. @xmath125 ( solid lines ) and @xmath126 ( dashed lines ) . in the left panel , @xmath127 is assumed , so that the number of solutions for @xmath122 is either 0 or 1 depending on the value of @xmath128 . in all the cases considered , the inferred @xmath122 strongly increases with @xmath120 , because @xmath121 decreases with both @xmath129 and @xmath130 for @xmath131 ( see right panels ) . for high enough values of @xmath132 , there exists a @xmath122 solution for @xmath125 ( see e.g. open pink square for @xmath123 and open green square for @xmath133 ) . for lower values of @xmath132 , there is no @xmath129 solution if @xmath125 , because the zero - bias conductance of the device can not reach @xmath121 ( see blue dotted line for @xmath124 and red dot - dashed line for @xmath134 ) . however , it is possible to find a @xmath129 solution at @xmath135 finite because @xmath121 increases with @xmath135 for @xmath131 ( see e.g. blue triangle in the bottom right panel , for @xmath124 and @xmath136 ) . in some cases , a finite @xmath4 can explain the considered values of @xmath121 and @xmath132 even in the absence of polarization ( see blue and red solid circles in the left panel ) . ] we now focus on the zero - voltage conductance @xmath137 of the qpc , which writes , from eqs . ( [ gfns ] ) and ( [ thesingle]),@xmath138}\leq4 \label{g0}%\ ] ] in an experimental context , one may hope to determine the value of the polarization @xmath139 from the zero voltage conductance and the high voltage conductance @xmath119 , with @xmath140 . we define @xmath141 with @xmath142 for @xmath143 , the polarizations @xmath144 with @xmath145 will be solutions of the problem provided @xmath146 is real and @xmath147 . thus , depending on the values of @xmath132 , @xmath121 and @xmath4 , there can be either 0 , 1 or 2 solutions for @xmath129 . importantly , the inferred values of @xmath129 depend not only on the zero voltage and normal conductances , but also strongly on @xmath4 . for simplicity , we will consider in the following the situation @xmath127 , for which one has @xmath148 and @xmath149 , so that there is either no solution for @xmath122 or one solution @xmath150 if @xmath151 . this situation is illustrated in fig . [ pola ] , whose left panel shows the calculated @xmath129 as a function of @xmath135 for @xmath152 and different values of @xmath132 , with @xmath127 ( the use of @xmath120 is due to the fact that @xmath129 is an even function of @xmath4 ) . for the largest values of @xmath132 used in fig . [ pola ] , left panel ( see pink solid curve and green double dot - dashed curve ) , a finite @xmath129 is found for @xmath125 ( see pink and green open squares ) . however , the calculated @xmath129 can also be larger if one uses a finite @xmath120 . this can be understood by noting that , for the relatively low values of @xmath132 used in this figure and @xmath153 , @xmath121 decreases monotonically with both @xmath129 and @xmath154 . an increase in @xmath120 thus compensates an increase in @xmath122 ( see upper right panel of fig . [ pola ] ) . for the lowest values of @xmath132 used in fig . [ pola ] , left panel ( see blue dotted curve and red dot - dashed curve ) , there is no @xmath129 solution if @xmath155 , because the zero - bias conductance of the device can not reach the considered value @xmath152 . however , as long as @xmath121 is not too large , it is possible to find a @xmath129 solution at @xmath128 finite because the zero bias conductance of the qpc increases with @xmath120 for @xmath156 ( see e.g. blue triangle in the bottom right panel of fig . [ pola ] ) . in this case , the calculated value of @xmath129 also increases with @xmath120 , for the same reason as previously . note that due to the continuity of the equations , there exists a limiting case where a finite @xmath4 can explain the considered values of @xmath121 and @xmath132 in absence of polarization ( these points are indicated with solid circles in fig . [ pola])@xcite . for @xmath157 , @xmath121 is not always a monotonic function of @xmath129 , so that there can be 2 solutions @xmath158 and @xmath159 for @xmath129 in some situations ( not shown ) . we conclude that even in the single - channel case , knowing the zero and high voltage conductances of the qpc is not sufficient to determine @xmath160@xcite . we will thus consider below the full voltage dependence of @xmath161 , which brings more information on the system . we will also show that noise measurements prove to be very useful to characterize unambiguously the properties of the qpc . the top panels of figs . [ panellowt ] and [ panelhight ] show the voltage dependence of @xmath58 for low and high values of @xmath132 , respectively . in this paragraph , we comment on the results for @xmath57 only ( black full lines ) . for @xmath125 , the conductance shows peaks at @xmath162 , because for @xmath163 , one has @xmath164 @xmath117 , so that the multiple reflection paths between @xmath2 and @xmath1 interfere constructively . between these peaks , the conductance reaches a minimum of @xmath165^{2}$ ] at @xmath166 . the existence of a finite sdips can strongly modify this behavior . indeed , for @xmath6 and a small enough value of @xmath132 , the resonance peaks of @xmath161 are shifted to lower voltages @xmath167 with @xmath168 ( fig . [ panellowt ] , top middle panel ) . for @xmath169 , these peaks are merged into a single peak and @xmath161 is maximum at @xmath166 ( fig . [ panellowt ] , top right panel ) . for high values of @xmath132 , no clear subgap conductance peaks occur , but the curvature of the sub - gap @xmath161 characteristic can be inverted by a finite sdips ( fig . [ panelhight ] , top panels ) . the @xmath161 curve is independent from @xmath4 only if interferences are suppressed for one spin direction , i.e. @xmath170 or @xmath171 . above the gap , in any case , the conductance of the device drops to its normal - state value @xmath172 which does not depend on the sdips . ( top panels ) and differential fano factor @xmath173 ( bottom panels ) as a function of the bias voltage @xmath9 for @xmath174 , different temperatures ( black solid lines @xmath57 , blue dashed lines @xmath175 ) , and different values of @xmath176 ( left panels : @xmath125 , middle panels @xmath126 , right panels @xmath169 ) . ] the current noise through the @xmath0 qpc can provide more information on its properties , which is of particular interest since this device is characterized by a larger number of parameters than in the spin - degenerate case . according to eqs . ( [ gfns ] ) and ( [ sfns ] ) , the conductance and the noise are not mathematically equivalent and the noise can thus provide information complementary to the conductance . the bottom panels of figure [ panellowt ] and [ panelhight ] show the voltage dependence of the differential fano factor @xmath177 for low and high values of @xmath132 , respectively . this quantity can be measured directly ( see e.g. ref . ) or obtained from a @xmath178 measurement . in this paragraph , we comment on the results for @xmath57 only . one can first note that @xmath179 is an odd function of @xmath9 due to @xmath104 . for a low @xmath132 and @xmath125 , @xmath180 shows subgap plateaus at values @xmath181^{2}$ ] for @xmath182 , with @xmath183 . for @xmath6 , a dip / peak appears on these plateaus , at the resonance voltages @xmath184 , again due to constructive quasiparticle interferences ( fig . [ panellowt ] , bottom middle panel ) . for higher values of @xmath132 , the dips and peaks are smoothed , but the shape of @xmath179 remains sensitive to @xmath185 as long as @xmath186 ( fig . [ panelhight ] ) . above the gap , in any case , @xmath180 drops to the normal state value @xmath187)/(\sum\nolimits_{\sigma}t_{\sigma } ) $ ] which does not depend on the sdips . one can thus determine the polarization @xmath160 of the tunnel rates by using@xmath188 then , the sdips parameter @xmath4 and the bcs gap @xmath91 can be determined from the voltage dependence of the conductance and noise curves . we now comment on the finite temperature @xmath161 and @xmath179 curves , obtained by differentiating eqs . ( [ igal ] ) and ( [ sgal ] ) with respect to @xmath9 ( see blue dashed lines in figs . [ panellowt ] and [ panelhight ] ) . at first glance , these curves simply seem to be thermally rounded . however , the finite temperature expressions of @xmath161 and @xmath179 involve elements of the @xmath102 matrix which do not appear in the zero - temperature expressions , thus the finite temperature curves sometimes display features not predictable from the zero temperature curves . for instance , in the bottom left panel of fig . [ panelhight ] , a slight dip[peak ] appears in the @xmath179 curve at @xmath189\delta / e$ ] for @xmath88 , an effect not present for @xmath57 . ( top panels ) and differential fano factor @xmath180 ( bottom panels ) as a function of the bias voltage @xmath9 for @xmath190 , different temperatures ( black solid lines @xmath57 , blue dashed lines @xmath175 ) , and different values of @xmath4 ( left panels : @xmath125 , middle panels @xmath126 , right panels @xmath169 ) . ] in the case of a weakly transparent contact , since the andreev resonance peaks of the conductance are shifted to lower voltages for @xmath6 , one can obtain , for @xmath88 , conductance curves similar to those obtained for @xmath125 and a reduced value of @xmath91 . the determination of the quantum point contact properties from the @xmath161 curves alone can thus be difficult at finite temperatures . figure [ comparaisons ] shows two examples where measuring the voltage dependence of the noise can clearly bring more information on the system . the left panel presents two cases where the @xmath161 curves are extremely close , one case with @xmath6 and one case with @xmath125 and a reduced gap value . the corresponding @xmath179 curves have a strong quantitative difference , which can help to discriminate the two cases . the right panels of fig . [ comparaisons ] present two cases where the @xmath161 curves are qualitatively similar , one case with @xmath6 and one case with @xmath125 and a reduced gap value . the corresponding @xmath179 curves have a strong _ qualitative _ difference , which can help to discriminate the two cases again . in the case @xmath6 , the differential fano factor shows a secondary peak[dip ] at @xmath191 [ @xmath192 . this effect , which can also be seen in the bottom middle panel of fig . [ panellowt ] , never occurs for @xmath125 . ( top panels ) and differential fano factor @xmath180 ( bottom panels ) as a function of the bias voltage @xmath9 , for different sets of parameters . the left panels compare a case with a finite sdips ( pink solid lines ) and a case with no sdips and a smaller gap value @xmath91 ( blue dash - dotted lines ) . the conductance curves obtained in the two cases ( top left panels ) are very close , but the differential fano factor curves ( bottom left panels ) have a strong quantitative difference . the right panels also compare a case with a finite sdips ( red solid lines ) and a case with no sdips and a smaller @xmath91 ( black dashed lines ) . in these two cases , the conductance curves ( top right panel ) are qualitatively similar but the fano factor curves ( bottom right panel ) show a strong qualitative difference : in the case of a finite sdips , @xmath179 shows a secondary peak[dip ] at @xmath193 [ @xmath194 . this feature never appears in the absence of a sdips . note that the high voltage limits of the two @xmath179 curves shown in the bottom left ( bottom right ) panel are different , but that this difference is not visible on the scale shown in the figure . ] before concluding this section , we point out that in most cases , experimental results on @xmath0 qpcs were interpreted with various models inspired from the btk approach@xcite . the effect of the sdips was not studied in these works . however the description of a @xmath0 interface with a delta - function barrier naturally takes into account the sdips . we have checked that the spin - dependent btk - like model corresponds to a particular case of the scattering model of sec . ii ( generalized to the multichannel case ) , with parameters given in appendix c. in this paper , we have studied the conductance @xmath58 and noise @xmath2 of a superconducting / ferromagnetic ( @xmath0 ) single channel quantum point contact ( qpc ) as a function of the qpc bias voltage @xmath9 , using a scattering approach . we have shown that the spin - dependence of interfacial phase shifts ( sdips ) acquired by electrons at the @xmath0 interface strongly modifies these signals . in particular , for a weakly transparent contact , the sdips produces unusual sub - gap resonances or dips in the @xmath161 and @xmath178 curves . we have shown that measuring the noise should help to gain information on the system . this fact is well illustrated e.g. by cases where the sdips modifies qualitatively the zero - frequency noise but not the conductance . one should note that , so far , experiments on @xmath0 quantum point contacts were performed in the multichannel limit . although single - channel contacts were already realized in the @xmath80 case@xcite , this seems more difficult in the @xmath0 case with the present fabrication techniques . a multichannel theory should thus be very useful . for a multichannel disordered @xmath80 qpc , ref . has shown that , due to time reversal symmetry , the eigenstates the qpc normal state scattering matrix are also those of @xmath102 . as a consequence , the multichannel generalization of @xmath161 and @xmath178 just requires a summation on the normal channels index . in the @xmath0 case , the calculation should be more complicated because time reversal symmetry breaking suppresses the normal channels independence @xcite . so far , the multichannel case has been addressed with the btk model , in which the hypothesis of a specular interface allows one to consider the normal channels as independent in spite of the time reversal symmetry breaking@xcite . however , this approach lacks of generality since it imposes a particular relation between the transmission probabilities and the sdips parameters , and it is not valid for disordered interfaces . interestingly , refs . reports on the ab - initio modelization of a particular type of disordered multichannel @xmath0 interface . however , to our knowledge , the multichannel @xmath0 case has not been studied with a more general approach so far . our work suggests that such a study should take into account the sdips . considering that , even in the single channel case , the conductance and noise of the device vary nonlinearly with the sdips parameter ( see e.g. eq . [ thesingle ] ) , there is no reason to expect that the effects of the sdips average out in the multichannel case , even in the extreme case of a sdips parameter @xmath195 randomly distributed with the channel index @xmath196 . we would like to acknowledge useful discussions with c. bruder , b. douot and t. kontos . this work was financially supported by the swiss national science foundation , the dfg through sfb 513 and sfb 767 , the priority programm _ semiconductor spintronics _ , and the landesstiftung baden - wrttemberg through the kompetenznetzwerk _ this appendix shows that the property @xmath197 , with @xmath198 and @xmath61 , used to derive eqs . ( [ gfns ] ) and ( [ sfns ] ) is a general property which stems from the symmetries of the bdg equations . we first consider the eigenstates of the bdg equations in a bulk s. an eigenstate with energy @xmath18 in the subspace @xmath199 has electron and hole components @xmath200 and @xmath201 such that @xmath202{cc}% h_{\sigma } & \delta\\ \delta^{\ast } & -h_{-\sigma}^{\ast}% \end{array } \right ] \left [ \begin{array } [ c]{c}% u_{\sigma}\\ v_{-\sigma}% \end{array } \right ] = \varepsilon\left [ \begin{array } [ c]{c}% u_{\sigma}\\ v_{-\sigma}% \end{array } \right]\ ] ] with @xmath203 the normal state hamiltonian for electrons with spin @xmath13 . this equation can be rewritten as @xmath202{cc}% h_{-\sigma } & \delta\\ \delta^{\ast } & -h_{\sigma}^{\ast}% \end{array } \right ] \left [ \begin{array } [ c]{c}% -v_{-\sigma}^{\ast}\\ u_{\sigma}^{\ast}% \end{array } \right ] = -\varepsilon\left [ \begin{array } [ c]{c}% -v_{-\sigma}^{\ast}\\ u_{\sigma}^{\ast}% \end{array } \right ] ~\text{.}%\ ] ] this shows that @xmath204 $ ] is another solution of the bdg equations , with energy @xmath17 in the @xmath205 subspace . this property is also valid in a bulk @xmath1 ( in this case @xmath206 and @xmath207 ) . we now consider the scattering processes at a @xmath0 interface . from the definition of @xmath102 , and outgoing wave @xmath208 with energy @xmath18 in lead @xmath209 writes@xmath210{cc}% \mathcal{s}_{m_{1},m_{2}}^{(e,\sigma),(e,\sigma)}(\varepsilon ) & \mathcal{s}_{m_{1},m_{2}}^{(e,\sigma),(h,-\sigma)}(\varepsilon)\\ \mathcal{s}_{m_{1},m_{2}}^{(h,-\sigma),(e,\sigma)}(\varepsilon ) & \mathcal{s}_{m_{1},m_{2}}^{(h,-\sigma),(h,-\sigma)}(\varepsilon ) \end{array } \right ] \psi_{m_{2},\sigma}^{i}%\ ] ] with @xmath211 $ ] the incoming state with energy @xmath18 in the @xmath29 space of lead @xmath212 . this induces@xmath213{cc}% \left ( \mathcal{s}_{m_{1},m_{2}}^{(h,-\sigma),(h,-\sigma)}(\varepsilon ) \right ) ^{\ast } & -\left ( \mathcal{s}_{m_{1},m_{2}}^{(h,-\sigma ) , ( e,\sigma)}(\varepsilon)\right ) ^{\ast}\\ -\left ( \mathcal{s}_{m_{1},m_{2}}^{(e,\sigma),(h,-\sigma)}(\varepsilon ) \right ) ^{\ast } & \left ( \mathcal{s}_{m_{1},m_{2}}^{(e,\sigma),(e,\sigma ) } ( \varepsilon)\right ) ^{\ast}% \end{array } \right ] \phi_{m_{2},\sigma}^{o } \label{r}%\end{aligned}\ ] ] with @xmath214}=^{t}[-(b_{m_{k},\sigma}^{o[i]})^{\ast } , ~(a_{m_{k},\sigma}^{o[i]})^{\ast}]$ ] for @xmath215 . from the previous paragraph , @xmath214}$ ] corresponds to an outgoing [ incoming ] state with energy @xmath17 in the @xmath205 subspace of lead @xmath216 . equation ( [ r ] ) thus gives @xmath217 and @xmath218 , which generalizes eq . ( c7 ) of ref . to the spin - dependent case . the relation @xmath219 follows straightforwardly . when a @xmath0 interface can be modeled as a ballistic @xmath80 interface in series with a dirty @xmath81 interface , with the thickness of @xmath82 tending to zero , the scattering matrix @xmath22 of the @xmath0 interface can expressed in terms of the scattering matrix @xmath221 of electrons with spins @xmath13 on the @xmath81 interface and of the andreev reflection amplitude @xmath36 defined in section ii.b . one finds eqs . ( [ s2]-[s4 ] ) , and @xmath222\nonumber\]]@xmath223@xmath224@xmath225@xmath226@xmath227 with @xmath228^{1/2}$ ] the andreev transmission amplitude . the eight missing elements of @xmath106 can be obtained from the above eqs . by replacing @xmath14 by @xmath12 and vice versa in the upper indices of the @xmath102 elements and by doing the permutations @xmath229 and @xmath230 at the right hand sides of the equations . in some works about qpcs @xcite , the data were interpreted in terms of a generalization of the btk model to the @xmath0 case . in this approach , no fictitious @xmath82 layer is used . the @xmath2 and @xmath1 leads are described with bdg equations@xcite , and the @xmath0 interfacial scattering is attributed to a delta - function barrier @xmath231 . a multichannel description is generally used , where channel @xmath196 corresponds to the @xmath232 transverse mode of the device , for which quasiparticles have a spin - dependent wavevector @xmath233 in @xmath1 and a spin - independent wavevector @xmath234 in @xmath2 with @xmath235^{1/2}$ ] , @xmath236^{1/2}$ ] , @xmath237 the fermi level in @xmath238 , @xmath239 the exchange field in @xmath1 and @xmath240 the energy of the @xmath232 transverse mode@xcite . due to the spectular nature of the @xmath0 interface , the scattering matrix @xmath22 associated to the btk - like model does not connect the different transverse modes of the device . consequently , the conductance and noise of the qpc can be obtained by summing the expressions introduced in this article on the channel index @xmath196 . the scattering parameters associated to channel @xmath196 are the transmission probability @xmath241\ ] ] and the sdips parameter @xmath242\ ] ] with @xmath243 , @xmath244 the effective mass of the particles , and @xmath245 interestingly , from the above eqs . , one can notice that , in principle , @xmath195 can be finite even if @xmath246 is not spin - dependent , provided @xmath247 is spin - dependent and @xmath246 is not too large . m. n. baibich , j. m. broto , a. fert , f. nguyen van dau , and f. petroff , p. etienne , g. creuzet , a. friederich , and j. chazelas , phys . lett . * 61 * , 2472 ( 1988 ) ; g. binasch , p. grnberg , f. saurenbach , and w. zinn , phys . rev . b * 39 * , 4828 ( 1989 ) . a. brataas , y. v. nazarov , and g. e. w. bauer , phys . 84 * , 2481 ( 2000 ) ; d. huertas hernando , yu . v. nazarov , a. brataas , and g. e. w. bauer , phys . b * 62 * , 5700 ( 2000 ) ; a. brataas , y. v. nazarov and g. e. w. bauer , eur . j. b * 22 * , 99 ( 2001 ) . a. millis , d. rainer , and j. a. sauls , phys . b * 38 * , 4504 ( 1988 ) ; m. fogelstrm , _ ibid . _ * 62 * , 11812 ( 2000 ) ; j. c. cuevas and m. fogelstrm , ibid . * 64 * , 104502 ( 2001 ) ; n. m. chtchelkatchev , w. belzig , yu . v. nazarov , and c. bruder , jetp lett . * 74 * , 323 ( 2001 ) ; m. eschrig , j. kopu , j. c. cuevas , gerd schn , phys . lett . * 90 * , 137003 ( 2003 ) ; j. kopu , m. eschrig , j. c. cuevas , and m. fogelstrm , phys . b * 69 * , 094501 ( 2004 ) ; m. s. kalenkov and a. d. zaikin , phys . rev . b * 76 * , 224506 ( 2007 ) ; physica e * 40 * , 147 ( 2007 ) ; j. linder and a. sudb , phys . rev . b * 75 * , 134509 ( 2007 ) ; m. eschrig , t. lofwander , t. champel , j. c. cuevas , j. kopu , and gerd schn , j. low temp . phys . * 147 * , 457 , ( 2007 ) ; e. zhao and j. a. sauls , phys . * 98 * , 206601 ( 2007 ) ; m. eschrig , t. lofwander , nat . * 4 * , 138 ( 2008 ) . d. huertas - hernando , yu . v. nazarov , and w. belzig , phys . 88 * , 047003 ( 2002 ) ; d. huertas - hernando , yu . v. nazarov , and w. belzig , cond - mat/0204116 ; d. huertas - hernando and yu . v. nazarov , eur . j. b * 44 * , 373 ( 2005 ) . r. j. soulen , jr . , j. m. byers , m. s. osofsky , b. nadgorny , t. ambrose , s. f. cheng , p. r. broussard , c. t. tanaka , j. nowak , j. s. moodera , a. barry , and j. m. d. coey , science * 282 * , 85 ( 1998 ) ; s.k . upadhyay , a. palanisami , r.n . luoie , and r.a . burhman , phys . lett . * 81 * , 3247 ( 1998 ) ; b. nadgorny , r. j. soulen , jr . , m. s. osofsky , i. i. mazin , g. laprade , r. j. m. van de veerdonk , a. a. smits , s. f. cheng , e. f. skelton , and s. b. qadri , phys . b * 61 * , 3788(r ) ( 2000 ) ; y. ji , g. j. strijkers , f. y. yang , and c. l. chien , phys . b * 64 * , 224425 ( 2001 ) ; y. ji , g. j. strijkers , f. y. yang , c. l. chien , j. m. byers , a. anguelouch , gang xiao , and a. gupta , phys . * 86 * , 5585 ( 2001 ) ; g. j. strijkers , y. ji , y. yang , c.l . chien and j. m. biers , phys .. rev . b * 63 * , 104510 ( 2001 ) ; j. s. parker , s. m. watts , p. g. ivanov , and p. xiong , phys lett . * 88 * , 196601 ( 2002 ) ; p. raychaudhuri , a. p. mackenzie , j. w. reiner , and m. r. beasley , phys . rev . b * 67 * , 020411 ( 2003 ) ; r. p. panguluri , g. tsoi , b. nadgorny , s. h. chun , n. samarth , and i. i. mazin , phys . b * 68 * , 201307 ( 2003 ) ; z. jiang , j. aumentado , w. belzig , v. chandrasekhar , _ transport through ferromagnet / superconductor interfaces _ , proceedings of the st . petersburg nato conference on quantum transport , st . petersburg , august , 2003 ( arxiv : cond - mat/0311334 ) . we are only aware of a brief zero - voltage btk - like calculation of the noise through a @xmath0 qpc@xcite and a two - dimensional tight - bindind calculation of the current cross - correlations between two @xmath1 leads connected through a @xmath2 electrode@xcite . in the model introduced in section ii.b , the voltage drop due to @xmath9 can be considered as located on the @xmath248 interface , so that the scattering matrix of the fictitious @xmath249 interface does not depend on @xmath9 . then , if @xmath101 , @xmath3 and @xmath99 are assumed to be independent from @xmath18 and @xmath9 , @xmath250 , does not depend on @xmath9 . we will not discuss on how to relate the polarization @xmath252 of the density of states of @xmath1 to the polarisation @xmath160 of the transmissions discussed by us . it has been shown that this is not straightforward @xcite . i. zutic and o. t. valls , phys . b * 60 * , 6320 ( 1999 ) ; _ ibid . _ * 61 * , 1555 ( 2000 ) ; j .- x . zhu , b. friedman and c. s. ting , phys . b * 59 * , 9558 ( 1999 ) ; s. kashiwaya , y. tanaka , n. yoshida and m. r. beasley , phys . rev . b * 60 * , 3572 ( 1999 ) ; r. mlin , europhys . lett . * 51 * , 202 ( 2000 ) ; j. cayssol and g. montambaux , phys . b * 71 * , 012507 ( 2005 ) .	we study the conductance and current noise of a superconductor / ferromagnet ( @xmath0 ) single channel quantum point contact ( qpc ) as a function of the qpc bias voltage , using a scattering approach . we show that the spin - dependence of interfacial phase shifts ( sdips ) acquired by electrons upon scattering by the qpc can strongly modify these signals . for a weakly transparent contact , the sdips induces sub - gap resonances in the conductance and differential fano factor curves of the qpc . for high transparencies , these resonances are smoothed , but the shape of the signals remain extremely sensitive to the sdips . we show that noise measurements could help to gain more information on the device , e.g. in cases where the sdips modifies qualitatively the differential fano factor of the qpc but not the conductance .
coronal mass ejections ( cmes ) are large - scale solar eruption activity , releasing a vast amount of plasma and magnetic field into the interplanetary space . they may severely affect the space environment around the earth ( e.g. , * ? ? ? * ; * ? ? ? coronagraph observations show that cmes generally have a three - part structure : a bright leading front ( lf ) , a dark cavity , and an inner bright core ( e.g. , * ? ? ? the cavity is usually believed to be a helical flux rope ( e.g. , * ? ? ? * ; * ? ? ? the bright core is thought to be cool filament / prominence matter that is suspended in magnetic dips of a flux rope configuration ( e.g. , * ? ? ? * ; * ? ? ? * ) . the flux rope structure has been revealed by faraday rotation observations that can measure cme magnetic field ( e.g. , * ? ? ? * ) , and recently validated by well - separated multi - spacecraft in situ measurements @xcite . the flux rope structure is also inferred from frequent appearance of sigmoids in euv and x - ray observations of the solar corona @xcite . these sigmiod structures are composed of two bundles of opposite j - shaped loops , and then form a twisted configuration by magnetic reconnection ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? many authors successfully simulated the formation of the flux rope . @xcite imposed a slow converging motion toward the polarity inversion line ( pil ) to the footpoints of field lines to produce the flux rope . using flux cancellation in a bald - patch separatrix , @xcite transformed the preexisting sheared arcades into a flux rope . @xcite and @xcite did a theoretical analysis for the flux rope and found that the kink and/or torus instability of the flux rope can initiate a cme . @xcite also found that the eruption of the flux rope can be fully driven by a partial torus instability ( see also , * ? ? ? * ) . the flux rope can also be formed in a nonlinear force - free field model based on the observed photospheric vector magnetic fields ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? another key issue in understanding cmes is the relationship between cmes and their associated flares . @xcite studied the kinematics of flare - associated cmes and proposed a scenario of three phases of cme evolution : a slow initiation phase , an impulsive acceleration phase , and then a propagation phase . further , they found that the three kinematic phases are closely related to the three phases of the associated flare : the pre - flare phase , flare rise phase , and decay phase , respectively ( see also , * ? ? ? * ; * ? ? ? @xcite and @xcite studied the temporal correlation between cme acceleration and the flare hard x - ray ( hxr ) flux and found that they coincide in time very well . these results suggest that the cmes are coupled together with the associated flares by the same physical mechanism , especially during the main energy release phase , presumably via magnetic reconnection @xcite . recently , @xcite provided possible observational evidence for magnetic reconnection that may form the flux rope fully in the impulsive acceleration phase of a cme . they found that a plasma blob appeared in the atmospheric imaging assembly ( aia ; * ? ? ? * ) passbands of high temperatures ( 94 and 131 ) , rose from the solar limb , then rapidly moved outward , and finally developed into a flux rope - like structure ( see also , * ? ? ? * ; * ? ? ? while in the cooler aia passbands ( 171 and 211 ) , the background magnetic field and plasma were compressed to form the cme lf . this resulted in that the cme in the inner corona is composed of the multi - temperature structures : the hot flux rope - like channel and the cool lf @xcite . because the associated source region was partially blocked by the solar limb , @xcite was not able to determine the cme structure prior to the eruption . in order to resolve this issue , @xcite further studied one cme originating on the solar disc . they found that a twisted channel first appeared in the low corona before the cme eruption and was visible in the hot 131 and 94 passbands . the twisted hot structure initially lay close to the surface and along the pil . once the impulsive acceleration phase began , it erupted explosively outward and developed into the flux rope - like structure . @xcite thus concluded that the flux rope , as observed as the hot channel , exists prior to the eruption . in this paper , we investigate the kinematics of two cmes in detail , especially looking into the relations between the observed hot channel and the lf . we find strong observational evidence that the cme hot channel acts as the engine that drives cme formation and acceleration , at least in the early acceleration phase . in section 2 , we describe the data . in section 3 , we present the observations and results , which are followed by a summary and discussion in section 4 . aia on board _ sdo _ has four telescopes that observe the solar atmosphere almost simultaneously from the photosphere up to the corona using ten narrow uv and euv passbands . in this paper , we mainly use five passbands including 131 ( fe viii , fe xxi ) , 94 ( fe xviii ) , 211 ( fe xiv ) , 171 ( fe ix ) , and 304 ( he ii ) ; they provide a range of effective temperatures from 0.05 to 20 mk ( see * ? ? ? * for response functions of aia channels ) . aia s capability of multi - temperature coverage allows to image different temperature structures of cmes . during the impulsive phase of a cme , the 131 and 94 passbands observe the hot plasma structures with a temperature range of @xmath0711 mk , the 211 and 171 passbands primarily observe the cme cavity and lf with a temperature range of @xmath00.63 mk @xcite , and the 304 passband reveals the associated filament material . note that , the aia 131 passband also includes the emission from the cool plasma in the transition region in addition to the high temperature plasma from the corona . aia s temporal cadence , pixel size and field of view ( fov ) are 12 seconds , 0.6@xmath1 , and 1.3@xmath2 , respectively . the multi - temperature capability and super - high cadence provide a unique opportunity to study cme formation in the low corona in detail . coronagraphs on board @xmath3 ( cor1 and cor2 ; * ? ? ? * ) and @xmath4 ( lasco ; * ? ? ? * ) image cmes in white light . in addition , _ goes _ x - ray data reveal the temporal profile of the soft x - ray flux for associated flares . the @xmath5 @xmath6 @xmath7 @xmath8 @xmath9 @xmath10 @xmath11 ( @xmath12 ; * ? ? ? * ) provides the hard x - ray spectrum and imaging of the flares . on 2011 march 7 and 8 , two limb solar eruptions took place and were observed simultaneously by _ sdo _ , _ stereo _ , and _ soho_. each of these two events was associated with an m - class flare from noaa active regions ( ars ) 11164 and 11171 , respectively . aia 94 and 131 ( @xmath0711 mk ) images show that a twisted hot channel appeared along the pil before the onset of the flare . @xcite presented the evolution of the flux rope - like hot channel for 2011 march 8 event . here we quantitatively analyze the full evolution of the cme hot channel and lf for both events with much more details . the aia 94 and 131 images of the hot channels of the 2011 march 7 and 8 cmes are shown in figures [ f1 ] and [ f2 ] . for both events , a hot channel appeared prior to the onset of corresponding flare by several minutes . as for the 2011 march 7 event , an interesting feature is that the cool filament material appeared along the magnetic structure of the hot channel , which is identified in the 304 passband ( figure [ f1](g i ) ) . from the 304 images , one also notices a brightening that started to appear after @xmath019:20 ut . the brightening occurred inside the filament , which indicates that magnetic reconnection initially took place there . the reconnection gradually heated the plasma in the filament and made the intensity in 304 passband increase . with the heating ongoing , the brightening was further enhanced and extended upward . at the same time , the whole filament structure started to rise slowly . by inspecting visually the movies of the 131 and 304 passbands ( available in the online version ) , we infer possible existence of an x - type reconnection point inside the filament structure , and the location of the brightening is almost coincident with that of the x - type point , which is shown in the small box of figure [ f1](c ) and ( f ) . in order to compare the emission produced at the x - type point with the @xmath13 soft x - ray 18 flux , we plot the temporal profiles of the euv intensity for all aia euv passbands ( figure [ f3](a ) ) , integrated over the brightening region shown by the small box in figure [ f1 ] . the temporal profiles of the intensity in the whole active region are also plotted . one can see that the intensity of the brightening in all aia euv passbands rapidly increase about 10 minutes prior to the beginning of the associated flare ( @xmath019:43 ut ) . however , for the whole ar no apparent increase of the emission intensity appears , the evolution is basically consistent with that of the @xmath13 soft x - ray 18 flux . similarly , for the 2011 march 8 cme , the hot channel appearing in the 131 and 94 passbands is shown in figure [ f2](a f ) . it became first visible at @xmath003:30 ut ( figure [ f2](a ) and ( d ) ) . with time elapsing , it took on a twisted shape with a double elbow above at @xmath003:34 ut . then , the twisted hot channel rose slowly with the southern elbow field slipping down ( figure [ f2](b ) and ( e ) ) . just prior to the onset of the associated flare , the hot channel apparently presented a twisted shape and had fully detached from the elbow field lines ( figure [ f2](c ) and ( f ) ) . by checking the euvi - b 195 images , we find that the 2011 march 8 event originated from a forward sigmoid active region ( figure [ f2](g)(i ) ) . the hot channel lay along the pil , which can be indicated by the connection line of the two footpoints of the channel ( black arrows in figure [ f2](i ) ) . it is worth noting that , for this case , there is no filament material associated with the hot channel structure . this may be the reason why the hot channel is only visible in the 131 and 94 passbands but not the 304 passband . we also compare the temporal evolution of the integrated flux intensity of the whole active region with the @xmath13 soft x - ray 18 flux . from figure [ f3](b ) we see that , except the 335 emission , the aia euv intensity has started to increase slightly along with the @xmath13 soft x - ray flux prior to the beginning of the associated flare ( @xmath003:37 ut ) . the decrease of the intensity in 335 passband is probably caused by the heating moving the plasma out of the temperature response range sensitive to this passband . here , we also study the property of the magnetic field associated with the eruptions . it is generally accepted that the decay index of the background magnetic field ( @xmath14 ; b denotes the background magnetic field strength at the geometrical height h above the solar surface ) controls the torus instability of the flux rope ; the torus instability will take place if the decay index @xmath15 is greater than a threshold value @xcite . here , the decay index of the background field is calculated from the potential field model based on the line - of - sight magnetograms observed by the helioseismic and magnetic imager ( hmi ; * ? ? ? * ) onboard _ sdo _ ( figure [ f4](a ) and ( b ) ) . note that , in calculating the decay index , only the transverse component of coronal magnetic field is used , since the line - of - sight component does not contribute to the downward constraining force exerting onto the flux rope . the final decay index is an average value over the main pils ( red lines ) of ar . the decay index distributions with the height for 2011 march 7 and 8 events are displayed in figure [ f4](c ) and ( d ) , respectively . through the aia 131 images , we find that the hot channels initially located at the height of @xmath0 20 mm over the ar ( see figure [ f7](a ) , [ f8](a ) , [ f9](b ) , and [ f10](b ) ) , where the decay indexes are always smaller than 1.5 ( a nominal critical value of the torus instability ; * ? ? ? * ) , thus may not be able to trigger the torus instability . taking the pre - flare brightening into account , we argue that it is most likely that the magnetic reconnection drives the slow rise phase of the two hot channels . this reconnection during the initiation phase should be similar to the suggested first step reconnection in a tether - cutting model which helps the formation and rise of the flux rope @xcite . furthermore , we study the transition of the hot channel from the slow rise phase to the impulsive acceleration phase . at the end of the slow rise phase , the 2011 march 7 ( 8) hot channel obtained the velocity of @xmath0200 km s@xmath16 ( @xmath0100 km s@xmath16 ) and attained the height of @xmath050 mm ( indicated by the arrows in figure [ f7](a ) and [ f8](a ) and the vertical dashed lines in figure [ f9](b ) and [ f10](b ) ) , where the decay index of the background field is greater or only slightly greater than 1.5 ( figure [ f4](c ) and ( d ) ) , thus possibly triggering the torus instability and playing an important role in the transition from the slow rise phase to the impulsive acceleration phase . for the 2011 march 8 hot channel , the transition occurred exactly at the height of @xmath050 mm ; while for the 2011 march 7 one , the transition possibly took place at the lower height , such as at @xmath030 mm , where the decay index has been larger than the critical value 1.5 . moreover , we note that the time of the magnetograms we used is several days before or after that of the cme eruption and the exact location of the hot channel in the ar is unknown , which may have some influence on the calculation of the decay index and the judgment of the torus instability occurrence . when the associated flare starts to brighten in soft x - ray or hard x - ray , a cme often impulsively accelerates outward . for the two cmes on 2011 march 7 and 8 , the cme hot channel started to rapidly rise and expand immediately at the time of the flare onset ( figure [ f5](a)(c ) and figure [ f6](a)(c ) ) . in order to clearly present the kinematic evolution of the cme hot channel , we take slices from the images along a direction of the rising motion ( figure [ f5](b ) and figure [ f6](b ) ) and show the time evolution of the hot channel brightness along these slices ( figure [ f7](a ) and figure [ f8](a ) ) . for the 2011 march 7 event , the hot channel quickly expanded and seemed to split into two parts : the hot channel top and bottom . the bottom part is actually related to the cool filament . the quick rise and expansion of the hot channel also occurred in the 2011 march 8 event without , however , the splitting of the structure ( figure [ f8](a ) ) . from figure [ f5](d)(f ) and figure [ f6](d)(f ) , we find that the overlying field of the hot channel gradually formed an euv cavity resembling a bubble . the temporal evolution of the bubble brightness along the same slices is shown in figure [ f7](b ) and figure [ f8](b ) . one can see that the cme bubble experienced internal expanding and also external compressing of the surrounding field and plasma to produce the cme lf with enhanced plasma density , especially for the 2011 march 8 event . note that , in the two cases we do not find any evident features of the euv shocks , which usually is driven by the early expansion of the cme bubble and appears ahead of the cme lf with a small standoff distance ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? by tracking the hot channel in the 131 images , we measure the height - time variations of the hot channel top and bottom , as shown by the crosses in figure [ f5](a)(c ) and figure [ f6](a)(c ) . the distance between the hot channel top and bottom is considered as its width . in order to obtain the height - time measurements of the cme lf , we fit the cme bubble as a circle and take the height of the bubble top as the height of the lf . the radius of the circle is regarded as the size of the cme bubble . the fitting circles are shown in figure [ f5](d)(f ) and figure [ f6](d)(f ) , in which the bubble center is indicated by a cross . one can find that the circle fits the upper part of the bubble very well in the fov of aia . note that all heights are measured from the surface of the sun . we first study the expansion properties of cme structures through the measurement of aspect ratios . the aspect ratio of the hot channel is the ratio between its top height and width , while that of the bubble is the ratio between the height of its center and its radius . the temporal evolution of these aspect ratios are shown in figure [ f9](a ) and figure [ f10](a ) . for the 2011 march 7 cme , the aspect ratio of the hot channel kept almost a constant value for @xmath04 minutes after the flare onset at @xmath019:43 ut and then decreased from @xmath04.0 to 3.0 in the next @xmath03 minutes . the aspect ratio of the bubble changed more rapidly . it continuously decreased from @xmath02.2 to 1.2 during the 6-minute of measurement . while for the 2011 march 8 cme , the aspect ratio of the bubble kept a constant value of @xmath01.1 with a slight decrease after @xmath03 minutes of the onset , the aspect ratio of the hot channel monotonically deceased from @xmath05.0 to 3.0 during the 6-minute of measurement . these results indicate that the growth rate of the width of the hot channel is generally larger than that of its height during the initial rise phase of the flares . it implies that the hot channel always undergoes an overexpansion ( here an overexpansion is defined such that the expansion speed is larger than the bulk propagation speed ) . for the cme bubble , the evolution is different from case to case . apparently , for the 2011 march 7 cme , the bubble undergoes an overexpansion ; while for the 2011 march 8 event , it mainly has a self - similar expansion . however , note that , the earlier overexpansion of the 2011 march 8 cme bubble than @xmath003:39 ut may be missed because of the indiscernibility of the initial cme bubble . furthermore , we study the kinematics of cme structures . the temporal variations of the height and width of the cme hot channel and bubble are plotted in figure [ f9](b ) and figure [ f10](b ) , from which one can see that the cme hot channel is catching up with the lf in the fov of aia . specifically , the distance between the hot channel top and the cme lf decreases . in order to obtain the temporal variations of their speed and acceleration , the height measurements usually need to be smoothed first in order to reduce the uncertainty in the measurement . here we use a cubic spline smoothing method to reduce uncertainty , and then calculate the speed with a piece - wise numerical derivative method , i.e. , the lagrangian interpolation of three adjacent points ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the deduced speed - time profiles are plotted in figure [ f9](c ) and figure [ f10](c ) . note that the uncertainty in the speed mainly results from the error in the height measurements , which is estimated to be 2 pixels for aia observations . with the same method the cme acceleration can be further derived from the speed , as shown in figure [ f9](d ) and figure [ f10](d ) . for both events , the speed of the cme hot channel is larger than that of the cme lf during the early acceleration phase . it is likely that the hot channel drives the cme bubble upward . for the 2011 march 7 event , the hot channel ascended with an average speed of @xmath0100 km s@xmath16 prior to the flare beginning ( @xmath019:43 ut ) . after the onset of the flare , the speed quickly increased from @xmath0200 km s@xmath16 to @xmath0800 km s@xmath16 within @xmath06 minutes . at all times , the speed of the hot channel was @xmath0100 km s@xmath16 greater than that of the lf . the speed evolutions of the cme hot channel and lf are synchronized in time with the @xmath13 18 soft x - ray flux profile ( figure [ f9](c ) ) . the results also hold for the 2011 march 8 event . the hot channel rose slowly with an average speed of @xmath060 km s@xmath16 in the initiation phase . subsequently , at each instant the speed of the hot channel was @xmath050 km s@xmath16 greater than that of the lf ( figure [ f10](c ) ) . in addition , we find that the onset of the acceleration of the hot channels is always earlier than the onset of the hard x - ray flux of the flares , as well as the first appearance of the lf . the acceleration evolutions of the cme hot channel and lf , as well as the rhessi 1525 kev hard x - ray curves of the associated flares , are displayed in figure [ f9](d ) and figure [ f10](d ) . one can see that the hot channel on 2011 march 7 had begun to accelerate at @xmath019:41 ut while both the onset of the 1525 kev hard x - ray flux of associated flare and the first appearance of the lf at later @xmath019:43 ut . for the 2011 march 8 cme , the acceleration of the hot channel took place at @xmath003:35 ut , whereas the 1525 kev hard x - ray flux of associated flare begins at @xmath003:37 ut and the lf appears first at @xmath003:38 ut . the leading time is @xmath02 minutes for both events . the results imply that the cme hot channel plays an important role in inducing the energy releases of flares and in forming the cme lf , also imply that the transition of the hot channel from the slow rise to the impulsive acceleration probably take place earlier than the flare onset . nevertheless , due to that it is very difficult to determine the exact onset of the impulsive acceleration caused by the torus instability , we thus use the flare onset as an approximation of the transition onset in previous section . the speed evolution of the cme hot channel and the lf is generally consistent in time with the variation of the soft x - ray flux although the hot channel acts as a driver in the cme formation and eruption . after the cmes have left the fov of aia , they are well observed by cor1 , cor2 and lasco . since the two events are close to the solar limb , the height - time measurements from lasco are only slightly affected by the projection effect . as for cor1 and cor2 observations , we use the sccmeasure " routine in the ssw package to estimate the three - dimension heights ( e.g. , * ? ? ? all the height - time data are plotted in figure [ f9](e ) and figure [ f10](e ) including the measurements from aia . the corresponding speed is plotted in figure [ f9](f ) and figure [ f10](f ) . for the 2011 march 7 cme , the speed profile of the cme is coincident in time very well with the variation of the @xmath13 soft x - ray 18 flux . the cme speed increases to @xmath02000 km s@xmath16 at @xmath020:10 ut and then decreases to @xmath01500 km s@xmath16 at @xmath021:10 ut . the 2011 march 8 cme seems to keep a constant speed of @xmath01100 km s@xmath16 after the peak time of the flare . due to the lack of high - cadence measurements during the later rise phase of the associated flare , we can not determine exactly ( i.e. , in one minute of accuracy ) when the cme get its maximum speed . note that for cor1 , cor2 , and lasco observations , the uncertainty of measuring the lf height is estimated as 4 pixels . the graduated cylindrical shell ( gcs ) model is proposed by @xcite and represents cmes as flux rope - like structures . it consists of a tubular section forming the main body and two cones rendering the legs " of the cme . the geometry is shown in figure 1 of @xcite . the model is controlled by six free parameters : carrington longitude ( @xmath17 ) and latitude ( @xmath18 ) of the source region , the tilt angle ( @xmath19 ) of the flux rope , the height ( @xmath20 ) of the cme lf , the half - angle ( @xmath21 ) between the two legs of the flux rope , and the aspect ratio ( @xmath22 ) of the cme flux rope . note that @xmath22 is defined by @xcite as the ratio between the minor radius ( @xmath23 ) of the flux rope and the height ( @xmath20 ) . it is different from the definitions of aspect ratios of the cme hot channel and bubble in this paper . but , the two kinds of definitions can be converted to each other ( see * ? ? ? the gcs model has been implemented in many aspects , e.g. , fitting the cme cavity @xcite , determining flux rope orientation and comparing the results with in situ measurements @xcite , and studying the kinematics and expansion of cmes @xcite . in this section , we test whether it can reproduce the morphology of the 2011 march 7 and 8 cmes and then determine the orientation of the flux ropes in white - light observations , comparing with that in euv observations . using the euvi 195 images , we first estimate @xmath17 and @xmath18 through the location of the ar . then we vary @xmath21 , @xmath22 , @xmath19 , and @xmath20 until the best approximations to cor2 and c2 images are obtained . in fact , @xmath21 is usually kept as a constant value ; @xmath22 and @xmath19 only change slightly when the cme is propagating in the fov of cor2 . the final positioning and model parameters of the two cme flux ropes are listed in table [ tb ] . a wireframe rendering is shown in figures [ f11 ] and [ f12 ] . one can see that the flux rope model reconstructs the cme images very well . for the 2011 march 7 cme , the flux rope propagated toward the northwest in the fovs of stereo - b and lasco , and toward the northeast in the fov of stereo - a . in the meantime , it expanded self - similarly ( @xmath22=0.4 ) . at 21:08 ut , it had appeared as a partial halo cme , which was possibly preceded by a shock ( figure [ f11](d)(f ) ) . the 2011 march 8 cme also expanded self - similarly ( @xmath22=0.3 ) and may also have caused a shock surrounding the cme ( figure [ f12](d ) and ( f ) ) , but it had a propagation direction toward the southwest in the fov of stereo - b and toward southeast in the fovs of lasco and stereo - a ( figure [ f12 ] ) . moreover , by inspecting the orientation of the flux rope , we find that the 2011 march 7 ( 8) cme have a counter - clockwise ( clockwise ) rotation while rising . assuming that the initial orientation of the cme is the orientation of the axis of the hot channel , as shown in figures [ f5](b ) and [ f6](b ) , the rotation angle is @xmath030@xmath24 ( 25@xmath24 ) when out of the fov of cor2 . this result is well compatible with the hemispheric preference of the cme rotation ( counter - clockwise ( clockwise ) rotation in the northern ( southern ) hemisphere ; * ? ? ? in this paper , we investigate in detail the formation and evolution of different structural components of two well - observed cmes . in the initiation phase , the pre - eruption structure appears as a twisted hot channel , having an orientation consistent with that of the pil and showing a shear with the overlying field . the twisted channel is most likely heated by the magnetic reconnection during the initiation phase , making it visible in the high temperature passbands , e.g. , 131 and 94 . the reconnection location may be inside or around the hot channel and the corresponding reconnection rate is relatively small , being different from the impulsive phase reconnection underneath the cme flux rope ( e.g. , * ? ? ? as for the physical mechanism of the slow rise phase of the flux rope - like hot channel , we examine the decay index of the background field with height and find that the decay index is small enough at the initial height of the hot channel ( @xmath020 mm ) , which possibly does not allow the occurrence of the torus instability . we thus speculate that the mechanism of driving the slow rise of the hot channel is most likely to be magnetic reconnection , but not the one producing x - ray flares . such reconnection should be similar to the coronal slip - running reconnection in the simulation of the flux rope @xcite . nevertheless , with the reconnection ongoing , the hot channel rises gradually with a speed of 50100 km s@xmath16 and reach a critical height finally , such as @xmath050 mm for the two cmes studied in this paper , where the background field declines rapidly enough , thus being able to trigger the torus instability , resulting in the onset of the hot channel impulsive acceleration phase and the flare rise phase ; such a scenario has been proposed in earlier numerical simulations by @xcite . in the acceleration phase , the speed of the cme hot channel is larger than that of the cme lf , at least in the fov of aia . in a classic eruptive flare model @xcite , when the explosive magnetic reconnection underneath the cme flux rope is triggered , the anchored overlying magnetic fields are quickly stretched upward by the erupting flux rope , and the magnetic field underneath is pushed together and forms a current sheet ( cs ) . the reconnection in the cs converts more and more poloidal magnetic fluxes into the flux rope . the enhanced poloidal flux in the flux rope would essentially increase the upward lorentz self - force and thus further accelerate the flux rope outward . this is similar to the flux injection model of @xcite , although his model is not reconnection driven but the flux comes from below the photosphere . in response to the fast escaping of the flux rope , a plasma inflow would form , which in turn makes more ambient magnetic fields reconnect in the cs . thus , it is this positive feedback reconnection process that impulsively accelerates the cme flux rope . it is found that the hot channel studied here behaves just like the flux rope in the classic model . after the associated flare begins , the hot channel rapidly rises and expands through the positive feedback reconnection process into a full - fledged large - scale flux rope with the cs underneath and two footpoints still anchored at the ends of the main pil . in the meantime , the flux rope compresses the overlying field and plasma that successively stack into the cme lf . since the hot channel drives the formation of the cme lf , it is reasonable that the speed of the hot channel is larger than that of the lf in the early acceleration phase . combining with the fact that the observed morphological evolution resembles that of the magnetic flux rope in theoretical models and 3d numerical simulations , we thus argue that the hot channel is possibly to be the magnetic flux rope of cmes , which acts as a central engine to drive the cme formation and eruption ( see also , * ? ? ? however , it is worth noting that the flux rope top may catch up and merge with the cme lf quickly ( e.g. , the 2011 march 8 hot channel ; figure [ f8](b ) ) , making the lf sharper and brighter . the main flux rope body takes the form of the dark cavity in white - light coronagraphs . a detailed study of building the connection between the cme cavity in white - light observations to the hot channel and bubble in the euv passbands is under the way . in addition , we have studied the aspect ratio ( height to width ) of the hot channel and find that it tends to decrease during the impulsive acceleration phase . this means that the expansion speed of the hot channel is larger than its bulk propagation speed , which is defined as an overexpansion of the hot channel in this paper . the overexpansion possibly results from the magnetic reconnection in the acceleration phase , which rapidly transfers the ambient fields into the hot channel and causes the increase of the poloidal flux . an ideal mhd process can also result in the overexpansion of the flux rope - like hot channel . during the rise phase of the hot channel , the poloidal flux in the channel decreases because its twist does not change . in response to the decreased polodial flux , the channel will expand laterally @xcite . in addition , the internal pressure ( thermal plus magnetic ) of the hot channel may be much larger than its external pressure , which can also drive the lateral expansion . we believe that the lateral expansion of the hot channel also drives the formation of the cme bubble in the early acceleration phase . for the 2011 march 7 and 8 cmes , it is found that the aspect ratio of the cme bubble gradually or slightly decreases after the beginning of the flare , indicating that the cme bubble also undergoes an overexpanding process like the hot channel ( also see , * ? ? ? * ; * ? ? ? however , our observational results suggest that the overexpansion of the cme bubble may be the result of the overexpansion of the hot channel ; when the small - scale hot channel ( the nascent flux rope ) is developing into the large - scale cme flux rope structure , it may result in the lateral expansion of surrounding magnetic loops , quickly forming the cme bubble and driving the bubble expansion in the low corona . therefore , the overexpansion of the cme bubble is actually a manifestation of the dynamics of the hot channel . we thank the anonymous referee for her / his constructive comments that have significantly improved this manuscript . we are grateful to o. olmedo , q. r. chen , and s. patsourakos for valuable discussions . sdo is a mission of nasa s living with a star program . x.c . , and m.d.d . are supported by nsfc under grants 10673004 , 10828306 , and 10933003 and nkbrsf under grant 2011cb811402 . . is also supported by the scholarship granted by the china scholarship council ( csc ) under file no . is supported by nsf grant atm-0748003 , ags-1156120 and nasa grant nng05gg19 g . cccccccc + event & time & lon ( @xmath17 ) & lat ( @xmath18 ) & tilt ( @xmath19 ) & h ( @xmath20 ) & ratio ( @xmath21 ) & half - angle ( @xmath22 ) + & ( ut ) & ( deg ) & ( deg ) & ( deg ) & ( r@xmath25 ) & ( r / h ) & ( deg ) + 2011 - 03 - 07 cme & 20:08 & 156.5 & 29.6 & 77.1 & 4.4 & 0.4 & 15.6 + & 21:08 & 156.5 & 29.6 & 86.6 & 15.2 & 0.4 & 40.2 + 2011 - 03 - 08 cme & 05:08 & 30.2 & 10.6 & 43.0 & 6.7 & 0.3 & 32.7 + & 06:08 & 30.2 & 10.6 & 45.8 & 11.1 & 0.3 & 35.7 + & 07:08 & 30.2 & 10.6 & 54.2 & 15.3 & 0.3 & 35.7 +	recent _ solar dynamic observatory _ observations reveal that coronal mass ejections ( cmes ) consist of a multi - temperature structure : a hot flux rope and a cool leading front ( lf ) . the flux rope first appears as a twisted hot channel in the atmospheric imaging assembly 94 and 131 passbands . the twisted hot channel initially lies along the polarity inversion line and then rises and develops into the semi - circular flux rope - like structure during the impulsive acceleration phase of cmes . in the meantime , the rising hot channel compresses the surrounding magnetic field and plasma , which successively stack into the cme lf . in this paper , we study in detail two well - observed cmes occurred on 2011 march 7 and 2011 march 8 , respectively . each of them is associated with an m - class flare . through a kinematic analysis we find that : ( 1 ) the hot channel rises earlier than the first appearance of the cme lf and the onset of the associated flare ; ( 2 ) the speed of the hot channel is always faster than that of the lf , at least in the field of view of aia . thus , the hot channel acts as a continuous driver of the cme formation and eruption in the early acceleration phase . subsequently , the two cmes in white - light images can be well reproduced by the graduated cylindrical shell flux rope model . these results suggest that the pre - existing flux rope plays a key role in cme initiation and formation . online - only material : animations , color figures
the interest to the @xmath5 reaction at relativistic energies is mostly due to the possibility to observe the manifestation of the non - nucleonic degrees of freedom and relativistic effects in the simplest bounded system . large amount of the polarization data in deuteron breakup obtained at a zero degree last years can be interpreted from the point of view @xmath6 configurations in the deuteron , where relativistic effects are taken into account by the minimal relativization scheme with the dependence of the deuteron structure on single variable @xmath7 . in addition the considering of multiple scattering is required to obtain the agreement with the data@xcite . on the other hand , it was shown that @xmath8 data for the pion - free deuteron breakup process @xmath9 in the kinematical region close to that of backward elastic @xmath10 scattering depended on the incident deuteron momentum in addition to @xmath7@xcite . the recent measurements of the tensor analyzing power @xmath0 of deuteron inclusive breakup on nuclear targets @xcite have demonstrated a significant dependence on the transverse secondary proton momentum @xmath11 being plotted at a fixed value of the longitudinal proton momentum . this forces one to suggest that description of this quantity requires an additional independent variable , aside from @xmath7 . in this report the angular dependence of @xmath0 in deuteron inclusive breakup on hydrogen and carbon at 9 gev / c are presented . the results are compared with the relativistic calculations using paris , cd - bonn and karmanov s deuteron wave functions ( dwfs ) . the experiment has been performed using a tensorially polarized deuteron beam of the dubna synchrophasotron and the sphere setup described elsewhere @xcite . the tensor polarization of the beam has been measured from the asymmetry of protons from the deuteron breakup on nuclear targets , @xmath12 , at a zero angle and the momenta @xmath13 @xcite . the vector polarization of the beam has been measured from the asymmetry of quasi - elastic @xmath14 scattering on @xmath15 target @xcite . the tensor and vector polarizations , @xmath16 and @xmath17 , were @xmath18 , @xmath19 , @xmath20 and @xmath21 , respectively . a slowly extracted deuteron beam with a typical intensity of @xmath22 @xmath23/spill was directed onto a liquid hydrogen target of 30 cm length or onto carbon targets with varied length . the data at 9 gev / c of the deuteron initial momentum were obtained at secondary proton emission angles of 85 , 130 and 160 mr and proton momenta between 4.5 and 7.0 gev / c on hydrogen and carbon . the separation of the protons and inelastically scattered deuterons was done by the measurements of their time - of - flight ( tof ) over a base line of @xmath24 m. the residual background was completely eliminated by the requirement that particles are detected at least in two prompt tof windows . the results on @xmath0 versus the momentum of the secondary protons are presented in fig . [ fig1 ] by the solid and open symbols for carbon and hydrogen targets , respectively . the circles are the data of this experiment , while the triangles represent the data obtained earlier@xcite . the dashed , dash - dotted and solid lines are the relativistic calculations using paris@xcite , cd - bonn@xcite and karmanov@xcite dwfs . the calculations performed in the framework of the light - front dynamics @xcite with the use of paris and cd - bonn dwfs fail to reproduce the @xmath0 data , while the use of karmanov dwf depending on two internal variables , @xmath7 and @xmath1 , is in a reasonable agreement with the data obtained at 85 mr . @xmath0 data obtained at fixed proton momenta of 6.0 , 6.5 and 7.0 gev / c are plotted versus transverse proton momenta @xmath1 . the symbols and curves are the same as in fig.[fig1 ] . again the use of dwf depending on two variables@xcite gives better agreement with the data .	the @xmath0 data for deuteron inclusive breakup off hydrogen and carbon at a deuteron momentum of 9.0 gev / c and large @xmath1 of emitted protons are presented . the large values of @xmath0 independent of the target mass reflect the sensitivity of the data to the deuteron spin structure . the data obtained at fixed @xmath2 and plotted versus @xmath3 clearly demonstrate the dependence of the deuteron spin structure at short internucleonic distances on two variables . the data are compared with the calculations using paris , cd - bonn and karmanov s deuteron wave functions . * tensor analyzing power @xmath0 in deuteron inclusive breakup at large @xmath3 and spin structure of deuteron at short internucleonic distances.@xmath4 * v.p.ladygin , l.s . azhgirey , s.v . afanasiev , v.v . arkhipov , v.k . bondarev , yu.t . borzounov , l.b . golovanov , a.yu . isupov , a.a . kartamyshev , v.a . kashirin , a.n . khrenov , v.i . kolesnikov , v.a . kuznezov , a.g . litvinenko , s.g . reznikov , p.a . rukoyatkin , a.yu . semenov , i.a . semenova , g.d . stoletov , a.p . tzvinev , v.n . zhmyrov and l.s . zolin _ joint institute for nuclear researches , 141980 dubna , russia _ g.filipov + _ institute of nuclear research and nuclear energy , 1784 sofia , bulgaria _ n.p . yudin + _ moscow state university , 117234 moscow , russia _
angular momentum is among the most important quantities determining the size and shape of galaxies , and yet a detailed understanding of its origins remains a missing ingredient in the theory of galaxy formation . @xcite was apparently the first astrophysicist who discussed the source of galaxy rotation in the framework of a theory of gravitational instability . he explained galaxy rotation as arising from gravitational coupling with the surrounding matter . alternative theories of the origin of galaxy rotation due to primordial turbulence and vorticity were also discussed at that time @xcite , but such theories were subsequently ruled out by the fact that velocities not arising from gravitation decay in an expanding universe , and also because of improving constraints on velocities derived from cosmic microwave background anisotropies . hierarchical clustering of cold dark matter ( cdm ) @xcite is now widely believed to describe the origin of structure in the universe , and in this scheme the angular momentum of dark matter halos and eventually the rotation of galaxies is usually thought to be produced by gravitational tidal torques in the course of the growth of rotation - free initial perturbations . this is a well - developed @xcite and widely accepted theory , and there is little doubt that the tidal torques mechanism is connected with the generation of angular momentum . to some degree the model is supported by n - body cosmological simulations . for example , if one considers all the mass which at @xmath6 ends up inside a given halo , then the angular momentum of that mass typically increases linearly with time at early stages of the collapse , just as predicted by the tidal torque theory ( e.g. , * ? ? ? * ) . at later times , again in agreement with predictions , the growth in angular momentum slows down @xcite . however , existing models and approximations that try to implement the tidal torque scenario do not agree in detail with the results of fully nonlinear n - body simulations . the predicted angular momentum of halos in the tidal torque model is typically overestimated by a factor of @xmath7 compared to the results of simulations , with a large scatter of the same magnitude @xcite . this over - prediction factor can be lessened if one assumes that the angular momentum stops its growth earlier than the turnaround moment @xcite because of nonlinear effects , but the scatter still remains very large . the direction of the spin parameter predicted by the theory also has large errors . @xcite give the average error in the direction of the spin of about 57@xmath8 , which agrees with @xcite who find a mean misalignment of 52@xmath8 with a scatter @xmath9 . the source of the errors is still under debate , but it seems that the main effect is due to nonlinear effects which are difficult to take into account in the tidal torque model @xcite . we also address another limitation of the tidal torque picture : predictions for the spin parameter in this framework are for _ all _ the matter which at @xmath6 is found in a given halo . at high redshifts this mass is of course not in a single halo . in other words , the tidal torque theory does not predict the rotation of any particular progenitor of the halo in question . in this paper , we will pursue an alternative approach , that of tracing the evolution of the angular momentum of the most massive progenitor of the present - day halo . this alternative approach is much more useful for semi - analytic modeling of galaxy evolution ( e.g. * ? ? ? the angular momentum can be expressed in terms of the dimensionless spin parameter , which is defined as @xmath10 here @xmath11 is the angular momentum , @xmath12 is the total energy , and @xmath13 is the mass of a halo . the value of the spin parameter roughly corresponds to the ratio of the angular momentum of an object to that needed for rotational support ( e.g. , * ? ? ? for example , a spin parameter of @xmath14 implies very little systematic rotation and negligible rotational support . typical values of the spin parameter of individual halos in simulations are @xmath15 to @xmath16 @xcite . the distribution of spin parameters in n - body simulations is well described by the log normal distribution : @xmath17 the parameters for the log - normal distribution were found to be @xmath18 and @xmath19 for standard cdm and various variants ( e.g. * ? ? ? * ; * ? ? ? * ) . for the @xmath1cdm cosmology with matter density @xmath20 , @xmath21 , @xmath22 , and @xmath23 , the log - normal parameters were found to be @xmath24 and @xmath25 @xcite . note that the @xmath26 parameter of the log - normal distribution ( 2 ) is not equal to the mean of @xmath0 ; rather , @xmath27 for @xmath28 . in order to study the evolution of the angular momentum of a dark matter halo it is important to find correlations of the spin parameter with other parameters of the halo and with its environment . correlation in the directions of spins of nearby halos or galaxies is interesting as an indicator of the strength of large - scale correlations and how much the next infalling satellite knows about the previous one . because the tidal torques are due to perturbations in the gravitational potential , one naively expects that there are long range correlations in spins of halos and , consequently , in the angular momenta of galaxies . indeed , there are claims that such correlations exist in both observational data and in n - body simulations . but the correlations of galactic rotation axes are measured to be small . if @xmath29 is a unit vector in the direction of the spin axis , then @xcite find that the correlation function of directions @xmath30 is less than 0.05 for distances @xmath31 , and is even smaller at larger distances . even at small separations the effect is small and is consistent with no correlations . results of n - body simulations also indicate that spin correlations are weak at best @xcite . @xcite found the spin - spin correlation function of halos is @xmath32 . @xcite found no correlation at distances larger than @xmath33 at late stages of evolution ( @xmath6 ) . these very small correlations also suggest that material which is accreted by a growing halo has very little memory : accretion is mostly random . the spin parameter of halos in n - body simulations appears to be a very stable statistic and has been shown to be independent of most physical parameters . no dependence has been found on the cosmological model @xcite , on halo environment or on halo mass @xcite . the only correlation that has been shown to exist is with the time of the last major merger : halos that experienced a recent merger have larger spin @xcite . in this paper we investigate the amount of angular momentum that is brought in by the random uncorrelated accretion of satellites . because there is no systematic rotation of the incoming satellites , one might naively expect that their contribution to the rotation of the accreting object will be small . but at the same time , each act of accretion contributes a large orbital angular momentum . thus the halo typically acquires a substantial rotation from each large merger , which may later be diluted by many uncorrelated small mergers . the outcome of the process was not obvious in advance , and to our surprise we found that the resulting rotation is quite significant . we begin , in 2 , with an analysis of n - body simulations . there are two reasons for this . first , we study the spin histories of several n - body halos in order to provide a comparative test for our model predictions . second , we need to characterize the distributions of velocities and angular momenta of infalling satellites , because these are essential inputs for our model . in 3 we present our random walk model for the build - up of angular momentum in halos . results of the random walk approximation are presented in 4 . we discuss our results and the general implications of our model in 5 . we use three n - body simulations of a low - density flat @xmath1cdm model with the following parameters : @xmath20 , @xmath34 , and @xmath22 . the simulations were done with the adaptive refinement tree ( art ) code @xcite . the first simulation , @xcite , was done only for three halos , but the resolution was extremely high : the mass per particle was @xmath35 and the ( formal ) force resolution was @xmath36pc . here we used the power spectrum normalization @xmath37 . at redshift @xmath6 the three halos had virial masses @xmath38 , corresponding to @xmath39 particles per halo . in order to find statistics of velocities and orbital angular momenta of satellites needed for the random walk model , we used two additional simulations , which do not have as high resolution as the first one but have many more halos . one simulation used @xmath40 dark matter particles within a cubic box of comoving size @xmath41mpc with mass per particle @xmath42 and with a ( formal ) force resolution @xmath43kpc . for this simulation we used @xmath44 . this simulation has 15500 dark matter halos at redshift @xmath6 with masses from @xmath45 to @xmath46 . another simulation used @xmath47 dark matter particles within a cubic box of comoving size @xmath48mpc . the mass per particle was @xmath49 . this simulation was run only to @xmath5 . it had 62,000 halos with mass larger than @xmath50 . this simulation was used to study statistics of large mergers . in our simulations the halos are identified by the bounded - density - maxima ( bdm ) algorithm @xcite . for distinct halos ( halos that are not inside larger halos ) the algorithm defines halos as spherical objects with average virial overdensity . halos inside larger halos are defined as gravitationally bound lumps of dark matter . figure [ fig : tracks ] shows the evolution of the mass and the spin parameter of the major progenitors of our three @xmath51 high resolution halos . the spin parameter of the halos clearly changes with time . even for the most quiet halo @xmath52 the spin parameter changed by a factor 1.5 since redshift 3 . for the other two halos the change was even larger . note that there is no steady increase of the spin with time ; instead , there are large sudden changes in @xmath0 , which clearly correlate with jumps in the masses of the halos mostly associated with large mergers . the jumps in @xmath0 are of both signs : most large merger events are associated with large spin increases , but halo @xmath53 had a relatively large merger event at @xmath54 , which decreased its spin by about a factor of two . there is also a tendency of @xmath0 to decline during periods of gradual mass accumulation , which is clearly observed in halos @xmath52 and @xmath55 at later stages of evolution . the halos accrete satellites from all directions . figure [ fig : directions ] shows directions of orbital angular momenta of satellites which are being accreted by the halos at two redshifts . we show satellites that at a given redshift are found in a spherical shell radius between 0.9 and 1.1 times the virial radius . only satellites that move toward the center are considered . the distribution is not totally random : the satellite spins are clustered . this clustering of orbital spins comes from real - space clustering : there are larger satellites that are surrounded by smaller ones . when a group of satellites falls in to the central halo , they have almost the same angular momentum . besides the clustering , there is no discernible pattern in the directions of the satellites . for example , halo @xmath52 at @xmath6 has a hole around the northern pole ( direction of rotation of the central halo ) . this would indicate that orbits of satellites are far from the plane of rotation of the central halo . at the same time , halo @xmath55 has an excess of satellites with the same direction of orbital motion . these three halos were chosen to reside in a @xmath56 mpc filament bordering a large void . this is done to roughly mimic the environment of the local group . the filament appeared at very high redshifts . the three halos were always in the filament , growing by merging with smaller halos . a simple naive expectation is to see a lump of satellites in panels in figure [ fig : directions ] . its direction would indicate the direction of the filament and the rotation of the halo relative to the filament . we do not find this : there is no obvious large single lump in the diagrams . the problem with the naive expectation is that one imagines a thin tube with the halo inside it . even qualitatively this is a wrong picture . the filament is much wider ( 1 - 2 mpc ) than the virial radii of the three halos ( @xmath57 kpc ) . when the satellites eventually fall in to the halo , they come from different directions in the filament , not from the main axis of the filament . because of the finite thickness of the filament , one may even expect to find anti - correlation of orbital spins of satellites . satellites moving parallel to the axis bring opposite signs of the angular momentum when they move above or below the axis . figure [ fig : directions ] also shows that there is almost no correlation with redshift for the same halo . patterns are not reproduced with time . statistical analysis of the directions confirms our visual impression . the measured distribution is consistent with an uncorrelated random distribution . the data still allow some degree of correlation or anti - correlation because of the small number of satellites and consequent statistical uncertainties . the total number of satellites in our simulations was large , about 300 per central halo at the end of evolution , yet the number of satellites in a shell around the virial radius was much smaller , about 40 for each halo . with the total number of 120 satellites , statistical uncertainties still allow correlation or anti - correlation on the level of 10 - 20% . altogether , the evolution of the spin parameter and the directions of accreted satellites indicate that the accumulation of the angular momentum happens mostly in a random , uncorrelated fashion . we used the second simulation with a large sample of halos to study the correlations of jumps in @xmath0 and in mass . the analyses presented here are based on a new structural catalog of halos from this simulation , developed as part of the study of the relationship of the structure of halos to their merging history in @xcite and wechsler et al . ( 2002 ) , where further details can be found . briefly , all separate bound virialized halos with more than 20 particles are cataloged at each of 36 stored time - steps from this simulation , and nfw fits @xcite are obtained for all halos with more than 200 particles . these catalogs were then compared in order to determine the merging histories of all halos . the statistics presented here are for distinct halos , i.e. those which are not subhalos of another halo . the simulations indicate that spin parameters of individual halos change very substantially during the growth of halos : there are sharp increases in @xmath0 which correlate with periods of fast mass increase , while periods of slow mass increase tend to correlate with declines in the spin parameter . thus the features illustrated for three halos in figure [ fig : tracks ] are in fact quite general . in figure [ fig : lambda_mm ] we present statistics of the change in the spin parameter in major mergers , while figure [ fig : lambda_change ] includes these statistics for a range of changes in the mass of the most massive progenitor . in these figures , we actually plot the changes not in @xmath0 but rather in @xmath58 , defined as @xcite @xmath59 where @xmath60 is the circular velocity at the virial radius , @xmath61 , for a halo with virial mass @xmath62 ( see 3.1 for more details ) . the spin parameters @xmath58 and @xmath0 are approximately equal for typical nfw halos @xcite , but @xmath58 is easier to measure in simulations . for figure [ fig : lambda_mm ] we identified all halos with two progenitors of mass ratio @xmath63 in the previous stored time - step , and investigate the distribution of @xmath64 in these merging events . on average , we find that @xmath58 increases by about 25% in major merger events ( figure [ fig : lambda_mm ] , upper left panel ) , though there is a wide distribution . not surprisingly , the change in @xmath58 is more severe for cases in which the initial @xmath58 was low ; in these cases an incoming satellite is likely to increase the spin regardless of its impact parameter . halos which have low initial spin values can thus experience jumps of up to a factor of ten in a single merging event . when the initial spin is high , however , merger events can serve to decrease the angular momentum if they come in at opposite direction to the spin . for very high initial values , @xmath64 decreases slightly on average , occasionally as much as a factor of five . major mergers are clearly important since they can bring in a large amount of angular momentum at once , but minor mergers also play a large role in influencing halo spin . in figure [ fig : lambda_change ] , we show the distribution of @xmath64 for a variety of initial spin parameters and changes in halo mass . we see that the spin of halos with very low initial spin values is likely to increase with any incoming material , even a small amount . for halos with intermediate initial spin values , only major mergers are going to have a large effect on their spins , and for halos with very large initial spin values , any incoming mass is likely to decrease the spin . thus although major mergers produce the largest jumps in @xmath58 , we find that the entire mass accretion history is important for its evolution . in this section we study the infall velocities of satellites into larger halos in the simulations in order to provide inputs to the random walk model . we show that the distribution of infall velocities can be well parameterized by a characteristic gaussian velocity dispersion and a radial anisotropy parameter . in addition , we find that the velocity anisotropy changes as a function of the mass ratio of satellite to major progenitor , @xmath65 : large satellites tend have more radial orbits . since we would like to include at least the basic trend with @xmath65 in our model , we make the following simplified division : major mergers satisfy @xmath63 and minor mergers satisfy @xmath66 . we characterize the velocity ellipsoids of each of these populations separately . we sorted all halos by mass and selected pairs of halos starting with the most massive ones . because we are interested in studying satellites at the time of accretion , we selected only pairs separated by approximately the virial radius @xmath67 of the larger halo in each pair . three criteria were used : ( 1 ) the distance between a pair of halos was required to be @xmath68 , ( 2 ) the relative velocity of halos had to be less than @xmath69 , where @xmath70 is the maximum circular velocity of the larger halo , and ( 3 ) the halos had to have negative relative radial velocities ( corresponding to inward motion ) , i.e. the angle between the line from the smaller to the larger halo and their relative velocity vector should be between @xmath71 and @xmath72 . the large sample of halos in our @xmath73 simulation allows us to study the statistics of relative velocities and angular momenta of satellites falling onto larger halos . we did so by averaging results for expansion factors @xmath74 and @xmath75 . numerical effects force and mass resolution and specifics of our halo finder result in a rather complicated selection of mergers of different mass ratios . the limited force and mass resolution result in destruction of most of satellites of small halos . this is the so called overmerging problem . because of this problem the galaxy - size halos in this simulation can not be used for studies of minor mergers , since their satellites were not resolved . small subhalos were resolved for large groups and clusters , which had dozens or hundreds of them @xcite . thus , we studied minor mergers for host halos with masses larger than @xmath76 . overall , there were 543 satellites at radius @xmath77 of the virial radius of the host of which 301 were falling onto the host and the rest were moving out . merging with large satellites was studied mostly at high redshifts . this was done using the @xmath73 simulation at @xmath78 and @xmath5 . we also used the @xmath79 simulation at @xmath5 to probe statistics of satellites with different masses . figure [ fig : catl60 ] shows several statistics for the minor mergers in our catalog . the upper right panel shows the distribution of infall angles , @xmath80 , of the smaller halos , where @xmath80 is the angle between the velocity and radius vectors of the satellite . the distribution of relative infall velocities in units of @xmath70 of the larger halo is shown in the lower left , and the distribution of the relative specific angular momenta of satellites is shown in the bottom right , plotted in units of @xmath81 of the larger halo , where @xmath82 . the error bars represent poisson errors . in the top left panel we show the distribution of the relative velocity as a function of @xmath80 . there is a small bias in the distribution of velocity vs. angle , with higher infalling velocities corresponding to more radial infall directions . the distribution of the angles in the top right panel shows that there are more infalling halos with radial than with tangential motions . in general , the magnitudes of the infall velocities are distributed around the circular velocity of the massive progenitor ( bottom left panel ) . the distribution of the velocity can be fit with a 3d maxwell - boltzmann distribution ( @xmath83 , @xmath84 ) : @xmath85 this fit provides a rough characterization of the distribution , but does not fully capture its nature . this is likely because the the velocities are anisotropic , and the maxwell - boltzmann distribution describes a system with isotropic velocities . the distribution of satellite angular momenta ( bottom right panel ) , or , equivalently , tangential velocities , is described reasonably well by a ( 2d ) maxwell - boltzmann shape : @xmath86 with @xmath87 , @xmath88 . a good fit is obtained because the tangential velocities are fairly isotropic . these results suggest that the velocities of satellites are well - described by a 3d gaussian distribution with a substantial radial velocity anisotropy . we further have checked the distribution of each component of velocity and found them to be roughly gaussian . based on this , it is justified to summarize the velocity distribution of infalling satellites with two parameters defining a velocity ellipsoid , one describing the velocity anisotropy , @xmath89 and the other describing the radial gaussian velocity dispersion , @xmath90 . for example , in the case of isotropic orbits @xmath91 and @xmath92 , and in the case of nearly radial orbits @xmath93 and @xmath94 . from the 60 @xmath95 mpc numerical simulation , the velocity anisotropy and tangential velocity dispersion were found to depend on the ratio @xmath65 of the mass of the satellite to that of the host galaxy . for major mergers @xmath96 and @xmath97 , and for minor mergers @xmath98 and @xmath99 . these values are in good agreement with @xcite . note that major mergers are significantly more radial than minor mergers and bring in a factor of 1.6 less specific angular momentum . we also studied the velocities of satellites in the simulation of the three galaxy - size halos . the statistics of satellites in each halo are much better in this simulation . we can also easily track the evolution of each halo with redshift . the results are consistent with what we found for the larger simulation . for minor mergers the average velocity anisotropy was @xmath100 at @xmath6 and @xmath101 at @xmath78 , however , the difference in this value is not statistically significant , and on average the value remains unchanged ( @xmath102 ) . the three halos in the first simulation have masses of @xmath103 . this is significantly smaller than @xmath104 in the case of the large - box second simulation where , as we explained above , most of the satellites were coming from group - size central halos . thus , the results for the three halos suggest that there is no ( significant ) trend of @xmath105 with the mass of the central halo . the dependence of the angular momentum of infalling satellites on the mass of the satellites is very important for accurate estimates of the spin parameter @xmath0 . we use all available simulations to measure this dependence . because the mass of a satellite is not always a well defined quantity ( e.g. , it depends on an uncertain truncation radius and on details of removal of unbound particles ) , we prefer to use a more reliable characteristic the maximum circular velocity . figure [ fig : sigma ] shows the dependence of the specific angular momentum of a satellite @xmath106 on the ratio of the satellite maximum circular velocity @xmath107 to the maximum circular velocity of the host halo @xmath108 . different markers show results for different simulations and redshifts . error bars show @xmath109 shot noise estimates . figure [ fig : sigma ] indicates that the specific angular momentum of small satellites is larger than that of the large ones . it also shows that there is no dependence of @xmath110 on the redshift . the same type of analysis shows that the radial velocity dispersion does not depend on the mass of the satellite : @xmath111 . thus , from eq . [ eq : beta ] , velocity anisotropy @xmath105 varies from @xmath112 for the small satellites to @xmath113 for the major mergers . we use the following approximation for the rms of the specific angular momentum : @xmath114 figure [ fig : sigma ] shows that this is a reasonable summary of the simulation data . the main idea of our random walk model is that halo angular momentum is built up from the summed contributions of uncorrelated orbital angular momenta of accreted objects during the process of halo assembly . the model needs several ingredients . first , we need to have the mass accretion history for each halo ( a merger tree ) that tells us how many satellites of what mass are accreted by the major progenitor as a function of redshift . for this , we use the extended press - schechter ( eps ) approximation @xcite and the eps merger - tree method of @xcite . second , we need to know the position of each satellite at the moment of accretion . the position is assumed to be randomly distributed on a sphere whose radius is the virial radius of the main progenitor , with the main progenitor in the center . finally , we need to know the velocities of accreted satellites . the three components of the velocities of each satellite are assumed to be gaussian , with a velocity anisotropy described by eq . we use @xmath115 regardless of the mass of the satellite . the tangential velocity dispersion is assigned in the following way . for a given satellite mass we estimate its concentration parameter ( see eqs . ( 16 - 17 ) below ) . the virial mass and the concentration define the maximum circular velocity @xmath116 . we then use eqs . ( 6 - 7 ) to find the tangential velocity dispersion . a realization drawn from the gaussian distributions gives the three components of the velocity of the satellite . the model described above provides all of the information needed to calculate the spins of an ensemble of halos . the spin parameter @xmath0 is determined by calculating the energy for an nfw halo in virial equilibrium and then using eq . ( 1 ) . for an nfw halo , @xmath117^{-1}$ ] , the virial concentration @xmath118 , where the virial radius @xmath119 is defined by the virial mass @xmath62 and by the parameters of the cosmological model : @xmath120 with @xcite @xmath121,\end{aligned}\ ] ] and @xmath122 . in the @xmath1cdm model @xmath123 at @xmath124 and @xmath125 at @xmath126 . the total energy of a halo truncated at the virial radius is @xmath127 here @xmath128 and @xmath129 the spin is then @xmath130^{1/2}}{f(c)}}.\ ] ] for the median concentration @xmath55 , we use the results of @xcite : @xmath131 in this paper we have assumed for simplicity that the halo mass and redshift uniquely determines its concentration . n - body simulations indicate that there is a scatter in @xmath55 at fixed mass of the order of 35% @xcite . in principle , we could include those deviations in our random walk model , but we decided to neglect them because the deviations in @xmath0 as a result of this scatter are less than @xmath132 percent . thus , knowing the mass and the angular momentum of a halo one can find the halo s concentration and its spin parameter . our main results are derived from two sets of monte - carlo merger trees for two different final masses , @xmath133 . each set of runs traced the merging history of the most massive progenitor back to @xmath134 or until its mass fell below @xmath135 . the first set had @xmath136 , @xmath137 , and a mass resolution ( minimum mass of a satellite ) of @xmath138 . the second set of trees was done with @xmath139 , @xmath140 , and @xmath141 . we used 300 mass tracks for each set and each track was run 100 times with different realizations of satellite velocities . unless otherwise stated , we use these two sets of merger trees . the evolution of the mass of two typical @xmath142 tracks as a function of the expansion parameter is shown on the top left panel in figure [ fig : track ] . the right top panel shows the ratio of the mass of the biggest satellite to the mass of the main progenitor . the mass of the main progenitor does not change smoothly , because we assume that all new incoming mass instantly merges with the main progenitor . when the mass of the merging object is significant , there are often dramatic changes in the specific angular momentum ( bottom left panel ) as well as in the spin parameter ( bottom right panel ) . this also was seen for simulated halos in 2 . of course , major mergers do not always increase the spin parameter ; depending on the velocity orientation , such mergers can also decrease the halo s spin . in spite of all these changes in the spins of individual halos , especially in the past when the number of major mergers was significant , we will show below that the _ distribution _ of the spin parameter does nt depend on time or on the mass of the main progenitor . these examples show that the angular momentum does nt increase linearly . whether there are increases or decreases in any given merging event depends on the mass of the satellite , its direction of motion , and its velocity . one can also see that at the later stages of the evolution of a dark matter halo , when the mass of the main progenitor does nt change much , the angular momentum does nt have dramatic changes either . this behavior is very similar to that found in the n - body simulations , although the comparison can not be made precise because of the finite differences ( up to @xmath143 gyr ) between the stored time - steps in the simulation compared to the instantaneous approximation used in our random walk model . we begin by illustrating that our random walk approximation naturally produces a log - normal distribution in @xmath0 . figure [ fig : mass_z ] shows a comparison between n - body results and our random walk model of spin acquisition . the simulation corresponds to @xmath1cdm ( @xmath20 , @xmath22 , @xmath23 ) . for the random walk model we used 200 monte carlo merger trees with a final ( @xmath6 ) mass of the main progenitor @xmath144 . each mass track was run 100 times with different realizations of satellite velocities . the distribution of the spin parameter from the n - body simulation is shown by dashed line while the distribution from the random walk model is shown by the solid curve . as one can see , there is a reasonably good agreement between the n - body results and the random walk model . to demonstrate that these distributions are well described by the log - normal function we also show a log - normal fit to the random walk data ( the dashed - dotted curve ) . parameters of the fit are given in table 1 , along with spin parameters measured from n - body simulations . as we discussed in 1 , n - body simulations indicate that the @xmath0 distribution does not vary as a function of redshift or halo mass . here we test our model against these results . we analyzed two different sets of merger trees : one with major progenitor mass @xmath145 at redshift @xmath6 , and the other with mass @xmath146 ( see 3 ) . for both cases the distributions of the spin parameters were examined at two different redshifts : @xmath6 and @xmath5 . these are shown in figure [ fig : redshift ] . the spin parameter distributions derived from the random walk model show no correlation with either the redshift or the halo mass ; the distributions overlap and each has mean value @xmath147 . as was just shown , the spin parameter distribution does nt depend on the redshift or on the mass . but figure [ fig : track ] indicates that @xmath0 of individual halos in the random walk model is sensitive to whether they have recently had major mergers , as we found in n - body simulations ( figures 1 , 3 ) . in this subsection we study the influence of these types of mergers on the distributions of the spin parameter . we study the effect of major mergers on the distribution of the spin parameter by splitting all halos into two groups : the halos that did nt have any major mergers in the past roughly 11 billion years ( @xmath148 ) , and the halos that did have a major merger during this period . since major mergers destroy galactic disks and produce spheroidal stellar systems ( see e.g. * ? ? ? * ) , while mergers with mass ratio less than 1/10 probably do not ( e.g. * ? ? ? * ) , and stellar disks are typically up to about 10 gyr old , only the halos that did not have a major merger since @xmath149 could host mature spiral galaxies . the distributions of @xmath0 for these two groups of halos are shown in figure [ fig : mmeger ] . the mean value of the spin parameter is @xmath150 and @xmath151 for no major mergers and major mergers since @xmath5 respectively . in other words , recent major mergers increase the spin parameter almost by 25% . thus , we find the same result that was found in the n - body simulations : the spin parameters are sensitive to major mergers . [ tab ] [ cols="<,^,^,^",options="header " , ] we have presented a new model for the origin of the angular momentum of dark matter halos in which angular momentum is built up in a random walk fashion by mass accretion events . the evolution of halo angular momentum in this picture is quite different from that inferred from the standard tidal torques argument , in which the angular momentum grows steadily at early times and its growth flattens out at late times . unlike this standard picture , we do not track the angular momentum of _ all _ matter that is in the present - day halo ; we track only the major progenitor of the halo . the evolution of the spin parameter @xmath0 of the major progenitor is a random process in which the the value of @xmath0 can vary by factors of a few over its history , with a tendency for sharp increases due to major mergers and steady decline during periods of gradual accretion of small satellites . these general properties of evolution , including large fluctuations in @xmath0 , are confirmed by n - body simulations . our model predicts that the distribution of the spin parameter is well approximated by the log - normal distribution , with mean and dispersion that do not depend on redshift or on the mass of the main progenitor . the model predicts that on average the spin parameters of halos that had major mergers after redshift @xmath5 should be considerably larger than the spin parameters of those halos that did not . perhaps surprisingly , this implies that halos which host later - forming elliptical galaxies should rotate faster than halos of mature spiral galaxies . there may actually be observational evidence of this . planetary nebulae in the two elliptical galaxies ngc 5128 ( radio continuum source cen a ) @xcite and ngc 1316 ( fornax a ) @xcite are rotating rapidly about their minor axes at large radii . the metal - rich globular clusters of ngc 5128 show rotation similar to that of the planetary nebulae . ngc 5128 is regarded as a prototypical remnant of the merger of two disks , and ngc 1316 is also considered a likely remnant of a fairly recent merger . from numerical simulations @xcite of mergers between disk galaxies with bulges , one indeed expects the angular momentum of the remnant to be concentrated in the outer regions . similar studies of the planetary nebulae of other elliptical galaxies such as ngc 4697 detect less evidence of strong rotation @xcite , and stellar spectra of only a relatively small fraction of elliptical galaxies show strong rotation at large radii ( with @xmath152 ) , e.g. ngc 1395 and ngc 1604 @xcite . the globular clusters of m87 and a few other giant elliptical galaxies are also rotating rapidly at large radii @xcite . galaxies like the cd m87 are thought to form from multiple mergers , and simulations of this process predict that while velocity dispersion should dominate in the central regions , rotation can grow to @xmath153 in outer regions @xcite . radial velocity measurements for the planetary nebulae and globular cluster systems of nearby elliptical galaxies should become much more feasible with the advent of multi - object spectrographs on 8-meter class telescopes . once data are available on a larger number of such systems , it will be interesting to see whether the statistics are consistent with predictions of the random walk model for the origin of galaxy rotation that we have proposed here . implementation of the random walk model is a rather complicated procedure if one wants to reproduce n - body results with an accuracy better than about 20 percent . it requires measurements of many parameters of halos and satellites ( e.g. , halo concentrations for different redshifts and merging trees ) . the most important and , unfortunately , the most uncertain are the parameters of the velocity ellipsoid of accreted satellites for different masses of the satellites . in principle , these parameters could vary as a function of redshift as well . estimates for the numerous minor mergers seem to be reasonably reliable ( see 2.3 ) . major mergers are rarer and therefore the estimates are more uncertain . generally we find that the trajectories of mergers with high @xmath65 ratios are more radial than their low @xmath65 counterparts , but it is difficult to measure the parameters of the velocity ellipsoid accurately . the model assumes that there is no net angular momentum of infalling satellites and that the satellites are accreted in random fashion . this picture is supported by results of n - body simulations ( 2.2 ) . yet the results allow some degree of either correlation or anti - correlation of orbital spins of the satellites . the allowed effects are rather small ( less than 10 - 20% ) , but they still may be important for accurate predictions of the angular momenta of halos . our colleagues @xcite have investigated a simplified version of what we call the random walk model for the origin of the angular momentum of dark matter halos , in comparison with a version of the tidal torques model more sophisticated than the standard one discussed briefly in 1 of the present paper . they find that both models can reproduce the log - normal distribution of halo spin parameters seen in simulations , with appropriate tuning of the model parameters . their work is thus complementary to that presented here . we expect that our random walk model for the origin of halo angular momentum through the accretion of satellites will be useful both conceptually and practically . in addition to giving a much better fit to the distribution of halo spin parameters than the tidal torque theory , it also accords better with the hierarchical nature of the process by which halos grow in the cold dark matter model . moreover , because it is based on the same extended press - schechter formalism used in semi - analytic approaches and because it is so simple to implement , we expect that it will prove invaluable in improving the treatment of angular momentum in modeling the formation and evolution of galaxies . for these same reasons , it will be worthwhile to test and improve the random walk model . we conclude with some comments about the implications of our model for disk galaxy formation and the angular momentum problem in disks . this random walk picture is somewhat at odds with the formalism made popular by fall & efstathiou ( 1980 ) in which the gas is initially well mixed within a smoothly rotating halo , and subsequently falls in to form an angular momentum supported disk . in our random walk model , the halo angular momentum obtains its smoothly varying distribution through a series of clumpy merger events and subsequent relaxation . unlike a picture in which tidal torques determine the spin of a halo at very early times , when the gas and dark matter was well - mixed , our scenario would suggest that the angular momentum distribution of gas could be considerably different than that of the dark matter . the merging satellite galaxies are likely to be strongly affected by the tidal field of the central object , and their gas is likely to shock against the gas in and around the host . since it is known that the angular momentum distribution in dark halos is quite unlike that in disk galaxies @xcite , generally with an excess of low - spin material , our random walk picture might provide an interesting avenue for exploring solutions of this angular momentum distribution problem and also of the well - known angular momentum overcooling problem ( e.g. navarro & steinmetz 1997 ) . work in this direction is under way . we acknowledge support from nasa and nsf grants at nmsu , ohio state , and ucsc , and we thank avishai dekel , tsafrir kolatt , ari maller , and rachel somerville for helpful discussions . is grateful to the institute of astronomy at cambridge for hospitality and financial support during his extended visit to cambridge university . was supported by nasa through a hubble fellowship grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , under nasa contract nas5 - 26555 47 natexlab#1#1 arnaboldi , m. , et al . 1998 , , 507 , 759 barnes , j.e . , & efstathiou , g. 1987 , , 319 , 575 barnes , j.e . 1992 , , 393 , 84 barnes , j.e . 1999 , in asp conf . 187 , the evolution of galaxies on cosmological timescales , ed . j.e beckman & t.j . mahoney ( san francisco : asp ) , 293 blumenthal , g.r . , faber , s.m . , primack , j.r . , & rees , m. 1984 , nature , 311 , 517 blumenthal , g.r . , faber , s.m . , flores , r. , & primack , j.r . 1986 , , 301 , 27 bond , j.r . , cole , s. , efstathiou , g. , & kaiser , n. 1991 , , 379 , 440 bower , r. 1991 , mnras , 248 , 232 bryan , g.l . , & norman , m.l . 1998 , , 495 , 80 bullock , j.s . , dekel , a. , kolatt , kravtsov a.v . , klypin , a.a . , porciani , c. , & primack , j.r . , 2001b , apj , 555 , 240 bullock , j.s . , kolatt , t.s . , sigad , y. , somerville , r.s . , kravtsov a.v . , klypin , a.a . , primack , j.r . , & dekel , a. 2001a , mnras , 321 , 559 . cohen , j.g . 2000 , , 119 , 162 cole , s. , & lacey , c. 1996 , mnras , 281 , 716 colin , p. , klypin , a.a . , & kravtsov a.v , 2000 , apj , 539 , 561 ct , p. , et al . 2001 , , 559 , 828 dekel , a. , bullock , j.s . , porciani , c. , kravtsov , a.v . , kolatt , t.s . , klypin , a.a . , & primack , j.r . , 2001 , in asp conf . ser . 230 , galaxy disks and disk galaxies ( san francisco : asp ) , eds . funes , s.j . , and e.m . corsini , astro - ph/0011002 doroshkevich , a.g . 1970 , astrofizika , 6 , 581 fall , s.m . , & efstathiou , g. 1980 , mnras , 193 , 189 flores , r. , primack , j.r . , blumenthal , g.r . , & faber , s.m . 1993 , , 412 , 443 franx , m. , illingworth , g.d . , & heckman , t. 1989 , , 344 , 613 gamow , g. , 1952 , phys . rev , 86 , 251 gardner , j.p . , 2001 , apj , 557 , 616 hernquist , l. 1993 , , 409 , 548 hoyle , f. , 1949 , _ problems of cosmical aerodynamics _ dayton : central air documents office . hui , xiaohui , ford , h.c . , freeman , k.c . , & dopita , m.a . 1995 , , 449 , 592 kissler - patig , m. , & gebhardt , k. 1998 , aj , 116 , 2237 klypin , a.a . , holtzman , j. , 1997 , astro - ph/9712281 klypin , a.a . , kravtsov , a.v . , bullock , j.s . , & primack , j.r . 2001 , apj , 554 , 903 kravtsov , a.v . , klypin , a.a . , & khokhlov , a.m. , 1997 , apjs , 111 , 73 lacey , c. , & cole , s. 1993 , mnras , 263 , 627 lee , j. , & pen , u. 2000 , , 532 , l5 lee , j. , pen , & u. , 2001 , , 555 , 106 lemson , g. , kauffmann , g. 1999 , , 302 , 111 maller , a.h . , dekel , a. , & somerville , r.s . 2002 , , 329 , 423 mendez , r.h . , et al . 2001 , , 563 , 135 mo , h .- j . , mao , s. , & white , s.d.m . 1998 , , 297 , l71 ; mnras , 295 , 319 navarro , j.f . , frenk , c.s . , & white , s.d.m . 1996 , , 462 , 563 navarro , j.f . , frenk , c.s . , & white , s.d.m . 1997 , apj , 490 , 493 navarro , j.f . , & steinmetz , m. 1997 , apj , 478 , 13 padmanabhan , t. , 1993 , _ structure formation in the universe _ , cambridge : cambridge univ . press peebles , p.j.e . 1969 , apj , 155 , 393 pen , u. , lee , j. , & seljak , u. 2000 , , 543 , l107 porciani , c. , dekel , a. , & hoffman , y. 2002 , , 332 , 325 press , w.h . , & schechter , p. 1974 , apj , 187,425 ryden , b.s . 1988 , apj , 329 , 589 somerville , r.s . , & kolatt , t.s . 1999 , mnras , 305 , 1 somerville , r.s . , & primack , j.r . 1999 , mnras , 310 , 1087 somerville , r.s . , lemson , l.g . , kolatt , t.s . , & dekel , a. 2000 , mnras , 316 , 479 steinmetz , m. , & bartelmann , m. 1995 , mnras , 272 , 570 sugerman , b. , summers , f.j . , kamionkowski , m. , 2000 , mnras , 311,762 van den bosch , f. 1998 , apj , 507 , 601 van den bosch , f.c . , burkert a. , & swaters r.a . 2001 , mnras , 326 , 1205 walker , i.r . , mihos , j.c . , & hernquist , l. 1996 , , 460 , 121 warren , m.s . , quinn , p.j . , salmon , j.k . , & zurek , w.h . 1992 , apj , 399 , 405 wechsler , r.h . 2001 , phd dissertation , university of california , santa cruz wechsler , r.h . , bullock , j.s . , primack , j.r . , kravtsov , a.v . , & dekel , a. 2002 , apj , 568 , 52 weil , m.l . , & hernquist , l. 1996 , , 460 , 101 wieszacker , c.f . von , 1951 , apj , 114 , 165 white , s.d.m . 1984 , apj , 286 , 38	we propose a new explanation for the origin of angular momentum in galaxies and their dark halos , in which the halos obtain their spin through the cumulative acquisition of angular momentum from satellite accretion . in our model , the build - up of angular momentum is a random walk process associated with the mass assembly history of the halo s major progenitor . we assume no correlation between the angular momenta of accreted objects . the main role of tidal torques in this approach is to produce the random tangential velocities of merging satellites . using the extended press - schechter approximation , we calculate the growth of mass , angular momentum , and spin parameter @xmath0 for many halos . our random walk model reproduces the key features of the angular momentum of halos found in @xmath1cdm n - body simulations : a log - normal distribution in @xmath0 with an average of @xmath2 and dispersion @xmath3 , independent of mass and redshift . the evolution of the spin parameter in individual halos in this model is quite different from the steady increase with time of angular momentum in the tidal torque picture . we find both in n - body simulations and in our random walk model that the value of @xmath0 changes significantly with time for a halo s major progenitor . it typically has a sharp increase due to major mergers , and a steady decline during periods of gradual accretion of small satellites . the model predicts that on average the @xmath0 of @xmath4 halos which had major mergers after redshift @xmath5 should be substantially larger than the @xmath0 of those which did not . perhaps surprisingly , this suggests that halos that host later - forming elliptical galaxies should rotate faster than halos of spiral galaxies .
the first - generation ground - based detector laser interferometer gravitational - wave observatory ( ligo ) has successfully undergone major technical upgrades in the past few years that have led to a significant increase of the volume of the observable universe.[multiblock footnote omitted ] the detector has already been operative for a few months and _ observation run 1 _ ( o1 ) is being analyzed as these lines are written . in this configuration , ligo can observe binaries of stellar - mass black holes ( bhs)one of the most interesting sources to be detected of masses @xmath4 out to a distance of @xmath5 gpc ( a redshift of @xmath6 ) , and with the final advanced ( aligo ) configuration about a factor 3 farther away ( see @xcite and @xcite ) . on the other hand , the virgo interferometer is currently undergoing an upgrade program to improve its strain sensitivity , and will be online in the advanced configuration in the next future . the first direct detection of gws is imminent , provided the number of sources in the observable volume of aligo is big enough . the fundamental low - frequency limitations of the second - generation detectors are given by thermal , gravity gradient , and seismic noise . to circumvent these problems , yet a third generation of gravitational - wave ( gw ) interferometers to be operated underground is currently being proposed . the einstein telescope will be a 10 km laser - interferometer with a sensitivity 100 times better than that of the current detectors , which expands the observable volume of the universe by a factor of a million @xcite . moreover it will cover the frequency range between 1 hz and @xmath7 hz . for the search , the availability of accurate waveform models for the full merger is pivotal . numerical relativity succeeded ten years ago in simulating the late inspiral , merger and ringdown @xcite . together with post - newtonian modeling of the inspiral phase , data analysts can do faithful searches for binaries of bhs of comparable - mass and with mass ratios up to about 10 ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? nonetheless , the high cost of the development of the many ( @xmath8 ) waveforms necessary for a matched - filtering search represents a problem , if not a true limitation . hence , to speed up the production of waveforms , certain values are assumed for some of the physical parameters . these values are not chosen randomly , but rely on electromagnetic observations or theoretical modeling , which represents an emergent symbiotic relationship between astrophysics and gw searches we reduce the spectrum of the parameter space based on our best understanding of the astrophysics of the system and , once the first detections arrive , they will help us to understand our astrophysical models better . following this line of thought , in this paper we show that there is a correlation between the mass of bhs with large masses , which we call `` hyperstellar '' ( henceforth hsb ) ) , ( b ) stellar - mass ( @xmath9 ) or ( c ) intermediate - massive ( imbh , @xmath10\,m_{\odot}$ ] ) . the range of masses we are interested in is @xmath11 . since these bhs are too heavy compared to the nominal mass of an stellar - mass bh but well below the mass of an imbh , we choose the name `` hyperstellar '' to avoid confusion . ] , and their spin . hsbs are expected to form in low - metallicity environments following the collapse of very - massive stars , as in the works of @xcite , and have been suggested to be the central engines of those ultra - luminous x - ray sources found in star - forming galaxies ( @xcite , who call this sort of bhs `` massive stellar black holes '' ) . _ we show that bhs with masses _ @xmath12 _ will be detected with typically low spin values _ ( @xmath13 ) _ and on basically circular binary orbits_. binaries of compact objects in the relativistic regime detectable by ground - based interferometers like aligo can , in principle , form in two different ways ( see e.g. the living review paper by @xcite and references therein ) : ( i ) in galactic plane or bulge as the remnants of massive binary stars ( `` field binaries '' ) , or ( ii ) in dense stellar - systems , such as globular and young clusters , or those nuclear star clusters at the centers of galaxies , via dynamical interactions ( henceforth `` dynamical binaries '' ) . to have two bhs in a field binary , we need a binary formed of two stars that are both massive enough ( e.g. * ? ? ? since we need them bound , the natal kick that the bhs receive during their formation can not exceed the break - up velocity of the stellar binary . this favors the formation of binaries of hsbs because ( i ) high - mass bhs receive relatively small natal kicks @xcite and ( ii ) a higher mass requires a larger break - up velocity . for these reasons , we expect the ground - based gw experiments to detect mostly hsbs ( see * ? ? ? * for a more quantitative evaluation ) . dynamical binaries will naturally tend to form with large masses . this is so because the timescale for an object to sink towards the center of a stellar system via dynamical friction is the relaxation time divided by its mass more massive stars will sink first into a dense environment . moreover , binaries of larger masses are more difficult to separate in three - body interactions because of the larger binding energy . in particular , @xcite found with monte - carlo simulations bh masses well exceeding the nominal @xmath1 in @xmath14 of the binaries in globular clusters , and @xcite estimated with semi - analytical arguments and scattering experiments that there is a strong tendency for the merging bhs in nuclear star clusters to be biased toward high masses . this bias of forming massive binaries due to the existence of hsbs has been confirmed using direct - summation @xmath15body simulations of young clusters @xcite . as a result of the above bias , ground - based observatories are more likely to detect hsb binaries in dense clusters than in the stellar - mass range ( i.e. with masses of about @xmath16 @xcite ) . in the field , the only way to produce a hsb binary is to form , in the first place , a stellar binary of two wolf - rayet ( wr ) stars ( with stellar mass @xmath17 , following the definition of * ? ? ? * ) . an environment with low metallicity is more favorable . for example , in a solar - metallicity environment , a @xmath18 wr star produces a bh of only @xmath19 @xcite , but in an environment with @xmath20 solar metallicity the same star can produce a @xmath21 hsb ( e.g. * ? ? ? it is interesting to note that for extremely metal - poor environments ( which are uncommon in the redshift range of our interest , @xmath22 , @xcite ) , hsbs in the mass range @xmath23 are prevented from forming because of a pair - instability process during the supernova phase @xcite . because of angular - momentum loss by stellar wind , wr stars are slow rotators . the rotational velocity at the stellar surface drops from an initial value of @xmath24 to below @xmath25 during the stellar evolution @xcite . as a result , the compact remnant which is sitting at the core of a wr star retains very little of the initial angular momentum . for example , when @xmath26which is the initial stellar mass relevant to hsb formation the current stellar evolution models predict that the specific angular momentum measured at the edge of the stellar remnant will drop from @xmath27 in the initial configuration to as small as @xmath28 , and even smaller for more massive stars @xcite . since the remnant of a wr star has a low specific angular momentum , the hsb must be a slow rotator . even under the assumption that no angular momentum is lost during the collapse of the remnant and the growing of the initial seed bh that eventually becomes a hsb , the spin parameter of the final hsb is : @xmath29 where @xmath30 is the speed of light . note that because of the reason we just argued , this is an upper limit , and in a more realistic estimation , the value will be even lower . also , it is interesting to note that equation ( [ eq.spin_param ] ) gives the same spin range as what we have observed currently for the stellar - mass bhs ( @xmath31 ) in the local group @xcite . to spin up the slowly rotating hsb that we have predicted , the only two possibilities are ( i ) merger with another hsb because of a dynamical interaction , such as a three - body encounter , or ( ii ) accretion of enough material onto the hsb , about an amount of @xmath32 to achieve a result of @xmath33 . the first possibility can indeed enhance the spin value of the merger product to @xmath33 . however , this process can not take place in the field , because the stellar relaxation timescale ( i.e. encounter timescale ) exceeds by many orders of magnitude the hubble time . moreover , in the next section we will see that the outcome will not be observed because of a dynamical selection effect . regarding ( ii ) , there are two sources that can possibly provide material to feed the hsb , either the interstellar medium ( ism ) surrounding our field binary , or the companion star , i.e. the other wr star in the binary which may not have collapsed into a bh yet . the bondi - hoyle accretion rate from the ism is @xmath34 where @xmath35 is the proton density in ism and @xmath36 the velocity dispersion of stars @xcite . this feeding rate is too low to spin up the hsb , because it would require a time longer than the hubble time . on the other hand , the life span of the wr companion star is of @xmath37 myr , which is too short compared to the minimum timescale for a bh to double its mass by accreting at the maximum ( eddington ) rate . this is the salpeter timescale , which is about @xmath38 myr assuming that @xmath39 of the rest mass is converted into radiation ( see sec . 6.3 in * ? ? ? we hence deduce that _ field hsbs must be slow rotators , with lower values than the upper limit shown in eq . [ eq.spin_param ] because of loss of angular momentum during the remnant collapse_. for this section , we assume that the hsb binary forms dynamically from two hsbs that were born isolated ( if this is not the case , then the previous result applies ) and then later formed a binary via three - body interactions . there are three possible channels to form an isolated hsb : ( i ) the collapse of an isolated wr star , ( ii ) the coalescence of two less massive bhs ( e.g. @xmath40 ) and ( iii ) by accreting background stars . channel ( i ) : as we have seen in section [ sec : field ] , hsbs with masses higher than @xmath21 have typically low spins , @xmath2 . this hsb receives a natal kick typically of @xmath41^{-1}~{\rm km~s^{-1}}$ ] @xcite and is hence kept in the host cluster , because the velocity dispersion is of order @xmath42 . lighter bhs will not be retained . since we want to build binaries , it is important to retain them in the cluster . hence , most bhs with masses lighter than @xmath21 are typically bound to leave the cluster at birth , without a chance of forming a binary . channel ( ii ) : although we have just shown that lighter bhs typically abandon the host cluster at birth , there is sufficient margin so that some remain ( see e.g. the work of @xcite , although the accretion rate is so low that it is difficult to assess whether the bhs are @xmath43 or @xmath44 ) , and so we also address this channel . we point out two particular properties of dynamical binaries that will be important in determining the spin parameter , @xmath45 , of the final merged bh . ( a ) on average , the two members of a dynamical bh binary very likely will have similar masses because the three - body interaction tends to retain equal - mass binaries . this is particularly true for bhs , since they belong to the heaviest objects in clusters , and so they have lower probabilities to be ionized or replaced by another interloper ( see e.g. * ? ? ? ( b ) the majority of bh - bh mergers have a zero - eccentricity orbit , as we will see in section [ sec : ecc ] . as a result of these two points , the binary has a significant amount of angular momentum before it merges , and hence preferentially produces a high - spin hsb by the end of the merger . to prove this , we calculate the distribution of @xmath45 for different mass ratios ( @xmath46 , with @xmath47 the more massive bh ) , using the method developed by one of us and presented in @xcite . the algorithm chooses a random orientation for the spin axis of each progenitor , and then , given the mass ratio @xmath48 ( @xmath49 in our model ) of the two bhs , computes @xmath45 and the relativistic recoiling velocity , @xmath50 , according to the empirical fitting formulae derived from numerical relativity simulations @xcite . the top panel of figure [ fig : merger ] shows our results , where we have assumed a random distribution for the spin magnitudes in the range @xmath51 $ ] for the two progenitors . we can see that when @xmath52 , the value of @xmath53 is most likely greater than @xmath54 , which correlates into a very large relativistic recoiling velocity ( see lower panel of fig . [ fig : merger ] ) , that , for all purposes , is much higher than the average dispersion velocity of the host cluster , typically of @xmath42 . the excess of recoiling velocity relative to the velocity dispersion of the host stellar system is also true even for nuclear star clusters ( nsc , dense stellar clusters found at the center of galaxies of all hubble types , see e.g. @xcite ) , whose dispersion velocities are around @xmath55 , see e.g. @xcite . hence , even if the few remaining lighter bhs led to the formation of high - spin hsbs in clusters , these are immediately kicked out of the host environment because of the relativistic kick , and do not stand a chance of forming a binary . channel ( iii ) : by accreting background stars , a bh with a mass initially well below @xmath21 can not become a hsb . this is because the accretion rate from other stars in a very dense environment , such as a nsc , is @xmath56 where @xmath57 is the critical radius for the bh to tidally disrupt a solar - type star and @xmath58 is the spatial density of background stars . hence , in this channel we can not form hsbs . therefore hsbs retained in globular clusters or nsc predominantly have slow spin ( @xmath59 ) , and hence _ dynamical bh binaries detected with ground - based detectors will be hsbs with low spin . _ field binaries will have low eccentricities when they enter the detector band because their orbits circularize during the binary - star evolution due to roche lobe overflow , common envelope phase and tidal synchronization ( see e.g. * ? ? ? * ; * ? ? ? * ) and , much later in the evolution , due to gw radiation . for dynamical binaries , however , the eccentricities in the detector band are more uncertain , because of dynamical interactions with other stars . for instance , a binary - binary interaction can lead to the existence of a third body gravitationally bound to and interacting with the binary ( though the probability to form a triple via single - binary interaction is very low , at least for bodies with the same masses , see @xcite and references therein ) . for this particular situation , the interaction can lead to a large oscillation of the orbital eccentricity of the binary because of the kozai - lidov resonance ( see e.g. * ? ? ? * ) . in this case , as many as @xmath60 of mergers in the detector band could have an eccentricity larger than @xmath61 @xcite . these results hold for bound triple systems in which all bodies have a similar or equal mass in the stellar - mass bh range ( i.e. around @xmath1 ) . however , direct - summation @xmath15body simulations that do not make a priori assumptions about mass ratios and bound systems , find that bh - bh mergers are preferentially circular ( @xmath62 ) and driven by interactions with unbound small bodies ( which is referred to as hardening in stellar dynamics ) @xcite . unfortunately , such direct - summation simulations are very expensive and not suited for statistical studies . hence , to statistically address the evolution of the eccentricities of the hsb binaries during three - body interactions , we use the numerical tool fewbody @xcite . moreover , we have modified it by adding relativistic corrections to the force term , which might be important in the kind of interactions that we are interested in ( hsb binaries ) . for this , we have chosen the centre - of - mass frame of the binary for the post - newtonian ( @xmath63 hereafter ) expressions , which is equivalent to the center - of - mass hamiltonian in adm ( arnowit , deser and misner ) coordinates thanks to a transformation of the particles variables , following the expressions given by @xcite . this means that the relative acceleration in the center - of - mass frame has the form @xmath64 + { \cal o}\left ( \frac{1}{c^7 } \right)\ ; , \label{eq.blanchet}\ ] ] where the relative separation of the binary is @xmath65 , @xmath66 and @xmath67 ; the @xmath68 and @xmath69 are given by the expressions ( 3.10a ) and ( 3.10b ) in @xcite . we truncate the series and neglect all terms of order higher than 2.5@xmath63 , since e.g. the 3@xmath63 correction requires a very expensive computation and provides us only with a correction which is negligible for the purpose of this study . on the other hand , the 3.5@xmath63 term , even if it is significantly less challenging , would provide us only with a rough ( @xmath70 ) estimate of the gravitational recoil velocity . in other words , we focus in this work only on the relativistic corrections corresponding to periastron shift ( 1@xmath63 and 2@xmath63 ) and the energy loss in the form of gw radiation ( 2.5@xmath63 ) . the first implementation of relativistic terms in a dynamical code was done by one of us and they have been well tested in a number of different works that feature different orders in the expansion ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the initial conditions for all experiments are a binary of two hsbs plus an unbound interloper . we note that during the interactions , there are long - lived hierarchical triples forming , which are hence also included in the experiments . we assume a relative velocity corresponding to the velocity dispersion of a nsc , @xmath71 . initially the code calculates @xmath72 , the impact parameter corresponding to a classical , point - mass closest approach of @xmath73 . it then performs scattering experiments with @xmath74 increasing from near zero , following a distribution uniform in area , @xmath75 . it continues performing classical , point - mass closest approach of @xmath76 , where @xmath77 is the initial semimajor axis of the hsb binary . it then performs scattering experiments with @xmath74 increasing from near zero . it continues performing experiments as long as @xmath74 is less than @xmath72 , or b is less than two times the last @xmath74 for which the encounter was either a recordable event or a resonant encounter . the reason for the latter constraint on @xmath74 is to ensure that any interesting encounters that might occur at large @xmath74 are not missed . the masses of the hsbs are set to @xmath78 . in a dense stellar environment such as a nsc , a bh binary becomes hard when its kinetic energy per unit mass , @xmath79 ( with @xmath80 the reduced mass of the binary ) , becomes greater than the kinetic energy per unit mass of the field stars , @xmath81 . this definition of a hard binary gives a critical semi - major axis , @xmath82 , which is about @xmath83 au when @xmath84 and @xmath85 . a genuinely hard binary ( i.e. one with binding energy @xmath86 , with @xmath87 the average thermal energy of stars in the system , see e.g. @xcite ) should have a semi - major axis @xmath77 about ten times smaller than the above critical value . hence , for the simulations we consider only @xmath88 au . for each binary we assume a single encounter with an interloper . the reason why we do not consider successive encounters can be seen from eq . ( [ eq.gamma ] ) : the encounter rate for a hsb binary ( of total mass @xmath89 and a semimajor axis of 0.1 au ) with another star is @xmath90 , while the rate to interact with an hsb interloper is @xmath91 . i.e. for a hubble time the binary will interact with 2 stars and with no other hsb . to first order , the initial mass function ( imf ) can be approximated by two well - separated mass scales . we hence consider interlopers of masses @xmath92 ( representing within order or magnitude main sequence stars , white dwarfs and neutron stars ) and @xmath93 ( stellar - mass black holes ) . the relative abundance ( i.e. the number fraction ) of objects in these mass ranges after a relaxation time is dominated by the lighter stars @xcite , and hence we use a number fraction of heavy mass particles of @xmath94 , with @xmath95 the abundance of light stars . we ran @xmath7 samples for each of the experiments to do a statistical study . the code scans different orbital parameters , such as the initial relative velocity and impact parameter , and integrates the system until it has been dynamically solved . this means that the three - body interaction has finished and we are then left with either a new binary with new orbital elements and one of the particles has been ejected , or the binary has coalesced during the three - body interaction because it was set on a very radial orbit and the pericenter was smaller than the semi - major axis defined for coalescence ( see below ) . we then filter the results looking for binaries that are already in the detector band . for instance , for aligo this means that we look for systems with an orbital period of @xmath96 seconds so that the gravitational wave frequency ( the double of the orbital frequency ) , should be @xmath97 . for those systems we calculate the eccentricities at both detector - entrance and coalescence . we assume as a first - order approximation that the systems coalesce if @xmath98 , where @xmath99 is the schwarzschild radius ( for instance , for a @xmath100 , @xmath101 ) . even if from a dynamical point of view the three - body interaction has finished before the binary enters the detector band , we want to know how many will be in the detector band within a hubble time . for this , we evolve the orbital parameters from the last snapshot in the evolution that we have from the numerical code , and we evolve them with an approximation of keplerian orbits , as in @xcite . the equations evolve a system under the assumption that gravitational radiation is the only source of shrinkage of the semi - major axis of the binary , which is valid in our situation , since we do not allow the binary to interact with any other interloper after the first three - body scattering process . we then evolve them until they reach @xmath102 , provided this happens in within a hubble time . of those that enter into the detector band , we record the eccentricity . the calculations were halted whenever the orbital speed of binary reached @xmath103 c ( the speed of light ) , since the @xmath63 expansion is not valid for higher velocities . furthermore , once that point in the evolution is reached , the binary will always coalesce . the description of the parameter space of the simulations given below corresponds to that point , which we refer to hereafter as the _ merger point _ in the evolution of the binary , even if it does not strictly corresponds to the real coalescence of the objects . to speed up calculations , the code analytically treats weakly tidally perturbed binaries : any hierarchies are analytically treated whenever they are tidally perturbed less than the tidal perturbation tolerance . this means that in a hierarchical triple in which the eccentricity of the outer binary is very large , the inner binary could be treated analytically at each apocenter passage in the orbit provided it is not strongly tidally perturbed ( for more details , see section 3.3.4 of * ? ? ? * ) . to make this feature work with @xmath63 gravity , we added the proviso that a weakly tidally perturbed binary must also have orbital speed at pericenter smaller than @xmath104 to be treated analytically . we typically take @xmath105 , which provides a good compromise between accuracy and computational speed . in fig . [ fig.hist_circular ] we display the distribution of eccentricities for the scattering experiments at detector entrance . we can readily see that there is a trend towards higher eccentricites as the binary is tighter because the tighter the binary , the shorter the time to coalesce , and the shorter time for gw radiation to reduce the eccentricity . we also explore binaries that initially are more eccentric , as displayed in figs . ( [ fig.hist_ecc_a1e-3 ] , [ fig.hist_ecc_a1e-4 ] ) , initially starting with eccentricities of 0.3 , 0.5 , 0.7 , and 0.9 . for these , only a negligible fraction of the experiments with a high initial eccentricity and a very tight semimajor axis achieves a final eccentricity of about @xmath106 and a very few events ( about @xmath107 of the binaries with initial @xmath108 which are very tight ) enter the detector band with @xmath109 , to completely circularise a bit before coalescence . all other results stay well below @xmath110 . the relative fraction of eccentric binaries is so low that they can be ignored even for the overoptimistic upper limit cases . current search strategies are based on templates of circular binaries , so that the size of the parameter space will be manageable . although this blinds them to most eccentric mergers @xcite , and in principle to the richer information contained in the gravitational wave signals associated with those mergers , these events are so rare that they can be safely ignored . the same applies to field binaries because , as explained before , they will also be circular . in this work we have addressed the formation channels for bh binaries in the range of observation of ground - based detectors . we find that in the two cases ( binaries forming isolated or via dynamical interactions ) the most likely is that the binary members are two bhs in the the mass range of @xmath111 , which we call `` hyperstellar black holes '' , produced by massive wr stars . we show that the two binary members , again in the two different formation channels , will be slow rotators , meaning a spin magnitude of @xmath2 . we show that there is ( so far ) no possible physical process capable of spinning a hsb up to values @xmath33 in a binary . we carry out a statical study of the eccentricity distribution for a binary of two hsbs in a dense stellar environment . for this we use a numerical newtonian code in which we have implemented relativistic correcting terms . we prove that hsb binaries will predominantely have almost circular orbits at detector entrance , so that eccentric binaries can be ignored in the searches . field bh binaries formed in isolation have been shown to also have almost circular orbits at aligo frequencies . moreover , these binaries should have similar masses . in section [ sec : cluster ] we gave reasons based in relaxation and we referenced as well the work of @xcite for the case of dynamical binaries . field binaries will also very likely have similar masses , as has been shown with binary evolution and population synthesis codes @xcite . we therefore predict that hsb binaries will be observed by ground - based detectors with similar masses , low spin magnitudes and almost zero eccentricities , regardless of where they have formed . this has the potential to speed up the searches . moreover , a binary of two @xmath100 bhs , i.e. a hsb binary , will be seen at farther distance than a binary of two @xmath1 bhs . so the increased volume will ensure that many more massive hsbs will be detected . although there is so far no evidence for the existence of hsbs , we deem the non - detection to be related to the limitation of conventional observations in the electromagnetic wavebands . stellar - origin bhs are discovered whenever a bh is accreting material from a companion star @xcite . for hsbs , the lifespan in such a configuration is short : the companion star is very likely also a wr star , since they should have similar masses , as we have seen . since the lifetime of a wr star is only 5 myr , the time window to observe a hsb binary is very short ( even shorter if the two wr stars collapse each to a hsb simultaneously ) . hence , hsbs stay virtually always dark in the electromagnetic domain . with the first detections being imminent , soon the symbiotic relation between astrophysics and data analysis that we mentioned in the introduction will be fulfilled . the upcoming detections will either confirm our prediction or rule it out , and we will hence obtain information about the birth and evolution of stellar black holes , as well as about their environments information that is virtually unaccessible to our old friend the photon . we thank craig heinke for discussions on observational aspects related to the determination of the spin , and douglas heggie for discussions about binaries . we are thankful to david j. vanecek for his advice and assistance in c , and cristin maureira - fredes for his help with python . we are indebted with bruce allen for granting access to the atlas cluster , where the simulations were performed , and to carsten aulbert for his administration . xc is supported by conicyt - chile through anillo ( act1101 ) . we thank matthew benacquista for comments on the manuscript . pas is thankful to ladislav ubr and the astronomical institute of the charles university in prague for a visit , in which this paper was finished .	gravitational waves are a prediction of general relativity , and with ground - based detectors now running in their advanced configuration , we will soon be able to measure them directly for the first time . binaries of stellar - mass black holes are among the most interesting sources for these detectors . unfortunately , the many different parameters associated with the problem make it difficult to promptly produce a large set of waveforms for the search in the data stream . to reduce the number of templates to develop , one must restrict some of the physical parameters to a certain range of values predicted by either ( electromagnetic ) observations or theoretical modeling . in this work we show that `` hyperstellar '' black holes ( hsbs ) with masses @xmath0 , i.e black holes significantly larger than the nominal @xmath1 , will have an associated low value for the spin , i.e. @xmath2 . we prove that this is true regardless of the formation channel , and that when two hsbs build a binary , each of the spin magnitudes is also low , and the binary members have similar masses . we also address the distribution of the eccentricities of hsb binaries in dense stellar systems using a large suite of three - body scattering experiments that include binary - single interactions and long - lived hierarchical systems with a highly accurate integrator , including relativistic corrections up to @xmath3 . we find that most sources in the detector band will have nearly zero eccentricities . this correlation between large , similar masses , low spin and low eccentricity will help to accelerate the searches for gravitational - wave signals . [ firstpage ]
let us consider an entangled state described by a density matrix @xmath1 on a bipartite hilbert space @xmath2 . a measurement performed on one subsystem reduces the other one according to the projection postulate . each possible outcome , say @xmath3 , occurs with probability @xmath4 , and corresponds to a different conditional state @xmath5 @xmath6\ : , \qquad \varrho_x = \frac1{p_x } \hbox{tr}_1 \left [ r \ : \pi_x \otimes i_2\right ] \label{cond}\:.\end{aligned}\ ] ] @xmath7 is the probability measure ( povm ) of the measurement ( acting on the hilbert space of the first subsystem ) and @xmath8 the identity operator on the second hilbert space . @xmath9 $ ] denotes full trace , whereas @xmath10 $ ] , @xmath11 denotes partial traces . ( [ cond ] ) shows that entanglement and conditional measurements can be powerful resources to realize ( probabilistically ) nonlinear dynamics that otherwise would not have been achievable through hamiltonian evolution in realistic media . since entanglement may be shared between two distant users ( the sender performing the measurement , and the receiver observing the conditional output ) , the inherent nonlocality of entangled states permits the _ remote preparation _ of the conditional states @xmath5 , a protocol that may be used to exchange quantum information between the two parties sending only classical bits @xcite . a different kind of remote state preparation is teleportation @xcite , where the measurement depends on an unknown reference state which may be recovered at the receiver location _ independently _ on the outcome of the measurement . in this paper , we focus our attention on continuous variable ( cv ) remote state preparation . in particular , we analyze in detail an optical scheme for remote preparation of squeezed states by realistic ( noisy ) conditional homodyining . our analysis is based on a phase - space approach , and this is motivated by the following reasons : i ) entanglement in optical cv quantum information processing is provided by the so - called twin - beam ( twb ) state of two field modes @xmath12 , @xmath13 ; the corresponding wigner function is gaussian ; ii ) trace operation corresponds to overlap integral @xcite , and the wigner function of ( realistic ) homodyne povm is also a gaussian . by wigner calculus we will be able to derive simple analytical formulas for conditional outputs , also in the case of noisy measurement at the sender location . in addition , we will show that phase - space approach is a convenient framework to describe cv teleportation as a conditional measurement , and to evaluate relevant degrading effects , such the finite amount of entanglement , the losses along the transmission channel , and the nonunit quantum efficiency at the sender location . twb is the maximally entangled state ( for a given , finite , value of energy ) of two modes of radiation . it can be produced either by mixing two single - mode squeezed vacuum ( with orthogonal squeezing phases ) in a balanced beam splitter @xcite or , from the vacuum , by spontaneous downconversion in a nondegenerate parametric optical amplifier ( nopa ) @xcite . the evolution operator of the nopa reads as follows @xmath14}$ ] where the `` gain '' @xmath15 is proportional to the interaction - time , the nonlinear susceptibility , and the pump intensity . we have @xmath16 , whereas the number of photons of twb is given by @xmath17 . in view of the duality squeezing / entanglement via balanced beam - splitter @xcite the parameter @xmath15 is sometimes referred to as the squeezing parameter of the twin - beam . throughout the paper we will refer to mode @xmath18 as `` mode 1 '' and to mode @xmath19 as `` mode 2 '' . the wigner function @xmath20(x_1,y_1;x_2,y_2)$ ] of a twb is gaussian , and is given by ( we omit the argument ) @xmath21=\left(2\pi \sigma_+^2 \ : 2\pi \sigma_-^2\right)^{-1}\ : \exp\left[-\frac{(x_1+x_2)^2}{4\sigma_+^2 } -\frac{(y_1+y_2)^2}{4\sigma_-^2 } -\frac{(x_1-x_2)^2}{4\sigma_-^2 } -\frac{(y_1-y_2)^2}{4\sigma_+^2}\right]\ ] ] where the variances are given by @xmath22 specializing eq . ( [ cond ] ) for @xmath23 we have @xmath24\nonumber \\ \varrho_x & = & \frac1{p_x } \hbox{tr}_1 \left[\|\lambda\rangle\rangle\langle\langle \lambda\|\ : \pi_x\otimes i_2\right ] \label{cond1}\;,\end{aligned}\ ] ] where , in the expression of @xmath4 , we have already performed the trace over the hilbert space @xmath25 . in the following , the partial traces in eq . ( [ cond1 ] ) will be evaluated as overlap integrals in the phase space . the wigner function of a generic operator @xmath26 is defined as the following complex fourier transform @xmath27(\alpha ) & = & \int \frac{d^2\gamma}{\pi^2 } e^{\alpha\bar\gamma - \bar\alpha\gamma}\ : \hbox{tr}\left[o\ : d(\gamma)\right ] \label{wigs}\;,\end{aligned}\ ] ] where @xmath28 is a complex number , and @xmath29 is the displacement operator . the inverse transformation reads as follows @xcite @xmath30 ( \alpha)\ : e^{-2 \|\alpha\|^2 } \ : e^{2\alpha a^\dag } \left(-\right)^{a^\dag a } e^{2\bar\alpha a } \label{invwig}\;\end{aligned}\ ] ] using the wigner function the trace between two operators can be written as @xmath31 & = & \pi \int d^2\beta\:w[o_1](\beta)\:w[o_2](\beta ) \label{trace}\;.\end{aligned}\ ] ] let us consider the optical scheme depicted in fig . [ f : rsp ] . a twb is produced by spontaneous downconversion in a nopa , and then homodyne detection is performed on one of the two modes , say mode @xmath32 . the povm of the measurement , assuming perfect detection _ i.e. _ unit quantum efficiency , is given by @xmath33 @xmath34 s being eigenstates of the quadrature operator @xmath35 . the wigner function of the povm @xmath7 is a delta function @xmath36 ( x_1 ) = \delta ( x_1-x ) \label{pomx}\;,\end{aligned}\ ] ] whereas that of the term @xmath37 in the first of eqs . ( [ cond1 ] ) is given by @xmath38 ( x_1,y_1 ) & = & ( 2\pi\sigma^2)^{-1 } \exp\left\{-\frac{x_1 ^ 2+y_1 ^ 2}{2\sigma^2}\right\ } \label{lam}\:,\end{aligned}\ ] ] where the variance @xmath39 depends on the number of photons of the twb @xmath40 . using eqs . ( [ pomx ] ) and ( [ lam ] ) it is straightforward to evaluate the probability distribution @xmath41(x_1,y_1;x_2,y_2 ) w[\pi_x ] ( x_1 ) \nonumber \\ & = & ( 1-\lambda^2 ) \!\!\int\!\!\int\!\!\ dx_1 dy_1 \ : w[\lambda^{a^\dag a}](x_1,y_1)\:w[\pi_x ] ( x_1 ) \nonumber \\ & = & ( 2\pi\sigma^2)^{-1/2 } \exp\left\{-\frac{x^2}{2\sigma^2}\right\ } \label{px}\;,\end{aligned}\ ] ] and the wigner function of the conditional output state @xmath42 ( x_2,y_2 ) & = & \!\!\int\!\!\!\!\int\!\ ! dx_1 dy_1 \ : w[\hbox{\footnotesize twb}](x_1,y_1;x_2,y_2 ) w[\pi_x ] ( x_1 ) \nonumber \\ & = & ( 2\pi\sigma_1 ^ 2 \ : 2\pi\sigma_2 ^ 2)^{-1/2 } \exp\left\{-\frac{(x_2-a_x)^2}{2\sigma_1 ^ 2 } -\frac{y_2 ^ 2}{2\sigma_2 ^ 2 } \right\ } \label{condx}\;.\end{aligned}\ ] ] the parameters in eq . ( [ condx ] ) are given by @xmath43 eqs . ( [ condx ] ) and ( [ parsx ] ) say that @xmath5 is a squeezed - coherent minimum uncertainty state of the form @xmath44 _ i.e. _ a state squeezed in the direction of the measured quadrature @xmath45 , with squeezing parameter given by @xmath46 . notice that this result is valid for any quadrature @xmath47 , and therefore the present scheme , by tuning the phase of the local oscillator in the homodyne detection , is suitable for the remote preparation of squeezed states with any desired phase of squeezing . of course , we have squeezing for @xmath5 if and only if @xmath48 _ i.e. _ if and only if entanglement is present . a question arises whether or not the remote preparation of squeezing is possible with realistic homodyne detection , _ i.e. _ with noisy measurement of the field quadrature . the povm of a homodyne detector with quantum efficiency @xmath49 is a gaussian convolution of the ideal povm @xmath50 with @xmath51 @xcite . the corresponding wigner function is given by @xmath52 ( x_1 ) = ( 2\pi\sigma_\eta^2)^{-1/2 } \exp\left\{-\frac{(x_1-x)^2}{2\sigma_\eta^2 } \right\ } \label{wighometa}\;.\end{aligned}\ ] ] using ( [ wighometa ] ) one evaluates the probability distribution and the wigner function of the conditional output state , one has @xmath53^{-1/2 } \exp\left\{-\frac{x^2}{2(\sigma^2+\sigma_\eta^2)}\right\ } \\ \fl w[\varrho_{x\eta } ] ( x_2,y_2 ) = ( 2\pi\sigma_{1\eta}^2 \ : 2\pi\sigma_{2\eta}^2)^{-1/2 } \exp\left\{-\frac{(x_2-a_{x\eta})^2}{2\sigma_{1\eta}^2 } -\frac{y_2 ^ 2}{2\sigma_{2\eta}^2 } \right\}\label{condxeta0}\;,\end{aligned}\ ] ] where @xmath54 as a matter of fact , the conditional output @xmath55 is no longer a minimum uncertainty state . however , for @xmath49 large enough , it still shows squeezing in the direction individuated by the measured quadrature _ i.e. _ @xmath56 . in order to obtain the explicit form of the conditional output state from the wigner function @xmath57 ( x_2,y_2)$ ] of eq . ( [ condxeta0 ] ) we use eq . ( [ invwig ] ) arriving at @xmath58 where @xmath59^p \ : given by @xmath60}{1+\eta n}}-1\right\ } \label{ntheta}\;,\end{aligned}\ ] ] and the squeezing parameter is given by @xmath61 we have squeezing in @xmath55 if @xmath62 , and this happens for @xmath63 independently on the actual value @xmath3 of the homodyne outcome . the values of efficiency that can be currently realized in a quantum optical lab is far above the @xmath0 limit , and thus we conclude that conditional homodyning on twb is a robust scheme for the remote preparation of squeezing . the scheme for optical cv teleportation is depicted in fig . [ f : tel ] . one part of a twb is mixed with a given reference state @xmath39 in a balanced beam splitter , and two orthogonal quadratures @xmath64 , @xmath65 are measured on the outgoing beams by means of two homodyne detectors with local oscillators phase - shifted by @xmath66 . the other part of the twb is then displaced by an amount @xmath67 that depends on the outcome of the measurements , and the resulting state ( averaged over the possible outcomes ) is the teleported state . overall , the measurement performed on the twb is a generalized double homodyne detection @xcite ( equivalent to generalized heterodyne ) , which can be described by the povm @xcite @xmath68 @xmath69 denoting transposition . therefore , using eq . ( [ cond1 ] ) , one has @xmath70\nonumber \\ \varrho_\alpha & = & \frac1{p_\alpha } d(-\alpha)\hbox{tr}_1 \left[\|\lambda\rangle\rangle\langle\langle \lambda\|\ : d(\alpha ) \sigma^t d^\dag ( \alpha ) \otimes i_2\right ] d^\dag ( -\alpha ) \label{condalpha}\;,\end{aligned}\ ] ] while the teleported state is given by @xmath71 d^\dag ( -\alpha ) \label{teleported}\;.\end{aligned}\ ] ] using wigner functions and taking into account that for any density matrix @xmath72(x , y ) = w[\varrho](x ,- y ) \\ w[d(\alpha)\varrho d^\dag ( \alpha)](x , y ) = w[\varrho](x - x_\alpha , y - y_\alpha ) \label{rems}\;,\end{aligned}\ ] ] with @xmath73 $ ] and @xmath74 $ ] , one has @xmath75(x_2,y_2 ) = \nonumber \\ \fl=\!\!\int\!\!\!\!\int\!\ ! dx_1 dy_1 \!\ ! dx_\alpha dy_\alpha \ : w[\hbox{\footnotesize twb}](x_1,y_1;x_2+x_\alpha , y_2+y_\alpha ) \ : w[\sigma ] ( x_1-x_\alpha ,- y_1-y_\alpha ) \nonumber \\ \fl=\!\!\int\!\!\!\!\int\!\ ! dx_1 dy_1\ : w[\sigma ] ( x_1,y_1 ) \int\!\!\!\!\int\!\ ! dx_\alpha dy_\alpha \ : w[\hbox{\footnotesize twb}](x_1+x_\alpha ,- y_1-y_\alpha;x_2+x_\alpha , y_2+y_\alpha ) \nonumber \\ \fl= \!\!\int\!\!\!\!\int\!\ ! \frac{dx_1 dy_1}{\pi \kappa_r^2 } \exp\left\{-\frac{(x_1-x_2)^2+(y_1-y_2)^2}{\kappa^2_r}\right\}\ : w[\sigma ] ( x_1,y_1 ) \nonumber \\ \fl= \!\!\int\!\!\!\!\int\!\ ! \frac{dx_1 dy_1}{\pi \kappa_r^2 } \exp\left\{-\frac{x_1 ^ 2+y_1 ^ 2}{\kappa^2_r}\right\}\ : w[d(\alpha_1)\sigma d^\dag ( \alpha_1 ) ] ( x_2,y_2)\label{wigtel}\;,\end{aligned}\ ] ] with @xmath76 and @xmath77 . from eqs . ( [ wigtel ] ) and ( [ invwig ] ) one has that the teleported state is given by @xmath78 which coincides with the input state only in the limit @xmath79 _ i.e. _ for infinite energy of the twb . ( [ tel ] ) shows that cv teleportation with finite amount of entanglement is equivalent to a thermalizing channel with @xmath80 thermal photons : this results has been obtained also with other methods @xcite . however , the present wigner approach may be more convenient in order to include other degrading effects such the nonunit quantum efficiency at the sender location and the losses along the transmission channel . nonunit quantum efficiency at the homodyne detectors affects the povm of the sender , which become a gaussian convolution of the ideal povm @xmath81 @xmath82 with @xmath83 @xcite . on the other hand , losses along the line degrade the entanglement of the twb supporting the teleportation . the propagation of a twb inside optical media can be modeled as the coupling of each part of the twb with a non zero temperature reservoir . the dynamics can be described in terms of the two - mode master equation @xmath84 \varrho_t + \gamma ( 1+m ) l[b ] \varrho_t + \gamma m l[a^\dag ] \varrho_t + \gamma m l[b^\dag ] \varrho_t \label{master } \end{aligned}\ ] ] where @xmath85 , @xmath86 denotes the ( equal ) damping rate , @xmath87 the number of background thermal photons , and @xmath88 $ ] is the lindblad superoperator @xmath89 \varrho_t = o \varrho_t o^\dag - \frac{1}{2 } o^\dag o \varrho_t - \frac{1}{2 } \varrho_t oo^\dag\:.$ ] the terms proportional to @xmath90 $ ] and @xmath91 $ ] describe the losses , whereas the terms proportional to @xmath92 $ ] and @xmath93 $ ] describe a linear phase - insensitive amplification process . this can be due either to optical media dynamics or to thermal hopping ; in both cases no phase information is carried . of course , the dissipative dynamics of the two channels are independent on each other . the master equation ( [ master ] ) can be transformed into a fokker - planck equation for the two - mode wigner function of the twb using the differential representation of the superoperators in eq . ( [ master ] ) the corresponding fokker - planck equation reads as follows @xmath94 w_\tau \label{fp}\ : , \end{aligned}\ ] ] where @xmath95 denotes the rescaled time @xmath96 , and @xmath97 the drift term . the solution of eq . ( [ fp ] ) can be written as @xmath98(x^{\prime}_1,y^{\prime}_1;x^{\prime}_2,y^{\prime}_2)\ : \prod_{j=1}^2 g_\tau(x_j\|x^{\prime}_j ) g_\tau(y_j\|y^{\prime}_j ) \ : \label{conv}\end{aligned}\ ] ] where @xmath20 $ ] is initial wigner function of the twb , and the green functions @xmath99 are given by @xmath100 \;,\quad d^2=\frac{1}{4\gamma } ( 1-e^{-\gamma \tau } ) \;. \label{green}\end{aligned}\ ] ] the wigner function @xmath101 can be obtained by the convolution ( [ conv ] ) , which can be easily evaluated since the initial wigner function is gaussian . the form of @xmath101 is the same of @xmath20 $ ] with the variances changed to @xmath102 inserting the wigner functions of the blurred povm @xmath103 and of the evolved twb in eqs . ( [ condalpha ] ) and ( [ teleported ] ) we obtain the teleported state in the general case , which is still given by eq . ( [ tel ] ) , with the parameter @xmath80 now given by @xmath104 eqs . ( [ tel ] ) and ( [ kappar ] ) summarize all the degrading effects on the quality of the teleported state . in the special case of coherent state teleportation @xmath105 ( which corresponds to original optical cv teleportation experiments @xcite ) the fidelity @xmath106 can be evaluated straightforwardly as the overlap of the wigner functions . since @xmath107(\alpha)=2/\pi\ : e^{-2\|\alpha -z\|^2}$ ] is the wigner function of a coherent state we have @xmath108 the condition on the fidelity , in order to assure that the scheme is a truly nonlocal protocol , is given by @xmath109 @xcite , _ i.e. _ @xmath110 therefore , the bound on the quantum efficiency to demonstrate quantum teleportation is given by @xmath111 if the propagation induces low perturbation _ i.e. _ if @xmath112 and @xmath113 we have @xmath114 , which ranges from @xmath115 to @xmath32 , and represents the range of `` useful '' values for the quantum efficiency . if @xmath86 and @xmath87 are not negligible then , for the same initial squeezing , we need a larger value of the quantum efficiency . moreover , since quantum efficiency should be lower or equal to unit @xmath116 , we may derive a bound on the initial squeezing that allows to demonstrate quantum teleportation . this reads as follows @xmath117 . remarkably , if the number of thermal photons is zero , _ i.e. _ if the twb is propagating in a zero temperature environment , then any value of the initial squeezing parameter make teleportation possible , of course if the quantum efficiency at the receiver location satisfies @xmath118 . a method for the remote preparation of squeezed states by conditional homodyninig on a twb has been suggested . the scheme has been studied using wigner function , which is the most convenient approach to describe effects of nonunit quantum efficiency at homodyne detectors . the method is shown to provide remote squeezing if the quantum efficiency is larger than @xmath0 . since downconversion correlates pair of modes at any frequencies @xmath119 and @xmath120 satisfying @xmath121 , @xmath122 being the frequency of the pump beam , the present method can be used to generate squeezing at frequencies where no media for degenerate downconversion are available @xcite . phase - space approach has been also used to analyze cv teleportation as a conditional generalized double homodyning on a twb . also in this case the use of wigner functions represents a powerful tool to evaluate the degrading effects of finite amount of entanglement , losses along the transmission channel , and nonunit quantum efficiency at sender location . a bound on the value of quantum efficiency needed to demonstrate quantum teleportation has been derived . this work has been sponsored by the infm through the project pra-2002-clon , by miur through the prin projects _ decoherence control in quantum information processing _ and _ entanglement assisted high precision measurements _ , and by eec through the project ist-2000 - 29681 ( atesit ) . mgap is research fellow at _ collegio alessandro volta_. 25 a. k. pati , phys . rev . a * 63 * ( 2001 ) 014302 ; hoi - kwong lo , phys . a * 62 * , 012313 ( 2000 ) . bennett et al . , phys . lett . * 70 * , 1895 ( 1993 ) . a. furusawa et al . , science * 282 * , 706 ( 1998 ) . o. aytur , p. kumar , phys . lett . * 65 * , 1551 ( 1990 ) . k. cahill , r. glauber , phys . rev . * 177 * , 1857 ( 1969 ) ; _ ibidem _ pag 1882 . m. g. a. paris , phys . a * 225 * , 28 ( 1997 ) . g. m. dariano and m. f. sacchi , n. cim . b * 112 * , 881 ( 1997 ) . u. leonhardt and h. paul , phys . rev . a * 48 * , 4598 ( 1993 ) ; g. m. dariano et al . a * 198 * , 286 ( 1995 ) . m. g. a. paris , a. chizhov , o. steuernagel , opt . * 134 * , 117 ( 1997 ) . p. busch , p. lahti , riv . nuovo cim . * 18 * , 1 ( 1995 ) . m. g a paris , in _ vii international conference on squeezed states and uncertainty relations _ , e - book http://www.wam.umd.edu/~ys/boston.html . h. f. hofmann et al . , a 62 , 062304 ( 2000 ) ; masashi ban et qal . , preprint quant - ph/0202172 . a. andreoni et al . journ d * 13 * , 415 ( 2001 ) .	continuous variable remote state preparation and teleportation are analyzed using wigner functions in phase space . we suggest a remote squeezed state preparation scheme between two parties sharing an entangled twin beam , where homodyne detection on one beam is used as a conditional source of squeezing for the other beam . the scheme works also with noisy measurements , and provide squeezing if the homodyne quantum efficiency is larger than @xmath0 . phase space approach is shown to provide a convenient framework to describe teleportation as a generalized conditional measurement , and to evaluate relevant degrading effects , such the finite amount of entanglement , the losses along the line , and the nonunit quantum efficiency at the sender location .
the parton evolution at small values of @xmath2 is believed to be best described by the ccfm evolution equation , which for @xmath3 is equivalent to the bfkl evolution equation and for large @xmath2 reproduce the standard dglap equations . the ccfm evolution equation takes coherence effects of the radiated gluons into account via angular ordering . on the basis of this evolution equation , the monte carlo program smallx has been developed already in 1992 . in 1997 the linked dipole chain was developed as a reformulation of the original ccfm equation . predictions of the ccfm equation for hadronic final state properties were studied in @xcite paying special attention to non - leading effects . all approaches found a good description of @xmath0 but failed completely to describe the forward jet data of hera experiments , which are believed to be a signature for new small @xmath2 parton dynamics . in the following i discuss the treatment of the non - sudakov form factor @xmath1 as well as the effects of the so - called consistency constraint " , which was found to be necessary to include non - leading contributions to the bfkl equation @xcite . i show that a good description of @xmath0 and the forward jet data can be achieved . the implementation of the ccfm parton evolution in the forward evolution monte carlo program smallx is described in detail in . here i only concentrate on the basic ideas and discuss the treatment of the non - sudakov form factor . the initial state gluon cascade is generated in a forward evolution approach from a starting distribution of the @xmath4 unintegrated gluon distribution according to : @xmath5 where @xmath6 is a normalization constant and @xmath7 gev@xmath8 . gluons then can branch into a virtual ( @xmath9-channel ) gluon @xmath10 and a final gluon @xmath11 according to the ccfm splitting function : @xmath12 with @xmath13 being the rescaled transverse momentum , @xmath14 , @xmath15 being a collinear cutoff to avoid the @xmath16 singularity and @xmath17 being the sudakov form factor : @xmath18 which at an inclusive level cancels against the @xmath16 collinear singularity and is used to generate @xmath19 . the gluon splitting function @xmath20 is given by : @xmath21 with the non - sudakov form factor @xmath1 being defined as : @xmath22 which gives for the region @xmath23 : @xmath24 the constraint @xmath23 is often referred to as the consistency constraint " @xcite . however the upper limit of the @xmath25 integral is constraint by the @xmath26 functions in eq.([ns ] ) by : @xmath27 , which results in the following form of the non - sudakov form factor @xcite : @xmath28 where @xmath29 giving no supression in the region @xmath30 we have @xmath31 . the monte carlo program smallx has been modified to include the non - sudakov form factor according to eq.([ns_new ] ) and the scale in @xmath32 was changed to @xmath33 in the cascade and in the matrix element . to avoid problems at small @xmath34 , @xmath35 is restricted to @xmath36 . the consistency constraint " was introduced to account for next - to - leading effects in the bfkl equation , and which was found @xcite to simulate about 70% of the full next - to - leading corrections to the bfkl equation . since in lo bfkl the true kinematics of the branchings are neglected , they can be interpreted as next - to - leading effects , and this constraint is often also called kinematic constraint " . in the ccfm equation energy and momentum conservation is already included at lo , and it is not clear , whether the arguments coming from bfkl also apply to ccfm . in the following the effects of the @xmath16 terms and the consistency constraint " are studied in more detail . with the modifications on the treatment of the non - sudakov form factors described above , the predictions of the ccfm evolution equation , as implemented in the program smallx , for the structure function @xmath37 are shown in fig . [ f2_1 ] without applying any additional consistency constraint " . -15 mm -10 mm here the following parameters were used : @xmath38 gev , @xmath39 gev , @xmath40 and the masses for light ( charm ) quarks were set to @xmath41 gev . the scale of @xmath32 in the off - shell matrix element was set to @xmath34 . with this parameter settings are very good description of @xmath0 over the range @xmath42 and @xmath43 gev@xmath8 is obtained . in fig . [ fwd_jets_1 ] the prediction for the forward jets using the same parameter setting is shown . the data are nicely described . -12 mm -8 mm the effect of the @xmath16 term in the splitting function is shown in figs . [ f2_1 ] and [ fwd_jets_1 ] separately with the dashed line . it is obvious that these terms are important for a reasonable description of @xmath0 and the forward jet data . including the consistency constraint " the @xmath2 dependence of the cross section changes , but a similarly good description of @xmath0 is obtained by changing @xmath44 to @xmath45 gev . however the forward jet cross section becomes smaller , as shown with the dotted line in figs . [ f2_1 ] and [ fwd_jets_1 ] . it is interesting to note , that this prediction is very similar to the one obtained from the bfkl equation as shown in @xcite for @xmath34 as the scale in @xmath32 . i am very grateful to b. webber for providing me with the code of smallx . i am grateful to b. andersson , g. gustafson , h. kharraziha , j. kwiecinski , l. lnnblad , a. martin , s. munier , r. peschanski and g. salam for many very helpful discussions about ccfm . 10 m. ciafaloni , _ nucl . phys . * b * _ * 296 * ( 1988 ) 49 + s. catani , f. fiorani , g. marchesini , _ phys . lett . * b * _ * 234 * ( 1990 ) 339 and _ nucl . phys . * b * _ * 336 * ( 1990 ) 18 + g. marchesini , _ nucl . phys . * b * _ * 445 * ( 1995 ) 49 . e. kuraev , l. lipatov , v. fadin , _ sov . * 44 * ( 1976 ) 443 and _ sov . . jetp _ * 45 * ( 1977 ) 199 + y. balitskii , l. lipatov , _ sov . _ * 28 * ( 1978 ) 822 . g. marchesini , b. webber , _ nucl . phys . * b * _ * 349 * ( 1991 ) 617 and _ nucl . phys . * b * _ * 386 * ( 1992 ) 215 . b. andersson , g. gustafson , j. samuelsson , _ nucl . * b * _ * 467 * ( 1996 ) 443 + b. andersson , g. gustafson , h. kharraziha , j. samuelsson , _ z. phys . * c * _ * 71 * ( 1996 ) 613 + g. gustafson , h. kharraziha , l. lnnblad , _ jhep _ * 98003:006 * ( 1998 ) . g. bottazi , g. marchesini , g. salam , m. scorletti , _ jhep _ * 9812:11 * ( 1998 ) , hep - ph/9810546 . j. kwiecinski , a. martin , j. outhwaite , ( 1999 ) , hep - ph/9903439 . j. kwiecinski , a. martin , p. sutton , _ phys . rev . * d * _ * 52 * ( 1995 ) 1445 . , _ nucl . * b * _ * 470 * ( 1996 ) 3 . , _ nucl . phys . * b * _ * 538 * ( 1999 ) 3 . ; j. breitweg et al . j. * c * _ * 6 * ( 1999 ) 239 .	predictions of the ccfm evolution equation for @xmath0 and forward jets at hera energies are obtained from a modified version of the monte carlo program smallx . the treatment of the non - sudakov form factor @xmath1 is discussed as well as the effect of the so called consistency constraint " . for the first time a good description of @xmath0 and the forward jet data is obtained from the ccfm equation .
there is growing evidence for the existence of so - called magnetars , supermagnetized neutron stars with the magnetic fields of @xmath0 g ( e.g. , @xcite ) . although such stars are estimated to be only a subclass ( @xmath1 ) of the canonical neutron stars with @xmath2 g @xcite , much attention has been paid to the objects because they pose many astrophysically exciting but unresolved problems . giant flaring activities in the soft gamma repeaters observed for the last two decades have given us good opportunities to study the coupling of the interior to the magnetospheric structures of magnetars @xcite , but the relationship between the crustal fraction and the subsequent star quarks has not yet been clarified ( see references in @xcite ) . the origin of the large magnetic field is also a big problem , whether generated at post - collapse in the rapidly rotating neutron star @xcite or descended from the main sequence stars @xcite . assuming large magnetic fields before core - collapse , researches have recently performed extensive magnetohydrodynamic ( mhd ) stellar collapse simulations @xcite in order to understand the formation mechanism of magnetars , with different levels of sophistication in the treatment of equations of state ( eoss ) , the neutrino transport , and general relativity ( see @xcite for a review ) . here it is worth mentioning that the gravitational waves from magnetars could be a useful source of information about magnetar interiors @xcite . from a microscopic point of view , the effects of magnetic field larger than the so - called qed limit of @xmath3 , on the eoss ( e.g. , @xcite ) and the radiation processes have been also extensively investigated ( see @xcite for a review ) . for the understanding of the formation and evolution of the magnetars , it is indispensable to unify these macroscopic and microscopic studies , albeit not an easy job . when one tries to study the above interesting issues , the construction of an equilibrium configuration of magnetars may be one of the most fundamental problems . in spite of the many magnetized star investigations that have so far been carried out , with different levels of sophistication in the treatment of the magnetic field structure , equations of state ( eoss ) , and the general relativity @xcite , there still remains room for more sophistication in the previous studies in which stellar structures are obtained within the framework of newtonian dynamics . since we are interested in magnetar structures , the general relativistic treatment due to the strong gravity of magnetars is required in constructions of a magnetar model . however , a fully general relativistic means constructing a stellar model with arbitrarily magnetic structure are still not available . in most studies on the structures of relativistic magnetized stars , weak magnetic fields and/or purely poloidal magnetic fields have been assumed . bocquet et al . @xcite and cardall et al . @xcite have treated relativistic stellar models containing purely poloidal magnetic fields . konno et al . @xcite have analyzed similar models using a perturbative approach . ioka and sasaki have investigated structures of mixed poloidal - toroidal magnetic fields around a spherical star by using a perturbative technique @xcite . as shown in core - collapse supernova mhd simulations ( see , e.g. , @xcite ) , the toroidal magnetic fields can be easily amplified because of the winding up of the initial seed poloidal fields as long as the core is rotating differentially . after core bounce , as a result , the toroidal fields dominate over the poloidal ones in some proto - neutron star models even if there is no toroidal field initially . this winding up amplification of the toroidal magnetic fields therefore indicates that some neutron stars could have toroidal fields much higher than poloidal ones . as mentioned before , however , no effect of a strong toroidal magnetic field has been taken into account in relativistic magnetized stellar models . in order to elucidate the effects of strong toroidal magnetic fields on neutron star structures , in this paper , we concentrate on magnetic effects due to purely toroidal fields on structures of relativistic stars as a first step . within the framework of the newtonian dynamics , miketinac has obtained highly magnetized stars containing purely toroidal magnetic fields @xcite . in this study , we extend miketinac s models so as to include the effects of general relativity and rotation . first of all , we derive master equations for obtaining relativistic rotating stellar models having purely toroidal magnetic fields . the master equations obtained are converted into integral expressions and solved numerically with a self - consistent field scheme similar to the komatsu - eriguchi - hachisu scheme for equilibrium structures of relativistic rotating stars @xcite . using the numerical scheme written in the present study , we construct a large number of equilibrium stars for a wide range of parameters . as shown by trehan & uberoi @xcite , the toroidal magnetic field tends to distort a stellar shape prolately . this is because the toroidal magnetic field lines behave like a rubber belt , pulling in the matter around the magnetic axis . such a prolate - shaped neutron star may be an optimal source of gravitational waves if the star rotates @xcite . in this gravitational wave emission mechanism , one important parameter is degree of prolateness , which determines the amount of gravitational radiation . thus , we pay attention to the deformation of the star . we also construct sequences of the equilibrium stars along which the total baryon rest mass and/or the magnetic flux with respect to the meridional cross - section keep constant . these equilibrium sequences are available for exploring evolutionary sequences because the total baryon rest mass and the magnetic flux are well conserved in quasi - steady evolution of isolated neutron stars . this paper is organized as follows . the master equations for obtaining equilibrium configurations of rotating stars with purely toroidal magnetic fields are derived in sec . [ sec : basic ] . the numerical scheme for computing the stellar models is briefly described in sec . [ sec : numsch ] . [ sec : result ] is devoted to showing numerical results . summary and discussion follow in sec . [ sec : summary ] . in this paper , we use geometrical units with @xmath4 . the rotating relativistic stars containing purely toroidal magnetic fields are considered in this paper . assumptions to obtain the equilibrium models are summarized as follows ; ( 1 ) equilibrium models are stationary and axisymmetric , i.e. the spacetime has the time killing vector , @xmath5 , and the rotational killing vector , @xmath6 , and lie derivatives of the physical quantities of the equilibrium models with respect to these killing vectors vanish . ( 2 ) the matter source is approximated by a perfect fluid with infinite conductivity . ( 3 ) there is no meridional flow of the matter . ( 4 ) the equation of state for the matter is barotropic . by virtue of assumption ( 3 ) , the matter four - velocity is reduced to @xmath7 where @xmath8 is the angular velocity of the matter . we further assume that the magnetic field of the star has toroidal parts only , so that the faraday tensor , @xmath9 , obeys the conditions , given by @xmath10 note that since eq . ( [ eq : fraday ] ) includes five independent conditions , we see that the faraday tensor has one independent component only . thanks to the conditions assumed above , one of the maxwell equations and the perfect - conductivity condition , @xmath11 } = 0,\label{eq : maxwel}\\ & & f_{\mu\nu}u^\nu = 0,\label{eq : mhd}\end{aligned}\ ] ] are automatically satisfied , and we need not consider them any further . the stress - energy tensor of the perfectly conductive fluid is given by @xmath12 , \label{eq : stress - tensor}\end{aligned}\ ] ] where @xmath13 is the rest energy density , @xmath14 the specific internal energy , @xmath15 the pressure , and @xmath16 the inverse of the metric @xmath17 . here , we have defined the magnetic field measured by the observer with the fluid four - velocity @xmath18 as @xmath19 where @xmath20 stands for the contravariant levi - civita tensor with @xmath21 . here @xmath22 denotes the determinant of the metric . by using eqs . ( [ eq:4-vel])([eq : fraday ] ) , ( [ eq : mhd])([eq : stress - tensor ] ) , and ( [ eq : bfield ] ) , we can prove that the circularity condition , i.e. , @xmath23 } = \varphi_\alpha t^{\alpha[\beta } t^\gamma \varphi^{\delta ] } = 0 $ ] , is satisfied in the present situation @xcite . note that a similar argument about the circularity has been given by oron @xcite . therefore , there exists a family of two - surfaces everywhere orthogonal to the plane defined by the two killing vectors @xmath5 and @xmath6 @xcite , and the metric can be written , following @xcite , in the form where the metric potentials , @xmath25 , @xmath26 , @xmath27 , and @xmath28 , are functions of @xmath29 and @xmath30 only . we see that the non - zero component of @xmath9 in this coordinate is @xmath31 . it is convenient to define the determinants of two two - dimensional subspaces for later discussion , where @xmath39 is the specific enthalpy and we have assumed the uniform rotation , i.e. , @xmath40 . note that although the extension to the case of the differential rotation is straightforward , we take the uniform rotation law for the sake of simplicity . integrability of eq . ( [ eq : euler2 ] ) requires with the metric ( [ eq : metric ] ) , the four - velocity ( [ eq:4-vel ] ) , the proper velocity of the matter with respect to a zero angular momentum observer ( zamo ) , @xmath47 , the components of the magnetic field ( [ eq : comp - mag1])-([eq : comp - mag2 ] ) , and the hydrostatic equilibrium equation ( [ eq : ber ] ) can be , respectively , written as @xmath48 following @xcite , we may write the general relativistic field equations determining @xmath26 , @xmath25 , and @xmath28 as @xmath49 = s_\rho(r,\mu),\label{eq : basic - rho}\\ & & \left ( \nabla^2 + \frac{1}{r}\partial_r - \frac{\mu}{r^2}\partial_\mu \right ) [ \gamma { \rm e}^{\gamma/2 } ] = s_\gamma ( r,\mu),\label{eq : basic - gam}\\ & & \left ( \nabla^2 + \frac{2}{r}\partial_r - \frac{2\mu}{r^2}\partial_\mu \right ) [ \omega { \rm e}^{(\gamma-2\rho)/2 } ] = s_\omega ( r,\mu ) , \label{eq : basic - omg}\end{aligned}\ ] ] where @xmath50 is the laplacian of flat 3-space , and @xmath51 . here , @xmath52 , @xmath53 , and @xmath54 are given by @xmath55 \big \},\label{eq : soc - rho}\\ & & s_\gamma(r,\mu ) = { \rm e}^{\gamma/2 } \left\ { 16 \pi { \rm e}^{2\alpha } \left ( p + \frac{\|b\|^2}{8\pi } \right ) + \frac{\gamma}{2 } \left [ 16 \pi { \rm e}^{2\alpha } \left ( p + \frac{\|b\|^2}{8\pi } \right ) - \frac{1}{2}\gamma_{,r}^2 - \frac{1-\mu^2}{2r^2}\gamma_{,\mu}^2\right ] \right \},\label{eq : soc - gam}\\ & & s_\omega(r,\mu ) = { \rm e}^{(\gamma-2\rho)/2 } \big\ { - 16\pi { \rm e}^{2\alpha } \frac{(\omega-\omega)(\rho_0 + \rho_0 e + p ) } { 1-v^2 } \nonumber\\ & & + \omega \big [ - 8 \pi { \rm e}^{2\alpha } \left ( \frac{(\rho_0 + \rho_0 e + p ) ( 1 + v^2 ) } { 1-v^2 } - p - \frac{\|b\|^2}{8\pi } \right ) - \frac{1}{r } \left ( 2 \rho_{,r } + \frac{1}{2}\gamma_{,r}\right ) + \frac{\mu}{r^2 } \left ( 2 \rho_{,\mu } + \frac{1}{2 } \gamma_{,\mu } \right ) \nonumber\\ & & + \frac{1}{4 } ( 4\rho_{,r}^2 - \gamma_{,r}^2 ) + \frac{1-\mu^2}{4r^2 } ( 4\rho_{,\mu}^2 - \gamma_{,\mu}^2 ) - r^2 ( 1 - \mu^2 ) { \rm e}^{-2\rho } \left ( \omega_{,r}^2 + \frac{1-\mu^2}{r^2}\omega_{,\mu}^2 \right ) \big ] \big\ } , \label{eq : soc - omg}\end{aligned}\ ] ] where @xmath56 denotes the magnitude of the magnetic field , given by latexmath:[\ ] ] are satisfied because of the maxwell equation and infinite conductivity of the fluid , we have @xmath160 where @xmath161 stands for the lie differentiation with respect to @xmath18 , and @xmath27 denotes an arbitrary function . this implies that the flux of @xmath9 is conserved along any tube generated by curves parallel to @xmath18 . in other words , the integral of @xmath9 , defined as @xmath162 where @xmath163 is a two - dimensional surface co - moving with the fluid flow parametrized by its proper time @xmath164 , is conserved in the sense that @xmath165 in the present situation , we can take as @xmath163 the two - surface orthogonal to the plane defined by the two killing vectors parametrized by @xmath29 and @xmath30 . for the present situation , we can then reduce @xmath88 into @xmath166 which is frequently called the magnetic flux .	we investigate the effects of the purely toroidal magnetic field on the equilibrium structures of the relativistic stars . the master equations for obtaining equilibrium solutions of relativistic rotating stars containing purely toroidal magnetic fields are derived for the first time . to solve these master equations numerically , we extend the cook - shapiro - teukolsky scheme for calculating relativistic rotating stars containing no magnetic field to incorporate the effects of the purely toroidal magnetic fields . by using the numerical scheme , we then calculate a large number of the equilibrium configurations for a particular distribution of the magnetic field in order to explore the equilibrium properties . we also construct the equilibrium sequences of the constant baryon mass and/or the constant magnetic flux , which model the evolution of an isolated neutron star as it loses angular momentum via the gravitational waves . important properties of the equilibrium configurations of the magnetized stars obtained in this study are summarized as follows ; ( 1 ) for the non - rotating stars , the matter distribution of the stars is prolately distorted due to the toroidal magnetic fields . ( 2 ) for the rapidly rotating stars , the shape of the stellar surface becomes oblate because of the centrifugal force . but , the matter distribution deep inside the star is sufficiently prolate for the mean matter distribution of the star to be prolate . ( 3 ) the stronger toroidal magnetic fields lead to the mass - shedding of the stars at the lower angular velocity . ( 4 ) for some equilibrium sequences of the constant baryon mass and magnetic flux , the stars can spin up as they lose angular momentum .
in this result , we develop the dipsersive estimates used to prove stability of solitons for a focusing , saturated nonlinear schrdinger equation ( nls ) in @xmath6 : @xmath7 where @xmath8 , @xmath9 for all @xmath10 , @xmath11 has a specific structure outlined in the following definitions : [ def : non1 ] saturated nonlinearities of type @xmath12 are of the form @xmath13 where @xmath14 and @xmath15 for @xmath16 and @xmath17 for @xmath18 . [ def : non2 ] saturated nonlinearities of type @xmath19 are of the form @xmath20 where @xmath21 , @xmath22 . in both cases , for @xmath23 large , the behavior is @xmath24 subcritical and for @xmath23 small , the behavior is @xmath24 supercritical . for definition [ def : non1 ] , @xmath25 is chosen much larger than the @xmath24 critical exponent , @xmath26 in order to allow sufficient regularity when linearizing the equation . in the sequel , we assume that @xmath27 and @xmath28 , or in other words , @xmath29 has finite variance . for this initial data , from the spatial and phase invariance of nls , we have many the following conserved quantities : conservation of mass ( or charge ) : @xmath30 and conservation of energy : @xmath31 where @xmath32 we also have the pseudoconformal conservation law : @xmath33 where @xmath34 note that @xmath35 is the hamilton flow of the linear schrdinger equation , so the above identity relates how the solution to the nonlinear equation is effected by the linear flow . detailed proofs of these conservation laws can be arrived at easily using energy estimates or noether s theorem , which relates conservation laws to symmetries of an equation . global well - posedness in @xmath24 of ( nls ) with @xmath11 of type @xmath12 or @xmath19 for finite variance initial data follows from standard theory for @xmath24 subcritical monomial nonlinearities . proofs of the above results can be found in numerous excellent references for ( nls ) , including @xcite and @xcite . this paper is a result of a thesis done under the direction of daniel tataru at the university of california , berkeley . it is fair to say this work would not exist without his assistance . the work was supported by graduate fellowships from the university of california , berkeley and nsf grants dms0354539 and dms0301122 . in addition , the author spent a semester as an associate member of msri during the development of these results . currently , the author is supported by an nsf postdoctoral fellowship . a soliton solution is of the form @xmath36 where @xmath37 and @xmath38 is a positive , radially symmetric , exponentially decaying solution of the equation : @xmath39 with this type of nonlinearity , soliton solutions exist and are known to be unique . existence of solitary waves for nonlinearities of the type presented in definitions [ def : non1 ] and [ def : non2 ] is proved by in @xcite by minimizing the functional @xmath40 with respect to the functional @xmath41 dx.\ ] ] then , using a minimizing sequence and schwarz symmetrization , one sees the existence of the nonnegative , spherically symmetric , decreasing soliton solution . for uniqueness , see @xcite , where a shooting method is implemented to show that the desired soliton behavior only occurs for one particular initial value . an important fact is that @xmath42 and @xmath43 are differentiable with respect to @xmath44 . this fact can be determined from the early works of shatah , namely @xcite , @xcite . by differentiating equation , @xmath45 and @xmath46 with respect to @xmath44 , we have @xmath47 numerics show that if we plot @xmath48 with respect to @xmath44 , we get a curve that goes to @xmath49 as @xmath50 and has a global minimum at some @xmath51 , see figure [ f : solcurves ] . we will explore this in detail in a subsequent numerical work @xcite . variational techniques developed in @xcite and @xcite tell us that when @xmath52 is convex , or @xmath53 , we are guaranteed stability under small perturbations , while for @xmath54 we are guaranteed that the soliton is unstable under small perturbations . for brief reference on this subject , see @xcite , chapter 4 . for nonlinear instability at a minimum , see @xcite . for notational purposes , we refer to a minimal mass soliton as @xmath55 . let us write down the form of nls linearized about a soliton solution . first of all , we assume we have a solution @xmath56 . for simplicity , set @xmath57 . inserting this into the equation we know that since @xmath58 is a soliton solution we have @xmath59 by splitting @xmath58 up into its real and imaginary parts , then doing a taylor expansion . hence , if @xmath60 , we get @xmath61 where @xmath62 where @xmath63 and @xmath64 [ spec : defn1 ] a hamiltonian , @xmath2 is called admissible if the following hold : + 1 ) there are no embedded eigenvalues in the essential spectrum , + 2 ) the only real eigenvalue in @xmath65 $ ] is @xmath66 , + 3 ) the values @xmath67 are not resonances . + [ spec : defn2 ] let ( nls ) be taken with nonlinearity @xmath11 . we call @xmath11 admissible if there exists a minimal mass soliton , @xmath55 , for ( nls ) and the hamiltonian , @xmath2 , resulting from linearization about @xmath55 is admissible in terms of definition [ spec : defn1 ] . the spectral properties we need for the linearized hamiltonian equation in order to prove stability results are precisely those from definition [ spec : defn1 ] . notationally , we refer to @xmath68 and @xmath69 as the projections onto the discrete spectrum of @xmath2 and onto the continuous spectrum of @xmath2 respectively . analysis of these spectral conditions will be done both numerically and analytically in the forthcoming work @xcite . we review here the disersive estimates from @xcite . let @xmath70 be the schwartz class of functions . then , we have the following results : [ thm : tdec ] given an admissible hamiltonian @xmath2 , @xmath69 the projection on the continuous spectrum of @xmath2 , for initial data @xmath71 , we have @xmath72 [ thm : wlindec ] let @xmath2 be an admissible hamiltonian as defined above . let @xmath73 be the associated distorted fourier basis . assume @xmath74 and @xmath75 for multi - indices @xmath76 , @xmath11 such that @xmath77 , where @xmath78 then , @xmath79 for any @xmath80 . from theorems [ thm : tdec ] and [ thm : wlindec ] , we have the following results : [ thm : disp1 ] let @xmath69 and @xmath68 be projections onto the continuous and discrete spectrum of @xmath2 respectively . then , @xmath81 [ thm : strich1 ] for @xmath25 and @xmath82 such that @xmath83 , with @xmath84 , and @xmath85 , the transformation @xmath86 maps continuously @xmath87 into @xmath88 and @xmath89 [ thm : strich2 ] for every @xmath90 and every admissible pair @xmath91 , the function @xmath92 belongs to @xmath93 , and there exists a constant @xmath94 depending only on @xmath95 such that @xmath96 it should be noted that similar estimates to those in theorems [ thm : tdec ] and [ thm : wlindec ] were proven in the works @xcite and @xcite , where in the first the techniques used were more along the lines of resolvent estimates and in the second the fact that the nonlinearities of interest were of even integer powers was crucial to the argument . however , from the scattering theory point of view taken in @xcite , we are able to define the orthogonality condition in order to generalize the results and more easily prove the weighted estimates . to begin , we define the function space @xmath97 with norm given by @xmath98 we similarly define the function space @xmath99 with norm given by @xmath100 in this result , we seek to prove that minimal mass solitons for nonlinear schrdinger equations in three dimensions have stable perturbations for long times . these minimal solitons are unstable as discussed below . the main goal of this thesis is to prove the following three theorems : [ thm : mainls ] take the equation in @xmath101 @xmath102 where @xmath11 is an admissible saturated nonlinearity of type @xmath12 . for any @xmath103 , equation has a solution @xmath104 for @xmath105 of the form @xmath106 where @xmath55 is the minimal mass soliton in definition [ spec : defn1 ] and @xmath107 as @xmath5 . for equation in @xmath1 of type @xmath12 , we have @xmath108 . [ thm : mainhls ] take equation , where @xmath11 is an admissible saturated nonlinearity of type @xmath12 . for any @xmath103 , equation has a solution @xmath104 for @xmath109 of the form @xmath110 where @xmath55 is the minimal mass soliton in definition [ spec : defn1 ] and @xmath111 as @xmath5 . in this theorem , for equation in @xmath1 of type @xmath12 , we have @xmath112 . [ thm : longtime ] given equation , where @xmath11 is an admissible saturated nonlinearity of type @xmath19 , for any @xmath113 with @xmath114 , equation has a solution for @xmath115 of the form @xmath110 where @xmath55 is the minimal mass soliton in definition [ spec : defn1 ] and @xmath116 in theorems [ thm : mainls ] and [ thm : mainhls ] , the stable perturbations can be shown to live on a finite codimension manifold for @xmath25 large enough compared to @xmath117 . this will be explored further below . the class of functions @xmath118 for @xmath119 will be developed throughout the course of this work . they will result from projecting onto a distorted fourier basis for the linearized problem . for a further discussion these topics and the notion of distorted fourier basis , see @xcite and the refences contained within . we wish to construct a contraction argument similar to that presented in @xcite in the case where we have a more general nonlinearity . in particular , we have the equation @xmath120 where @xmath121 is chosen to be of type @xmath12 or type @xmath19 . many estimates that hold for the @xmath24 critical equation hold at that soliton because they share the property that @xmath122 where @xmath44 is the soliton parameter and @xmath45 is the @xmath24 mass . as this is a minimal mass soliton , there are many possible perturbations . one could perturb onto the manifold of stable solitons , onto the manifold of unstable solitons , or in fact , reduce the @xmath24 energy so that solitons no longer formed . unfortunately , due to a lack of scaling and general difficulties , very little is known about stable perturbations to such a soliton . also , it is a major question whether or not we have dispersion and scattering for initial data with @xmath24 mass below the minimal soliton mass . we hope to address this in future work , but for now we wish to prove the existence of stable solutions to the minimal mass soliton . we may assume that the minimal mass soliton occurs at @xmath123 . in other words , if @xmath124 is the desired soliton , we seek a solution of the form @xmath125 where @xmath126 ; x)$ ] and @xmath127 for some normed space @xmath128 and some large @xmath129 to be determined . the goal is to solve this problem for @xmath130 solving both @xmath131 as well as @xmath132 for @xmath2 the matrix hamiltonian that results from linearizing about the minimal mass soliton . let @xmath144 - [ f(r^2 ) + f'(r^2)r^2 ] , \\ b & = & f'(\|r+v_0\|^2)(r+v_0)^2 - f'(r^2 ) r^2 , \\ g(w ) & = & f(\|r+v_0+w\|^2 ) ( r+v_0+w ) - f(\|r + v_0\|^2)(r+v_0 ) \\ & - & [ f(\|r+v_0\|^2 ) + f'(\|r+v_0\|^2)\|r+v_0\|^2 ] w - f'(\|r+v_0\|^2)(r+v_0)^2 \bar{w}.\end{aligned}\ ] ] then , we have @xmath145 note that in both cases , @xmath153 and in the second case , @xmath154 . however , we can define @xmath155 . this is @xmath156 at @xmath157 in both cases . to see this , note @xmath158 hence at @xmath157 , the terms resulting in exponential growth from @xmath159 are controlled . in the resulting taylor expansion , we see @xmath160 let us make the assumption that we are working with type @xmath12 nonlinearities with @xmath161 . the author believes that similar results should hold even if @xmath25 is not restricted to allow more regularity of the nonlinearity , however for the expansions in the sequel to be accurate , we must restrict the nonlinearities to have sufficient regularity , as well as to provide sufficient decay in @xmath162 . let us explore the behaviors of the the above functions . for simplicity , let @xmath163 . to begin , @xmath164 a similar calculation gives that @xmath165 , we have latexmath:[\ ] ] hence , by a continuity argument , @xmath148 exists on @xmath290 $ ] with @xmath291 h^2 } \leq \delta^{3+\frac{1}{8}}$ ] . since we assume @xmath292 , there exists a @xmath293 independent of @xmath58 such that after each time step @xmath294 , the initial perturbation @xmath295 is of size @xmath296 in @xmath297 . hence , we can continue this iteration from @xmath294 to @xmath298 for any @xmath299 . thus , there exists a small perturbative solution of the minimal mass soliton , however we are unable to show the lower order terms disperse as @xmath5 . if by using the ideal scaling of @xmath300 , then we see we have existence on a time scale of @xmath301 and @xmath302 . once again , due to the gain in @xmath303 , this argument can be continued for all @xmath162 , though we can not prove dispersion of @xmath148 . . _ semilinear schrodinger equations , _ courant lecture notes in mathematics , * 10*. new york university , courant institute of mathematical sciences , new york ; american mathematical society , providence , ri ( 2003 ) .	in this result , we develop the techniques of @xcite and @xcite in order to determine a class of stable perturbations for a minimal mass soliton solution of a saturated , focusing nonlinear schrdinger equation @xmath0 in @xmath1 . by projecting into a subspace of the continuous spectrum of @xmath2 as in @xcite , @xcite , we are able to use a contraction mapping similar to that from @xcite in order to show that there exist solutions of the form @xmath3 where @xmath4 disperses as @xmath5 . hence , we have long time persistance of a soliton of minimal mass despite the fact that these solutions are shown to be nonlinearly unstable in @xcite .
at a distance of only 6.7 pc , the primary of the @xmath0 boo binary system , @xmath0 boo a ( g8 v ) , is one of the two brightest `` moderately active '' coronal x - ray sources in the sky . the only star of this coronal activity level with a comparable x - ray flux at earth is @xmath3 eri ( k2 v , @xmath4 pc ) . as such , these two stars are of crucial importance for establishing the coronal properties of stars that are significantly more active than the sun , but which are not in the extremely active regime that would be represented by the rs cvn binaries or very young stars with rotation periods of only a few days . @xcite find that the coronal emission measure ( em ) distributions of @xmath3 eri and @xmath0 boo are similar to those of bright active regions on the sun . they note that the x - ray emission from these stars can therefore be explained by stellar surfaces that are completely covered with solar - like active regions , suggesting that @xmath3 eri and @xmath0 boo a represent a maximum of solar - like activity for coronal stars , whereas stars with even higher x - ray luminosities and higher coronal temperatures represent a completely different regime of coronal activity . this interpretation places @xmath3 eri and @xmath0 boo a at an interesting coronal transition point . further evidence that @xmath3 eri and @xmath0 boo a lie near a transition in the nature of coronal activity comes from stellar wind measurements , which are based on analyses of h i ly@xmath5 absorption from wind - ism interaction regions @xcite . at low activity levels mass loss increases with activity , with @xmath3 eri having a mass loss rate 30 times that of the sun ( i.e. , @xmath6 , where @xmath7 m@xmath8 yr@xmath9 ) . however , the somewhat more active @xmath0 boo system has a surprisingly weak wind with only @xmath10 , leading to the suggestion that the observed increase of mass loss with activity ends abruptly at an x - ray surface flux of about @xmath11 ergs @xmath12 s@xmath9 , with @xmath3 eri and @xmath0 boo on opposite sides of this dividing line @xcite . the wind measurements suggest a slight modification of the proposal that @xmath3 eri and @xmath0 boo a represent an extreme of truly solar - like activity . they suggest that @xmath3 eri is the true extreme , while @xmath0 boo a actually lies beyond it at the low end of the non - solar - like activity regime . evidence that @xmath0 boo a s magnetic topology is fundamentally different from the sun comes from spectropolarimetric measurements of photospheric fields on @xmath0 boo a. @xcite find evidence for both a global dipole field component much stronger than that of the sun ( @xmath13 g ) , and a strong toroidal field component ( @xmath14 g ) that has no clear solar analog . we also note that the activity level of @xmath0 boo a is roughly where one starts to see polar starspots @xcite , which also have no solar analog . @xcite find evidence for high latitude spots on @xmath0 boo a , though not necessarily `` polar spots . '' perhaps the strong large - scale fields found by @xcite envelope the star and inhibit wind flow , explaining why @xmath0 boo s wind is surprisingly weak . consistent with this idea , @xcite present evidence that as stellar rotation and magnetic activity increase , stars tend to store a larger fraction of their magnetic energy in large scale fields , which could in principle inhibit mass loss . with this background in mind , we here present an analysis of the x - ray spectrum of @xmath0 boo a based on new observations from _ chandra _ , with the goal of seeing whether the apparent changes in coronal character noted above find clear expression in the coronal spectrum of @xmath0 boo a , and coronal properties inferred from it . a similar spectrum of @xmath3 eri has already been analyzed , so a central goal of this paper is to compare the @xmath0 boo a spectrum with that of @xmath3 eri and similar stars ( * ? ? ? * hereafter wl06 ) . other coronal spectra of @xmath0 boo have been obtained by previous x - ray and euv missions ( e.g. , * ? ? ? * ; * ? ? ? * ) , but the new _ chandra _ data are superior in many ways . one of them is that _ chandra_/letgs observations allow the x - ray spectrum to be observed at high spectral resolution over a very broad spectral range of @xmath15 . this allows the detection of many coronal emission lines that have never been previously detected before from @xmath0 boo a. another particulary important improvement is that _ chandra _ s superb spatial resolution allows us for the first time to truly address the issue of @xmath0 boo s binarity . for @xmath0 boo a has a companion , @xmath0 boo b ( k4 v ) , which is only 6@xmath16 away , and _ no x - ray or uv observation of @xmath0 boo has ever truly resolved the binary . _ table 1 lists various properties of the two @xmath0 boo stars . the faster rotation and larger size of the primary lead to the expectation that @xmath0 boo a should dominate the coronal emission from the system . detailed analyses of iue data and a rosat / hri image of @xmath0 boo do indeed suggest that @xmath0 boo a dominates the binary s uv and x - ray emission @xcite , so the usual assumption of simply assigning all flux to the primary should be a decent approximation . however , @xmath0 boo b s spectral type and rotation period are not that different from the x - ray luminous @xmath3 eri ( k2 v ; @xmath17 days ) . thus , @xmath0 boo b should be contributing a non - negligible flux to the system s total x - ray emission . the older observations simply do not have sufficient spatial resolution to quantify how much . in short , the new _ chandra _ data presented here allow the first x - ray detection of @xmath0 boo b , and they ensure that the brighter spectrum of @xmath0 boo a is uncontaminated by any emission from the companion . the @xmath0 boo binary was observed on 2007 may 2 for 97.2 ksec with _ chandra _ s letgs configuration , combining the low energy transmission grating ( letg ) and the component of the high resolution camera detector intended for use in spectroscopy ( hrc - s ) . the zeroth - order letgs image of @xmath0 boo is displayed in figure 1 , showing how cleanly the stars are separated in the _ chandra _ data , representing the first resolution of the binary in x - rays . we measure a position angle and stellar separation for the binary of @xmath18 and @xmath19 , respectively , in reasonable agreement with the @xmath20 and @xmath21 values expected from the system s orbital elements @xcite . the zeroth order image provides the simplest way to measure the contribution of each star to @xmath0 boo s total x - ray emission . the primary dominates the emission from the system , accounting for 88.5% of the total counts . we also use the zeroth order image to assess the variability of the two stars during the course of the daylong observation . figure 2 shows light curves for the two @xmath0 boo components , indicating significant variability for both stars . for @xmath0 boo a , there appears to be a long duration flare near the beginning of the observation , and a much shorter duration event later on . in extracting spectra from the raw _ chandra _ data , we initially tried to separate flaring and quiescent times . however , we were unable to discern any clear differences between the resulting noisy flare spectrum and that of quiescence . thus , in the final spectral extraction procedure described below , we simply extract a single spectrum for each star , including both quiescent and flaring times . the standard pipeline processing of letgs data does not properly separate the spectra of the two stars , so the spectral extraction must be performed in a more manual fashion . in a previous paper we have described in detail how we extracted separate letgs spectra of stars in two other binary systems , 36 oph and 70 oph ( wl06 ) . we follow a very similar procedure here , which is now described in abbreviated form . the data are processed using version 4.1.2 of the ciao software . the processing includes a removal of background counts using the standard `` light '' pulse height background filter available in ciao , but we also account for time - dependent gain corrections using software outside of ciao , corrections which would later be added to a subsequent official ciao release ( version 4.2 ) . by reducing background counts from our data , this filtering significantly improves the signal - to - noise of our final spectra . the processing initially yields a zeroth order image of the target at the aimpoint , which we have already discussed ( see fig . 1 ) , and two essentially identical spectra dispersed in opposite directions along the long axis of the hrc - s detector , a plus order and a minus order spectrum . the spatial resolution in the cross - dispersion direction worsens at long wavelengths where the spectrum is furthest from the focal point . we use a conservatively broad source extraction window with a width of 25 pixels centered on each stellar spectrum for @xmath22 . the window is broadened for @xmath23 . to avoid overlapping source windows , for each star we expand the window only in the direction away from its companion star by 30 pixels . at the longest wavelengths , we have to accept that there is some small degree of unresolvable blending . we use 90 pixel background windows on each side of the stellar spectra to determine a wavelength dependent background to subtract from the spectra of the two stars . finally , the plus and minus order spectra are coadded to yield the final product shown in figure 3 . the clear separation of the @xmath0 boo binary allows the first precise and unambiguous measurement of the x - ray luminosity of both stars . the system s x - ray luminosity reported in the rosat all - sky survey is @xmath24 @xcite . dividing this emission between @xmath0 boo s two stars according to the contribution ratio suggested by figure 1 yields the rosat x - ray luminosities reported in table 1 . this new knowledge of the coronal activity of both @xmath0 boo stars has interesting ramifications for the interpretation of @xmath0 boo s wind measurement . as mentioned in 1 , @xmath0 boo has a measured wind strength of @xmath10 , but the measurement is for the combined wind of the system , and there is no way to tell from the observations the contribution of each star to this total @xcite . figure 4 plots mass loss rate as a function of coronal x - ray emission for all main sequence stars with measured winds , analogous to a figure from @xcite . but in this figure we plot separate points for @xmath0 boo a and @xmath0 boo b , for three different assumptions about how the collective wind is divided between the two stars . we now know for a fact that @xmath0 boo b is very similar to @xmath3 eri , 36 oph , and 70 oph in terms of its coronal activity . but these stars all have rather high mass loss rates . the only way that @xmath0 boo b can be consistent with even the weakest of these winds , that of 36 oph , is if @xmath0 boo b accounts for practically all of @xmath0 boo s wind . this would leave the more active @xmath0 boo a with a wind that is possibly weaker than that of the sun , despite its active corona , strengthening the case for the existence of a wind dividing line near an x - ray surface flux of @xmath11 ergs @xmath12 s@xmath9 ( see fig . 4 ) , where stronger large scale magnetic fields begin to inhibit mass loss . even if @xmath0 boo a accounts for nearly all of the binary s wind , it would still be weak compared to that of @xmath3 eri , and then one would have to explain why @xmath0 boo b apparently has a wind over an order of magnitude weaker than that of many other stars of equal activity . we come to the surprising conclusion that the most likely interpretation of the available data is that @xmath0 boo b dominates the wind of the @xmath0 boo system . this conclusion is also consistent with the idea that a wind dividing line exists at an x - ray surface flux of about @xmath11 ergs @xmath12 s@xmath9 ( see 1 ) , separating less active stars with strong winds and more active stars with weaker winds @xcite . this activity dependent wind dividing line is not to be confused with the spectral type dependent wind - corona dividing line long known to exist for red giant stars @xcite . it must be noted that our conclusions rely on a very limited number of wind detections . further support for these interpretations should be sought from more wind measurements of main sequence stars , particularly for active stars . we use version 6.0 of the chianti atomic database to identify emission lines in our spectra @xcite . the identified lines are shown in figure 3 and listed in table 2 . table 2 lists counts for the detected lines , measured by direct integration from the spectra . line formation temperatures are quoted in the third column of the table , based on maxima of contribution functions computed using the ionization equilibrium computations of @xcite . uncertainties quoted in table 2 are 1@xmath25 . for the fainter @xmath0 boo b source , only the strongest lines are detected , but we quote 2@xmath25 upper limits for the nondetections , based on summing in quadrature the uncertainties in 5 bins surrounding the location of the line , and then multiplying this sum by two . measurements are made for all lines that appear visually to be at least marginally detected . however , cases in which the uncertainty implies a @xmath26 detection must be considered questionable . after the emission measure analysis described in 4.4 is completed , we confirm _ a posteriori _ that these questionable detections are at least plausible based on the derived emission measure distributions and abundances , which are constrained primarily by the stronger emission lines in the analysis . many of the emission lines are blends . we list in table 2 all lines that we believe contribute a significant amount of flux to the feature based on the line strengths in the chianti database . this determination is reassessed after em distributions are estimated , and model spectra can be computed from them and compared with the data . a few of the features identified in figure 3 are blends of lines of different species ( see table 2 ) . although we measure counts for these blends and list them in the table , these measurements are not used in the emission measure analysis . coronal electron densities can be estimated for the two @xmath0 boo stars directly from the o vii @xmath2721.8 line measurements in table 2 . there are in fact several he - like triplets in the observed spectral range that are useful for density measurements : si xiii @xmath276.7 , mg xi @xmath279.2 , ne ix @xmath2713.6 , and o vii @xmath2721.8 . however , insufficient spectral resolution , low signal - to - noise , and blending issues complicate any attempt to use the si xiii , mg xi , and ne ix lines for this purpose @xcite . for this reason , we focus only on the stronger , better separated o vii lines here . the specific line ratio of interest is that of the forbidden line to the intersystem line : @xmath28 . using the results of collisional equilibrium models from @xcite , the measured line ratios for @xmath0 boo a and @xmath0 boo b translate to upper limits of @xmath29 and @xmath30 , respectively , as reported in table 3 . the @xmath0 boo a limit is consistent with values that @xcite estimate from euve spectra . densities of about @xmath31 are consistent with spectra of @xmath3 eri , 36 oph ab , and 70 oph ab as well , which have activity levels similar to that of the @xmath0 boo stars ( wl06 ) . thus , there is no clear evidence for coronal density differences among moderately active stars observed by _ chandra_. among moderately active gk dwarfs with intrinsic x - ray luminosities of @xmath32 , the two easiest to observe from earth are @xmath0 boo a ( g8 v ) and @xmath3 eri ( k2 v ) , due to their close proximity . we can see how the coronal temperatures and abundances of these two key stars compare simply by plotting ratios of their coronal line luminosities as a function of line formation temperature , as is done in figure 5 . the @xmath0 boo a line luminosities in the figure originate from the line counts measured in table 2 . the @xmath3 eri luminosities are computed from line counts similarly listed in wl06 . figure 5 shows that nearly all the line ratios are above one at all temperatures , indicative of @xmath0 boo a being the more active star . the ratios are clearly temperature dependent , with the ratio rising from @xmath33 at @xmath34 to @xmath35 at @xmath36 . this implies a hotter corona for the more active @xmath0 boo a. this is consistent with previous observations demonstrating correlations between coronal activity and coronal temperature @xcite . it is intriguing how smoothly and linearly the line ratios rise with temperature . the lines in figure 5 are divided into those with low first ionization potential ( fip ) and those with high fip . in the solar corona and solar wind , elemental abundances are found to be dependent on fip . relative to the photosphere , elements with low fip ( fe , mg , si , etc . ) are generally found to have coronal abundances that are enhanced relative to elements with high fip ( c , n , o , ne , etc . ) evidence for this effect has been found for some stars of low to moderate activity @xcite . however , on very active stars the fip effect tends to be either absent , or sometimes an inverse fip effect is observed , where low - fip elements have coronal abundances that are _ lower _ than photospheric @xcite . thus , for main sequence stars the picture is of the fip effect generally decreasing as activity increases @xcite . but there are exceptions to this trend , and figure 5 is itself inconsistent with a tight correlation between activity and fip effect , as the lower flux ratios of the high - fip ions in figure 5 demonstrate that @xmath0 boo a s corona exhibits a stronger fip effect than the less active @xmath3 eri . we will return to this issue in 4.6 . in order to quantify the coronal abundances and temperature distributions , we perform an emission measure analysis for @xmath0 boo a and b based on our measurements of their coronal emission lines . the em analysis uses version 2.6 of the pintofale software developed by @xcite . this analysis follows closely that of wl06 in our study of letgs spectra from @xmath3 eri , 36 oph , and 70 oph , so we do not describe it again in great detail here . figure 6 shows em distributions ( in units of @xmath37 ) for the two @xmath0 boo stars , and compares them with distributions that we previously computed for five other moderately active g8-k5 dwarfs ( wl06 ) . these distributions are computed using chianti emissivities @xcite , and ionization balance calculations from @xcite . our line measurements are corrected for interstellar absorption , but the interstellar h column densities towards these stars , including @xmath0 boo ( see table 1 ) , are low enough that this is not an important correction . for elements with detected lines we can compute coronal abundances in the em analysis . however , line measurements alone only allow _ relative _ abundances to be computed , and only the shape of the em distribution can be inferred as opposed to its absolute value . initially , we assume a solar photospheric abundance for fe from @xcite , and all other abundances are computed relative to fe . the derived @xmath0 boo abundance ratios are listed in table 3 . in order to compute an _ absolute _ fe abundance , and to properly normalize the em distribution , the line - to - continuum ratio must be assessed . figure 7 illustrates this process for the case of @xmath0 boo a. the figure shows a synthetic spectrum generated from the emission measure distribution in figure 6 , assuming an absolute fe abundance of @xmath38}=0.7{\rm [ fe / h]}_{\odot}$ ] , which is judged to lead to the best fit . this is the same as the ratio found for @xmath3 eri ( wl06 ) . for @xmath0 boo b , we find a similar value of @xmath38}=0.8{\rm [ fe / h]}_{\odot}$ ] ( see table 3 ) . these results are used to properly normalize the em distributions in figure 6 . figure 3 displays synthetic spectra of @xmath0 boo a and b based on the em distributions . in addition to the dominant first - order spectrum , we also take into account orders 25 in generating synthetic spectra . both figures 3 and 7 explicitly show the contributions of these higher orders to the total spectrum . in logarithmic terms , our measurement for @xmath0 boo a implies @xmath39}=-0.15 $ ] , where @xmath40 is here the solar photospheric abundance of fe . this measurement can be compared with the @xmath39}=0.1^{+0.2}_{-0.15}$ ] measurement of @xcite based on euve spectra . at this point it is worth noting that the contributions of weak , unresolved , and unidentified lines to the observed `` continuum '' could easily lead to underestimates of @xmath38}$ ] , so our @xmath38}$ ] measurements could be conservatively regarded as lower limits . given that our measurement is lower than that of @xcite , perhaps there are more lines missing and unaccounted for within the letgs spectral region than in the euve spectral region analyzed by @xcite , presumably leading to an underestimate of @xmath38}$ ] on our part . nevertheless , both our measurement and that of @xcite show a higher fe abundance in the corona than in the photosphere , where @xmath41}=-0.26 $ ] ( see table 1 ) , consistent with a solar - like fip effect ( see 4.3 ) . in following sections that focus on comparing the coronal abundances of different stars , we avoid the systematic errors involved in computing absolute abundances by comparing only relative abundances , specifically abundances relative to fe . the seven em distributions in figure 6 are superficially similar in appearance , with a dramatic rise at about @xmath34 . but perhaps the most interesting feature is that all of the distributions show an em peak near @xmath1 , a peak that is particularly convincing for the two brightest stars , @xmath0 boo a and @xmath3 eri . the coronal heating on these moderately active stars seems to favor this particular temperature . the em distribution that @xcite derived for @xmath0 boo a from euve data agrees nicely with our _ chandra _ distribution . their peak at @xmath1 is not quite as narrow or prominent , but this may just be a result of the ion - by - ion method that @xcite use to quantify their emission measures . as there is some overlap in the euve and _ chandra_/letgs spectra in the @xmath42 range , there are some lines in common to these two analyses . the most precise way to compare the activity level of @xmath0 boo a during the euve and letgs observations is to directly compare the line fluxes rather than the derived em distributions . the letgs / euve line flux ratios have a weighted mean and standard deviation of @xmath43 . an assessment of the relative flux calibration of euve and letgs by @xcite using white dwarf spectra suggests calibration errors in the opposite direction , with letgs / euve ratios of @xmath44 . another systematic error that works in the opposite direction is the fact that , unlike letgs , the euve fluxes will be contaminated by emission from @xmath0 boo b. thus , the below unity value of the average letgs / euve line ratio for @xmath0 boo a is likely indicative of genuine stellar variability , with the star being somewhat fainter during the letgs observation . however , the very modest level of variability suggested by the letgs / euve comparison is hardly surprising , as the coronal emissions of cool main sequence stars are known to be quite variable on all kinds of timescales . finally , we can compute broadband x - ray luminosities directly from the letgs spectra . we first subtract the higher order contributions to the spectra estimated from the em analysis ( i.e. , the green line in figure 7 ) . we can then combine the corrected spectra with the known first order effective area curve and exposure time to convert counts in our spectra to fluxes . from the flux spectra we can compute x - ray luminosities for any wavelength region within the spectrum . in table 1 , we list luminosities computed in this way for the rosat - like @xmath45 kev ( @xmath46 ) bandpass . the total @xmath0 boo system luminosity of @xmath47 computed in this way agrees reasonably well with the rosat all - sky survey measurement of @xmath24 @xcite . the first order spectra suggest @xmath0 boo a accounts for about 84.6% of @xmath0 boo s total x - ray flux , compared to the 88.5% contribution to counts in the zeroth order image . the discrepancy is due in part to the difference between count measurements and flux , the latter of which requires consideration of the effective area of the instrument . the discrepancy is also due in part to differences in the effective areas for the zeroth order image and first order spectra . in 4.4 , we compared the em distributions of seven moderately active g8-k5 dwarfs observed with letgs . we also wish to compare the coronal abundances of these seven stars , in order to first determine if a fip effect exists for these stars , and if so to then compare its magnitude within this stellar sample . our calculation of abundances in the em analysis assumes that abundances are the same at all temperatures throughout the corona . figure 5 suggests that this is probably a good assumption , as it shows that the differential fip effect between @xmath0 boo a and @xmath3 eri is the same throughout the @xmath48 temperature range that dominates the coronal emission measure distributions of these stars . figure 8 compares the coronal abundances of our sample of stars , plotting the abundances relative to fe , as listed in table 3 . but the search for a fip effect relies on knowing not just the coronal abundance by itself , but the coronal abundance relative to the photospheric abundance . thus , in figure 8 we subtract photospheric abundances from the coronal ones . however , because photospheric abundances are imperfectly known , taking them into account involves numerous assumptions , which we now describe . in table 1 , photospheric abundances measured for @xmath0 boo a are listed for c , o , mg , si , and fe . these abundances are quoted relative to solar . the @xcite analysis from which these abundances are obtained involves a line - by - line comparison of solar and stellar photospheric absorption lines , so these stellar abundances are _ fundamentally _ relative values . for the purposes of figure 8 , we require absolute photospheric abundances relative to fe , log [ x / fe]@xmath49 , so we have to combine the abundances in table 1 with some assumed solar abundances . in figure 8a , we use solar abundances from @xcite . there are two elements for which no stellar photospheric abundances are available for @xmath0 boo a : n and ne . since these are high - fip elements , we assume that their abundances relative to the sun are the same as that of another high - fip element , o. in other words , we assume @xmath50 = \log [ ne / ne_{\odot } ] = \log [ o / o_{\odot}]=-0.09}$ ] ( see table 1 ) . for @xmath0 boo b there are no independent photospheric abundance measurements , so we simply have to assume @xmath0 boo b is identical to @xmath0 boo a in this respect . similar assumptions have been made for the photospheric abundances of @xmath3 eri , 70 oph ab , and 36 oph ab ( see wl06 ) . the low - fip and high - fip data points in figure 8a are connected with separate dotted lines . these dotted lines indicate the following curious abundance patterns for all the stars : 1 . mg abundances are always higher than those of si , 2 . ne abundances are always higher than the other high - fip abundances , and 3 . c and n abundances are intermediate between those of o and ne . are these abundance patterns telling us something about how coronal fractionation is operating in these moderately active stars ? or are these patterns just a product of assumptions made about the photospheric abundances ? at least for the high - fip elements , the latter of these two explanations seems most likely . in figure 8b , we explore this further by changing our assumed photospheric abundances . this figure is like figure 8a , but instead of @xcite solar abundances , we use those of @xcite , with the exception of ne . photospheric ne abundances provide unique problems , because there are no photospheric ne absorption lines to measure , which means that solar ne abundance measurements actually have to rely on emission lines from the corona , where elements are subject to fractionation processes like those that yield the fip effect . in figure 8b , we change the assumed photospheric abundance of ne drastically by assuming that the average coronal ne / o ratio of @xmath51 measured by @xcite _ for active stars _ actually represents the real _ photospheric _ abundance of _ all _ cool main sequence stars , as opposed to the much lower solar coronal ne / o ratio of @xmath52 @xcite . in other words , rather than assuming that ne is fractionated from o in active stars but not in the sun , in figure 8b we assume that ne is being fractionated in the atmospheres of inactive stars like the sun , but not in active stars . theoretical support for this interpretation is provided by @xcite , and a higher solar photospheric ne abundance could also help resolve discrepancies between solar interior models and helioseismological observations @xcite . with the revised assumptions regarding photospheric abundances discussed above , the high - fip abundance patterns seen in figure 8a are largely absent in figure 8b , and the high - fip dotted lines are closer to horizontal . this result could be used as evidence in favor of the photospheric abundance assumptions being made in figure 8b , as opposed to the @xcite solar abundances being assumed to make figure 8a . although changing the assumed solar abundances seems to be able to significantly reduce the discrepancies among the high - fip elements , such is not the case for the low - fip elements , where there is little difference in the discrepancy between mg and si in figures 8a and 8b . thus , presumably coronal fractionation processes really are enhancing stellar mg abundances relative to si in the coronae of these moderately active stars . @xcite propose that coronal fractionation originates from ponderomotive forces induced by the propagation of alfvn waves through the chromosphere . these models demonstrate the potential for fractionating mg relative to si in a manner consistent with the observed mg / si ratios . in their sample of early g dwarfs , @xcite also generally find high mg / si ratios . but in a sample of m dwarfs , all of which happen to show inverse fip effects , @xcite find _ low _ mg / si ratios , suggesting this behavior may be spectral type dependent . for the high - fip dotted lines in figure 8 , the lower the line , the higher the coronal fe abundance is relative to the high - fip elements , and therefore the stronger the inferred fip effect . significant fip effect is apparent for many of the stars , with 70 oph a s corona possessing the strongest fip bias . but there is a significant amount of scatter in the level of fip effect found for this sample of stars . figure 8 suggests the following sequence of decreasing fip bias : 70 oph a , @xmath0 boo a , 36 oph b , 36 oph a , @xmath0 boo b , @xmath3 eri , 70 oph b. the last of these stars , 70 oph b , appears to actually have a slight _ inverse _ fip effect . there is some spread of x - ray luminosity within our sample of stars , but there is no correlation at all between x - ray luminosity and fip effect in this sample . the most luminous star , @xmath0 boo a , shows a strong fip effect , but the second most luminous , @xmath3 eri , shows only a weak fip effect at most . the relatively strong fip effect of the most active star in our sample also runs contrary to the general anti - correlation between stellar activity and fip effect that has been reported in the past ( see 4.3 ) . since the fip effect on the sun is latitude dependent , with the effect being strong near the equator and minimal in the polar regions , one might imagine that perhaps stellar orientation could be a factor in the fip bias observed from a star , assuming coronal abundances on these more active stars also exhibit some sort of latitude dependence . however , since coronal emission is optically thin , at least half of a star s total emission should be visible from a star regardless of orientation , so it is hard to imagine orientation effects explaining the roughly factor - of-3 variation in fip bias seen in figure 8 . in our previous paper ( wl06 ) we have emphasized the remarkable difference between 70 oph a and 70 oph b , which represent the fip effect extremes in figure 8 despite being similar stars in the same binary system . one possible explanation that we proposed is that perhaps the fip effect is a time - dependent phenomenon . the solar example provides a precedent for this , with newly emerged active regions having close to photospheric abundances and only acquiring a fip bias on timescales of days @xcite . thus , for the stars in figure 8 with little or no apparent fip bias ( e.g. , 70 oph b , @xmath3 eri ) , perhaps the visible surface of the star at the time of observation was dominated by young active regions , whereas the stars with strong observed fip effects ( e.g. , 70 oph a , @xmath0 boo a ) were observed with older active regions . the time - dependent interpretation can only be tested by observing these stars again , preferably more than once . but the limited amount of evidence available so far does _ not _ support this interpretation . abundances computed from euve spectra taken at different times from our letgs data seem to be very consistent with our letgs results , showing a strong fip effect for @xmath0 boo a , and only a weak fip effect , if any , for @xmath3 eri @xcite . @xcite looked for abundance variations in multiple observations of the very active k dwarf ab dor , with the observations encompassing a range of activity levels , and found no such variability . there does not seem to be a strong case for activity dependence , time dependence , or orientation effects being responsible for the variation in fip bias seen in figure 8 . there does seem to be a case for there being a spectral type dependence . the two earliest type stars ( @xmath0 boo a and 70 oph a ) have the strongest solar - like fip effect , while the three latest type stars ( 70 oph b , @xmath0 boo b , and @xmath3 eri ) have the weakest . this is explored further in figure 9 , where we first compute the average of the high - fip dotted lines in figure 8b , a quantity we simply call the `` fip bias , '' and then plot this quantity versus spectral type . in order to broaden the spectral range under consideration , in figure 9 we supplement our measurements with those of early g stars from @xcite and m dwarfs from @xcite . in utilizing measurements from these other papers , we were careful to compute the `` fip bias '' in the same manner that we have for our sample of stars , with the same assumptions about solar photospheric abundances as in figure 8b . for the @xcite sample , stellar photospheric abundances are available from @xcite , which we take into account , but for the m dwarfs of @xcite there are no such measurements and we have no recourse but to assume that the photospheres of these stars possess solar photospheric abundances . finally , we have added a point for the sun , using abundances listed in @xcite . figure 9 shows a remarkably tight correlation of fip bias with spectral type , with the earlier type stars having solar - like fip effects and the later stars having inverse fip effects , though the lack of photospheric abundance measurements for the m dwarfs means that conclusions about them are more uncertain . the reversal point seems to happen near spectral type k5 v. to the best of our knowledge , this is the first time a strong spectral type dependence of fip bias has been claimed . why has this seemingly strong correlation not been recognized in the past ? answering this question requires clearly defining what the sample of stars in figure 9 represents . in short , these are inactive to moderately active main sequence stars with x - ray luminosities of @xmath53 . for solar - like main sequence stars such stars have rotational periods of @xmath54 days and ages of @xmath55 gyr @xcite most surveys of coronal x - ray emission are dominated by interacting binaries , evolved stars , and other targets that represent extremes of coronal activity , which are commonly observed by x - ray telescopes because they are so bright and easy to detect . but by avoiding such extremes and focusing solely on more normal stars , we find a very different empirical view of the fip effect . gone is the kind of activity dependence on which previous authors have focused ( see 4.3 ) , replaced instead with a strong spectral type dependence . in order to emphasize the lack of activity dependence in figure 9 , consider the two least active stars in the sample : the sun ( g2 v ) and proxima cen ( m5.5 ve ) . the two g1 stars right next to the sun in the figure have x - ray luminosities almost two orders of magnitude higher than the sun . but there is no difference in `` fip bias '' . likewise , three of the four stars right next to proxima cen in the figure have x - ray luminosities about two orders of magnitude higher than it . once again , no difference in fip bias . there are two stars in the @xcite sample that are not included in figure 9 , because they have @xmath2 and therefore violate our policy of avoiding extremes . these two stars are 47 cas b and ek dra , which both have rotation periods of under three days . this rapid rotation will probably exist for all main sequence stars with @xmath2 . if we plotted points for 47 cas b and ek dra in figure 9 , they would lie well above the relation defined by the other stars . rather than proposing that magnetic activity is starting to operate somewhat differently at the activity level of 47 cas b and ek dra , thereby yielding a different fip bias , we instead propose that the changes in fip effect are ultimately driven by changes to fundamental stellar properties induced by the rapid stellar rotation . rotation as rapid as that possessed by 47 cas b and ek dra will presumably affect the atmospheric and convection zone properties of the star to some extent ( e.g. , * ? ? ? interacting binaries and evolved stars will also have substantial differences between their atmospheric properties and those of main sequence stars with identical spectral types . figure 9 suggests to us that fip bias is tied directly to basic stellar properties of the star , and not any simple coronal property like total x - ray flux or magnetic field strength . within the theoretical paradigm of @xcite , in which alfvn wave propagation through the chromosphere drives coronal abundance anomalies , it is possible to imagine that changes in basic stellar properties such as radius , surface gravity , or convective motions could affect these waves and thereby affect the coronal fip bias . _ in short , we propose that the spectral type dependence of fip bias in figure 9 provides a more fundamental insight into the nature and cause of the fip effect than the activity trends that have been explored in the past . _ we have used _ chandra_/letgs observations to study the coronae of the @xmath0 boo binary , and we have compared our results with a previous study of five other moderately active gk dwarfs : @xmath3 eri , 70 oph ab , and 36 oph ab . our findings are summarized as follows : 1 . the two stellar components of @xmath0 boo are resolved for the first time in x - rays . we find that @xmath0 boo a accounts for 88.5% of the counts in the zeroth order image . for comparison , spectral analysis implies an 84.6% contribution to the system s x - ray flux in the rosat - like 0.1 - 2.4 kev bandpass . 2 . with @xmath0 boo b detected for the first time in x - rays , we now know it has an x - ray luminosity comparable to stars known to have strong stellar winds , leading to the conclusion that @xmath0 boo b may account for most of the modest wind observed from the system , with the more active @xmath0 boo a possessing a surprisingly weak wind despite its active corona . this strengthens the case for the presence of a `` wind dividing line '' separating moderately active stars with strong winds and very active stars with weak winds , possibly due to the increasing strength of large scale global magnetic fields that inhibit mass loss . 3 . our emission measure analysis of @xmath0 boo a and b finds evidence for a narrow peak in em at @xmath1 for both stars , a peak also seen for the other moderately active g8-k5 dwarfs in our sample , indicating that the coronal heating mechanism operating on these stars favors this temperature . the emission measure analysis provides us with coronal abundances relative to fe . looking for coronal abundance anomalies requires correcting for stellar photospheric abundances , which in turn involves assuming solar photospheric abundances , since the stellar abundances we use are measured relative to solar . when we simply assume solar abundances from @xcite , for our sample of stars we find that the high - fip element abundances are inconsistent with each other . this inconsistency is decreased significantly for all our stars if we instead assume a photospheric ne abundance consistent with stellar coronal measurements from @xcite , and solar abundances from @xcite for other elements . this is an argument in favor of these latter photospheric abundances , unless there are actual coronal fractionation processes taking place among the high - fip elements that we are erroneously removing . 5 . among low - fip ions , the coronal mg / si ratio is consistently higher than the photospheric ratio for all of our g8-k5 stars , presumably an effect of actual coronal fractionation . an enhancement of mg relative to si in the corona is predicted by many of the models of @xcite , which are designed to explain solar / stellar coronal abundance anomalies . however , although @xcite also finds high mg / si ratios for early g dwarfs , coronal mg / si ratios appear to be _ lower _ than photospheric for m dwarfs @xcite , suggesting a spectral type dependence for this coronal abundance ratio . 6 . with regards to the fip effect , our sample of stars exhibit a range of fip bias , with 70 oph a and @xmath0 boo a having the strongest solar - like fip effect , @xmath3 eri showing little or no fip effect , and 70 oph b possessing a slight inverse fip effect . expanding our sample of stars even further to include early g stars from @xcite and m dwarfs from @xcite , we find that the fip bias correlates beautifully with spectral type , but only if extremely active stars with @xmath2 are excluded . . the tight correlation of fip bias with spectral type suggests to us that the nature and causes of the fip effect can ultimately be traced directly to fundamental surface properties . among normal main sequence stars , these change in a very predictable manner with spectral type , explaining the tight correlation in figure 9 . no such correlation exists for extremely active stars , because many are evolved stars with disparate photospheric characteristics , and in other cases the stellar atmospheres will be affected by rapid rotation or the effects of close binarity . support for this work was provided by the national aeronautics and space administration through chandra award number go8 - 9003z issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics and space administration under contract nas8 - 03060 . allende prieto , c. , barklem , p. s. , lambert , d. l. , & cunha , k. 2004 , a&a , 420 , 183 antia , h. m. , & basu , s. 2005 , apj , 620 , l129 arnaud , m. , & rothenflug , r. 1985 , a&as , 60 , 425 asplund , m. , grevesse , n. , sauval , a. j. , & scott , p. 2009 , ara&a , 47 , 481 . audard , m. , behar , e. , gdel , m. , raassen , a. j. j. , porquet , d. , mewe , r. , foley , c. r. , & bromage , g. e. 2001 , a&a , 365 , l329 audard , m. , gdel , m. , sres , a. , raassen , a. j. j. , & mewe , r. 2003 , a&a , 398 , 1137 bahcall , j. n. , basu , s. , & serenelli , a. m. 2005 , apj , 631 , 1281 ball , b. , drake , j. j. , lin , l. , kashyap , v. , laming , j. m. , & garca - alvarez , d. 2005 , apj , 634 , 1336 barnes , t. g. , evans , d. s. , & moffett , t. j. 1978 , mnras , 183 , 285 beuermann , k. , burwitz , v. , & rauch , t. 2006 , a&a , 458 , 541 brinkman , a. c. , et al . 2001 , a&a , 365 , l324 brown , b. p. , browning , m. k. , brun , a. s. , miesch , m. s. , & toomre , j. 2008 , apj , 689 , 1354 dere , k. p. , landi , e. , mason , h. e. , monsignori fossi , b. c. , & young , p. r. 1997 , a&as , 125 , 149 dere , k. p. , landi , e. , young , p. r. , del zanna , g. , landini , m. , & mason , h. e. 2009 , a&a , 498 , 915 dorren , j. d. , gdel , m. , & guinan , e. f. 1995 , apj , 448 , 431 drake , j. j. , & kashyap , v. 2001 , apj , 547 , 428 drake , j. j. , laming , j. m. , & widing , k. g. 1997 , apj , 478 , 403 drake , j. j. , peres , g. , orlando , s. , laming , j. m. , & maggio , a. 2000 , apj , 545 , 1074 drake , j. j. , & testa , p. 2005 , nature , 436 , 525 feldman , u. , & laming , j. m. 2000 , phys . , 61 , 222 grevesse , n. , & sauval , a. j. 1998 , space sci . rev . , 85 , 161 gdel , m. 2004 , astron . astrophys . , 12 , 71 gdel , m. , et al . 2001 , a&a , 365 , l336 gdel , m. , guinan , e. f. , & skinner , s. l. 1997 , apj , 483 , 947 hartmann , l. , schmidtke , p. c. , davis , r. , dupree , a. k. , raymond , j. , & wing , r. f. 1979 , apj , 233 , l69 huenemoerder , d. p. , canizares , c. r. , drake , j. j. , & sanz - forcada , j. 2003 , apj , 595 , 1131 huenemoerder , d. p. , canizares , c. r. , & schulz , n. s. 2001 , apj , 559 , 1135 kashyap , v. & drake , j. j. 1998 , apj , 503 , 450 kashyap , v. , & drake , j. j. 2000 , bull . india , 28 , 475 laming , j. m. 2004 , apj , 614 , 1063 laming , j. m. 2009 , apj , 695 , 954 laming , j. m. , & drake , j. j. 1999 , apj , 516 , 324 laming , j. m. , drake , j. j. , & widing , k. g. 1996 , apj , 462 , 948 liefke , c. , ness , j. -u . , schmitt , j. h. m. m. , & maggio , a. 2008 , a&a , 491 , 859 linsky , j. l. , & haisch , b. m. 1979 , apj , 229 , l27 maggio , a. , sciortino , s. , vaiana , g. s. , majer , p. , bookbinder , j. , golub , l. , harnden , f. r. , jr . , & rosner , r. 1987 , apj , 315 , 687 mazzotta , p. , mazzitelli , g. , colafrancesco , s. , & vittorio , n. 1998 , a&as , 133 , 403 mckenzie , d. l. 1987 , apj , 322 , 512 ness , j. -u . , schmitt , j. h. m. m. , burwitz , v. , mewe , r. , raassen , a. j. j. , van der meer , r. l. j. , predehl , p. , & brinkman , a. c. 2002 , a&a , 394 , 911 noyes , r. w. , hartmann , l. w. , baliunas , s. l. , duncan , d. k. , & vaughan , a. h. 1984 , apj , 279 , 763 pandey , j. c. , & srivastava , a. k. 2009 , apj , 697 , l153 perryman , m. a. c. , et al . 1997 , a&a , 323 , l49 petit , p. , et al . 2005 , mnras , 361 , 837 petit , p. , et al . 2008 , mnras , 388 , 80 porquet , d. , mewe , r. , dubau , j. , raassen , a. j. j. , & kaastra , j. s. 2001 , a&a , 376 , 1113 sanz - forcada , j. , affer , l. , & micela , g. 2009 , a&a , 505 , 299 sanz - forcada , j. , maggio , a. , & micela , g. 2003 , a&a , 408 , 1087 schmelz , j. t. , nasraoui , k. , roames , j. k. , lippner , l. a. , & garst , j. w. 2005 , apj , 634 , l197 schmitt , j. h. m. m. 1997 , a&a , 318 , 215 schmitt , j. h. m. m. , & liefke , c. 2004 , a&a , 417 , 651 sderhjelm , s. 1999 , a&a , 341 , 121 strassmeier , k. g. 2002 , astronomische nachrichten , 323 , 309 telleschi , a. , gdel , m. , briggs , k. , audard , m. , ness , j. -u . , & skinner , s. l. 2005 , apj , 622 , 653 toner , c. g. , & gray , d. f. 1988 , apj , 334 , 1008 widing , k. g. , & feldman , u. 2001 , apj , 555 , 426 wood , b. e. 2004 , living rev . solar phys . , 1 , 2 , url : http://www.livingreviews.org/lrsp-2004-2 wood , b. e. , & linsky , j. l. 2006 , apj , 643 , 444 ( wl06 ) wood , b. e. , mller , h. -r . , zank , g. p. , linsky , j. l. , & redfield , s. 2005a , apj , 628 , l143 wood , b. e. , redfield , s. , linsky , j. l. , mller , h. -r . , & zank , g. p. 2005b , apjs , 159 , 118 young , p. r. 2005 , a&a , 444 , l45 lccc other name & hd 131156 & hd 131156b & + spect . type & g8 v & k4 v & + dist . ( pc ) & 6.70 & 6.70 & 1 + radius ( r@xmath8)&0.83 & 0.61 & 2 + p@xmath56 ( days ) & 6.2 & 11.5 & 3 + @xmath57 & 28.86 & 27.97 & 4 + @xmath57 & 28.91 & 28.17 & + @xmath58 ( @xmath59 ) & 5 & 5 & 5 + @xmath60 & 17.92 & 17.92 & 6 + @xmath61}$ ] & @xmath62 & ... & 7 + @xmath63}$ ] & @xmath64 & ... & 7 + @xmath65}$ ] & @xmath66 & ... & 7 + @xmath67}$ ] & @xmath62 & ... & 7 + @xmath68}$ ] & @xmath66 & ... & 7 + lcccc si xiii & 6.648 & 6.99 & @xmath69 & @xmath70 + si xiii & 6.688 & 6.99 & & + si xiii & 6.740 & 6.99 & & + mg xii & 8.419 & 7.11 & @xmath71 & @xmath72 + mg xii & 8.425 & 7.11 & & + mg xi & 9.169 & 6.80 & @xmath73 & @xmath74 + mg xi & 9.231 & 6.80 & & + mg xi & 9.314 & 6.79 & & + ne x & 12.132 & 6.87 & @xmath75 & @xmath76 + ne x & 12.138 & 6.87 & & + fe xvii & 12.264 & 6.62 & @xmath77 & @xmath78 + fe xxi & 12.285 & 6.98 & & + ne ix & 13.447 & 6.58 & @xmath79 & @xmath80 + ne ix & 13.553 & 6.58 & & + ne ix & 13.699 & 6.58 & @xmath81 & @xmath82 + fe xviii & 14.203 & 6.74 & @xmath83 & @xmath84 + fe xviii & 14.208 & 6.74 & & + fe xvii & 15.015 & 6.59 & @xmath85 & @xmath86 + fe xvii & 15.262 & 6.59 & @xmath87 & @xmath88 + o viii & 15.176 & 6.65 & & + fe xix & 15.198 & 6.83 & & + o viii & 16.006 & 6.63 & @xmath89 & @xmath90 + fe xviii & 16.005 & 6.73 & & + fe xviii & 16.072 & 6.73 & & + fe xvii & 16.778 & 6.58 & @xmath91 & @xmath92 + fe xvii & 17.053 & 6.58 & @xmath93 & @xmath94 + fe xvii & 17.098 & 6.58 & & + o vii & 18.627 & 6.34 & @xmath95 & @xmath96 + o viii & 18.967 & 6.59 & @xmath97 & @xmath98 + o viii & 18.973 & 6.59 & & + o vii & 21.602 & 6.32 & @xmath99 & @xmath100 + o vii & 21.807 & 6.32 & @xmath101 & @xmath102 + o vii & 22.101 & 6.31 & @xmath103 & @xmath104 + n vii & 24.779 & 6.43 & @xmath105 & @xmath106 + n vii & 24.785 & 6.43 & & + c vi & 33.734 & 6.24 & @xmath107 & @xmath108 + c vi & 33.740 & 6.24 & & + s xiii & 35.667 & 6.43 & @xmath109 & @xmath110 + si xi & 43.763 & 6.25 & @xmath111 & @xmath112 + si xii & 44.019 & 6.44 & @xmath113 & @xmath114 + si xii & 44.165 & 6.44 & @xmath115 & @xmath116 + si xii & 45.521 & 6.44 & @xmath117 & @xmath118 + si xii & 45.691 & 6.44 & @xmath119 & @xmath120 + fe xvi & 46.661 & 6.43 & @xmath121 & @xmath122 + fe xvi & 46.718 & 6.43 & & + si xi & 49.222 & 6.24 & @xmath123 & @xmath124 + fe xvi & 50.361 & 6.43 & @xmath125 & @xmath126 + fe xvi & 50.565 & 6.43 & @xmath127 & @xmath128 + si x & 50.524 & 6.15 & & + si x & 50.691 & 6.15 & @xmath129 & @xmath130 + si xi & 52.298 & 6.24 & @xmath131 & @xmath132 + fe xv & 52.911 & 6.32 & @xmath133 & @xmath134 + fe xvi & 54.127 & 6.43 & @xmath135 & @xmath136 + fe xvi & 54.710 & 6.43 & @xmath137 & @xmath138 + mg x & 57.876 & 6.22 & @xmath139 & @xmath140 + mg x & 57.920 & 6.22 & & + fe xv & 59.405 & 6.32 & @xmath141 & @xmath142 + fe xvi & 62.872 & 6.42 & @xmath143 & @xmath144 + mg x & 63.295 & 6.21 & @xmath145 & @xmath146 + fe xvi & 63.711 & 6.42 & @xmath147 & @xmath148 + fe xvi & 66.249 & 6.42 & @xmath149 & @xmath128 + fe xvi & 66.357 & 6.42 & & + fe xv & 69.682 & 6.32 & @xmath150 & @xmath151 + fe xv & 69.941 & 6.32 & @xmath152 & @xmath140 + fe xv & 69.987 & 6.32 & & + fe xv & 70.054 & 6.32 & & + mg ix & 72.312 & 6.02 & @xmath153 & @xmath154 + fe xv & 73.472 & 6.32 & @xmath155 & @xmath156 + fe xvi & 76.497 & 6.42 & @xmath157 & @xmath158 + ne viii & 88.082 & 5.96 & @xmath159 & @xmath160 + ne viii & 88.120 & 5.96 & & + fe xviii & 93.923 & 6.68 & @xmath161 & @xmath162 + ne viii & 98.116 & 5.94 & @xmath163 & @xmath164 + ne viii & 98.260 & 5.94 & @xmath165 & @xmath166 + fe xix & 101.550 & 6.80 & @xmath167 & @xmath168 + fe xviii & 103.937 & 6.68 & @xmath169 & @xmath170 + fe xix & 108.355 & 6.79 & @xmath171 & @xmath164 + fe xxii & 117.180 & 7.03 & @xmath172 & @xmath173 + fe xx & 121.845 & 6.88 & @xmath174 & @xmath175 + fe xxi & 128.752 & 6.95 & @xmath176 & @xmath177 + fe xx & 132.840 & 6.88 & @xmath178 & @xmath179 + fe xxiii & 132.906 & 7.12 & & + fe ix & 171.073 & 5.95 & @xmath180 & @xmath181 + cccc @xmath182 & @xmath183 & @xmath184 & 9.4 + @xmath185 & 0.7 & 0.8 & 1 + @xmath186}$ ] & @xmath187 & @xmath188 & @xmath189 + @xmath190}$ ] & @xmath191 & @xmath192 & 1.02 + @xmath193}$ ] & @xmath194 & ... & 0.42 + @xmath195}$ ] & @xmath196 & @xmath197 & 1.33 + @xmath198}$ ] & @xmath199 & @xmath200 & 0.58 + @xmath201}$ ] & @xmath202 & @xmath203 & 0.08 + @xmath204}$ ] & @xmath205 & @xmath206 & 0.05 + @xmath207}$ ] & @xmath208 & ... & @xmath209 +	on 2008 may 2 , _ chandra _ observed the x - ray spectrum of @xmath0 boo ( g8 v+k4 v ) , resolving the binary for the first time in x - rays and allowing the coronae of the two stars to be studied separately . with the contributions of @xmath0 boo a and b to the system s total x - ray emission now observationally established ( 88.5% and 11.5% , respectively ) , consideration of mass loss measurements for gk dwarfs of various activity levels ( including one for @xmath0 boo ) leads to the surprising conclusion that @xmath0 boo b may dominate the wind from the binary , with @xmath0 boo a s wind being very weak despite its active corona . emission measure distributions and coronal abundances are computed for both stars and compared with _ _ measurements of other moderately active stars with g8-k5 spectral types , all of which exhibit a narrow peak in emission measure near @xmath1 , indicating that the coronal heating process in these stars has a strong preference for this temperature . as is the case for the sun and many other stars , our sample of stars shows coronal abundance anomalies dependent on the first ionization potential ( fip ) of the element . we see no dependence of the degree of `` fip effect '' on activity , but there is a dependence on spectral type , a correlation that becomes more convincing when moderately active main sequence stars with a broader range of spectral types are considered . this clear dependence of coronal abundances on spectral type weakens if the stellar sample is allowed to be contaminated by evolved stars , interacting binaries , or extremely active stars with @xmath2 , explaining why this correlation has not been recognized in the past .
a heterogeneous material ( medium ) is one that composed of domains of different materials or phases ( e.g. , a composite ) or the same material in different states ( e.g. , a polycrystal ) . such materials are ubiquitous ; examples include sandstones , granular media , animal and plant tissue , gels , foams and concrete . the microstructures of heterogeneous materials can be only characterized statistically via various types of @xmath1-point correlation functions @xcite . the effective transport , mechanical , and electromagnetic properties of heterogeneous materials are known to be dependent on an infinite set of correlation functions that statistically characterize the microstructure @xcite . _ reconstruction _ of heterogeneous materials from a knowledge of limited microstructural information ( a set of lower - order correlation functions ) is an intriguing inverse problem @xcite . an effective reconstruction procedure enables one to generate accurate structures and subsequent analysis can be performed on the image to obtain macroscopic properties of the materials ; see , e.g. , ref . this provides a nondestructive means of estimating the macroscopic properties : a problem of important technological relevance . another useful application is reconstruction of a three - dimensional structure of the heterogeneous material using information extracted from two - dimensional plane cuts through the material @xcite . such reconstructions are of great value in a wide variety of fields , including petroleum engineering , biology and medicine , because in many cases one only has two - dimensional information such as a micrograph or image . generating realizations of heterogeneous materials from a set of hypothetical correlation functions is often referred to as a _ construction _ problem . a successful means of construction enables one to identify and categorize materials based on their correlation functions . one can also determine how much information is contained in the correlation functions and test realizability of various types of hypothetical correlation functions . furthermore , an effective ( re)construction procedure can be employed to investigate any physical phenomena where the understanding of spatiotemporal patterns is fundamental , such as in turbulence @xcite . a popular ( re)construction procedure is based on the use of gaussian random fields : successively passing a normalized uncorrelated random gaussian field through a linear and then a nonlinear filter to yield the discrete values representing the phases of the structure . the mathematical background used in the statistical topography of gaussian random fields was originally established in the work of rice @xcite . many variations of this method have been developed and applied since then @xcite . the gaussian - field approach assumes that the spatial statistics of a two - phase random medium can be completely described by specifying only the volume fraction and standard two - point correlation function @xmath0 , which gives the probability of finding two points separated by vector distance @xmath2 in one of the phases @xcite . however , to reproduce gaussian statistics it is not enough to impose conditions on the first two cumulants only , but also to simultaneously ensure that higher - order cumulants vanish @xcite . in addition , the method is not suitable for extension to non - gaussian statistics , and hence is model dependent . recently , torquato and coworkers have introduced another stochastic ( re)construction technique @xcite . in this method , one starts with a given , arbitrarily chosen , initial configuration of random medium and a set of target functions . the medium can be a dispersion of particle - like building blocks @xcite or , more generally , a digitized image @xcite . the target functions describe the desirable statistical properties of the medium of interest , which can be various correlation functions taken either from experiments or theoretical considerations . the method proceeds to find a realization ( configuration ) in which calculated correlation functions best match the target functions . this is achieved by minimizing the sum of squared differences between the calculated and target functions via stochastic optimization techniques , such as the simulated annealing method @xcite . this method is applicable to multidimensional and multiphase media , and is highly flexible to include any type and number of correlation functions as microstructural information . it is both a generalization and simplification of the aforementioned gaussian - field ( re)construction technique . moreover , it does not depend on any particular statistics @xcite . there are many different types of statistical descriptors that can be chosen as target functions @xcite ; the most basic one is the aforementioned two - point correlation function @xmath0 , which is obtainable from small - angle x - ray scattering @xcite . however , not every hypothetical two - point correlation function corresponds to a realizable two - phase medium @xcite . therefore , it is of great fundamental and practical importance to determine the necessary conditions that realizable two - point correlation functions must possess @xcite . shepp showed that convex combinations and products of two scaled autocovariance functions of one - dimensional media ( equivalent to two - point correlation functions ; see definition below ) satisfy all known necessary conditions for a realizable scaled autocovariance function @xcite . more generally , we will see that a hypothetical function obtained by a particular combination of a set of realizable scaled autocovariance functions corresponding to @xmath3-dimensional media is also realizable . in this paper , we generalize shepp s work and argue that given a complete two - point correlation function space , @xmath0 of any statistically homogeneous material can be expressed through a map on a selected set of bases of the function space . we collect all known necessary conditions of realizable two - point correlation functions and formulate a new conjecture . we also provide new examples of realizable two - point correlation functions and suggest a set of analytical basis functions . we further discuss an exact mathematical formulation of the ( re)construction problem and show that @xmath0 can not completely specify a two - phase heterogeneous material alone , apart from the issue of chirality . moreover , we devise an efficient and isotropy - preserving construction algorithm to generate realizations of materials from their two - point correlation functions . subsequent analysis can be performed on the generated images to estimate desired macroscopic properties that depend on @xmath0 , including both linear @xcite and nonlinear @xcite behavior . these developments are integrated here into a general scheme that enables one to model and categorize heterogeneous materials via two - point correlation functions . although the general scheme is applicable in any space dimension , we will mainly focus on two - dimensional media here . in the second part of this series of two papers @xcite , we will provide algorithmic details and applications of our general scheme . the rest of this paper is organized as follows : in sec . ii , we briefly introduce the basic quantities used in the description of two - phase random media . in sec . iii , we gather all the known necessary conditions for realizable two - point correlation functions and make a conjecture on a new possible necessary condition based on simulation results . in sec . iv , we propose a general form through which the scaled autocovariance functions can be expressed by a set of chosen basis functions and discuss the choice of basis functions . in sec . v , we formulate the ( re)construction problem using rigorous mathematics and show that @xmath0 alone can not completely specify a two - phase random medium . thus , it is natural to solve the problem by stochastic optimization method ( i.e. , simulated annealing ) . the optimization procedure and the lattice - point algorithm are also discussed . in sec . vi , we provide several illustrative examples . in sec . vii , we make concluding remarks . the ensuing discussion leading to the definitions of the @xmath1-point correlation functions follows closely the one given by torquato @xcite . consider a realization of a two - phase random heterogeneous material within @xmath3-dimensional euclidean space @xmath4 . to characterize this binary system , in which each phase has volume fraction @xmath5 ( @xmath6 ) , it is customary to introduce the indicator function @xmath7 defined as @xmath8 where @xmath9 is the region occupied by phase @xmath10 and @xmath11 is the region occupied by the other phase . the statistical characterization of the spatial variations of the binary system involves the calculation of @xmath1-point correlation functions : @xmath12 where the angular brackets @xmath13 denote ensemble averaging over independent realizations of the random medium . for _ statistically homogeneous _ media , the @xmath1-point correlation function depends not on the absolute positions but on their relative displacements , i.e. , @xmath14 for all @xmath15 , where @xmath16 . thus , there is no preferred origin in the system , which in eq . ( [ eq1002 ] ) we have chosen to be the point @xmath17 . in particular , the one - point correlation function is a constant everywhere , namely , the volume fraction @xmath5 of phase @xmath10 , i.e. , @xmath18 and it is the probability that a randomly chosen point in the medium belongs to phase @xmath10 . for _ statistically isotropic _ media , the @xmath1-point correlation function is invariant under rigid - body rotation of the spatial coordinates . for @xmath19 , this implies that @xmath20 depends only on the distances @xmath21 ( @xmath22 ) . for @xmath23 , it is generally necessary to retain vector variables because of chirality of the medium . the two - point correlation function @xmath24 defined as @xmath25 is one of the most important statistical descriptors of random media . it also can be interpreted as the probability that two randomly chosen points @xmath17 and @xmath26 both lie in phase @xmath10 . for _ statistical homogeneous _ and _ isotropic _ media , @xmath27 only depends on scalar distances , i.e. , @xmath28 where @xmath29 . global information about the surface of the @xmath10th phase may be obtained by ensemble averaging the gradient of @xmath7 . since @xmath30 is different from zero only on the interfaces of the @xmath10th phase , the corresponding specific surface @xmath31 defined as the total area of the interfaces divided by the volume of the medium is given by @xcite @xmath32 note that there are other higher - order surface correlation functions which are discussed in detail by torquato @xcite . the calculation of higher - order correlation functions encounters both analytical and numerical difficulties , and very few experimental results needed for comparison purposes are available so far . however , their importance in the description of collective phenomena is indisputable . a possible pragmatic approach is to study more complex lower - order correlation functions ; for instance , the two - point cluster function @xmath33 defined as the probability that two randomly chosen points @xmath17 and @xmath26 belong to the same cluster of phase @xmath10 @xcite ; or the lineal - path function @xmath34 defined as the probability that the entire line segment between points @xmath17 and @xmath26 lies in phase @xmath10 @xcite . @xmath33 and @xmath34 of the reconstructed media are sometimes computed to study the non - uniqueness issue of the reconstruction @xcite . the task of determining the necessary and sufficient conditions that @xmath35 must possess is very complex . in the context of stochastic processes in time ( one - dimensional processes ) , it has been shown that the autocovariance functions must not only meet all the necessary conditions we will present in this section but another condition on `` corner - positive '' matrices @xcite . since little is known about corner - positive matrices , this theorem is very difficult to apply in practice . thus , when determining whether a hypothetical function is realizable or not , we will first check all the necessary conditions collected here and then use the construction technique to generate realizations of the random medium associated with the hypothetical function as further verification . here we collect all of the known necessary conditions on @xmath36 @xcite . for a two - phase statistically homogeneous medium , the two - point correlation function for phase 2 is simply related to the corresponding function for phase 1 via the expression @xmath37 and the _ autocovariance _ function @xmath38 for phase 1 is equal to that for phase 2 . generally , for @xmath39 , @xmath40 and in the absence of any _ long - range _ order , @xmath41 an important necessary condition of realizable @xmath35 for a two - phase statistically homogeneous medium with @xmath3 dimensions is that the @xmath3-dimensional fourier transform of the autocovariance function @xmath42 , denoted by @xmath43 must be non - negative for all wave vectors @xcite , i.e. , for all @xmath44 @xmath45 this non - negativity result is sometimes called the wiener - khintchine condition , which physically results since @xmath43 is proportional to the scattered radiation intensity . the two - point correlation function must satisfy the following bounds for all @xmath2 @xmath46 and the corresponding bounds on the autocovariance function are given by @xmath47 a corollary of eq . ( [ eq113 ] ) recently derived by torquato @xcite states that the infimum of any two - point correlation function of a statistically homogeneous medium must satisfy the inequalities @xmath48 \le \phi_i^2.\ ] ] another necessary condition on @xmath35 in the case of statistically homogeneous and isotropic media , i.e. , when @xmath35 is dependent only the distance @xmath49 , is that its derivative at @xmath50 is strictly negative for all @xmath51 : @xmath52 this is a consequence of the fact that slope at @xmath53 is proportional to the negative of the specific surface @xcite . taking that it is axiomatic that @xmath54 is an even function , i.e. , @xmath55 , then it is non - analytic at the origin . a lesser - known necessary condition for statistically homogeneous media is the so - called `` triangular inequality '' that was first derived by shepp @xcite and later rediscovered by matheron @xcite : @xmath56 where @xmath57 . note that if the autocovariance @xmath58 of a statistically homogeneous and isotropic medium is monotonically decreasing , nonnegative and convex ( i.e. , @xmath59 ) , then it satisfies the triangular inequality eq . ( [ eq115 ] ) . the triangular inequality implies several point - wise conditions on the two - point correlation function . for example , for statistically homogeneous and isotropic media , the triangular inequality implies the condition given by eq . ( [ eq114 ] ) , the fact that the steepest descent of the two - point correlation function occurs at the origin @xcite : @xmath60 and the fact that @xmath61 must be convex at the origin @xcite : @xmath62 torquato @xcite showed that the triangular inequality is actually a special case of the more general condition : @xmath63 where @xmath64 ( @xmath65 and @xmath66 is odd ) . note that by choosing @xmath67 ; @xmath68 , @xmath69 , eq . ( [ eq115 ] ) can be rediscovered . if @xmath67 ; @xmath70 are chosen instead , another `` triangular inequality '' can be obtained , i.e. , @xmath71 where @xmath57 . equation ( [ eq1151 ] ) was first derived by quintanilla @xcite . equation ( [ eq118 ] ) is a much stronger necessary condition that implies that there are other necessary conditions beyond those identified thus far . however , eq . ( [ eq118 ] ) is difficult to check in practice , because it does not have a simple spectral analog . one possible method is to randomly generate a set of @xmath66 points and compute the value of @xmath72 . among these values of @xmath73 , select the largest @xmath66 ones and set their coefficients @xmath74 equal to @xmath75 . thus , we have @xmath66 equations for @xmath66 @xmath76 s . then we can substitute the solved @xmath76 s into eq . ( [ eq118 ] ) and check the inequality . if the inequality holds , then we can generate several different sets of random points and test the inequality in the same way . @xmath77{hypo_s2.eps}\\ \mbox{\bf ( a ) } \\\\\\ \includegraphics[width=5cm , keepaspectratio]{hypo_config.eps}\\ \mbox{\bf ( b ) } \end{array}$ ] besides the aforementioned explicit necessary conditions , we find in two - dimensional simulations that the value of the second peak of a nonmonotonic @xmath78 for a statistically homogeneous and isotropic two - dimensional medium is always smaller than that of the medium composed of circular disks on a triangular lattice at fixed volume fractions , i.e. , @xmath79 where @xmath80 and @xmath81 denote the positions of the second peak for the two media , respectively and the superscript `` @xmath82 '' denotes the medium composed of disks on a triangular lattice . a hypothetical damped - oscillating two - point correlation function with an artificially higher second peak than that of the medium composed of disks on a triangular lattice at fixed volume fractions is tested by the construction algorithm . the results in fig . [ fig1 ] show that the structure for which @xmath83 ( `` @xmath84 '' denotes the black phase ) best matches the target function indeed has its `` particles '' arranged on triangular lattice while the second peak of the target still can not be reached . here we make the conjecture that for any @xmath3-dimensional statistically homogeneous and isotropic medium with a two - point correlation function @xmath78 that is nonmonotonic in @xmath85 , the value of the first peak of its @xmath61 away from origin is bounded from above by the value of the first peak of the two - point correlation function associated with the densest packings of @xmath3-dimensional identical hard spheres at fixed a volume fraction , i.e. , @xmath86 where the superscript `` @xmath87 '' denotes closest packings of spheres , for the first three dimensions , they are regular array of hard rods , hard disks on triangular lattice , and hard spheres on face - centered cubic lattice , respectively . for @xmath88 and @xmath89 , the densest packings are believed to be four- and five - dimensional checkerboard lattice packing , respectively @xcite . note that this conjectured condition could be a corollary of eq . ( [ eq118 ] ) or some other unknown necessary conditions . it is first shown by shepp @xcite that the convex combination and product of two realizable _ scaled autocovariance functions _ for _ one - dimensional _ statistically homogeneous media satisfy all known necessary conditions , i.e. , @xmath90 where @xmath91 ( @xmath92 ) , @xmath93 and the superscripts `` @xmath94 '' and `` @xmath95 '' denote `` combination '' and `` product '' , respectively . the scaled autocovariance function @xmath96 of a statistically homogeneous material is defined as @xcite @xmath97 the necessary conditions for realizable scaled autocovariance function @xmath96 can be obtained from eq . ( [ eq120 ] ) and the equations by which the necessary conditions for realizable two - point correlation function @xmath35 are given . from eqs . ( [ eq115 ] ) and ( [ eq1151 ] ) , we can obtain the triangular inequalities for @xmath96 , respectively , @xmath98 @xmath99 moreover , the bounds of @xmath96 become @xmath100 \le f({\bf r } ) \le 1,\ ] ] and the corollary eq . ( [ eq1132 ] ) is equivalent to @xmath101 \le f_{\inf } \le 0,\ ] ] where @xmath102 is the infimum of @xmath96 . our focus in this paper will be hypothetical continuous functions @xmath103 that are dependent only on the scalar distance @xmath104 which could potentially correspond to statistically homogeneous and isotropic media without long range order , i.e. , @xmath105 @xmath103 is also absolutely integrable so that the fourier transform of @xmath103 exists and is given by @xmath106 where @xmath107 and @xmath108 is the bessel function of order @xmath109 . generalization of eq . ( [ eq125 ] ) to higher dimensions is straightforward . ( @xmath111 ) are the scaled autocovariance functions for @xmath3-dimensional statistically homogeneous and isotropic media , then the convex combination @xmath112 and product @xmath113 defined as @xmath114 satisfy all known necessary conditions , where @xmath115 ( @xmath111 ) and @xmath116 . equation ( [ eq126 ] ) is of great fundamental and practical importance . on the one hand , it enables us to construct new realizable two - point correlation functions with properties of interest , corresponding to structures of interest , from a set of known functions . thus , one can categorize microstructures with the set of known functions and the proper combinations . on the other hand , suppose that we can find a `` full '' set of those _ basis _ scaled autocovariance functions @xmath117 , then the scaled autocovariance function of any statistically homogeneous and isotropic medium can be expressed in term of the combinations of the basis functions , i.e. , @xmath118 \equiv \wp [ f_1(r),f_2(r), ... ,f_m(r)],\ ] ] where @xmath119 denotes a map composed of convex combinations and products of @xmath117 . for example , for @xmath120 , a _ possible _ explicit form for @xmath119 is @xmath121 = \alpha_1 f_1(r ) + \alpha_2 [ \beta_1 f_2(r ) + \beta_2 f_3(r ) ] + \alpha_3 [ f_4(r)f_5(r)],\ ] ] where @xmath122 and @xmath123 ( @xmath124 ) . once the scaled autocovariance function ( or equivalently the two - point correlation function ) of a medium is known , an effective reconstruction procedure enables one to generate accurate structures at will , and subsequent analysis can be performed on the image to obtain desired macroscopic properties of the medium . in other words , the medium is actually _ modeled _ by a set of basis scaled autocovariance functions @xmath117 and a particular map @xmath125 $ ] . there could be different choices of the basis functions ( like different basis choices of a hilbert space ) , and we would like the basis functions to have nice mathematical properties , such as simple analytical forms . let @xmath126 denotes our choice of the basis functions . thus , the media can be represented merely by different maps @xmath127 s . note that a hypothetical two - point correlation function corresponds to a hypothetical map @xmath128 and effective construction algorithms can be used to test the realizability of @xmath128 . a systematic way of determining the basis functions @xmath129 is not available yet . here we take the first step to determine the bases by considering certain known realizable analytical two - point correlation functions and the corresponding scaled autocovariance functions . for convenience , we categorize these functions into three families : ( @xmath10 ) _ monotonically decreasing _ functions ; ( @xmath130 ) _ damped - oscillating _ functions ; and ( @xmath131 ) functions of _ known constructions_. the family of monotonically decreasing functions includes the simple exponentially decreasing function introduced by debye @xcite and polynomial functions . the former is given by @xmath132 where @xmath133 is a correlation length , corresponding to structures in which one phase consists of `` random shapes and sizes '' @xcite ( shown in fig . [ fig2 ] ) . it is now known that certain types of space tessellations have autocovariance functions given by eq . ( [ eq129 ] ) @xcite . we have referred to this class of structures as _ debye random media _ @xcite . @xmath77{debye_obj2.eps}\\ \mbox{\bf ( a)}\\\\\\ \includegraphics[width=5cm , keepaspectratio]{debye_config.eps}\\ \mbox{\bf ( b ) } \end{array}$ ] another example of monotonically decreasing functions is the family of polynomials of order @xmath1 ( @xmath15 ) given by @xmath134 where @xmath133 is the correlation length . the polynomial function of order 1 is shown to be realizable only for a statistically homogeneous two - phase medium in one dimension @xcite . we have constructed for the first time realizations of random media in dimensions @xmath135 that correspond to the polynomial function of order @xmath1 with very high numerical precision using the yeong - torquato construction technique @xcite . however , for dimensions higher than @xmath1 , eq . ( [ eq130 ] ) violates eq . ( [ eq124 ] ) . we will henceforth assume that the polynomial functions of order @xmath1 are realizable in dimensions @xmath135 . an example of the family of damped - oscillating functions is given by @xcite @xmath136 where the parameters @xmath137 and @xmath138 ( @xmath139 ) control the amplitude of the @xmath140 profile , @xmath141 is the wavenumber and @xmath142 is the phase angle . in general , the wavenumber should be a function of @xmath85 which could correspond to the different distances between successive neighbor shells of a crystalline material . note that @xmath141 and @xmath142 need to be carefully chosen such that @xmath143 satisfies all known necessary conditions . @xmath77{od_obj2.eps}\\ \mbox{\bf ( a)}\\\\\\ \includegraphics[width=5cm , keepaspectratio]{od_config.eps}\\ \mbox{\bf ( b ) } \end{array}$ ] the family of functions of known constructions includes scaled autocovariance functions of @xmath3-dimensional identical overlapping spheres @xcite and symmetric - cell materials @xcite . for overlapping spheres of radius @xmath144 , the scaled autocovariance function for the particle phase ( spheres ) is given by @xmath145-\phi_1 ^ 2}{\phi_1\phi_2},\ ] ] where @xmath146 and @xmath147 are volume fractions of the spheres and matrix respectively , @xmath148 is the number density of spheres , and @xmath149 is the union volume of two spheres of radius @xmath144 whose centers are separated by @xmath85 . for the first three space dimensions , the latter is respectively given by @xmath150 @xmath151\theta(2r - r),\ ] ] @xmath152\theta(2r - r),\ ] ] where @xmath153 is the heaviside function , and @xmath154 is the volume of a @xmath3-dimensional sphere of radius @xmath144 given by @xmath155 where @xmath156 is the gamma function . for @xmath157 and @xmath158 , @xmath159 and @xmath160 , respectively . @xmath77{cb_obj2.eps}\\ \mbox{\bf ( a)}\\\\\\ \includegraphics[width=5cm , keepaspectratio]{config14.eps}\\ \mbox{\bf ( b ) } \end{array}$ ] two - phase symmetric - cell materials are constructed by partitioning space into cells of arbitrary shapes and sizes , with cells being randomly designated as phase 1 and 2 with probability @xmath146 and @xmath147 from a uniform distribution @xcite . for such a statistically homogeneous and isotropic medium , the scaled autocovariance function is given by @xmath161 where @xmath162 is the probability that two points separated by distance @xmath85 are in the same cell . this quantity is only a function of cell shapes and sizes , depending on a dimensionless size - averaged intersection volume of two cells defined by @xmath163 where @xmath144 is the size parameter for each cell , @xmath164 is the size - averaged intersection volume of two cells whose centers are separated by @xmath85 and @xmath165 is the size - averaged single - cell volume . the random checkerboard is a very useful model of symmetric - cell material because its @xmath166 is known analytically for @xmath167 and @xmath168 @xcite . for @xmath167 , it is easy to verify that the probability of finding two points in the same one - dimensional cell is given by @xmath169 where @xmath133 is the length of the side of a square cell . note that eq . ( [ eq139 ] ) is just the polynomial function of order one [ cf . ( [ eq130 ] ) ] for one - dimensional homogeneous media . for @xmath170 , @xmath162 is given by @xmath171 , \quad\quad 0\le r\le a , } \\\\ \displaystyle { 1 - \frac{1}{\pi}\left[{2 + \left({\frac{r}{a}}\right)^2}\right ] + \frac{4}{\pi}\left[{\sqrt{\left({\frac{r}{a}}\right)^2 - 1 } - \cos^{-1}\left({\frac{a}{r}}\right)}\right ] , ~ a \le r \le \sqrt{2}a,}\\\\ 0 , \quad\quad\quad r \ge \sqrt{2 } a. \end{array}}\right.\ ] ] for @xmath172 , @xmath162 is given by @xmath173 where @xmath174[1-r\sin\theta\sin\phi][1-r\sin\theta\cos\phi]\\\quad\quad\quad\times\theta(1-r\cos\theta)\theta(1-r\sin\theta\sin\phi)\theta(1-r\sin\theta\cos\phi ) . \end{array}\ ] ] note that @xmath162 for @xmath172 does not have a simple analytical form . there are some other known scaled autocovariance functions for different types of materials ( e.g. , @xmath3-dimensional identical hard spheres @xcite ) and realizable hypothetical functions ( e.g. , complementary error function , see appendix ) . however we do not include these functions in our basis function set because they do not have simple analytical mathematical forms . note that some of the basis functions are dimension - dependent ( e.g. , @xmath140 and @xmath175 ) , and thus the proper forms of basis functions should be used for different dimensions . in the following discussion , without further specification , we will focus on two - dimensional cases . consider a digitized ( i.e. , pixelized ) representation of a heterogeneous material . different colored pixels ( in a discrete coloring scheme ) may have numerous interpretations . the image can reflect different properties , such as the geometry captured by a photographic image , topology of temperature and scalar velocity fields in fluids , distribution of magnitudes of electric and magnetic fields in the medium , or variations in chemo - physical properties of the medium . in the last case , typical examples are composite materials in which the different phases may have different thermal , elastic or electromagnetic properties , to name a few . our focus in this paper is two - dimensional , two - phase statistically homogeneous and isotropic random media composed of black and white pixels . such a system can be represented as a two - dimensional array , i.e. , @xmath176,\ ] ] where the integer @xmath177 categorizes the linear size of the system ( @xmath178 is the total number of pixels in the system ) and the entries @xmath179 ( @xmath180 ) can only take the value of @xmath181 or @xmath182 , which correspond to the white and black phases , respectively . note that eq . ( [ eq201 ] ) is only an abstract representation and the real morphological configuration of the medium also depends on the choice of _ lattices_. for example , as shown in fig . [ fig5 ] , the isotropic medium composed of overlapping disks generated on a square lattice and the anisotropic medium composed of orientated ellipses generated on a triangular lattice have the same array presentation . a vector distance in the digitized medium can be uniquely expressed as @xmath183 where @xmath184 and @xmath185 are lattice vectors for the particular lattice and @xmath186 , @xmath187 are integers . for example , for square lattice , @xmath188 , @xmath189 , where @xmath190 , @xmath191 are unit vectors along horizontal and vertical directions , respectively ; while for triangular lattice @xmath192 , @xmath189 . @xmath193{new2.eps } & \includegraphics[width=4cm , keepaspectratio]{new1.eps } \\ \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ] without loss of generality , we choose the black phase to be the phase of interest and assume _ periodic boundary condition _ is applied , which is commonly used in computer simulations . the two - point correlation function @xmath0 of the black phase can be calculated based on its probabilistic nature , i.e. , the probability of finding two points separated by the vector distance @xmath2 in the same phase . the value of two - point correlation function for a particular @xmath194 is given by @xmath195 where @xmath179 are entries of @xmath196 defined in eq . ( [ eq201 ] ) , @xmath186 and @xmath187 are integers satisfying @xmath197 $ ] due to _ minimum image distance _ convention . for statistically isotropic media , the two - point correlation function only depends on the magnitude of @xmath2 , i.e. , @xmath198 , thus we have @xmath199}{\omega n^2}},\ ] ] where @xmath200}\right\},\ ] ] and @xmath201 is the number of elements of set @xmath202 . it is well known that the two - point correlation function can not completely specify a two - phase heterogeneous material alone . here , we provide a proof of this statement for two - dimensional statistically homogeneous digitized media . note that the proof trivially extends to any dimension . suppose we already know the value of @xmath0 for every vector distance @xmath2 , using eq . ( [ eq203 ] ) , we can obtain a set of equations of @xmath179 , i.e. , for each @xmath194 @xmath203 since the digitized medium is represented by @xmath196 , once we obtain all the entries @xmath179 ( the unknowns in eq . ( [ eq206 ] ) ) , the medium is ( re)constructed . in eq . ( [ eq206 ] ) , the number of unknowns @xmath204 equals @xmath178 and the number of equations @xmath205 equals the number of all possible vector distances in the digitized medium . to calculate @xmath205 , all possible different combinations of integers @xmath186 and @xmath187 subjected to @xmath197 $ ] need to be considered . this quantity is given by @xmath206 ^ 2 - 8[n/2 ] + 6 $ ] , which is smaller than the number of unknowns @xmath207 for normal - sized systems ( i.e. , @xmath208 ) . similar proof is applied to the case when we average over the angles of vector @xmath2 to yield @xmath78 that depends only on the radius distance @xmath85 . the angle - averaged equations of @xmath179 are given by @xmath209 - \omega n^2 s_2(r ) = 0,\ ] ] where @xmath202 is given by eq . ( [ eq205 ] ) . the number of equations is even smaller for the angle - averaged case , while the number of unknowns does not change . the analysis shows that one could never find a unique solution of @xmath179 from neither eq . ( [ eq206 ] ) nor eq . ( [ eq207 ] ) unless some assumptions have been made that reduce @xmath204 such that @xmath210 . for example , in an interesting model to study spatial distribution of algae , the algae are put on top of each other in order to reduce the unknowns @xcite . in general cases , one could use the stochastic optimization procedure ( i.e. , simulated annealing ) @xcite to find solutions of eq . ( [ eq206 ] ) and eq . ( [ eq207 ] ) . note that although the aforementioned proofs focus on two - dimensional media , they trivially extend to any dimensions ( e.g. , @xmath211 ) . generally , consider a given set of correlation functions @xmath212 of the phase of interest that provides partial information on the random medium . the index @xmath213 is used to denote the type of correlation functions . note that the set of @xmath214 should not be confused with the basis function set @xmath119 , the former contains correlation functions of different type , i.e. , two - point correlation function , lineal - path function , two - point cluster function , etc . , while the latter contains basis functions through which the scaled autocovariance function of the medium of interest can be expressed . the information contained in @xmath214 could be obtained either from experiments or it could represent a hypothetical medium based on simple models . in both cases we would like to generate the underlying micro - structure with a specified set of correlation functions . in the former case , the formulated inverse problem is frequently referred to as a `` reconstruction '' procedure , and in the latter case as a `` construction '' . as we have noted earilier , it is natural to formulate the construction or reconstruction problem as an optimization problem @xcite . the discrepancies between the statistical properties of the best generated structure and the imposed ones is minimized . this can be readily achieved by introducing the `` energy '' function @xmath215 defined as a sum of squared differences between target correlation functions , which we denote by @xmath216 , and those calculated from generated structures , i.e. , @xmath217 ^ 2.\ ] ] note that for every generated structure ( configuration ) , these is a set of corresponding @xmath218 . if we consider every structure ( configuration ) as a `` state '' of the system , @xmath215 can be considered as a function of the states . the optimization technique suitable for the problem at hand is the method of simulated annealing @xcite . it is a popular method for the optimization of large - scale problems , especially those where a global minimum is hidden among many local extrema . the concept of finding the lowest energy state by simulated annealing is based on a well - known physical fact : if a system is heated to a high temperature @xmath219 and then slowly cooled down to absolute zero , the system equilibrates to its ground state . at a given temperature @xmath219 , the probability of being in a state with energy @xmath215 is given by the boltzmann distribution @xmath220 . at each annealing step @xmath221 , the system is allowed to evolve long enough to thermalize at @xmath222 . the temperature is then lowered according to a prescribed annealing schedule @xmath222 until the energy of the system approaches it ground state value within an acceptable tolerance . it is important to keep the annealing rate slow enough in order to avoid trapping in some metastable states . in our problem , the discrete configuration space includes the states of all possible pixel allocations . starting from a given state ( current configuration ) , a new state ( new configuration ) can be obtained by interchanging two arbitrarily selected pixels of different phases . this simple evolving procedure preserves the volume fraction of all involved phases and guarantees ergodicity in the sense that each state is accessible from any other state by a finite number of interchange steps . however , in the later stage of the procedure , biased and more sophisticated interchange rules , i.e. , surface optimization , could be used to improve the efficiency . we choose the metropolis algorithm as the acceptance criterion : the acceptance probability @xmath223 for the pixel interchange is given by @xmath224 where @xmath225 . the temperature @xmath219 is initially chosen so that the initial acceptance probability for a pixel interchange with @xmath226 averages approximately @xmath227 . an inverse logarithmic annealing schedule which decreases the temperature according to @xmath228 would in principle evolve the system to its ground state . however , such a slow annealing schedule is difficult to achieve in practice . thus , we will adopt the more popular and faster annealing schedule @xmath229 , where constant @xmath230 must be less than but close to one . this may yield suboptimal results , but , for practical purposes , will be sufficient . the convergence to an optimum is no longer guaranteed , and the system is likely to freeze in one of the local minima if the thermalization and annealing rate are not adequately chosen . the two - point correlation function of a statistically homogeneous and isotropic medium is the focus of this paper . in this case , eq . ( [ eq208 ] ) reduces to @xmath231 ^ 2.\ ] ] since for every configuration ( structure ) , the corresponding two - point correlation function needs to be computed , the efficiency of the construction or reconstruction is mainly determined by the efficiency of the @xmath36-sampling algorithm . furthermore , the properties of generated configurations ( structures ) , i.e. , isotropy of the medium , is also affected by the @xmath36-sampling algorithm . one of the most commonly used and efficient @xmath36-sampling algorithms is the _ orthogonal - sampling _ algorithm introduced by yeong and torquato @xcite . due to the isotropic nature of the medium , every sampling direction should be equivalent . for simplicity , two orthogonal directions ( usually the horizontal and vertical directions of a square lattice ) are chosen and the two - point correlation function is sampled along these directions and averaged . at each pixel interchange , only the values of @xmath78 sampled along the rows and columns that contain the interchange pixels are changed . thus , the complexity of the algorithm is @xmath232 , where @xmath177 is the linear size of the system . however , for certain media with long - range correlations , the generated media have microstructures with two orthogonal anisotropic directions due to the biased sampling . modifications of the orthogonal - sampling algorithm to preserve the isotropy of the underlying medium have been proposed , such as adding more sampling directions and using more isotropic lattices @xcite . cule and torquato i ntroduced a new isotropy - preserving _ fast fourier transform _ ( fft ) algorithm @xcite . at each pixel interchange step , the two - point correlation function @xmath0 containing angle information is calculated in momentum space using an efficient fft algorithm . since information of all directions is considered , the generated media always have the required isotropy structures . however , since the complexity of fft is @xmath233 , the algorithm is relatively time consuming . we have developed an efficient and isotropy - preserving algorithm , namely , the _ lattice - point _ algorithm by considering the black pixels as hard `` particles '' on a particular lattice . the two - point correlation function is then computed in a similar way of obtaining the _ pair correlation function _ @xmath234 for an isotropic point process @xcite . at each monte - carlo step , the randomly selected `` particle '' ( black pixel ) is given a random displacement subjected to the nonoverlapping constraint and the distances between the moved `` particle '' and all the other `` particles '' need to be recomputed . thus the complexity of the algorithm is @xmath232 . since all directions are effectively sampled , constructions based on the angle - averaged @xmath78 well preserve isotropy of the media . a detailed discussion and applications of the algorithm will be included in the second paper of this series @xcite . in this paper , we only provide several illustrative examples generated from this algorithm . as illustrative examples , we use the aforementioned ( re)construction techniques to investigate both deterministic , crystal - like structures , and random systems . we also study a hypothetical medium with the two - point correlation function obtained from convex combination of known ones . in the case of completely deterministic structures , the algorithm produces almost perfect reconstructions . however , the optimization of disordered structures is significantly harder . furthermore , we will see that for the media with long - range correlations , e.g. , a damped - oscillating @xmath78 , the orthogonal - sampling algorithm may produce unexpected anisotropy , while the lattice - point algorithm well preserves isotropy of the media . @xmath235{lattice.eps } \end{array}$ ] first , we consider specific two - dimensional and two - phase structure composed of a square array of nonoverlapping disks , as shown in fig . this morphology may be viewed as a cross section of two - phase materials containing rod- or fiber - like inclusions . various transport properties of these materials have been well explored because of their practical and theoretical importance in materials science @xcite . the regular structure is discretized by introducing an @xmath236 square lattice . the volume fractions of black and white phases are @xmath237 , @xmath238 , respectively . the target two - point correlation of the digitized medium is sampled using both the orthogonal and the lattice - point algorithm for comparison purpose . the simulations start from random initial configurations ( i.e. , random collections of black and white pixels ) , at some initial temperature @xmath239 , with fixed volume fractions @xmath5 . at each monte - carlo ( mc ) step , when an attempt to exchange two randomly chosen pixels with different colors ( or to randomly displace a chosen black pixel ) is made , @xmath78 is efficiently recomputed by using the orthogonal - sampling algorithm ( or the lattice - point algorithm ) . the set of constants @xmath240 specifies the annealing schedule : at each temperature , the system is thermalized until either @xmath241 mc moves are accepted or the total number of attempts to change the original configurations reaches the value @xmath242 . subsequently , the system temperature is decreased by the reduction factor @xmath230 , i.e. , @xmath243 . @xmath193{lattice_lp.eps } & \includegraphics[width=4cm , keepaspectratio]{lattice_ot.eps } \\ \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ] the reconstruction results are shown in fig . both of the algorithms are able to reproduce the exact global square - array arrangement of clusters of black pixels . this implies that the two - point correlation function of regular configurations contains enough structural information to properly characterize the long - range correlations . however , it is clear that the structure generated by the lattice - point algorithm has a better local arrangement of the pixels ( i.e. , the shape of the particles ) than that generated by the orthogonal - sampling algorithm . this is because the orthogonal algorithm only uses structural information along two directions , which is not sufficient to reproduce detailed local structures , while the lattice - point algorithm efficiently uses information along all possible directions . in this example , we will generate two - dimensional statistically homogeneous and isotropic random media with long - range correlations ( i.e. , nontrivial inter - particle interactions ) . examples of this type of media include low - density fluids and amorphous materials ( i.e. , porous media , randomly polymerized plastics , glass , etc . ) . a meaningful , yet nontrivial , two - point correlation function capturing these features is @xcite @xmath244 where @xmath245 . here @xmath246 and @xmath247 are two characteristic length scales . the overall exponential damping is controlled by the correlation length @xmath246 , determining the maximum correlations in the system . the constant @xmath247 determines oscillations in the term @xmath248 which also decays with increasing @xmath85 , such that @xmath247 can reduce the effective range of @xmath246 . interestingly , this hypothetical function is _ not _ exactly realizable , because it violates the convexity condition eq . ( [ eq117 ] ) at the origin , or more generally the triangular inequality eq . ( [ eq115 ] ) . however , we mainly focus on its damped - oscillating property and we will see that the algorithms are robust enough to detect violation of the convexity condition . @xmath193{sine_0.2_ortho.eps } & \includegraphics[width=4cm , keepaspectratio]{sine_0.5_otho.eps } \\ \mbox{\bf ( a1 ) } & \mbox{\bf ( a2)}\\\\\\ \includegraphics[width=4cm , keepaspectratio]{sine_lp_0.2.eps } & \includegraphics[width=4cm , keepaspectratio]{sine_lp_0.5.eps } \\ \mbox{\bf ( b1 ) } & \mbox{\bf ( b2 ) } \end{array}$ ] for comparison purposes , both the orthogonal - sampling algorithm and the lattice - point algorithm are used in the construction , the results are shown in fig . [ fig8 ] . at a lower density of the black phase @xmath146 , @xmath247 is manifested as a characteristic repulsion among different elements with diameter of order @xmath247 . the repulsion vanishes beyond the length scale @xmath246 . at a higher density , both length scales @xmath247 and @xmath246 are clearly noticeable in the distribution of the black and white phases . note that the structures generated by the orthogonal - sampling algorithm exhibit some anisotropy features , i.e. , containing stripes along @xmath249 degree directions , which implies that the orthogonal - sampling algorithm should be used with care in the case where the medium has long - range correlations . @xmath77{dampsine_0.2.eps } \end{array}$ ] the target two - point correlation function @xmath250 for @xmath251 and @xmath78 sampled from generated structures are shown in fig . it can be seen clearly that @xmath250 is non - convex at the origin and the largest discrepancies between @xmath78 and @xmath252 occur around the origin because @xmath78 satisfies eq . ( [ eq117 ] ) . this implies that our algorithms are robust enough and can be used to test realizability of hypothetical functions . note that the following classes of functions @xcite @xmath253 \quad\quad \alpha > 1,\ ] ] and @xmath254^{\beta -1 } } \quad\quad \beta \ge d,\ ] ] can not correpond to a two - phase medium in @xmath3 dimensions also because of violation of triangular inequality eq . ( [ eq121 ] ) . in this last example , we study a hypothetical statistically homogeneous and isotropic medium whose scaled autocovariance function is a convex combination of debye random medium function @xmath255 and damped - oscillating function @xmath143 , i.e. , @xmath256 where @xmath257 . since both @xmath255 and @xmath143 are independent of volume fractions , the medium with scaled autocovariance function @xmath103 has _ phase - inversion symmetry _ @xcite , i.e. , the structures with volume fraction of black pixels @xmath251 is statistically the same with those having volume fraction of white pixels @xmath258 , if the colors of the two phases are inverted in the latter . @xmath77{comb_f_0.25.eps } \end{array}$ ] @xmath259{cb0.25_0.1.eps } & \includegraphics[height=3cm , keepaspectratio]{cb0.25_0.3.eps } & \includegraphics[height=3cm , keepaspectratio]{cb0.25_0.5.eps } \\ \mbox{\bf ( a ) } & \mbox{\bf ( b ) } & \mbox{\bf ( c ) } \end{array}$ ] different constant pairs @xmath260 can be used to construct @xmath103 with required properties . in particular , we choose two pairs : @xmath261 and @xmath262 . the construction results obtained by application of the lattice - point algorithm are shown in figs . [ fig10 ] , [ fig11 ] and figs . [ fig12 ] , [ fig13 ] . for @xmath263 , @xmath143 is dominant in the combination . at lower densities , the generated structures resemble those with `` pure '' damped oscillationg two - point functions , i.e. , dispersions of particles , altough they contain more clusters . at higher densities , some stripe - like structures and several ( almost equal - sized ) clusters can be identified . for @xmath264 , @xmath255 is dominant in the combined @xmath103 . clusters with considerable sizes form even at low densities , which is a consequence of the large effective correlation length in @xmath255 . however , no stripe - like structures can be identified in the generated structures , since the contribution of @xmath143 is significantly supressed by its damping nature and the small combination constants . @xmath77{comb_f_0.75.eps } \end{array}$ ] @xmath259{cb0.75_0.1.eps } & \includegraphics[height=3cm , keepaspectratio]{cb0.75_0.3.eps } & \includegraphics[height=3cm , keepaspectratio]{cb0.75_0.5.eps } \\ \mbox{\bf ( a ) } & \mbox{\bf ( b ) } & \mbox{\bf ( c ) } \end{array}$ ] the results imply that even a simple combination of two basis functions enables one to obtain scaled autocovariance functions with properties of interest and to generate a variety of structures with controllable morphological features , e.g. , local `` particle '' shape and cluster size . in this paper , we have provided a general rigorous scheme to model and categorize two - phase statistically homogeneous and isotropic media . in particular , given a set of basis functions , we have shown that the medium can be modeled by a map @xmath119 composed of convex combination and product operations . the basis functions should be realizable but , if they are not , they should at least satisfy all the known necessary conditions for a realizable autocovariance function . we have gathered all the known necessary conditions and made a conjecture on a possible new condition based on simulation results . a systematic way of determining basis functions is not available yet . we proposed a set of basis functions with simple analytical forms that capture salient microstructural features of two - phase random media . we give for the first time a rigorous mathematical formulation of the ( re)construction problem and showed that the two - point correlation function alone can not completely specify a two - phase heterogeneous material . moreover , we devised an efficient and isotropy - preserving ( re)construction algorithm , namely , the lattice - point algorithm to generate realizations of materials based on the yeong - torquato technique . we also provided an example of non - realizable yet non - trivial two - point correlation function and showed that our algorithm can be used to test realizability of hypothetical functions . an example of generating hypothetical random media with combined realizable correlation functions was given as an application of our general scheme . we showed that even a simple combination of two basis functions enables one to produce media with a variety of microstructures of interest and therefore a means of categorizing microstructures . we are investigating applications of our general scheme in order to model real materials . we are also developing more efficient ( re)construction algorithms . there is a need for a theoretical and numerical analysis of the _ energy threshold _ of the algorithm , which is the aforementioned `` acceptable tolerance '' . this quantity provides an indication of the extent to which the algorithms have reproduced the target structure and it is directly related to the non - uniqueness issue of reconstructions @xcite . moreover , additional realizable basis functions are needed to construct a complete basis set . such work will be reported in our future publications . where @xmath267 and @xmath133 is the effective correlation length . it is easy to check that @xmath265 satisfies the known necessary conditions collected in this paper except for eq . ( [ eq118 ] ) , which can only be checked for a finite number of cases . realizations of the random medium associated with @xmath265 have been constructed using the yeong - torquato technique with very high numerical precision . thus , we believe @xmath265 to be a valid candidate for a realizable scaled autocovariance function .	heterogeneous materials abound in nature and man - made situations . examples include porous media , biological materials , and composite materials . diverse and interesting properties exhibited by these materials result from their complex microstructures , which also make it difficult to model the materials . yeong and torquato [ phys . rev . e * 57 * , 495 ( 1998 ) ] introduced a stochastic optimization technique that enables one to generate realizations of heterogeneous materials from a prescribed set of correlation functions . in this first part of a series of two papers , we collect the known necessary conditions on the standard two - point correlation function @xmath0 and formulate a new conjecture . in particular , we argue that given a complete two - point correlation function space , @xmath0 of any statistically homogeneous material can be expressed through a map on a selected set of bases of the function space . we provide new examples of realizable two - point correlation functions and suggest a set of analytical basis functions . we also discuss an exact mathematical formulation of the ( re)construction problem and prove that @xmath0 can not completely specify a two - phase heterogeneous material alone . moreover , we devise an efficient and isotropy - preserving construction algorithm , namely , the lattice - point algorithm to generate realizations of materials from their two - point correlation functions based on the yeong - torquato technique . subsequent analysis can be performed on the generated images to obtain desired macroscopic properties . these developments are integrated here into a general scheme that enables one to model and categorize heterogeneous materials via two - point correlation functions . we will mainly focus on the basic principles in this paper . the algorithmic details and applications of the general scheme are given in the second part of this series of two papers .
charmonium physics is always one of the interesting and intriguing fields of particle physics . charmonium provides us an excellent and simple system to study qcd , the production and decay mechanisms of heavy quarkonia and light hadron spectra from its decays , and can be treated nonrelativistically and perturbatively . using @xmath0 sample collected with the beijing spectrometer ( bes ) at bepc , the recent @xmath4 , @xmath5 and @xmath3 results are presented . the bes detector is described in detail in ref . 0.1 cm out of the charmonium decays , there exists a mysterious and longstanding @xmath6 puzzle of @xmath7 and @xmath1 decays @xcite , first revealed by mark - ii collaboration . in qcd , both @xmath1 and @xmath8 decays are expected to be dominated by annihilations into three gluons . based on the similarity between the @xmath1 and @xmath8 wave functions and the slow running of the strong interaction coupling constant @xmath9 , the non - relativistic perturbative qcd predicts @xcite : the ratios of the decays widths of @xmath4 into hadrons to those of @xmath7 into hadrons are : @xmath10 named as the `` @xmath11 '' rule . mark ii first observed the vector - pseudoscalar ( v for vector , p pseudoscalar , b baryon , a axial - vector and t tensor ) @xmath6 and @xmath12 channels are highly suppressed w.r.t . the @xmath13 expectation known as the @xmath6 puzzle . bes has confirmed the puzzle with much lower upper limit by a suppression factor of @xmath14 and observed new highly suppressed decay modes , meanwhile bes presents many observations ( table [ hadronic ] ) . from table [ hadronic ] , a large number of @xmath4 decay branching ratios have been measured , and most for the first time @xcite . and it is found that some vp , vpp , vt , pbb and ap decay modes are suppressed w.r.t . the @xmath13 rule , where vt modes @xcite is the first evidence for suppression other than vp . while for @xmath4 decays into ap modes @xcite , the normal decay @xmath15 , enhanced mode @xmath16 and suppressed decay @xmath17 are observed . the @xmath13 rule holds for the radiative decays into vt modes @xcite within errors , but it is suppressed for the decays into vp modes@xcite . now the suppressed decays also extend to the 3-body decays , but the @xmath13 rule holds for the baryonic decays @xcite . quite a few models @xcite have been put forward to explain the puzzle , but none of them is satisfactory . from the measured results , it has also been observed that the * strong double ozi ( dozi ) violation * in @xmath18 and the isospin violation in the decays between @xmath19 and the isospin - conserving and su(3)-allowed decay @xmath20 , and between the charged and neutral decays @xmath21 and @xmath22 . fig . [ phif0 ] shows the @xmath23 and @xmath24 invariant masses for @xmath25 and @xmath26 @xcite . also bes first measures @xmath27 and has precisely determined @xmath28 @xcite with errors half of those in pdg98 . -1.2 cm -0.0 cm .[hadronic ] @xmath4 decay branching ratios ( the unpublished results indicated by * are preliminary ) . [ cols="<,^,^,^",options="header " , ] -1.2 cm using the @xmath29 @xmath4 data sets at bes , a large number of @xmath1 and @xmath2 results have been presented , many of them for the first measurements and/or with the unprecedented precision . bes has observed new suppressed decay modes and first observed enhanced decay from @xmath4 among a large number of the normal hadronic decays . some results are also compared with nrqcd calculations . we acknowledge the strong efforts of the bepc staff and the helpful service from the ihep computing center . the work of the bes collaboration is supported in part by the national natural science foundation of china under contract no . 19991480 and the chinese academy of sciences under contract kj95t-03 , 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 ) . 99 j. z. bai et al . , bes coll . , & 344&319(1994 ) . m. e. b. franklin et al . , & 51&963(1983 ) . t. applequist and d. politzer & 34&43(1975 ) . particle data group , r. m. barnet et al . j. * c3 * 1998 . j. z. bai et al . , bes coll . , & 81&5080(1998 ) . . @xmath1 ratiative decays into two pseudoscalars , to be submitted . @xmath1 decays into baryon - antibaryon pairs , to be submitted . for example , w. s. hou and a. soni , phys . lett * 50 * , 569(1983 ) ; s. j. brodsky , g. p. lepage and s. f. tuan , phys . lett * 59 * , 631(1987 ) ; x. q. li et al . , phys . rev . * d55 * , 1241(1997 ) s. j. brodsky and m. karliner , phys . 78,468(1997 ) ; y. q. chen and e. braaten , phys lett 80,5060(1998 ) . @xmath1 hadronic decays involving @xmath30 , to be submitted . . g. t. bodwin , e. braaten and g. p. lepage phys . rev.d46 * , 1914(1992 ) ; ibid * d51 * , 1125(1995 ) . j. bolz , p. kroll and g.a . schuler , & 392&198(1997 ) . j. * c2 * , 705 ( 1998 ) ; p. kroll , hep - ph/9709393 ; s. m. h. wong , hep - ph/9903236 .	based on @xmath0 data sets collected with the bes detector at bepc , the recent @xmath1 , @xmath2 and @xmath3 results from bes are presented . some results are compared with nrqcd . # 1&#2&#3(#4)#1#2 , # 3 ( # 4 ) -4 mm 0.2 cm
the electronic effects in mesoscopic rings were studied by using the non - relativistic quantum mechanics @xcite-@xcite based on the schr " odinger equation with additional terms describing the spin - orbit interaction @xcite-@xcite . however , there are nano - systems , as for example the graphenes , where several relativistic effects can be observed in the electronic transport . these can be satisfactory explained considering the electrons as massless dirac particles moving on honeycomb lattices @xcite-@xcite . other dirac materials are the topological insulators like @xmath1 and @xmath2 quantum wells with low density and high mobility , in which the quantum spin hall effect can be realized @xcite . consequently , many studies @xcite , @xcite-@xcite concentrate on the relativistic effects considering the electrons near the fermi surface as being described by the @xmath3-dimensional dirac equation corresponding to a restricted three - dimensional clifford algebra . however , in this manner one restricts simultaneously not only the orbital degrees of freedom but the spin ones too , reducing them to those of the @xmath4 symmetry . under such circumstances , we believe that there are situations when it is convenient to use the _ complete _ @xmath5-dimensional dirac equation restricting the orbital motion , according to the concrete geometry of the studied system , but without affecting the natural spin degrees of freedom described by the @xmath6 group . thus the polarization effects could be better pointed out . nevertheless , the complete dirac equation was only occasionally used for investigating some special problems of the fermions in external aharonov - bohm ( ab ) field as for example the spin effects in perturbation theory @xcite , the behaviour of the ab fermions in mit cylinders @xcite and even the ab dynamics using numerical methods . in the present report we discuss this topics focusing on the results obtained in refs . herein we have shown first that the solutions of the dirac equation in ab rings and cylinders may be determined as common eigenspinors of a complete systems of commuting operators including the energy , total angular momentum and a specific operator analogous to the well - known dirac spherical operator of the relativistic central problems @xcite . these solutions can be normalized with respect to the relativistic scalar product obtaining thus the system of normalized fundamental solutions that allow us to write down the exact expressions of the relativistic partial currents and derive the persistent ones . the relativistic partial currents we have obtained are related to the derivative of the relativistic energies as in the non - relativistic case but , in contrast with this , there appears a crucial difference , i. e. in the non - relativistic theory the partial currents are proportional to the angular momentum while in our approach the relativistic currents tend to _ saturation _ in the limit of high total angular momenta . for this reason we have reconsidered the problem of the relativistic persistent currents at @xmath0 proposing an approximative analytic formula that matches the numerical calculations with a satisfactory accuracy @xcite . this report is organized as follows . in the second section we present the relativistic theory of the fermions in ab rings based on a suitable restriction of the complete dirac equation , deducing the form of the normalized spinors , deriving the partial currents and writing down the formula of the persistent currents . the next section is devoted to the relativistic theory of the fermions in ab cylinders based on a restriction of the complete dirac equation appropriate to this case . the fundamental solutions on finite or infinite ab cylinders are derived and the partial current are calculated pointing out the saturation effect in both these cases . for the finite cylinders we apply boundary conditions of mit type that eliminate the longitudinal current but preserving the saturation properties of the circular one . finally , we derive the persistent circular currents at @xmath0 on finite ab cylinders discussing the case of very short ones and the non - relativistic limit . let us consider a dirac fermion of mass @xmath7 moving on a _ ideal _ ring of radius @xmath8 whose axis is oriented along the homogeneous and static external magnetic field @xmath9 given by the electromagnetic potentials @xmath10 and @xmath11 . the ideal ring is a one - dimensional manifold ( without internal structure ) embedded in the three - dimensional space according to the equations @xmath12 and @xmath13 , written in cylindrical coordinates @xmath14 with the @xmath15 axis oriented along @xmath9 . then , it is natural to assume that any field @xmath16 defined on this manifold depends only on the remaining coordinates @xmath17 such that @xmath18 and @xmath19 . these restrictions give the kinetic term , @xmath20-m\overline{\psi}\psi\right\}\,,\ ] ] of the dirac action @xmath21 in the mentioned external magnetic field , where @xmath22 and @xmath23 is depending on @xmath24 . the notation @xmath25 stands for the usual dimensionless flux parameter ( in natural units ) . from this action we obtain the correctly restricted dirac equation , @xmath26 , with the new self - adjoint dirac operator @xmath27 whose supplemental last term guarantees that @xmath28 . this operator commutes with the energy operator @xmath29 and the third component , @xmath30 , of the total angular momentum , formed by the orbital part @xmath31 the spin one @xmath32 . therefore , we have the opportunity to look for particular solutions of the form @xmath33 which satisfy the common eigenvalue problems , @xmath34 and @xmath35 laying out the energy @xmath36 and the angular quantum number @xmath37 whose values are determined by the condition @xmath38 . in this manner we separated the variables remaining with a system of algebraic equations that in the standard representation of the gamma matrices ( with diagonal @xmath39 ) reads @xmath40 this system has non - trivial solutions only for the discrete values of energy @xmath41^{\frac{1}{2}}\,,\ ] ] whose second terms encapsulate the ab effect . for each value @xmath42 we find two particular solutions for which @xmath43 where @xmath44 are the usual pauli spinors of polarization @xmath45 with respect to the @xmath15 axis , @xmath46 thus , we find that for @xmath47 the spinors ( [ psi ] ) take the form @xmath48 while for @xmath49 we obtain the solutions @xmath50 the normalization constants are fixed in accordance to the relativistic scalar product @xmath51 such that @xmath52 hence we obtained a pair of fundamental solutions of the same energy and total angular momentum but which are not eigenspinors of the operators @xmath53 or @xmath54 . therefore , we may ask how these solutions can be defined as different eigenspinors of a new operator . the answer is obvious if we observe that the desired operator is @xmath55 which satisfies @xmath56 . the conclusion is that the spinors @xmath57 are common eigenspinors of the complete set of commuting operators @xmath58 . the operator @xmath59 introduced above is the analogous of the spherical dirac operator @xmath60 that concentrates the angular variables of the dirac equation in external fields with central symmetry @xcite . note that the genuine three - dimensional operator @xmath61 can not be used here because of our dimensional reduction such that we must consider the simplified version @xmath59 . the eigenvalues of this operator give the polarization in the non - relativistic limit . for this reason we keep this terminology considering that the eigenvalues @xmath62 of the operator @xmath59 define the fermion polarization with respect to the direction of the magnetic field @xmath9 . using the above results we can calculate the exact relativistic expressions of the partial currents on quantum rings , pointing out the difference between the genuine relativistic theory and the non - relativistic one . we show that in the relativistic approach the partial current tends to saturation for increasing @xmath63 such that the persistent currents at @xmath0 will get new properties . let us start with the quantum rings where the states of the fermions of energy @xmath64 are described by the normalized linear combinations @xmath65 for which the expectation value of the polarization operator reads , @xmath66 the partial currents ( of given @xmath63 ) coincide in this case with their densities , @xmath67 , that can be calculated with the help of the matrix ( [ gamph ] ) . then , observing that @xmath68 we obtain the _ partial _ current of a fermion of energy @xmath42 as @xmath69 since @xmath70 . thus we find that the partial currents are independent on polarization being related to energies in a similar manner as in the non - relativistic theory . the exact relativistic expressions of the partial currents we obtained here depend only on two dimensionless parameters @xmath71 and @xmath72 ( or @xmath73 in usual units ) that are the arguments of the auxiliary function @xmath74 defined as @xmath75 this function has the remarkable asymptotic behaviour @xmath76 which shows that the relativistic partial currents tend to _ saturation _ for large values of @xmath63 . moreover , for small values of @xmath77 we can expand @xmath78 note that the non - relativistic limit recovers the well - known behaviours @xmath79 hereby we conclude that the principal difference is that the relativistic partial currents ( [ irel ] ) are saturated while in the non - relativistic case we do not meet this effect since the function @xmath80 is replaced then by the linear function @xmath81 that is just its tangent in @xmath82 as we deduce from eq . ( [ cucu ] ) . this result was previously outlined in ref . @xcite but based on the non - hermitian dirac equation of ref . obviously , the correct saturation effect is given by the expression of the partial currents ( [ irel ] ) derived here . note that the non - relativistic approximation can be used with a satisfactory accuracy only in the domain where the function @xmath80 is approaching to the linear function @xmath81 . our numerical evaluations show that in the domain @xmath83 the difference @xmath84 is satisfactory small remaining less than 0.05 . in addition , we estimate that for @xmath85 the current is approaching to its saturation value since @xmath86 . the above results allow us to derive the total persistent current at @xmath0 in a semiconductor ring of parameter @xmath87 having a even number of electrons @xmath88 fixed by the fermi - dirac statistics . for the mesoscopic rings with @xmath89 the parameter @xmath87 is of the order @xmath90 . for example , in a @xmath91 ring of this radius , the effective electron mass is @xmath92 @xcite such that @xmath93 . this seems to be the minimal value of @xmath87 obtained so far but it is possible to obtain smaller values in further experiments with mesoscopic rings with @xmath94 or even with nano - rings having @xmath95 . according to our estimation , the relativistic effects may be measurable for @xmath96 which means that the actual experiments are approaching to this threshold which could be reached soon . in all these cases the flux parameter @xmath97 remains very small ( less than @xmath98 ) such that we can neglect the terms of the order @xmath99 of the taylor expansions of our functions that depend on @xmath100 . the total persistent current at @xmath0 is given by the sum @xmath101\ ] ] over all the allowed polarizations , @xmath102 where @xmath103 . furthermore , by using the expansion @xmath104 where @xmath105 we arrive at the relativistic persistent currents , @xmath106 that can be calculated numerically on computer for any concrete value of @xmath87 . versus @xmath87 calculated for @xmath107 ( a ) and @xmath108 when the co - domain is very narrow , @xmath109 ( b ) . ] the function @xmath110 is simple reaching its maximal value 0.7698 for @xmath111 and @xmath112 and decreasing then monotonously to zero when @xmath87 and @xmath63 are increasing to infinity . this behaviour is a direct consequence of the saturation of the partial currents that compensate each other in the saturation zone where @xmath113 . these simple monotony and smoothness properties of the function @xmath110 lead to nice results concerning the values of the sum ( [ ic]c ) when we compute all the allowed contributions . our numerical examples show that when @xmath87 is increasing then the functions @xmath114 are monotonously decreasing tending to an asymptotic value ( as in fig . 1 ) . consequently , in the asymptotic zone , @xmath115 , we can use the following approximation @xmath116 giving the definitive formula of the relativistic persistent currents @xmath117 that reproduces the numerical results with a satisfactory accuracy ( under @xmath118 ) . note that the non - relativistic persistent current , that in our notation reads @xmath119 , represents a good approximation of eq . ( [ final ] ) only for small values of @xmath120 ( say @xmath121 ) for which we can use the approximation @xmath122 . we consider the motion of a dirac fermion of mass @xmath7 on an _ ideal _ cylinder of radius @xmath8 whose axis is oriented along the homogeneous and static external magnetic field @xmath9 given by the electromagnetic potentials @xmath10 and @xmath11 . this background is a two - dimensional manifold ( without internal structure ) embedded in the three - dimensional space obeying the simple equation @xmath12 in cylindrical coordinates @xmath14 with the @xmath15 axis oriented along @xmath9 . then , it is natural to assume that any field @xmath16 defined on this manifold depends only on the remaining coordinates @xmath123 such that we can put @xmath18 in the kinetic term of the lagrangian density . thus we obtain the action of the dirac fermion in the mentioned external magnetic field @xmath124 having the kinetic part @xmath125-m\overline{\psi}\psi\right\}\,,\end{aligned}\ ] ] where @xmath22 and @xmath126 the notation @xmath25 stands for the usual dimensionless flux parameter ( in natural units ) . from this action we obtain the correctly restricted dirac equation , @xmath26 , with the self - adjoint dirac operator @xmath127 whose supplemental third term guarantees that @xmath28 . this operator commutes with the energy operator @xmath29 , the momentum along to the third axis , @xmath128 , and the similar component , @xmath30 , of the total angular momentum , formed by the orbital part @xmath31 the spin one @xmath32 . under such circumstances , we have the opportunity to look for particular solutions of the form @xmath129 which satisfy the eigenvalue problems , @xmath130 laying out the energy @xmath36 and the quantum number @xmath131 of the total angular momentum . the normalization constant @xmath132 has to be determined after we solve the functions of @xmath15 from the remaining reduced equation that in the standard representation of the gamma matrices ( with diagonal @xmath39 ) reads @xmath133 this is in fact a system of linear differential equations allowing us to solve the functions of @xmath15 . the general solutions of this system can be obtained reducing the number of functions with the help of the last two equations that yield @xmath134 leading to the second order equations @xmath135 consequently , the solutions must be linear combinations of the form @xmath136 where @xmath120 is the fermion momentum along the @xmath15 axis . the concrete form of these solutions depends on the boundary conditions we chose for determining the integration constants @xmath137 and @xmath138 up to a normalization factor , @xmath132 . this last constant has to be determined by imposing the desired normalization condition with respect to the relativistic scalar product @xmath139 calculated on the domain @xmath140 of the entire cylinder . the simplest case is of the infinite cylinder , with @xmath141 , where the motion along its axis is a free one . we assume that the spin projections are measured just with respect to this axis such that we may chose @xmath142 where @xmath44 are the usual pauli spinors , @xmath46 of polarizations @xmath45 . then the functions @xmath143 can be derived from eq . ( [ g12 ] ) as @xmath144 while the energy that depends on @xmath120 and @xmath63 reads @xmath145^{\frac{1}{2}}\ ] ] as it results from eq . ( [ ene ] ) . we obtain thus a mixed energy spectrum whose ground level is given by @xmath146 and one of the values @xmath147 that minimizes the last term in eq . ( [ ekm ] ) , e. g. @xmath148 if @xmath149 . consequently , the spinor components have to be tempered distributions that may be normalized in the momentum scale . according to the above results we can write two types of fundamental solutions ( [ psi ] ) of the form @xmath150 corresponding to the polarizations @xmath151 . for @xmath152 we obtain @xmath153 assuming that the normalization factor depend on @xmath120 and @xmath63 . similarly , for @xmath49 we deduce @xmath154 it remains to calculate the normalization in the momentum scale finding that by fixing the values @xmath155 we obtain the desired ( generalized ) orthogonality relations @xmath156 these fundamental solutions are eigenspinors of the same set of commuting operators which seems to be incomplete as long as we can not distinguish between @xmath157 and @xmath158 . therefore we need to introduce a new operator for completing this set . a short inspection suggests that this must be an analogous of the dirac spheric operator @xcite that reads now @xmath159 giving the eigenvalues problems @xmath160 . the conclusion is that the fundamental solutions we derived above are eigenspinors of the complete set of commuting operators @xmath161 . the spinors describing physical states are square integrable packets of a given total angular momentum @xmath63 having the form @xmath162\ ] ] where the functions @xmath163 satisfy the condition @xmath164=1\ ] ] that assures the normalization condition @xmath165 . we say that the packets ( [ psilam ] ) describe the states @xmath166 in which the expectation value of the total angular momentum reads @xmath167 while the polarization degree can be defined as @xmath168\,.\ ] ] now we can derive the currents of the fermions in the states @xmath166 by using the components of the current density @xmath169 . we consider first the total circular current @xmath170 that can be calculated according to eqs . ( [ gamph ] ) , ( [ psiup ] ) and ( [ psium ] ) , that yield @xmath171 then , observing that the integral over the @xmath15 axis generates a @xmath172-function , we obtain the definitive closed form @xmath173\,.\ ] ] hereby we draw the conclusion that the circular current is stationary ( i. e. independent on @xmath174 ) depending only on the packet content as given by the arbitrary functions @xmath163 . nevertheless , it is remarkable that there are two important properties of the circular currents that are independent on the form of these functions . the first one is the saturation effect for increasing total angular momenta , @xmath175 = \pm \frac{1}{2\pi r}\,.\ ] ] on the other hand , bearing in mind that the expectation value of the energy in the state @xmath166 reads @xmath176\,,\ ] ] we recover the familiar formula @xmath177 that has the same form as in the case of the relativistic @xcite or non - relativistic ab rings . note that , in contrast to the circular current , the longitudinal one , @xmath178 depends on time , reflecting thus the propagation and dispersion of the packet along the @xmath15 axis . the general expression of this current is presented in the appendix a. another interesting problem is of a finite cylinder of length @xmath179 for which we must consider the boundary conditions @xmath180 . therefore , we may chose @xmath181 denoting now @xmath182 . thus we obtain the countable discrete energy spectrum @xmath183^{\frac{1}{2}}\,,\ ] ] corresponding to the square integrable spinors whose components are given by eq . ( [ fkn1 ] ) and eq . ( [ g12 ] ) that yields now @xmath184 then , according to eq . ( [ psi ] ) we can write down the form of two types of solutions corresponding to @xmath151 . for @xmath152 we obtain @xmath185 and similarly for @xmath49 , @xmath186 after a little calculation we find that by fixing the value of the normalization constants as @xmath187 we obtain the desired orthogonality relations @xmath188 the conclusion is that the above fundamental solutions are eigenspinors of the set of commuting operators @xmath189 . let us consider the fermions in the states @xmath190 given by the normalized linear combinations @xmath191 which satisfy @xmath192 . the constants @xmath193 give the polarization degree defined as in the previous case , @xmath194 obviously , the fermions are unpolarized when @xmath195 . with these ingredients we can calculate the quantities @xmath196 that help us to derive the definitive form of the circular currents in the states @xmath190 that read @xmath197 since the integral over @xmath15 vanishes the mixed terms ( [ usles ] ) while the constants @xmath193 satisfy eq . ( [ psicc ] ) . it is remarkable that this current is independent on polarization and has a similar form and relation with the energy as in the case of the ab rings @xcite . the difference is that now the energy ( [ enm ] ) depends on two quantum numbers , @xmath198 and @xmath63 , as well as on the length @xmath179 of the ab cylinder . the longitudinal current vanishes since we used boundary conditions that guarantee that @xmath199 in other words , our boundary conditions are of the mit type vanishing the currents but without canceling all the components of the dirac spinors on boundaries . the properties of the currents ( [ inm ] ) can be better understood by introducing the appropriate dimensionless parameters @xmath200 that allow us to write @xmath201 pointing out the function @xmath74 which gives the behavior of the circular currents . this function is smooths with respect to all of its variables increasing monotonously with @xmath63 from @xmath202 to @xmath203 since @xmath204 and vanishing for @xmath205 . this means that , as in previous case , for increasing total angular momenta , the circular current tends to the asymptotic saturation values @xmath206 just as it happens with the partial currents in ab rings @xcite . now we can use these properties for estimating the persistent current at @xmath0 in semiconductor ab cylinders where the electron discrete energy levels @xmath207 are given by eq ( [ enm ] ) . according to the fermi - dirac statistics , at @xmath0 the electrons occupy all the states @xmath190 which satisfy the condition @xmath208 where @xmath209 is the ( non - relativistic ) energy of the fermi level . therefore , the total number of electrons @xmath88 and the persistent current @xmath210 can be calculated as @xmath211 in practice the flux parameter @xmath97 remains very small ( less than @xmath98 ) such that we can neglect the terms of the order @xmath99 of the taylor expansions of our functions ( [ ichi ] ) . thus we can write @xmath212 where @xmath213 obtaining thus the expression of the relativistic persistent currents , @xmath214 the principal problem in evaluating such sums is the computation of the contributing states @xmath190 ( with @xmath215 and @xmath216 ) which satisfy the condition ( [ cond0 ] ) . we denote first by @xmath217 the greatest value of the quantum number @xmath198 and by @xmath218 the greatest value of @xmath219 for a given @xmath198 , assuming that the states @xmath220 and respectively @xmath221 ( with @xmath222 are very close to the fermi level , i. e. @xmath223 . in addition , we denote by @xmath224 the greatest value among the quantities @xmath218 . then we can rewrite eq . ( [ cond0 ] ) as @xmath225 obtaining the approximative identities @xmath226 that help us to estimate the numbers @xmath217 and @xmath218 . moreover , we observe that @xmath227 must be much greater than @xmath203 since otherwise we can not speak about statistics . then we can use the approximative formula ( [ bsum ] ) obtaining the compact results @xmath228 that represent a very good approximation for the systems with @xmath229 @xcite . it remains to calculate on computer the sum over @xmath198 or to consider the estimation ( [ bsum1 ] ) when @xmath230 . an interesting case is of the very short cylinders with @xmath231 whose quantum number @xmath198 can take the unique value @xmath232 in order to satisfy eq . ( [ cond ] ) that becomes now @xmath233 . consequently , the allowed states are @xmath234 which means that @xmath235 and @xmath236 however , for the very short cylinders with @xmath237 the identities ( [ ident ] ) do not make sense such that we need to rebuild the entire theory without motion along the @xmath15 axis ( @xmath146 ) , retrieving thus the case of the ideal ab rings @xcite for which we must substitute @xmath82 and @xmath238 in eq . ( [ ishort ] ) . finally , we note that for the non - relativistic short ab cylinders with @xmath239 we recover the well - known result , @xmath240 of the non - relativistic persistent current in ab rings . we presented the relativistic theory of the dirac fermions on ab cylinders based on the complete @xmath5-dimensional dirac equation with restricted orbital degrees of freedom but without affecting the spin ones . this can be achieved by using the method we proposed recently for the ab rings @xcite according to which the orbital restrictions must be imposes on the lagrangian density giving rise to a correct self - adjoint dirac operator . the results presented here point out two principal features of the circular currents on the ab cylinders . the first one is the relation between the circular current and the derivative of the energy with respect to the flux parameter that is the same for ab rings or cylinders either in the relativistic approach or in the non - relativistic one . in other words this property is universal for all the ab systems with cylindric symmetry . the second feature is specific only for the relativistic circular currents that tend to saturation in the limit of very large total angular momenta , in contrast with the non - relativistic ones that are increasing linearly to infinity . the saturation effect determines a specific form of the relativistic persistent current on finite ab cylinders that can be seen as a generalization of the persistent current in ab rings . obviously , the dependence on parameters is more complicated in the case of the finite cylinders but these models are closer to the real devices involved in experiments . on the other hand , we must specify that the closed formulas derived in section 3 are only approximations that must be used prudently and completed by numerical calculations on computers . on the infinite ab cylinders the longitudinal currents in the state @xmath166 are determined by the structure of the wave packet ( [ psilam ] ) as @xmath241\left[a_{+}^(k)a_{+}(k')+a_{-}^(k)a_{-}(k')\right]\right . \nonumber\\ & -&\left . \frac{i(\lambda+\beta)}{r}(e_{k}-e_{k'})\left[a_{+}^(k)a_{-}(k')+a_{-}^(k)a_{+}(k ' ) \right ] \right\}\,,\end{aligned}\ ] ] where the energies are given by eq . ( [ ekm ] ) and the arbitrary functions @xmath163 satisfy the condition ( [ calpha ] ) . since @xmath227 must be much greater than @xmath203 we can use the approximative formula @xmath242 that reproduces the numerical results with a satisfactory accuracy for @xmath229 @xcite . moreover , when @xmath243 we can evaluate @xmath244 n. byers , c. n. yang , _ phys . _ 7 ( 1961 ) 45 . y. imry , _ introduction to mesoscopic physics _ ( oxford university press , oxford 2002 ) . s. viefers , p. koskinen , p. singha deo and m. manninen , _ physica e _ 21 ( 2004 ) 1 . b. chen , x. dai and r. han , _ phys . a _ 302 ( 2002 ) 325 . m. szopa , m. marganska and e. zipper , _ phys . a _ 299 ( 2002 ) 593 . d. mailly , c. chapellier and a. benoit , _ phys . _ 70 ( 1993 ) 2020 . h. f. cheung , y. gefen , e. k. riedel and w. h. shih , _ phys . rev . b _ 37 ( 1988 ) 6050 . e. papp , c. micu , l. aur and d. racolta , _ physica e _ 36 ( 2007 ) 178 . f. carvalho dias , i. r. pimentel and m. henkel , _ phys . b _ 73 ( 2006 ) 075109 . m. koskinen , m. manninen , b. mottelson and s. m. reimann , _ phys . b _ 63 ( 2001 ) 205323 . e. i. rashba , _ sov . solid state _ 2 ( 1960 ) 1109 . m. v. moskalets , _ physica b _ 291 ( 2000 ) 350 . b. molnr , f. m. peeters and p. vasilopoulos , _ phys . b _ 69 ( 2004 ) 155335 . j. s. sheng and k. chang , _ phys . b _ 74 ( 2006 ) 235315 . t. w. chen , c. m. huang and g. j. guo , _ phys . b _ 73 ( 2006 ) 235309 . x. w. zhang and j. b. xia , _ phys . b _ 74 ( 2006 ) 075304 . k. s. novoselov , a. k. geim , s. v. morozov , d. jiang , y . zhang , s. v. dubonos , i. v. grigorieva , a. a. firsov , _ science _ * 306 * , ( 2004 ) 666 . k. s. novoselov , a. k. geim , s. v. morozov , d. jiang , m. i. katsnelson , i. v. grigorieva , s. v. dubonos and a. a. firsov , _ nature _ * 438 * ( 2005)197 . c. yannouleas , i. romanovsky , u. landman , _ phys . b _ * 89 * ( 2014 ) 035432 . c. yannouleas , i. romanovsky , u. landman,_j . phys . c _ * 119 * ( 2015 ) 11131 . b. bttner , c. x. liu , g. tkachov , e. g. novik , c. brne , h. buhmann , e. m. hankiewicz , p. recher , b. trauzettel , s. c. zhang , l. w. molenkamp , nature physics 7 , 418 ( 2011 ) ; b. a. bernevig , t. l. hughes , s.c . zhang , science 314 , 1757 ( 2006 ) ; m. knig , s. wiedmann , c. brne , a. roth , h. buhmann , l. w. molenkamp , x .- qi , s .- c . zhang , science 318 , 766 ( 2007 ) . c. g. beneventano and e. m. santangelo , _ j. phys . _ 39 ( 2006 ) 7457 . v. p. gusynin and s. g. sharapov , _ phys . _ 95 ( 2005 ) 146801 . d. v. khveshchenko , _ phys . _ 87 ( 2001 ) 206401 . s. g. sharapov , v. p. gusynin and h. beck , _ phys . b _ 69 ( 2004 ) 075104 . f. vera and i. schmidt , _ phys . 42 ( 1990 ) 3591 . m. boz and n. k. pak , _ phys . d _ 62 ( 2000 ) 045022 . a. albeed and m. s. shikakhwa , _ int . _ 46 ( 2007 ) 405 . m. de francia and k. kirsten , _ phys . d _ 64 ( 2001 ) 065021 . i. i. cotescu , d .- blteanu and i. cotescu jr . , _ int b _ vol 30 , 1550245 ( 2016 ) , arxiv:1502.07563 . i. i. cotescu , d .- blteanu and i. cotescu jr . , arxiv:1511.00585 , to appear in _ int . j. mod . s. ghosh , _ advances in condensed matter physics _ vol . 2013 , article i d 592402 . i. i. cotescu and e. papp , _ j. phys . : condens . matter _ 19 ( 2007 ) 242206 .	the exact solutions of the complete ( 1 + 3)-dimensional dirac equation of fermions moving in ideal aharonov - bohm ( ab ) rings and cylinders are used for deriving the exact expressions of the relativistic partial currents . it is shown that these currents can be related to the derivative of the fermion energy with respect to the flux parameter , just as in the non - relativistic case . however , a new and remarkable relativistic effect is the saturation of the partial currents for high values of the total angular momentum . based on this property , the total relativistic persistent currents at @xmath0 is evaluated for rings and cylinders obtaining approximative simple closed formulas . keywords : dirac equation ; aharonov - bohm ring ; aharonov - bohm cylinder ; persistent current ; saturation .
the ejecta structure of a type ii - p supernova ( sn ii - p ) may be sketchy portrayed as a combination of a hydrogen envelope and a `` mixed core '' . the latter , being a macroscopic mixture of newly synthesised elements , such as , he , o , @xmath1ni with a fraction of the hydrogen envelope , is responsible for the luminosity at the nebular epoch , i.e. , at the age @xmath2 d. the model of mixed core of sn ii - p is fundamental to the assessement of the nuclesynthesis yields , which is the subject of great interest . two recent nebular models of sn 1987a ( de kool , li & mccray @xcite ; kozma & fransson @xcite ) demonstrate that the basic physical processes in the mixed core are well understood to make feasible the composition analysis . however , to achieve this goal a nebular model of sn ii - p should , first , pass the rigorous test on reproducing fluxes and profiles of hydrogen lines , including h@xmath0 . here i argue that the model of the h@xmath0 line in sn ii - p at the early nebular epoch ( 100200 d after the outburst ) should essentially include effects of radiation damping wings always ignored before . the physical reason for the damping wings to play a role is a high degree of a non - thermal excitation of hydrogen in the mixed core of sn ii - p , which results in the large h@xmath0 optical depth at the early nebular epoch . in this situation to model h@xmath0 damping wing effects the sobolev ( local escape ) approximation must be abandoned . the radiation transfer in the physical and the frequency space simulated with the monte carlo method is applied to demonstrate expected effects of the h@xmath0 damping wings in sn ii - p ( section 2 ) . recently found in sn 1997d a disparity between the large asymmetry of h@xmath0 at the early nebular phase and the symmetric h@xmath0 profile at the late nebular epoch ( chugai & utrobin @xcite ) admits a simple explanation in terms of the damping wings ( section 3 ) . another motivation to address the issue of the h@xmath0 damping wings in sn ii - p is that this factor may provide an independent diagnostic tool for constraining parameters of the mixed core . the sobolev approximation widely used to analyse supernova optical spectra is justified for most lines because of a highly supersonic expansion , @xmath3 , ( @xmath4 and @xmath5 being the expansion and the thermal velocity , respectively ) . however , if the optical depth in damping wings is large , the `` supersonic '' criterion for the local escape of photons should be replaced by the inequality @xmath6 ( chugai @xcite ) , where @xmath7 is the voigt parameter , @xmath8 is the sobolev optical depth ( @xmath9 is the line integrated absorption coefficient , @xmath10 is the wavelength , @xmath11 is the expansion time ) . the h@xmath0 voigt parameter is @xmath12 , where @xmath13 is the einstein transition probability , @xmath14 is the thermal doppler width ( kinetic temperature of 5000 k is adopted ) . at the early nebular epoch ( @xmath15 d ) the sobolev optical depth of the mixed core in h@xmath0 may exceed @xmath16 ( xu et al . @xcite ) implying that @xmath17 , i.e. , is of the order of @xmath18 . in this case the local escape approximation is too crude , so that the full radiation transfer should be applied to compute the h@xmath0 line profile . one needs , first , to specify the frequency redistribution function for the h@xmath0 damping wings . let @xmath19 and @xmath20 be the incident and scattered photon frequencies in the atom frame . with lower and upper levels broadened by the radiation damping , the atomic angle - averaged frequency redistribution function in far wings ( @xmath21 ) may be written as @xmath22 ( oxenius @xcite ) , where @xmath23 is the normalized absorption coefficient and @xmath24 here @xmath25 and @xmath26 are voigt parameters for the lower and the upper level . thermal velocity effects in the wings are negligible and will be ignored here . this implies that the frequency redistribution function in the wings is the same for the atom and the fluid frame . the redistribution function may be essentially simplified to reduce the number of scattering in the monte carlo computations preserving good precision . since multiple local scattering in the doppler core ( @xmath27 ) eventually ends up with the photon re - emitted in the wings , the scatterings in the doppler core may be omitted by adopting the function @xmath28 equal to zero in the core @xmath29 . after the re - normalization this function is labeled @xmath30 . for instance , the function @xmath31 in the second term of eq.([eq : prob ] ) transforms into @xmath32 for @xmath33 and @xmath34 otherwise . results are not sensitive to the choice of the core width and @xmath35 is assumed . the h@xmath0 photon history is simulated as follows . the photon is launched with the distribution function @xmath36 . when propagating it experiences a redshift in the co - moving frame until it randomly scatters or escapes . if scattered , the frequency of the re - emitted photon is randomly choosen according to the modified distribution function @xmath30 . the process continues until the photon escapes . to demonstrate the role of h@xmath0 damping wings in sn ii - p let us consider a freely expanding envelope ( @xmath37 , @xmath4 being the velocity at the radius @xmath38 at the expansion time @xmath11 ) with a homogeneous line - emitting core ( @xmath39 ) and a scattering halo ( @xmath40 ) . the sobolev optical depth of the halo decreases outward as @xmath41 . the core , optically thin in the continuum , may emit a continuum radiation from homogeneously distributed sources . the velocity and the optical depth in the core ( subscript 1 ) and in the halo ( subscript 2 ) , respectively , are given in table 1 . the homogeneous core with the low optical depth in the model m1 produces a parabolic profile ( fig . 1a ) expected for the sobolev approximation . when the optical depth is high ( models m2 and m3 ) the extended wings emerge , while the line core becomes broader and gets redshifted . all the effects are caused by the radiation transfer in the damping wings . note , that in the case of the sobolev approximation the models m2 and m3 would have the same parabolic profile as that of the model m1 . a conspicuous kink at the blue slope in the model m3 is caused by a self - absorption due to non - local scattering . the models m4 and m5 show how the models m1 and m3 are changed when a scattering halo and a continuum core are present ( fig . the model m5 gives a realistic impression of damping wings effects in h@xmath0 at the early nebular epoch of sn ii - p . how would the inhomogeneity of the hydrogen distribution in the mixed core change these results ? generally , the optical depth in the damping wing depends not only on the local concentration of absorbers , but also on its filling factor ( @xmath42 ) , and , in lesser degree , both on the size of inhomogeneities , and on their topological properties . in the limit of small scale inhomogeneities , when the mean free path length in the far wing exceeds a typical size of the inhomogeneity , the optical depth in the wing may be reduced to a simple form @xmath43 , where @xmath44 is the sobolev optical depth and @xmath45 is the frequency shift in comoving frame at the length interval @xmath46 . under this approximation the h@xmath0 profile from the inhomogeneous core is the same as that from the homogeneous core with the `` effective optical depth '' @xmath47 , the analysis of the h@xmath0 damping wings thus permits us to estimate the h@xmath0 effective optical depth in the mixed core , or , equivalently , @xmath48 ( @xmath49 being the population of the second hydrogen level ) . it is noteworthy that the derived value of @xmath48 is independent of the adopted distance of a supernova . the spectra of the type ii - p supernova sn 1997d with a low expansion velocity have been obtained at the early ( @xmath50 d ) and the late ( @xmath51 d ) nebular epoch by turatto et al . ( @xcite ) . the nebular spectrum on day 300 was reproduced in the model of a spherically symmetric mixed core with the total ejecta mass of 6 @xmath52 and the kinetic energy of @xmath53 erg ( chugai & utrobin @xcite ) . however , it was stressed there that h@xmath0 on day 150 shows a significant redshift in a disparity with a spherically symmetric model . note , that the asphericity of the @xmath1ni distribution might produce the h@xmath0 asymmetry on day 150 likewise it was in sn 1987a . however , this reason for sn 1997d is discarded by the absence of h@xmath0 asymmetry on day 300 . we will see below that damping wings effects may resolve the revealed controversy . the assumed envelope model is similar to that of the section 2 , i.e. , it consists of a line - emitting homogeneous core and a scattering halo . for models considered here the core velocity is @xmath54 km s@xmath55 and the outer velocity of the scattering halo is @xmath56 km s@xmath55 ; the latter is consistent with the steep density drop in the range @xmath57 km s@xmath55 for the hydrodynamical model of sn 1997d ( chugai & utrobin @xcite ) . the sobolev optical depth in the halo is assumed to decrease outward as @xmath58 with @xmath59 . an additional factor implemented into the model is the thomson scattering , although it is of minor importance . the corresponding optical depth ( @xmath60 ) of the mixed core , estimated from the h@xmath0 luminosity ( distance @xmath61 mpc ) using data by turatto et al . ( @xcite ) , is 0.16 on day 150 and 0.08 on day 300 . bellow , i adopt @xmath62 for @xmath63 d and @xmath64 for @xmath65 d to roughly take into account that the hydrogen filling factor in the mixed core is essentially less than unity ( chugai & utrobin @xcite ) . in the late time spectrum on day 300 the model without damping wings is not very much different to the observed profile ( fig . it implies that the damping wings at this epoch are not significant , although , the model with the damping wings and @xmath66 fits the observed profile better . on day 150 the model without damping wings is strikingly dissimilar to the observed profile : the latter is broader and shows both significant redshift and red wing . the model with damping wings included and @xmath67 turns out to be successful in the description of the observed profile . note , that with the core velocity set by the late time profile , the primary fitting parameter is the effective optical depth in the mixed core . the found effective optical depth value being combined with the sobolev optical depth may be used to derive the hydrogen filling factor in the mixed core . to estimate the sobolev optical depth i adopt 0.002 @xmath52 of @xmath1ni ( turatto et al . @xcite ) in the mixed core ( @xmath68 km s@xmath55 ) . the density @xmath69 g @xmath70 is adopted on day 150 following the model of sn 1997d ( chugai & utrobin @xcite ) . the non - thermal excitation of hydrogen may be calculated assuming that all the energy of gamma - rays from the radioactive decay of @xmath1co is uniformly deposited in the mixed core , which is a sound approximation at the early nebular epoch when the mean free path for gamma - rays is small . the energy deposited in h - rich matter is spent ( apart from coulomb heating ) , eventually , on the excitation of the second level of hydrogen . with the depopulation rate controlled by the two - photon decay and the collisional de - excitation i find @xmath71 @xmath70 on day 150 , which leads to the sobolev optical depth of @xmath72 . the latter being combined with the effective optical depth ( @xmath73 ) results in the hydrogen filling factor of @xmath74 . the latter value is quite realistic , although it is a factor of 1.5 larger than the estimate from the nebular model on day 300 ( chugai & utrobin @xcite ) . the difference is not dramatic taking into account possible uncertainties of both models . the factor of ten lower value of the effective optical depth in h@xmath0 on day 300 compared to that on day 150 is consistent with the decrease of the sobolev optical depth in h@xmath0 expected for the mechanism of the non - thermal excitation ( e.g. , xu et al . @xcite ) . although describing an inhomogeneity of the excited hydrogen in the mixed core in terms of the filling factor seems sensible , a possibility exists that between the mixed core and unmixed hydrogen envelope there is a transition layer of strongly excited hydrogen with the filling factor of unity . to estimate qualitatively effect of this component i introduced in the model on day 150 an additional spherical layer in the velocity range 650700 km s@xmath55 with filling factor of unity , negligibly small net emission , and the sobolev optical depth being equal to the effective optical depth of the mixed core . such a choice of optical depth suggests factor of three lower hydrogen excitation in the transition layer than in the mixed core . the new model is found to produce the same h@xmath0 profile as in the old model provided the effective optical depth of the mixed core is of @xmath75 , i.e. , 20% lower compared to the old model . this implies that the value of the h@xmath0 effective optical depth found from the profile modelling is quite robust to reasonable edditing the geometry of the distribution of the excited hydrogen in the mixed core . summing up , the h@xmath0 damping wings effects provide a natural explanation for the dramatic discrepance between the model based on the sobolev approximation and the observed profile in sn 1997d on day 150 . this conclusion is strengthen by the absence of a reasonable alternative . at the early nebular epoch of sn ii - p the h@xmath0 emission should exhibit observational signatures of damping wings . although , the damping wings were never considered before as relevant for supernova optical spectra , the claimed effects are robust and may be easily identified . we see damping wings effects in sn 1997d spectrum , where they are responsible for the redshift , large width and extended red wing of h@xmath0 on day 150 . the detection of h@xmath0 damping wings effects in sn ii - p provides a direct estimate of the h@xmath0 effective optical depth ( i.e. the product of the sobolev optical depth and the hydrogen filling factor ) in a distance independent way . we thus obtain a reliable test for nebular models of sn ii - p . on the other hand , with a confident nebular model in hand , we may able to measure the hydrogen filling factor in the mixed core . the price for this knowledge is the abandoning the sobolev approximation and the need for the full radiation transfer treatment of the h@xmath0 radiation . needless to say that the simple model of the h@xmath0 profile used above for sn 1997d admits some improvements . for instance , one may take the radial dependence of hydrogen excitation and filling factor in the mixed core into account . another modification , might be the inclusion of the absorption of h@xmath0 quanta in the paschen continuum . this may be done if the radiation transfer and the hydrogen excitation are calculated simultaneously . more detailed description of the hydrogen inhomogeneities is also welcome , although the expected change of the value of the effective optical depth would be within 20% . to recover the damping wings in other sn ii - p it is necessary to use at least two nebular spectra : at the early epoch , ( around day 150 ) and at the late epoch , but before the dust formation , ( e.g. , between 300 d and 400 d ) . the early nebular epoch is most appropriate to extract the information about damping wings , while the late time nebular spectrum provides a template to set the core velocity and to secure results from a confusing effects of asymmetric @xmath1ni distribution . notably , it is @xmath1ni asymmetry that prevent us from using a simple symmetric model to study h@xmath0 damping wings effects in sn 1987a . lcccc model & @xmath76 & @xmath77 & @xmath78 & @xmath79 + & ( km s@xmath55 ) & & ( km s@xmath55 ) & + m1 & 2000 & 0.001 & 2000 & 0 + m2 & 2000 & 2 & 2000 & 0 + m3 & 2000 & 5 & 2000 & 0 + m4 & 2000 & 0.001 & 4000 & 10 + m5 & 2000 & 5 & 4000 & 10 +	it is predicted that the h@xmath0 emission line at the early nebular epoch of type ii - p supernovae may display robust observational effects of radiation damping wings . monte carlo simulations demonstrate that the h@xmath0 line with the optical depth typical of the early nebular epoch acquires extended wings and redshift and becomes broader compared to the sobolev approximation case . the strength of these effects may be used for constraining parameters of the line - emitting zone . the anomalous redshift , width and red wing of the h@xmath0 emission in the supernova sn 1997d on day 150 are explained in terms of damping wings effects .
hyperbolic metamaterials ( hmms ) , also known as indefinite media , enjoy a great deal of attention owing to their unique hyperbolic dispersion relations @xcite , with associated high - wavenumber propagating waves without upper limit . this leads to numerous applications , such as enhanced - light interactions @xcite , subwavelength imaging @xcite , negative refraction @xcite , and low - loss fiber cladding @xcite . the hmms are usually artificially made by periodic dielectric - metal structures , such as a 1d dielectric - metal bragg grating . to describe the optical properties of the hmms , it is common to employ the local - response approximation ( lra ) , _ i.e. _ with every location @xmath0 in the structure a certain value for the permittivity @xmath1 is associated . in the lra , the effective material parameters of hmms have been thoroughly studied @xcite . thanks to advances in nanofabrication , we witness a miniaturization of the feature size of metamaterials towards the deep nanoscale . the lra becomes more inaccurate , since the nonlocal response of free electrons starts to play a role @xcite . it is known that the nonlocal response causes a blueshift of the surface plasmon ( sp ) resonance of the metallic particle @xcite , and limits the sp field enhancement that in the lra sometimes diverges @xcite . we employ a simple generalization to the lra , namely the hydrodynamic drude model ( hdm ) @xcite , which takes the nonlocal response into account . new in this model , as compared to the lra , are longitudinal waves with sub - nanometer wavelengths , besides the usual transverse waves . incidentally , some hmms have been found to exhibit strong effective ` nonlocal response ' , even in studies that employ the local - response approximation @xcite . what is meant here is that the hmms , when considered as scatterers , can not be described in the single - scattering born approximation . this differs from the ( material ) nonlocal response of hmms that we consider here and in ref . @xcite , where already the scattering _ potentials _ associated with the metal constituents are nonlocal . we recently showed that material nonlocal response plays an important role on hmms in the limit of vanishing unit - cell size @xcite . in particular , the nonlocal response gives rise to a cutoff to the hyperbolic dispersion curve , and an associated finite but very large fundamental upper limit to the enhanced local optical density of states ( ldos ) , which in the lra is known to diverge in the limit of vanishingly small unit cells . now in realized hmms , it is predominantly the finite size of the unit cell that keeps the ldos finite . so it is an intriguing question , not explored yet as far as we know , whether nonlocal response can also have noticeable effects in state - of - the - art hmms . this theoretical paper describes our search for observable nonlocal effects in hmms with unit cells much smaller than an optical wavelength , but much larger than the wavelength of hydrodynamic longitudinal pressure waves . as the hmm we consider a 1d subwavelength dielectric - metal bragg grating , with a unit cell of thickness @xmath2 , and thicknesses @xmath3 and @xmath4 of the dielectric and metal layers , respectively . the permittivity of the dielectric layer is @xmath5 . the metal is described in the hdm as a free - electron plasma with @xcite @xmath6 where @xmath7 is the drude permittivity for the transverse electric fields as in the lra , while the wavevector dependence of the permittivity @xmath8 of the longitudinal electric fields is responsible for the nonlocal response . we apply the hydrodynamic generalization of the common transfer - matrix method for layered systems @xcite , and obtain the exact dispersion equation of the hmm @xmath9 + \frac{{k_{\scriptscriptstyle \parallel}({w_d } - { w_m})}}{{{z_d}}}\sin { \theta _ d}\nonumber\\ & \,&(1 - \cos { \theta _ m}\cos { \theta _ l})-\frac{1}{2}\big[\frac{k_{\scriptscriptstyle \parallel}^2}{k_{{\scriptscriptstyle \bot}{\rm m}}^{\rm l}}\frac{{{}{{({w_d } - { w_m})}^2}}}{{{z_d}{z_m } } } + k_{{\scriptscriptstyle \bot}{\rm m}}^{\rm l}\left(\tfrac{{{z_d}}}{{{z_m } } } + \tfrac{{{z_m}}}{{{z_d}}}\right)\big]\sin { \theta _ m}\sin { \theta _ l}\big\ } \nonumber\\ & \,&{\big[k_{{\scriptscriptstyle \bot } { \rm m}}^{\rm l}\sin { \theta _ l } - k_{\scriptscriptstyle \parallel}\frac{{({w_d } - { w_m})}}{{{z_m}}}\sin { \theta _ m}\big]^ { - 1}},\label{exact_dispersion}\end{aligned}\ ] ] where for convenience we introduced the dimensionless parameters @xmath10 and where @xmath11 represents the bloch wavevector in the direction of the periodicity , @xmath12 the wavevector along the layers , and @xmath13 the free - space wavevector . we also introduced the derived wavevectors @xmath14 , @xmath15 , and @xmath16 . we note that solving the hydrodynamic drude model requires boundary conditions in addition to the usual maxwell boundary conditions . the dispersion relation eq . ( [ exact_dispersion ] ) was derived using the continuity of the normal component of the free - electron current as the additional boundary condition . details can be found in refs . the exact dispersion relation for nonlocal response eq . ( [ exact_dispersion ] ) reduces to the exact dispersion relation in the lra by putting the longitudinal permittivity @xmath17 of eq . ( [ epstl ] ) equal to the transverse permittivity @xmath18 of eq . ( [ epstl ] ) , in other words by taking the nonlocal parameter @xmath19 to zero . as stated before , we consider sub - wavelength unit cells much smaller than optical wavelengths , but with metal layers much larger than the wavelengths of their longitudinal pressure waves . so we focus on the situation where the following two parameters are small , @xmath20 where @xmath21 is the longitudinal wavevector . furthermore , we will only consider the frequency range @xmath22 , where @xmath23 and the dispersion curve of the hmm could be a hyperbola in the lra . we now make a first - order taylor approximation in the small parameters of eq . ( [ condition ] ) to the exact dispersion relation eq . ( [ exact_dispersion ] ) , and obtain the approximate dispersion relation @xmath24 the two terms in the brace are two leading correction terms that originate from the nonlocal response of the free electrons and from the finite size of the unit cell , respectively . without them , _ i.e. _ when neglecting both the finiteness of the unit cells and nonlocal response , we have the well - known dispersion relation in the lra @xmath25 where @xmath26 and @xmath27 are the effective local permittivities @xcite @xmath28 obviously , when @xmath29 , eq . ( [ approx_dispersion2 ] ) describes a hyperbolic dispersion curve . however , we are now rather interested in nonlocal response effects in realistic hmms , so we investigate the importance of the terms in the brace in eq . ( [ approx_dispersion ] ) , which are given in terms of @xmath30 . \end{aligned}\ ] ] from the expression ( [ ehdm ] ) for @xmath31 , it follows that the local - response limit is found in the limit @xmath32 rather than by taking the size @xmath2 of the unit cell to zero , see also ref . @xcite on this point . in the following we will keep the first ( _ i.e. _ the nonlocal ) correction term , while neglecting the other correction term @xmath33 . the sole justification for doing so is that the second term turns out to be negligible in comparison to the first one for the hmms that we consider , as we illustrate numerically below . so , as an improved approximation to eq . ( [ approx_dispersion2 ] ) , we keep the correction term concerning @xmath34 , and then obtain the approximate nonlocal dispersion relation that is central to our present study , @xmath35 in terms of the new effective permittivity in the direction of the periodicity corrected by the nonlocal response @xmath36 the other tensor component @xmath26 of the dielectric tensor has no such nonlocal correction . since @xmath37 , we usually have @xmath38 , and then @xmath39 is a good approximation . however , for large @xmath27 and especially in the extreme case that @xmath40 , it may occur that @xmath41 becomes much larger than @xmath34 , so that @xmath42 by virtue of eq . ( [ nlocapmp ] ) . so in this case , the nonlocal response has the important role of replacing the infinite @xmath27 by the finite @xmath31 in the dispersion relation . but when does it occur that @xmath40 ? this immediately follows from eq . ( [ localmp ] ) , and occurs when @xmath43 vanishes . in particular , neglecting the drude damping @xmath44 of the metal , the frequency @xmath45 for which @xmath40 is given by @xmath46 let us also consider another extreme case , namely @xmath47 . this is achieved when @xmath48 , which happens at the frequency @xmath49 at this frequency @xmath50 , the nonlocal response changes the finite @xmath27 into the infinite @xmath51 , again a strong nonlocal effect . from the above analysis , it follows that nonlocal response is important for hmms near the frequencies @xmath45 and @xmath50 , as also evidenced numerically in the next section . the larger the difference between @xmath45 and @xmath50 , the broader the frequency range with noticeable nonlocal effects . as indicated by eqs . ( [ res1 ] ) and ( [ res2 ] ) , the nonlocal resonance frequency @xmath50 is blueshifted with respect to the local one @xmath45 . the relative blueshift @xmath52 is proportional to @xmath5 , and inversely proportional to the dimensionless parameter @xmath53 . the latter dependence is rather surprising , since @xmath54 is the longitudinal wavevector of the free electrons in the metal layers , whereas @xmath3 is the thickness of the dielectric layer ! we checked , also numerically , that our first - order taylor expansion of the dispersion relation in the small parameter @xmath55 indeed gives a relative blueshift @xmath56 , independent of the thickness @xmath4 of the metal layer . to numerically illustrate the effects of the nonlocal response on the hmm , we choose a specific example of the hmm with @xmath57 , @xmath58 , and @xmath59 . we choose the metal to be au , and describe it by only its free - electron response , with parameters @xmath60 , @xmath61 , and @xmath62 . and @xmath63 , ( b ) real part of @xmath64 , and ( c ) real parts of @xmath27 and @xmath51 , of the hmm . the unit cell of the hmm shown in the inset of ( a ) has @xmath57 , @xmath58 , @xmath59 , and the yellow metal layer is au.,scaledwidth=60.0% ] fig . [ fig2](a ) shows the value of @xmath65 and @xmath66 . clearly , in the frequency range @xmath67 , both @xmath65 and @xmath66 satisfy eq . ( [ condition ] ) . this indicates that we are in the regime where the analytical results of the above section are valid . [ fig2](b ) displays the real part of @xmath68 , and fig . [ fig2](c ) the real parts @xmath27 and @xmath51 . the two dashed vertical lines in fig . [ fig2](c ) mark the positions of @xmath45 and @xmath50 as calculated with the approximate but accurate eqs . ( [ res1 ] ) and ( [ res2 ] ) , respectively . with the damping loss of the metal , @xmath27 and @xmath51 show large but finite values near @xmath45 and @xmath50 , respectively . around @xmath45 and @xmath50 , @xmath27 and @xmath51 show a huge difference , indicating the importance of nonlocal response , consistent with the theoretical prediction in the previous section . far from @xmath45 and @xmath50 , the dielectric functions @xmath27 and @xmath51 are nearly equal , so that away from these resonances the nonlocal response is only a small perturbation . fig . [ fig3](a ) and ( b ) illustrate the dispersion curves at @xmath69 and @xmath70 , respectively . these frequencies are far away from the resonances @xmath45 and @xmath50 . the small damping loss is neglected for a clearer illustration of the dispersion curve . several observations can be extracted from fig . firstly , the exact local and nonlocal dispersion curves agree well with those based on the effective material parameters , which confirms the validity of the approximate theoretical results of sec . [ sec : effective ] . secondly , the local and nonlocal curves only show a slight difference , agreeing well with fig . [ fig2](c ) where we see @xmath71 both for @xmath69 and @xmath72 . and ( b ) @xmath73 . the hmm is as that in fig . [ fig2 ] except that the loss is neglected . red solid curves for the exact local dispersion , blue solid curves for the exact nonlocal dispersion , red dashed curves for the approximated local dispersion of eq . ( [ approx_dispersion2 ] ) , and blue dashed curves for the approximated nonlocal dispersion of eq . ( [ approx_dispersion3]).,scaledwidth=60.0% ] and ( b ) @xmath74 . the hmm is as that in fig . [ fig2 ] except that the loss is neglected . red solid curves denote the exact local dispersion , blue solid curves the exact nonlocal dispersion ; red dashed curves for the approximated local dispersion curves of eq . ( [ approx_dispersion2 ] ) , and blue dashed curves for the approximated dispersion curves of eq . ( [ approx_dispersion3]).,scaledwidth=60.0% ] fig . [ fig4](a ) and ( b ) show the on - resonance dispersion curves , namely for @xmath75 and for @xmath74 , respectively . loss is again neglected , as in fig . [ fig3 ] . at both frequencies , the nonlocal response modifies the dispersion curve noticeably . in particular , at @xmath76 , the local dispersion curve consists of two nearly flat lines of @xmath77 . with the nonlocal response , however , the dispersion curve becomes a closed ellipse ! this remarkable difference can be understood from the difference between @xmath27 and @xmath51 . in particular , @xmath27 diverges while @xmath78 stays finite in the lossless case . when including the loss , the effective parameters change to @xmath79 and @xmath80 . the loss would only slightly modify the dispersion curves of fig . [ fig4](a ) . in fig . [ fig4](b ) we display dispersion relations at @xmath81 . here the local dispersion curve is a hyperbola . with the nonlocal response , however , the dispersion curve becomes nearly flat lines , which can be attributed to the extremely large value of @xmath51 at this frequency @xmath50 . a hmm slab can operate as a superlens with subwavelength resolution , since the hyperbolic dispersion curve supports propagating waves with arbitrarily high wavevectors , which can transfer the evanescent information of the object . in the lra , it is known that the hmm with its flat dispersion curve at @xmath45 [ recall fig . [ fig4](a ) ] is especially favorable for subwavelength imaging @xcite . the reason is that the flat dispersion curve with the constant @xmath82 ensures that all plane - wave components experience the same phase changes after transmission through the hmm slab , at least if reflections can be neglected . actually , the reflections can be suppressed by appropriate choice of the thickness @xmath83 of the hmm lens , and even be made to vanish by choosing @xmath84 , where @xmath85 is a positive integer . in theory this could lead to a perfect image at @xmath86 . let us now investigate how nonlocal response may influence the subwavelength imaging characteristics of the hmm lens . as demonstrated in the above sections , the nonlocal response sets the infinite @xmath87 to a finite value and accordingly destroys the desired flat dispersion curve at the frequency @xmath45 where one would normally choose to operate for perfect imaging , based on the lra . nevertheless , at a blueshifted frequency @xmath50 , the local effective dielectric function @xmath27 may be finite , but the hydrodynamic drude model predicts instead that @xmath51 diverges , with the concomitant flat dispersion curve suitable for subwavelength imaging . thus nonlocal response is expected to strongly affect the performance of hmm superlenses for frequencies @xmath88 around @xmath45 and @xmath50 . positioned to the left of a hmm slab that has its left interface at @xmath89 and right interface at @xmath86 . for @xmath90 , panel ( a1 ) shows the intensity for local response , ( b1 ) for nonlocal response , and ( c1 ) the intensity along @xmath91 with @xmath86 . panels ( a2 ) , ( b2 ) , and ( c2 ) are the analogous graphs for @xmath92 . in panels ( c1 ) and ( c2 ) , the green dashed curves are for the case without the hmm slab , the red solid curves for the hmm with local response , and the blue solid curves for the hmm with nonlocal response . the hmm unit cell is as that in fig . [ fig2 ] , @xmath93 and @xmath94.,scaledwidth=60.0% ] as an example , we consider a hmm slab with @xmath94 , _ i.e. _ composed of @xmath95 unit cells , in a free - space background with the two boundaries at @xmath89 and @xmath86 . the unit cell is as in fig . [ fig2 ] , and is arranged in a symmetric sandwich structure with the metal layer at the center . a line dipole source is positioned to the left of the hmm slab , with @xmath96-coordinate @xmath97 and @xmath91-coordinate 0 , and is represented by @xmath98 . we choose the distance to the hmm slab to be @xmath93 . [ fig5](a1 ) and ( b1 ) demonstrate the transmitted electric - field intensity at @xmath99 for local and nonlocal response , respectively . metal loss is taken into account . in the local case , we have the known the dispersion of fig . [ fig4](a ) featuring flat lines . additionally , the reflection is nearly zero since @xmath100 , so all wave components experience the same phase change . accordingly , after propagation through the hmm from @xmath89 to @xmath86 , a subwavelength image is formed near @xmath86 , as seen in fig . [ fig5](a1 ) . by contrast , in the nonlocal case the dispersion curve turns into an ellipse . this closed dispersion curve sets an upper wavevector cutoff for the evanescent waves , and the wave components below the cutoff experience different phase changes . accordingly , the quality of the image becomes worse , see fig . [ fig5](a2 ) . in fig . [ fig5](c1 ) , we depict the electric - field intensity again at @xmath45 as a function of @xmath91 at the hmm boundary @xmath86 , and also for the case without the hmm slab . it is seen that the electric - field intensity is nearly vanishing in the absence of the hmm slab , since the evanescent components vanish after traveling a distance of @xmath83 , which is of the order of one free - space wavelength . with the hmm slab in place , the electric - field intensity in the local case shows a subwavelength image that peaks at @xmath101 , with a full width at half maximum ( fwhm ) of only @xmath102 . with nonlocal response , however , the electric - field intensity distribution for @xmath45 becomes flatter , with a double rather than a single peak , with peak intensities at @xmath103 . let us now turn to the other resonance frequency , namely @xmath104 of eq . ( [ res2 ] ) . [ fig5](a2 ) and ( b2 ) show the transmitted electric - field intensity for the local and nonlocal cases , respectively , at @xmath74 . at this frequency , the local dispersion curve is a hyperbola , while the nonlocal dispersion shows two flat lines . in the nonlocal case , we have @xmath105 indicating that reflections do exist and the wave components experience different phase changes . however , compared to the local case , the transmission coefficients for the different wave components in the nonlocal case vary more smoothly as a function of @xmath106 , thanks to the flat dispersion curve with the constant @xmath82 . as a result , the hydrodynamic drude model predicts a better focusing performance of the hmm slab at @xmath107 than does the local - response theory , compare fig . [ fig5](a2 ) and ( b2 ) . in fig . [ fig5](c2 ) , the electric - field intensity along @xmath91 at @xmath86 is shown . in the local case , the electric - field intensity shows several peaks with the strongest two at @xmath108 . in the nonlocal case , the electric - field intensity is peaked at @xmath101 with a subwavelength fwhm of only @xmath109 . fig . [ fig5 ] indicates the possibility of detecting the nonlocal response experimentally by measuring the transmitted near - field distribution at the surface of a hmm superlens , which in our setup would be its right interface at @xmath86 . the images in fig . [ fig5](c1,c2 ) correspond to hypothetical measurements with infinitely small detectors . in experiments , the measured near - field signal will rather be an area - averaged electric - field intensity @xmath110 where @xmath111 is the finite detection area of the detector . let us now assume that we have a near - field detector with detection area in the @xmath112-plane , with a square shape of size @xmath113 , touching the hmm interface at @xmath86 . for the same light source interacting with the hmm slab as in fig . [ fig5 ] , we depict in fig . [ fig6 ] the calculated @xmath114 as a function of the @xmath91-coordinate of the center of the detector . it is seen that local and nonlocal response models give significantly different predictions for the measured signal , and thus the differences between the two models survive detection area averaging . the single peaks at @xmath115 for local response and at @xmath116 for nonlocal response are naturally broader than in fig . [ fig5](c1,c2 ) , also due to the area averaging . interestingly , as the frequency increases from @xmath115 to @xmath116 , the distribution of @xmath114 becomes broader in the lra , but narrower in the hydrodynamic drude model . of the light emitted by the same source as in fig . [ fig5 ] and transmitted through the same hmm slab . the detector is square - shaped with size @xmath117 , touching the hmm right interface at @xmath86 . it scans along the @xmath91-direction , and the @xmath91-coordinate on the horizontal axes in panels ( a ) for the lra and ( b ) for the hdm corresponds to that of the center of the detector . [ fig6],scaledwidth=60.0% ] we investigated the effects of the hydrodynamic nonlocal response on hyperbolic metamaterials with periodicity in the subwavelength regime of the transverse optical wave , but much larger than the wavelength of the longitudinal hydrodynamic pressure waves . it is found that the nonlocal response corrects the effective permittivity tensor element @xmath27 in the periodicity direction to the new form @xmath51 of eq . ( [ nlocapmp ] ) . around the frequencies @xmath45 and @xmath50 corresponding to @xmath87 and @xmath118 , respectively , @xmath27 and @xmath51 show noticeable differences , even leading to completely different dispersion curves with and without the nonlocal response . we find that nonlocal response blueshifts the resonance frequency of hmms from @xmath45 [ eq . ( [ res1 ] ) ] to @xmath50 [ eq . ( [ res2 ] ) ] . the relative blueshift has an interesting simple form independent of the thickness of the metal layers . similar nonlocal blueshifts for single nanoplasmonic particles have been predicted before , and significant blueshifts have also been measured @xcite ; how much of these can be attributed to hydrodynamic effects is a hot topic @xcite . furthermore , we predict that nonlocal response shows its mark in the performance of a finite hmm slab as a focusing lens : when increasing the operating frequency from @xmath45 to @xmath50 , for local response the near - field distribution of the transmitted light shows an image that gets out of focus , whereas the focus would improve instead according to the nonlocal - response theory . we propose to test the blueshift and these contrary predictions experimentally , as a clear and interesting test whether nonlocal response can be observed in hyperbolic metamaterials . this work was financially supported by an h. c. rsted fellowship ( w.y . ) .	we investigate the effects of hydrodynamic nonlocal response in hyperbolic metamaterials ( hmms ) , focusing on the experimentally realizable parameter regime where unit cells are much smaller than an optical wavelength but much larger than the wavelengths of the longitudinal pressure waves of the free - electron plasma in the metal constituents . we derive the nonlocal corrections to the effective material parameters analytically , and illustrate the noticeable nonlocal effects on the dispersion curves numerically . as an application , we find that the focusing characteristics of a hmm lens in the local - response approximation and in the hydrodynamic drude model can differ considerably . interestingly , sometimes the nonlocal theory predicts significantly better focusing . thus , to detect whether nonlocal response is at work in a hyperbolic metamaterial , we propose to measure the near - field distribution of a hyperbolic metamaterial lens . 10 d. r. smith and d. schurig , `` electromagnetic wave propagation in media with indefinite permittivity and permeability tensors , '' phys . rev . lett . * 90 * , 077405 ( 2003 ) . i. i. smolyaninov , `` vacuum in a strong magnetic field as a hyperbolic metamaterial , '' phys . rev . lett . * 107 * , 253903 ( 2011 ) . h. n. s. krishnamoorthy , z. jacob , e. narimanov , i. kretzschmar , and v. m. menon , `` topological transitions in metamaterials , '' science * 336 * , 205209 ( 2012 ) . z. jacob , i. smolyaninov , and e. narimanov , `` broadband purcell effect : radiative decay engineering with metamaterials , '' appl . phys . lett . * 100 * , 181105 ( 2012 ) . a. n. poddubny , p. a. belov , g. v. naik , and y. s. kivshar , `` spontaneous radiation of a finite - size dipole emitter in hyperbolic media , '' phys . rev . a * 84 * , 023807 ( 2011 ) . t. tumkur , g. zhu , p. black , y. a. barnakov , c. e. bonner , and m. a. noginov , `` control of spontaneous emission in a volume of functionalized hyperbolic metamaterial , '' appl . . lett . * 99 * , 151115 ( 2011 ) . m. a. noginov , y. a. barnakov , g. zhu , t. tumkur , h. li , and e. e. narimanov , `` bulk photonic metamaterial with hyperbolic dispersion , '' appl . . lett . * 94 * , 151105 ( 2009 ) . p. belov and y. hao , `` subwavelength imaging at optical frequencies using a transmission device formed by a periodic layered metal - dielectric structure operating in the canalization regime , '' phys . rev . b * 73 * , 113110 ( 2006 ) . b. wood , j. pendry , and d. tsai , `` directed subwavelength imaging using a layered metal - dielectric system , '' phys . rev . b * 74 * , 115116 ( 2006 ) . x. li , s. he , and y. jin , `` subwavelength focusing with a multilayered fabry - perot structure at optical frequencies , '' phys . rev . b * 75 * , 045103 ( 2007 ) . a. salandrino and n. engheta , `` far - field subdiffraction optical microscopy using metamaterial crystals : theory and simulations , '' phys . rev . b * 74 * , 075103 ( 2006 ) . z. jacob , l. alekseyev , and e. narimanov , `` optical hyperlens : far - field imaging beyond the diffraction limit , '' opt . express * 14 * , 84278256 ( 2006 ) . z. liu , h. lee , y. xiong , c. sun , and x. zhang , `` far - field optical hyperlens magnifying sub - diffraction - limited objects , '' science * 315 * , 1686 ( 2007 ) . m. yan and n. a. mortensen , `` hollow - core infrared fiber incorporating metal - wire metamaterial , '' opt . express * 17 * , 1485114864 ( 2009 ) . j. elser , v. podolksiy , i. salakhutdinov , and i. avrutsky , `` nonlocal effects in effective - medium response of nanolayered metamaterials , '' appl . phys . lett . * 90 * , 191109 ( 2007 ) . a. chebykin , a. orlov , a. vozianova , s. maslovski , y. s. kivshar , and p. belov , `` nonlocal effective medium model for multilayered metal - dielectric metamaterials , '' phys . rev . b * 84 * , 115438 ( 2011 ) . a. chebykin , a. orlov , c. simovski , y. s. kivshar , and p. belov , `` nonlocal effective parameters of multilayered metal - dielectric metamaterials , '' phys . rev . b * 86 * , 115420 ( 2012 ) . l. shen , t. yang , and y. chau , `` effect of internal period on the optical dispersion of indefinite - medium materials , '' phys . rev . b * 77 * , 205124 ( 2008 ) . l. shen , t. yang , and y. chau , `` 50/50 beam splitter using a one - dimensional metal photonic crystal with parabolalike dispersion , '' appl . . lett . * 90 * , 251909 ( 2007 ) . f. bloch , `` bremsvermgen von atomen mit mehreren elektronen , '' z. phys . a * 81 * , 363376 ( 1933 ) . a. d. boardman , _ electromagnetic surface modes _ ( john wiley and sons , chichester , 1982 ) . w. l. mochn , m. castillo - mussot , and r. g. barrera , `` effect of plasma waves on the optical properties of metal - insulator superlattices , '' phys . rev . b * 15 * , 10881098 ( 1987 ) . c. david and f. j. garc ' ia de abajo , `` spatial nonlocality in the optical response of metal nanoparticles , '' j. phys . chem . c * 115 * , 1947019475 ( 2011 ) . f. j. garc ' ia de abajo , `` nonlocal effects in the plasmons of strongly interacting nanoparticles , dimers , and waveguides , '' j. phys . chem . c * 112 * , 1798317987 ( 2008 ) . s. raza , g. toscano , a .- p . jauho , m. wubs , and n. a. mortensen , `` unusual resonances in nanoplasmonic structures due to nonlocal response , '' phys . rev . b * 84 * , 121412(r ) ( 2011 ) . j. a. scholl , a. l. koh , and j. a. dionne , `` quantum plasmon resonances of individual metallic nanoparticles , '' nature * 483 * , 421427 ( 2012 ) . c. cirac ` i , r. t. hill , j. j. mock , y. urzhumov , a. i. fern ' andez - dom ' inguez , s. a. maier , j. b. pendry , a. chilkoti , and d. r. smith , `` probing the ultimate limits of plasmonic enhancement , '' science * 337 * , 10721074 ( 2012 ) . s. raza , n. stenger , s. kadkhodazadeh , s. v. fischer , n. kostesha , a .- p . jauho , a. burrows , m. wubs , and n. a. mortensen , `` blueshift of the surface plasmon resonance in silver nanoparticles studied with eels , '' arxiv:1210.2535v1 ( 2012 ) . g. toscano , s. raza , a .- p . jauho , m. wubs , and n. a. mortensen , `` modified field enhancement in plasmonic nanowire dimers due to nonlocal response , '' opt . express * 13 * , 41764188 ( 2012 ) . g. toscano , s. raza , s. xiao , m. wubs , a .- p . jauho , s. i. bozhevolnyi , and n. a. mortensen , `` surface - enhanced raman spectroscopy ( sers ) : nonlocal limitations , '' opt . lett . * 37 * , 25382540 ( 2012 ) . a. i. fern ' andez - dom ' inguez , a. wiener , f. j. garc ' ia - vidal , s. a. maier , and j. b. pendry , `` transformation - optics description of nonlocal effects in plasmonic nanostructures , '' phys . rev . lett . * 108 * , 106802 ( 2012 ) . w. yan , m. wubs , and n. a. mortensen , `` hyperbolic metamaterials : nonlocal response regularizes broadband supersingularity , '' phys . rev . b * 86 * , 205429 ( 2012 ) . p. jewsbury , `` electrodynamic boundary conditions at metal interfaces , '' j. phys . f : met . phys . * 11 * , 195206 ( 1981 ) . t. teperik , p. nordlander , j. aizpurua , and a. g. borisov , `` quantum plasmonics : nonlocal effects in coupled nanowire dimer , '' arxiv:1302.3339v1 ( 2013 ) .
we first assess the energetic description of au nws predicted by reaxff by comparing to dft and tb - sma . we report the energy for [ 100 ] oriented periodic fcc nws with diameters , d=1.1 , 1.5 , and 1.9 nm , as well as an elongation sequence of a periodic 0.85 nm nw . the elongation sequence is representative of the low coordination structures typically found in the necks of mechanically deformed nws . we perform dft calculations that make use the generalized gradient approximation by perdew , burke , ernzerhof ( pbe - gga ) and additionally include spin - orbital coupling ( see methods ) . all reported dft energy values are calculated using structures that have been fully and self - consistently energy - minimized . in [ fig : energy]a we plot the energy calculated with dft vs. the energy calculated with reaxff . in this plotting scheme , an ideal match between dft and reaxff occurs if the data point falls along the line @xmath2 . in [ fig : energy]a , filled symbols represent the energy calculated using the coordinates energy - minimized with dft ( i.e. , when calculating the reaxff energy , atoms are static and arranged according to the configuration generated with dft ) . the dft energy - minimized configurations for the elongation sequence are shown in figures c - g . we observe that reaxff is capable of accurately reproducing the per particle energy predicted by dft , where the magnitude of the difference is less than 0.05 ev per particle and all data points fall along the line @xmath2 . in [ fig : energy]a , data points with open symbols correspond to energy - minimized structures predicted by the two different treatments ( i.e. , the dft energy corresponds to the structure energy - minimized with dft and the reaxff energy corresponds to the structure energy - minimized with reaxff ) . the reaxff structures are generated starting from the dft optimized structures . the energy - minimized elongation sequence for reaxff is shown in figures h - l , where we observe minor changes as compared to the dft structures in figures c - g . note , the preferred interatomic spacings will vary slightly between methods , thus we use a two stage energy minimization for reaxff , first allowing the box volume to change to minimize pressure ( which scales atom positions ) , then allowing atom positions to change in order to minimize the potential energy . for minimized configurations , reaxff predicts a lower energy structure than dft for all statepoints . however the values are still close , where the magnitude of the deviation varies from 0.09 to 0.20 ev / particle . [ fig : energy]b is constructed in the same manner as [ fig : energy]a , but using tb - sma rather than reaxff . again , filled symbols represent the energy calculated using the dft energy - minimized structures . tb - sma fails to closely reproduce the energy , predicting lower values than dft for all structures . here the magnitude of the differences ranges from 0.17 to 0.41 ev per particle . open symbols again represent the energy of structures minimized with the two methods . we use a similar procedure for tb - sma as we did for reaxff . the energy minimized structures of tb - sma are shown in [ fig : energy]m - q , where we observe only minor differences when compared to either dft or reaxff structures . the most noticeable difference is observed in [ fig : energy]q , where the left hand side of the nw compresses into a more locally dense state . for energy - minimized structures , the energy predicted by tb - sma is shifted to even more negative values , where now the magnitude of the energy difference with dft ranges from 0.21 to 0.47 ev per particle . in addition to comparing the absolute energy , we examine the relative change in energy ranging from the highest coordination state ( d=1.9 nm nw ) to the lowest coordination state ( i.e. , the final elongated structure , shown in figures g , l and q ) . this can be assessed by examining the slope of the data points shown in figures a and b ; a slope of unity provides an ideal energy change with respect to dft . for the reaxff calculations in [ fig : energy]a , we observe a slope 0.99 for the static structure ( filled symbols ) and 1.11 for the energy - minimized structures ( open symbols ) , using linear regression . thus , even though the absolute energies are slightly different for the energy - minimized states ( open symbols ) , the relative change predicted by reaxff is in very good agreement with dft . for the tb - sma calculations in [ fig : energy]b , we observe a slope of 1.63 for static calculations ( filled symbols ) and a slope of 1.81 for energy - minimized structures ( open symbols ) . the strong deviation from unity indicates that tb - sma fails to properly predict the energy loss associated with a reduction in average coordination number , which may strongly influence the structural evolution of the nw . in previous work comparing common semi - empirical methods for the elongation of finite au nws , tb - sma was demonstrated to provide the best agreement with dft in terms of the relative energy change@xcite . thus it appears that reaxff is an improvement over tb - sma , providing a better description of both the relative energy change and absolute energy for au nws . 75@xmath3 of their true size with individual polytetrahedral structures highlighted within the nw and shown at the right . * ( c ) * rendering of a z12 structure , i.e. , an icosahedron , with an individual tetrahedral subunit highlighted . partial z12 clusters are shown as a function of coordination number , cn , varying from fully coordinated , cn=12 , to cn=6 . note , cn=6 shows two different partial clusters commonly observed . , width=316 ] we perform simulations of cylindrical au nws of length 3.6 nm , with diameters , d=1.1 , 1.5 and 1.9 nm ( 177 , 317 , and 589 atoms , respectively ) for a variety of statepoints , calculating 10 independent simulations for each parameter combination ( see methods for more details ) . here we focus on reaxff - based simulations but also include tb - sma - based simulations for comparison . in these reaxff simulations we observe that , as a typical au nw elongates , a small region begins to `` neck , '' becoming narrower than the original diameter . the neck forms in response to the applied tensile load , localizing many of the structural changes to this neck region , allowing the bulk of the nw to relax its configuration . within the necks of our simulated nws , we often observe the formation of non - crystalline structures with distinct ring- and crescent - like geometries . [ fig : polyhedra]a shows a representative reaxff simulation snapshot of a d=1.1 nm nw at t=298k , shortly after neck formation ; an individual ring - like structure is highlighted in the snapshot with an enlarged version , extracted from the system , shown to the right . [ fig : polyhedra]b shows a representative snapshot of a d=1.5 nm nw where two crescent - like structures are highlighted and extracted . these ring- and crescent - like structures are polytetrahedral in nature , where the local structure of the atoms can be decomposed solely into individual tetrahedral subunits@xcite . specifically , the highlighted structure in [ fig : polyhedra]a is best classified as an icosahedron . an icosahedron is constructed of 13 particles arranged as two splayed pentagonal pyramids ( composed of 20 total tetrahedral subunits ) , demonstrating 5-fold symmetry@xcite ; an ideal icosahedron is shown in [ fig : polyhedra]c . the two highlighted clusters in [ fig : polyhedra]b are partially coordinated icosahedral clusters , that maintain the same basic geometry of the full cluster , but with @xmath4 number of particles removed ( i.e. , several tetrahedral subunits removed)@xcite . ideal partial clusters of various coordination are shown in [ fig : polyhedra]c . tb - sma simulations also demonstrate polytetrahedral structures within the necks of the elongated nws , similar to figures a and b. to quantify the structures formed in our simulations and establish trends , we use the @xmath5 method to determine which atoms are in polytetrahedral local environments ( see methods ) . we use this quantitative information to construct three metrics : ( 1 ) the maximum number polytetrahedra that occur concurrently during the course of a simulation trajectory , ( 2 ) the number of polytetrahedra per timestep in a simulation ( calculated by normalizing the total number of polytetrahedra observed by the total number of statepoints considered ) , and ( 3 ) the coordination number ( cn ) histogram of polytetrahedral atoms . we first report the behavior of thin nws with d=1.1 nm elongated at a rate of 1 m / s for t=\{10 , 100 , 200 , 298 , 400 , 500 } k. figures a and b plot the maximum number of polytetrahedra and the polytetrahedra per timestep , respectively . in all cases , we report median values , where error bars correspond to the first and third quartiles , since the data is not necessarily well described by a gaussian distribution . reaxff simulations demonstrate a steady increase in both metrics as temperature is increased , i.e. , both the maximum size of the polytetrahedral domains and the likelihood of finding polytetrahedra increases as t increases . the polytetrahedral motif is not prominent for reaxff systems at low temperature , i.e. , t @xmath6 200k . while the behavior of the reaxff and tb - sma potentials converge for high temperature , tb - sma demonstrates markedly different trends and behavior . overall , we observe that tb - sma predicts a significantly larger value of both the maximum number of polytetrahedra and polytetrahedra per timestep than the reaxff simulations . tb - sma predicts two regimes with an approximate crossover of t=200k . tb - sma systems at t=10 and 100k tend to maintain a predominantly fcc structure , whereas higher temperature states transition into a3 core - shell helical structures @xcite upon moderate amounts of elongation . for example , at t=298k , 70@xmath3 of the tb - sma simulations transform into predominantly a3 structured nws by the time the nw has been elongated by 5 . the a3 structure is locally composed of tetrahedral subunits that closely match the polytetrahedral structures in our @xmath5 reference library . however , we note that even for t@xmath6200k , where a3 structures are not found with either potential , reaxff still forms fewer polytetrahedra than tb - sma , by a factor of @xmath010 for both metrics . the predominance of crystalline structures at low temperature predicted by reaxff is more consistent with results of zero temperature dft calculations performed for the elongation of [ 110 ] and [ 111 ] au nws @xcite . figures c and d plot the median value of the maximum number of polytetrahedra and polytetrahedra per timestep , respectively , for nws as a function of nw diameter ; these simulations are performed at t=298k , as this statepoint demonstrates appreciable numbers of polytetrahedra for both potentials and temperatures in this vicinity are common in experiment @xcite . the values reported in figures . c and d correspond to the aggregate of simulations performed for rates 0.1 to 5 m / s , as we did not observe strong variability with rate of elongation . we observe that the median value of both metrics decreases with increasing diameter for both reaxff and tb - sma . that is , the likelihood of forming polytetrahedra decreases with increasing nw diameter . in all cases , reaxff - based simulations predict smaller median values for the two metrics than tb - sma , with the largest differences occurring for the d=1.1 nm nws . the differences we observe between reaxff and tb - sma are likely related to the energy scaling we previously explored , where tb - sma overpredicts both the absolute magnitude and relative energy of lower - coordination states . [ fig : cn ] plots histograms of the coordination number of atoms identified as having polytetrahedral local ordering for nws at t=298k ( corresponding to the data presented in figures c and d ) . histograms of reaxff simulations demonstrate a diameter dependence ; d=1.1 nm nws show two key peaks at cn=6 and 8 , but the relative fraction of cn=6 strongly increases with increasing diameter . for d=1.9 nm nws simulated with reaxff , @xmath0 80 % of the observed polytetrahedral clusters have cn=6 . histograms generated from tb - sma simulations demonstrate two clear peaks at cn=6 and 8 , encompassing @xmath0 80 % of the total polytetrahedral structures , with much less dependence on diameter . we calculate the zero - bias conductance of several representative trajectories generated with reaxff and tb - sma in order to assess the impact of polytetrahedra , i.e. , is there a characteristic conductance `` fingerprint '' associated with their formation . as described in the methods section , we use a combined dft - non - equilibrium green s function technique to calculate the conductance@xcite , which we report in terms of @xmath7 , where @xmath8 is the charge of an electron and @xmath9 is planck s constant . in experiment , a typical conductance trace demonstrates a `` staircase '' behavior , with abrupt plateaus and drops as a function of elongation . the abrupt changes in the conductance coincide with structural rearrangements of the atoms , such as thinning of the nw neck . for very small cross - sections , the steps often occur at roughly integer multiples ( @xmath10 ) of @xmath11 . it is important to note that histograms of conductance tend to demonstrate a substantial spread around each of these integer peaks @xcite and , as a result , non - integer values are also regularly observed in experiment . recent work has correlated the formation of non - integer plateaus to the difference in dimensionality of the local structures within the nw neck ( i.e. , the difference between 3- , 2- , and 1-dimensional atomic arrangements)@xcite . 3@xmath12 are inset in each figure , where c and e each highlight a polytetrahedral structure in the neck ( red).,width=432 ] [ fig : conductance ] plots the conductance and number of polytetrahedra as a function of elongation for trajectories of d=1.1 nm nws elongated at 1 m / s , generated with both reaxff and tb - sma . we first consider a reaxff trajectory at t=298k that does not demonstrate any significant polytetrahedral ordering , as shown in figures a - b . we observe distinct step - wise behavior with conductance plateaus at roughly integer multiples of @xmath12 ( @xmath13 1,2 , 3 and 4 g@xmath14 , highlighted in [ fig : conductance]a with horizontal lines ) . additionally , the length of these plateaus range from 1.5 - 3.5 which matches the approximate plateau length of 1 - 4 seen in experiments at t=4k @xcite and t=298k @xcite . a snapshot of the nw at @xmath03@xmath12 is inset in [ fig : conductance]a showing a crystalline neck structure . figures c - d plot data for a nw trajectory generated with reaxff at t=10k . as before , we observe step wise behavior in the conductance . in contrast to the previous trajectory , we now observe the formation of an appreciable number of polytetrahedral structures , as shown in [ fig : conductance]d . the formation of polytetrahedra appears to have a distinct impact on the conductance . for example , while we observe plateaus at @xmath133 and 1 @xmath12 in [ fig : conductance]c , there is a gradual change in conductance between these two steps rather than a plateau at 2 @xmath12 . the onset of the formation of polytetrahedral structures at @xmath05 correlates with the start of the conductance plateau at 3@xmath12 . visually , we observe that at the start of the gradual change ( i.e , 3@xmath12 ) , the narrowest region of the neck is spanned by a polytetrahedral structure @xmath0 3 atoms wide ; a shapshot is inset in [ fig : conductance]c . we further assess the impact of polytetrahedra by examining a trajectory at t=10k generated with tb - sma , as shown in figures e - f . recall that tb - sma over - predicts the tendency to form polytetrahedra as compared to reaxff , making it ideal for assessing their impact . we still observe step - like behavior , but now only see plateaus at @xmath13 4 and 1 @xmath12 . again , we observe a rather gradual transition from 3@xmath12 to 1@xmath12 . while the onset of polytetrahedra appears at 2.5 of elongation , the most significant peak occurs at @xmath05 ( see [ fig : conductance]f ) , which correlates to the start of the gradual change in conductance . as was noted in [ fig : conductance]c , we also observe that at @xmath03@xmath12 ( i.e. , the start of the gradual change ) , the narrowest region of the neck is spanned by a polytetrahedral structure @xmath0 3 atom diameters across ( a shapshot is inset in [ fig : conductance]c ) . thus it appears that the gradual conductance changes we observe are associated with the formation of polytetrahedra that span the cross - section of the neck . this is likely related to different modes of mechanical deformation exhibited by tetrahedral structures verses crystals ; the mechanical behavior of related core - shell nws as compared to crystals has previously been explored in ref . @xcite . our results regarding the impact of polytetrahedral neck structures on the conductance is supported by recent simulations of tavazza , _ et al . _ that used dft to investigate the elongation of [ 110 ] and [ 111 ] au nws at zero temperature @xcite . in that work , nearly identical conductance behavior was observed for systems that demonstrated non - crystalline neck structures , i.e. , gradual changes from 3@xmath12 to 1@xmath12 @xcite . similar conductance traces were also observed in experiments at t=4k @xcite . additionally , it was noted that systems exhibiting this gradual conductance change occurred in the minority of simulations and experiments , suggesting that non - crystalline structures are rare at these very low temperatures@xcite . as previously shown in figures a and b , our reaxff simulations also predict that the non - crystalline polytetrahedral structures are a rare structural motif at low temperature . the formation of polytetrahedral structures within the necks of the elongated au nws may be surprising as icosahedra are not predicted to be the lowest energy structures @xcite . generally speaking , planar structures are predicted to be the minimal energy states for small , isolated clusters of au @xcite . however , this does not imply that icosahedra , or other low , but not minimal energy structures are inherently unstable . to assess the stability of the polytetrahedral structures , we perform conjugate gradient energy minimization using dft of the isolated icosahedral clusters shown in [ fig : polyhedra]c . starting from both ideal structures and those generated with reaxff , we find that clusters with cn=6 , 8 , 9 , and 12 maintain their original geometry at the end of the minimization , suggesting that these structures are locally stable . we also note that systems that start as partial icosahedra with cn=10 and 11 relax their structures into more `` flattened '' states that , in 3 out of 4 calculations , contain a distinct polytetrahedral pentagonal di - pyramind ( i.e. , a cn=6 polytetrahedral structure ) . recall that the cn=6 structure is the most common polytetrahedral configuration found in our simulations . furthermore , polytetrahedral structures of au are well reported in experiment and theory , including monolayer protected icosahedral clusters with cn=12@xcite , phosphone - thiolate - protected partial - polytetrahedral au structures@xcite , and bi - icosahedral clusters of both face - sharing @xcite and vertex - sharing icosahedra @xcite . it is important to note that the polytetrahedral structures we observe in our simulations are not isolated clusters , but rather occur within the context of a larger , mechanically deformed nw . the average coordination number of atoms in an isolated cluster will be lower than an equivalent configuration that occurs within a nw . additionally , we must also consider that these structures form as a result of mechanically deforming a nw via an applied tensile load . previous experiments of nws pulled at a constant rate demonstrate a saw - tooth like behavior in the mechanical force on the nw as a function of elongation @xcite and thus the local forces of the atoms may be far from equilibrium and constantly changing . the original structure of the nw will also dictate how it can respond to tensile loads @xcite . for example , it has been demonstrated that the likelihood of forming monatomic au chains in elongated au nws is influenced by the original orientation of the crystalline lattice ( i.e. , [ 111 ] vs. [ 110 ] vs. [ 100 ] ) @xcite . focusing only on minimal potential energy structural motifs also ignores the importance of thermal effects . recall that in figures a - b we observed that the formation of polytetrahedral structures is very unlikely at low temperature , but increases with temperature . previous work from our group proposed a universal energy release mechanism that outlines the strong role that temperature can play in determining how a nw responds to the applied tensile load @xcite . additionally , au is ductile , which manifests itself as a tendency for the nw to remain in a cylindrical , connected structure by thinning within the neck region . as the neck of the nw thins , polytetrahedral structures may become the optimal configurations , since they are able to span narrow , cylindrical - like geometries by forming face - sharing polytetrahedral helices @xcite or vertex - sharing structures @xcite . this was demonstrated for particles with van der waals interactions , where a transition was observed from fcc / hcp structures to polytetrahedral structures upon decreasing the diameter of the confining cylinder @xcite . similar behavior was also demonstrated for au , where tetrahedral helices were predicted to be the optimal structure for small diameter nws whereas fcc structures were optimal for large diameter nws @xcite . recent experimental work has demonstrated the synthesis of thin au / ag nws that are locally constructed of face - sharing icosahedra@xcite , additionally demonstrating this packing motif for metallic nws . furthermore , the formation of polytetrahedral structures within the necks of elongating nws is not without precedent . previous dft calculations of the mechanical deformation of [ 100 ] oriented bcc na nws demonstrated the formation of distorted full and partial icosahedra within the nw neck @xcite . this may suggest that the formation of polytetrahedra is a more universal relaxation mechanism for mechanically deformed nws . the formation of polytetrahedral structures may have important implications regarding the electronic properties of the nw . as we saw in [ fig : conductance ] , the formation of polytetrahedra diminishes the quantization of the conductance traces . combined with the increased likelihood of forming polytetrahedra as temperature increases , this may be an important to consider when engineering nanoelectronic devices designed to operate at , e.g. , room temperature . the impact of polytetrahedra may be even more substantial . the work of jiang , _ et al . _ predicts that thiolated vertex - sharing icosahedral au nws behave as a semiconductor for the -2 and + 2 charge states@xcite . this suggest that au nws elongated in the presence of a thiol solvent might , under certain conditions , change from behaving as metals to semiconductors as a result of the formation of polytetrahedral neck structures . in this work , we showed good agreement between reaxff and dft in terms of the absolute energy and relative energy scaling . we also showed that reaxff provides a better energetic description than the semi - empirical tb - sma potential , which has commonly been used to study au nw elongation . we further demonstrated the formation of polytetrahedral local structures within the necks of elongated au nws simulated with reaxff . we observed that the formation polytetrahedra is not a strong motif of structural relaxation for low temperature systems , but the likelihood of forming polytetrahedra increases with temperature . we also found that the likelihood of forming polytetrahedra increases as nw diameter decreases . polytetrahedral order was also observed for simulations performed with the tb - sma potential , however , tb - sma appears to significantly over - predict their formation as compared to reaxff , especially for small diameter nws and low temperature . we find that reaxff - based simulations produce trajectories that demonstrate clear step - wise behavior in conductance that closely matches experiment . we also observe that the formation of polytetrahedra ( i.e. non - crystalline structures ) within the nw necks dampens the quantization behavior of the conductance , closely matching the observations in ref . the results presented here suggest that the formation of polytetrahedral structures within the necks of the [ 100 ] orientated au nws is a ubiquitous and important mechanism for relieving the internal stress brought about by elongation , in particular , for small diameter nws at high temperature . _ acknowledgment : _ this work was funded by the us department of energy ( doe ) and supported by computational resources provided by the national energy research scientific computing center ( nersc ) of the doe under contract no . de - ac02 - 05ch11231 and the national institute for computational sciences , project - id ut - tnedu014 @xcite . wrf was supported under the u.s . department of education graduate assistance in areas of national need ( gaann ) fellowship under grant no . research by prck at the center for nanophase materials sciences was sponsored at oak ridge national laboratory by the office of basic energy sciences , doe . _ dft calculations : _ we perform dft calculations using the generalized gradient approximation of perdew , burke , ernzerhof ( pbe - gga)@xcite within the plane wave projector augmented wave formalism ( paw ) @xcite . our calculations were performed at the gamma point with a 230ev plane - wave energy cutoff . the paw potentials included the outermost d and s electrons in the valence , giving 11 electrons per atom . significantly , we fully included spin - orbit coupling in all our calculations of the energy and forces . all reported dft energy values are calculated using structures that have been fully and self - consistently energy - minimized . _ simulation methodology : _ our basic simulation procedure closely mimics the `` stretch and relax '' procedure of pu , _ et al . _ @xcite : we rigidly fix the last two rows of each end of the nw creating `` grip '' atoms at each end ; we allow the core of the system to relax for 100 ps ; after relaxation , we elongate the nw by displacing the grip atoms at one end by increments of 0.05 , allowing the system to relax for a prescribed time between displacements , repeating until the nw breaks . we explore elongation rates from 5 m / s ( 1 ps relaxation between displacements ) to 0.1 m / s ( 50 ps ) . this range encompasses typical elongation rates used to draw metallic whiskers with stm or afm @xcite . simulations are conducted between t=10 k and 500 k ; common experimental conditions are t=4 - 10 k @xcite and t@xmath0 300k @xcite . for each statepoint we perform 10 independent simulations to obtain sufficient statistics . reaxff simulations use the fitting parameters derived by keith , et al.@xcite . this particular fitting was chosen because it was parameterized using pbe - gga dft , which is well suited for systems with surfaces ( i.e. , nws ) , and the elastic moduli of the fitting closely match experiment@xcite . tb - sma simulations are performed using fitting parameters derived by cleri and rosato@xcite . tb - sma is used as previous comparisons with dft showed it provides a better match in terms of relative energy change than other common semi - empirical potentials@xcite . both reaxff and tb - sma simulations are performed using the lammps simulation package@xcite ( extended to incorporate tb - sma ) with the nose - hoover thermostat for systems without periodic boundary effects . as is common , tb - sma simulations use a timestep of 2 femtoseconds @xcite , whereas a smaller timestep of 0.5 femtoseconds is needed for reaxff . to simulate the same timescale , reaxff requires more walltime than tb - sma , by a factor of @xmath050 . _ @xmath5 method : _ to analyze the local configurations of atoms , we employ techniques borrowed from the field of _ _ shape matching__@xcite . we use the r@xmath1 scheme proposed in references @xcite , which allows us to objectively identify structural trends in an automated fashion . this method relies on creating a rotationally invariant spherical harmonics @xcite `` fingerprint '' of the first neighbor shell surrounding each atom . we compare the fingerprint of each local cluster to a library of reference configurations and determine the best match . here , we quantify the match , @xmath15 , using the euclidian distance , normalized such that it spans from 0 ( worst match ) to 1 ( perfect match ) . by definition , every cluster in this scheme has a `` best match '' even if that match is poor . to avoid misidentification , we define a cutoff such that we ignore matches where @xmath16 . this provides results consistent with visual inspection ; the choice of cutoff will impact the specific calculated values , however the overall trends are not strongly perturbed by this choice@xcite . our reference library includes full and partially coordinated clusters of type fcc , hcp , and icoshaedra , incorporating both ideal configurations and configurations relaxed using the reaxff and tb - sma potentials , where we only consider clusters of 6 @xmath17 cn @xmath17 12 . _ conductance measurement : _ conductance calculations are carried out in a combined dft - non - equilibrium green s function framework . a double - zeta numerical atomic orbital basis set is used ( 18 basis functions per au atom ) in the calculations . the zero - bias conductance is given by @xmath18 where @xmath19 is the transmission coefficient of the device and @xmath20 is the fermi energy . @xmath21 is reported in units of @xmath7 , where @xmath8 is the charge of an electron and @xmath9 is planck s constant . we validated our conductance measurements by investigating single point atom contacts , finding good agreement with previous theoretical calculations @xcite . further details of the implementation and its accuracy can be found in ref . @xcite . muller , c. ; van ruitenbeek j. m. , ; de jongh l. j. , conductance and supercurrent discontinuities in atomic - scale metallic constrictions of variable width . _ physical review letters _ * 1992 * , _ 69 _ , 140143 krans , j. m. ; van rutlenbeek , j. m. ; fisun , v. v. ; yanson , i. k. ; de jongh , l. j. the signature of conductance quantization in metallic point contacts . _ nature _ * 1995 * , _ 375 _ , 767769 ohnishi , h. ; kondo , y. ; takayanagi , k. quantized conductance through individual rows of suspended gold atoms . _ nature _ * 1998 * , _ 395 _ , 780783 yanson , a. i. ; bollinger , g. r. ; van den brom , h. e. ; agrait , n. ; van ruitenbeek , j. m. formation and manipulation of a metallic wire of single gold atoms . _ nature _ * 1998 * , _ 395 _ , 783785 guo , s. ; hihath , j. ; tao , n. breakdown of atomic - sized metallic contacts measured on nanosecond scale . _ nano letters _ * 2011 * , _ 11 _ , 927933 marszalek , p. ; greenleaf , w. ; li , h. atomic force microscopy captures quantized plastic deformation in gold nanowires . _ proceedings of the national academy of sciences _ * 2000 * , _ 97 _ , 62826286 coura , p. ; legoas , s. ; moreira , a. ; sato , f. on the structural and stability features of linear atomic suspended chains formed from gold nanowires stretching . _ nano letters _ * 2004 * , _ 4 _ , 11871191 da silva , e. ; da silva , a. ; fazzio , a. breaking of gold nanowires . _ computational materials science _ * 2004 * , _ 30 _ , 7376 gall , k. ; diao , j. ; dunn , m. l. the strength of gold nanowires . _ nano letters _ * 2004 * , _ 4 _ , 24312436 sato , f. ; moreira , a. s. ; coura , p. z. ; dantas , s. o. ; legoas , s. b. ; ugarte , d. ; galvo , d. s. computer simulations of gold nanowire formation : the role of outlayer atoms . _ applied physics a _ * 2005 * , _ 81 _ , 15271531 koh , s. j. a. ; lee , h. p. molecular dynamics simulation of size and strain rate dependent mechanical response of fcc metallic nanowires . _ nanotechnology _ * 2006 * , _ 17 _ , 345167 pu , q. ; leng , y. ; tsetseris , l. ; park , h. s. ; pantelides , s. t. ; cummings , p. t. molecular dynamics simulations of stretched gold nanowires : the relative utility of different semiempirical potentials . _ the journal of chemical physics _ * 2007 * , _ 126 _ , 144707 wang , d. ; zhao , j. ; hu , s. ; yin , x. ; liang , s. ; liu , y. ; deng , s. where , and how , does a nanowire break ? _ nano letters _ * 2007 * , _ 7 _ , 120812 pu , q. ; leng , y. ; cummings , p. t. rate - dependent energy release mechanism of gold nanowires under elongation . _ journal of the american chemical society _ * 2008 * , _ 130 _ , 1790712 tsutsui , m. ; shoji , k. ; taniguchi , m. ; kawai , t. formation and self - breaking mechanism of stable atom - sized junctions . _ nano letters _ * 2008 * , _ 8 _ , 3459 lagos , m. j. ; sato , f. ; bettini , j. ; rodrigues , v. ; galvao , d. s. ; ugarte , d. observation of the smallest metal nanotube with a square cross - section . _ nature nanotech . _ * 2009 * , _ 4 _ , 149152 tavazza , f. ; levine , l. ; chaka , a. elongation and breaking mechanisms of gold nanowires under a wide range of tensile conditions . _ journal of applied physics _ * 2009 * , _ 106 _ , 043522 barnett , r. n. ; landman , u. cluster - derived structures and conductance fluctuations in nanowires . _ nature _ * 1997 * , _ 387 _ , 788791 rego , l. g. c. ; rocha , a. r. ; rodrigues , v. ; ugarte , d. role of structural evolution in the quantum conductance behavior of gold nanowires during stretching . _ physical review b _ * 2003 * , _ 67 _ , 045412 lopez - acevedo , o. ; koudela , d. ; hkkinen , h. conductance through atomic point contacts between fcc(100 ) electrodes of gold . _ the european physical journal b _ * 2008 * , _ 66 _ , 497501 tavazza , f. ; smith , d. t. ; levine , l. e. ; pratt , j. r. ; chaka , a. m. electron transport in gold nanowires : stable 1- , 2- and 3-dimensional atomic structures and noninteger conduction states . lett . _ * 2011 * , _ 107 _ , 126802 french , w. r. ; iacovella , c. r. ; cummings , p. t. the influence of molecular adsorption on elongating gold nanowires . _ the journal of physical chemistry c _ * 2011 * , _ 115 _ , 1842218433 fioravante , f. ; nunes , r. w. semiconducting chains of gold and silver . _ applied physics letters _ * 2007 * , _ 91 _ , 223115 park , h. s. ; cai , w. ; espinosa , h. d. ; huang , h. mechanics of crystalline nanowires . _ mrs bulletin _ * 2009 * , _ 34 _ , 178183 ikeda , h. ; qi , y. ; agin , t. ; samwer , k. ; johnson , w. l. ; goddard , w. a. strain rate induced amorphization in metallic nanowires . _ physical review letters _ * 1999 * , _ 82 _ , 29002903 zhang , h. ; srolovitz , d. j. ; douglas , j. f. ; warren , j. a. grain boundaries exhibit the dynamics of glass - forming liquids . _ proceedings of the national academy of sciences _ * 2009 * , _ 106 _ , 77357740 velzquez - salazar , j. j. ; esparza , r. ; meja - rosales , s. j. ; estrada - salas , r. ; ponce , a. ; deepak , f. l. ; castro - guerrero , c. ; jos - yacamn , m. experimental evidence of icosahedral and decahedral packing in one - dimensional nanostructures . _ acs nano _ * 2011 * , _ 5 _ , 62726278 jiang , d .- e . ; nobusada , k. ; luo , w. ; whetten , r. l. thiolated gold nanowires : metallic versus semiconducting . _ acs nano _ * 2009 * , _ 3 _ , 23512357 , pmid : 19603760 iacovella , c. r. ; keys , a. s. ; horsch , m. a. ; glotzer , s. c. icosahedral packing of polymer - tethered nanospheres and stabilization of the gyroid phase . _ physical review e _ * 2007 * , _ 75 _ , 040801 keys , a. s. ; iacovella , c. r. ; glotzer , s. c. characterizing structure through shape matching and applications to self - assembly . _ annual review of condensed matter physics _ * 2011 * , _ 2 _ , 263285 keys , a. s. ; iacovella , c. r. ; glotzer , s. c. characterizing complex particle morphologies through shape matching : descriptors , applications , and algorithms . _ journal of computational physics _ * 2011 * , _ 230 _ , 6438 6463 hkkinen , h. ; barnett , r. n. ; scherbakov , a. g. ; landman , u. nanowire gold chains : formation mechanisms and conductance . _ the journal of physical chemistry b _ * 2000 * , _ 104 _ , 90639066 vlez , p. ; dassie , s. a. ; leiva , e. p. m. when do nanowires break ? a model for the theoretical study of the long - term stability of monoatomic nanowires . _ chemical physics letters _ * 2008 * , _ 460 _ , 261265 tavazza , f. ; levine , l. e. ; chaka , a. m. structural changes during the formation of gold single - atom chains : stability criteria and electronic structure . b _ * 2010 * , _ 81 _ , 112 wu , h. molecular dynamics study of the mechanics of metal nanowires at finite temperature . _ european journal of mechanics - a / solids _ * 2006 * , 370377 zhao , j. ; murakoshi , k. ; yin , x. ; kiguchi , m. ; , y. g. dynamic characterization of the postbreaking behavior of a nanowire . _ the journal of physical chemistry c _ * 2008 * , _ 112 _ , 2008820094 xiao , l. ; wang , l. from planar to three - dimensional structural transition in gold clusters and the spin - orbit coupling effect . _ chemical physics letters _ * 2004 * , _ 392 _ , 452455 keith , j. a. ; fantauzzi , d. ; jacob , t. ; van duin , a. c. t. reactive forcefield for simulating gold surfaces and nanoparticles . _ b _ * 2010 * , _ 81 _ , 235404 cleri , f. ; rosato , v. tight - binding potentials for transition metals and alloys . _ physical review b _ * 1993 * , _ 48 _ , 2233 jrvi , t. t. ; kuronen , a. ; hakala , m. ; nordlund , k. ; van duin , a. c. ; goddard , w. a. ; jacob , t. development of a reaxff description for gold . _ the european physical journal b - condensed matter and complex systems _ * 2008 * , _ 66 _ , 7579 , 10.1140/epjb / e2008 - 00378 - 3 gall , k. ; diao , j. ; dunn , m. l. ; haftel , m. ; bernstein , n. ; mehl , m. j. tetragonal phase transformation in gold nanowires . _ journal of engineering materials and technology _ * 2005 * , _ 127 _ , 417422 van duin , a. c. t. ; dasgupta , s. ; lorant , f. ; goddard , w. a. reaxff : a reactive force field for hydrocarbons . _ the journal of physical chemistry a _ * 2001 * , _ 105 _ , 93969409 plimpton , s. j. fast parallel algorithms for short - range molecular dynamic . _ journal of computational physics _ * 1995 * , _ 117 _ , 119 frank , f. ; kasper , j. complex alloy structures regarded as sphere packings . i. definitions and basic principles . _ acta crystallographica _ * 1958 * , _ 11 _ , 184190 iacovella , c. r. ; horsch , m. a. ; glotzer , s. c. local ordering of polymer - tethered nanospheres and nanorods and the stabilization of the double gyroid phase . _ the journal of chemical physics _ * 2008 * , _ 129 _ , 044902 steinhardt , p. j. ; nelson , d. r. ; ronchetti , m. bond - orientational order in liquids and glasses . _ physical review b _ * 1983 * , _ 28 _ , 784805 wang , b. ; yin , s. ; wang , g. ; buldum , a. ; zhao , j. novel structures and properties of gold nanowires . _ physical review letters _ * 2001 * , _ 86 _ , 20462049 driscoll , j. a. ; varga , k. convergence in quantum transport calculations : localized atomic orbitals versus nonlocalized basis sets . _ phys . b _ * 2010 * , _ 81 _ , 115412 costa - krmer , j. l. conductance quantization at room temperature in magnetic and nonmagnetic metallic nanowires . b _ * 1997 * , _ 55 _ , r4875r4878 li , x .- b . ; wang , h .- y . ; yang , x .- d . ; zhu , z .- h . ; tang , y .- j . size dependence of the structures and energetic and electronic properties of gold clusters . _ journal of chemical physics _ * 2007 * , _ 126 _ , 084505 gruber , m. ; heimel , g. ; romaner , l. ; brdas , j .- l . ; zojer , e. first - principles study of the geometric and electronic structure of @xmath22 clusters : importance of the prism motif . _ phys . rev . b _ * 2008 * , _ 77 _ , 165411 pundlik , s. s. ; kalyanaraman , k. ; waghmare , u. v. first - principles investigation of the atomic and electronic structure and magnetic moments in gold nanoclusters . _ the journal of physical chemistry c _ * 2011 * , _ 115 _ , 38093820 menard , l. d. ; xu , h. ; gao , s .- p . ; twesten , r. d. ; harper , a. s. ; song , y. ; wang , g. ; douglas , a. d. ; yang , j. c. ; frenkel , a. i. et al . metal core bonding motifs of monodisperse icosahedral au13 and larger au monolayer - protected clusters as revealed by x - ray absorption spectroscopy and transmission electron microscopy . _ the journal of physical chemistry b _ * 2006 * , _ 110 _ , 1456414573 walter , m. ; akola , j. ; lopez - acevedo , o. a unified view of ligand - protected gold clusters as superatom complexes . _ proceedings of the national academy of sciences _ * 2008 * , _ 105 _ , 91579162 pei , y. ; gao , y. ; zeng , x. c. structural prediction of thiolate - protected au38 : a face - fused bi - icosahedral au core . _ journal of the american chemical society _ * 2008 * , _ 130 _ , 78307832 shichibu , y. ; negishi , y. ; watanabe , t. ; chaki , n. k. ; kawaguchi , h. ; tsukuda , t. biicosahedral gold clusters [ au25(pph3)10(scnh2n+1)5cl2]2 + ( n = 218 ) : a stepping stone to cluster - assembled materials . _ the journal of physical chemistry c _ * 2007 * , _ 111 _ , 78457847 agrait , n. ; rubio , g. ; vieira , s. plastic deformation of nanometer - scale gold connective necks . _ physical review letters _ * 1995 * , _ 74 _ , 39953998 zheng , c. ; hoffmann , r. ; nelson , d. r. a helical face - sharing tetrahedron chain from irrational twist , stella quadrangula , and related matters . _ j am chem soc _ * 1990 * , _ 112 _ , 37843791 lidin , s. ; andersson , s. regular polyhedra helices . _ zeitschrift fur anorganische und allgemeine chemie _ * 1996 * , _ 622 _ , 164166 vetter , j. ; glassbrook , r. ; dongarra , j. ; schwan , k. ; loftis , b. ; mcnally , s. ; meredith , j. ; rogers , j. ; roth , p. ; spafford , k. et al . keeneland : bringing heterogeneous gpu computing to the computational science community . _ ieee computing in science and engineering _ * 2011 * , _ 13 _ , 9095 perdew , j. p. ; burke , k. ; ernzerhof , m. generalized gradient approximation made simple . lett . _ * 1996 * , _ 77 _ , 3865 kresse , g. ; furthmuller , j. efficient iterative schemes for ab initio total - energy calculations using a plane - wave basis set . b _ * 1996 * , _ 54 _ , 11169 kresse , g. ; joubert , d. from ultrasoft pseudopotentials to the projector augmented - wave method . _ b _ * 1999 * , _ 59 _ , 1758	we report comprehensive high - accuracy molecular dynamics simulations using the reaxff forcefield to explore the structural changes that occur as au nanowires are elongated , establishing trends as a function of both temperature and nanowire diameter . our simulations and subsequent quantitative structural analysis reveal that polytetrahedral structures ( e.g. icosahedra ) form within the `` amorphous '' neck regions , most prominently for systems with small diameter at high temperature . we demonstrate that the formation of polytetrahedra diminishes the conductance quantization as compared to systems without this structural motif . we demonstrate that use of the reaxff forcefield , fitted to high - accuracy first principles calculations of au , combines the accuracy of quantum calculations with the speed of semi - empirical methods . the observation of conductance quantization in atomic - scale junctions @xcite has fueled much interest in understanding their formation in mechanically deformed metallic nanowires ( nws ) @xcite , in large part due to potential applications in nano- and molecular - electronic devices . to this extent , quantifying the structural changes that occur as a nw elongates is important for understanding how and why parameters such as initial crystal structure @xcite , temperature @xcite , and rate of elongation @xcite affect the material properties ( e.g. , conductance @xcite , mechanical stability @xcite , or the opening of a bandgap @xcite ) . many studies have been performed that focus on identifying changes to the crystalline structure , such as slipping and reorientation of the crystalline lattice @xcite , and under what conditions single - atom wide monatomic chains are likely to form @xcite . however , to date , there has only been limited emphasis on quantifying the formation and structure of the `` amorphous '' or , more generally , non - crystalline domains that often appear within nws @xcite . the formation of these non - crystalline structures is likely to result in significant changes to the properties of the nws . for example , amorphous structures separating otherwise crystalline nw domains may exhibit unique dynamical properties akin to glass - forming liquids , similar to atoms at grain boundaries in bulk metals @xcite . moreover , non - crystalline structures may still demonstrate distinct local ordering , such as the distorted icosahedral structures predicted to form in the necks of na nws @xcite and the recent synthesis of bi - metallic nws with local icosahedral ordering @xcite . this is of particular consequence as thiolated icosahedral au nws are predicted to behave as semiconductors under certain charge states @xcite and non - crystalline nw neck structures have recently been demonstrated to produce gradual changes in the conductance , as opposed the more typical quantization , i.e. , `` steps '' and `` plateaus''@xcite . in this paper , we focus on identifying the amorphous / non - crystalline structures that form within simulated au nws undergoing tensile elongation . simulation is well suited to study this problem as it provides the full spatial and temporal coordinates of the atoms in the nw , allowing the structure to be quantified by the use of order parameters , such as those recently adapted from the field of shape matching @xcite . much of the previous simulation work dealing with the elongation of au nws has relied on either quantum mechanical @xcite or , more commonly , semi - empirical methods @xcite , requiring a compromise between the accuracy of the interatomic interactions and the accessible system size and timescale . quantum mechanical methods , such as density functional theory ( dft ) , are widely considered to provide a highly accurate description of the interactions between au atoms , in particular , capturing the stability of 2-d planar geometries for low coordination clusters @xcite . however , the high computational cost of quantum mechanical methods often limits both the total number of independent statepoints that can be efficiently considered and the total size of the nw ( typically @xmath0100 atoms or less @xcite ) , both of which may inadvertently bias the results . semi - empirical methods , such as the second - moment approximation of the tight - binding ( tb - sma ) scheme @xcite , are many orders of magnitude faster than quantum mechanical methods , allowing for the efficient exploration of large numbers of statepoints and systems sizes more representative of experiment @xcite . while methods such as tb - sma have been shown to qualitatively match the behavior of experiment in side - by - side studies @xcite , semi - emperical methods are often fitted to bulk systems , and thus may produce results that are quantitatively inconsistent with theory or experiment@xcite , especially for low coordination atomic - scale contacts @xcite . in particular , the minimal energy 2-d planar structures of au are not typically stable for semi - empirical methods @xcite which may incorrectly bias nw simulations towards forming 3-d isomers . in order to bridge the gap between semi - empirical and quantum mechanical methods , we use the reaxff reactive forcefield @xcite which is designed to provide accuracy approaching quantum mechanical methods , but at a substantially reduced computational cost . reaxff forcefields are derived by fitting analytical functions to dft bonding curves@xcite . the parameters derived for au by keith , _ _ et al.__@xcite have been demonstrated to satisfactorily capture the expected behavior of both bulk and low coordination states @xcite . in particular , this parameterization has been shown to closely reproduce the energy and stability of low and minimal energy clusters of au@xcite . in this work , we show that reaxff is a substantial improvement over tb - sma for the simulation of nw structures by comparing the absolute and relative energy scaling to dft calculations . we perform comprehensive molecular dynamics ( md ) simulations@xcite with reaxff to investigate the behavior and properties of [ 100 ] orientated fcc au nws undergoing tensile elongation in vacuum , which we also compare to simulations performed using tb - sma . we report , for the first time , the formation of well - ordered polytetrahedral local structures ( e.g. , full and partially coordinated icosahedra @xcite ) within the non - crystalline domains of the nws . we quantify the formation of these structures by using the r@xmath1 shape matching method @xcite based on spherical harmonics@xcite , which we use to construct trends as a function of temperature and nw diameter . we further calculate the zero - bias conductance of our simulation trajectories to demonstrate the impact that the formation of polytetrahedra has on the conductance behavior .
the importance of the extragalactic background light ( ebl ) for cosmology has long been recognized ( e.g. ( * ? ? ? * partridge & peebles 1967 ) ) . the ebl at uv , optical and near infrared wavelengths consists of the integrated light of all unresolved galaxies along the line of sight plus any contributions by intergalactic gas and dust and by hypothetical decaying relic particles . a large fraction of the energy released in the universe since the recombination epoch is expected to be contained in the ebl . an important aspect is the balance between the uv - nir and the far infrared ebl : what is lost by dust obscuration in the uv - nir will re - appear in the fir . some central , but still largely open astrophysical problems to be addressed through ebl measurements include the formation and early evolution of galaxies and the star formation history of the universe . because of the foreground components , much larger than the ebl itself , the measurement of the ebl has turned out to be a tedious problem . as a consequence we lack a generally accepted measured value of the optical ebl . for a review see leinert et al . ( 1998 ) . recently , ( * ? ? ? * bernstein et al . ( 2002 ) ) have announced `` the first detection '' of the ebl at 300 , 550 , and 800 nm . they used a combination of space borne ( hst ) and ground based ( las campanas ) measurements in their method . however , ( * ? ? ? * mattila ( 2003 ) ) has argued that they neglected important effects of the atmospheric scattered light and had a problem with the inter - calibration of the two telescopes used . therefore , the claim for a detection of the ebl appeared premature . after reanalysis of their systematic errors ( * ? ? ? * bernstein et . ( 2005 ) ) and ( * ? ? ? * bernstein ( 2007 ) ) came to the same conclusions which they formulated as follows : `` ... the complexity of the corrections required to do absolute surface ( spectro)photometry from the ground make it impossible to achieve 1% accuracy in the calibration of the zl . '' and `` ... the only promising strategy ... is to perform all measurements with the same instrument , so that the majority of corrections and foreground subtractions can be done in a relative sense , before the absolute calibration is applied . '' in this situation it is highly desirable to obtain a measurement of the ebl with an independent method which utilises one and the same instrument for all sky components and is virtually free of the large foreground components , the airglow ( agl ) , zodiacal light ( zl ) and tropospheric scattered light . we have been developing over several years a method for the measurement of the ebl which utilises the screening effect of a dark nebula on the background light ( see mattila 1990 for a review and previous results ) . a differential measurement of the night - sky brightness in the direction of a high galactic latitude dark nebula and its surrounding area , which is ( almost ) free of obscuring dust , provides a signal that is due to two components only : ( 1 ) the ebl and ( 2 ) the diffusely scattered starlight from interstellar dust in the cloud ( and to smaller extent also in its surroundings ) . all the large foreground components , i.e. the zl , the agl , and the tropospheric scattered light , are completely eliminated ( see fig . the direct starlight down to @xmath321 - 23 mag can be eliminated by selecting the measuring areas with the help of deep images . at high galactic latitudes ( @xmath430 deg ) , the star density is sufficiently low to allow blank areas of sufficient size to be easily found . models of the galaxy show that the contribution from unresolved stars beyond this limiting magnitude is of minor or no importance ( mattila 1976 ) . if the scattered light from the interstellar dust were zero ( i.e. if the grain albedo @xmath5 ) , then the difference in surface brightness between a transparent comparison area and the dark nebula would be due to the ebl only , and an opaque nebula would be darker by the amount of the ebl intensity ( dashed line in fig . the scattered light is not zero , however . a dark nebula in the interstellar space is always exposed to the radiation field of the integrated galactic starlight , which gives rise to a diffuse scattered light ( shaded area in fig . 1c ) . because the intensity of this scattered starlight in the dark nebula is equal to or larger than the ebl , its separation is the main task in our method . the key issue in the dark cloud method is to get a reliable estimate for the scattered light . this can be achieved by means of spectroscopy . with intermediate - resolution spectroscopy ( @xmath6 ) it is possible to determine the strengths of the stronger fraunhofer lines in the spectrum of the excess surface brightness of the dark cloud . spectroscopic measurement of a faint surface brightness signal , only 1 10% above the dark sky level has been possible with the vlt / fors long slit spectroscopy . _ target cloud and positions . _ the high galactic latitude dark nebula lynds 1642 ( @xmath7 ) has been chosen as our primary target . it has a high obscuration ( @xmath8 mag ) in the centre and areas of good transparency ( @xmath9 mag ) in its immediate surroundings within @xmath10 . its declination of @xmath11 allows observations at low airmasses from paranal . our measuring positions ( @xmath12 ) for intermediate band photometry ( mattila and schnur 1990 ) are shown in fig . 2 as circles and the positions where photometry and vlt / fors long slit spectra ( 2x6.8 ) were taken , as squares . differential measurements with repeated on / off switching cycle of @xmath3 0.5 h were used to eliminate airglow variations . the spectral range was @xmath13 nm and the resolution @xmath14 nm . stamps of 10x10 size centered on the vlt / fors positions , adopted from dss blue plates , are shown in the margins : the dark central position # 8 ( @xmath8 mag ) in upper left , followed clockwise by the two intermediate opacity positions # 9 and 42 , and then the transparent off positions . two observed spectra `` dark nebula minus surrounding sky '' are shown in fig . 3 : one is for the opaque central position # 8 with @xmath8 mag , the other for the average of two intermediate opacity ( 0,0 ) ( 0,0 ) ( 0,0 ) ( 30,-60)(0,0 ) ebl measurement with ( 30,-63)(0,0 ) the dark cloud shadow method positions , # 9 and # 42 with @xmath15 1 mag . the crosses show the results of previous intermediate band photometry by mattila and schnur ( mattila 1990 ) with error bars and filter band widths indicated . the spectra have been re - scaled by a multiplicative factor to adjust them to the more reliably calibrated photometry . the intermediate opacity positions are used to validate the model of galactic starlight spectrum , needed to accurately separate the galactic contribution from the observed spectrum `` dark position minus clear sky '' . in combination with the dark central position they enable us to disentangle the effects of extinction and reddening by dust from the spectral shape of the galactic starlight . our separation method utilises the difference in the spectra of the ebl and the integrated galactic starlight ( see fig 1d ) . while the scattered galactic starlight spectrum has the characteristic stellar fraunhofer lines and the discontinuity at 400 nm the ebl spectrum is a smooth one without these features this can be understood because the radiation from galaxies and other luminous matter over a vast redshift range contributes to the ebl , thus washing out any spectral lines or discontinuities . the spectrum of the integrated galactic starlight can be synthesised by using the known spectra of representative stars of different spectral classes , as well as the ( 0,0 ) ( 0,0 ) ( 0,0 ) ( 50,-120)(0,0 ) the observed positions in the l 1642 cloud area . ( 50,-123)(0,0 ) superimposed on an iras 100 @xmath16 m map . for details see text . ( 0,0 ) ( 0,0 ) ( 0,0 ) ( 60,-196)(0,0 ) observed spectrum `` dark nebula minus surrounding sky '' for the central ( 60,-199)(0,0 ) opaque position # 8 and the mean of two translucent positions , # 9 and # 42 . ( 60,-199)(0,0 ) distribution of stars and dust in the solar neighbourhood . such synthetic spectra have been calculated using the model of mattila ( 1980 ) and the recent stelib library of stellar spectra ( le borgne et al . the resulting isl spectrum , mean over sky , is shown in fig . 4 ( the uppermost blue line ) . this spectrum , smoothed to our fors resolution of 1.1 nm , scaled and reddened to correspond to the dark cloud scattered light spectrum , shows several strong fraunhofer lines and the discontinuity at 400 nm . for the separation of the scattered light we can not , however , rely on the integrated starlight _ model _ alone . we have to test and validate this model by observations at intermediate opacity positions of the dark cloud where most of the ebl is transmitted through the cloud and contributes only little to the surface brightness difference `` dark cloud minus surroundings '' . the observed surface brightness difference , @xmath17 , `` dark nebula minus surrounding sky '' consists of the following main components ( followed by * + * or * - * depending on whether it makes a positive or negative contribution to @xmath17 ) : * ( 1 ) * starlight scattered by dust in the nebula * ( + ) * ; * ( 2 ) * starlight scattered by dust in the comparison fields of the surrounding sky * ( - ) * ; * ( 3 ) * diffuse line emission by ionised gas in the comparison fields beyond the distance of the dark cloud * ( - ) * ; * ( 4 ) * the extragalactic background light ( present in the comparison fields ) * ( -)*. in order to derive an estimate for the ebl it is necessary to separate the contribution of scattered light ( components 1 and 2 ) . when applying the dark cloud method to intermediate band photometry , we utilised the 400 nm discontinuity which is strong and well defined in the integrated starlight spectrum and is present also in the scattered light spectra as shown in fig . 3 however , its value as determined using intermediate bands is strongly influenced by the wavelength - dependent extinction and multiple scattering in the cloud which need to be modeled . we use our model isl spectra and the observed vlt / fors spectrum for the opaque central position # 8 to separate the scattered light component . the scattered light spectrum is assumed to be a copy of the isl spectrum but to be reddened by a factor linearly proportional to the wavelength . we show in fig . 4 an overall model fit for 360 - 550 nm assuming that @xmath18 cgs at off and = 0 at the # 8 on position . the `` on - position '' model spectrum ( blue line ) is the mean isl spectrum over the whole sky , suitably scaled ( by factor 0.18 at 400 nm ) and linearly reddened ( by 0.075 per 100 nm ) . the `` off - position '' model spectrum ( green line ) is the isl spectrum for @xmath19 deg , scaled according to the waveleghth dependent extinction ( determined from 2mass jhk stellar and isophot 200 @xmath16 m surface photometry ) . the on minus off model spectrum is shown as red line . the observed on minus off spectrum for pos # 8 ( as in fig . 3 ) is shown as black line . as can be seen the fit is almost perfect except for the balmer lines . these lines are contaminated by the excess emission of diffuse ionised gas in the off positions . at this stage we do not attempt to model this component ( shown as magenta line in fig . 4 on arbirtrary scale ) and exclude the balmer lines from our fitting procedure . although a good fit of the observed spectrum by scattered isl only can be achieved there are small differences in the depths of the fraunhofer lines ( other than the balmer lines ) indicating that @xmath20 . we have made fits over suitably selected narrow wavelength intervals . the ebl is assumed to be constant over the wavelength ( 50,-185)(0,0 ) fitting of the 390 - 407 nm spectral range ( 50,-188)(0,0 ) of the opaque position 8 spectrum with the integrated ( 50,-191)(0,0 ) starlight spectrum and different assumed values ( 50,-194)(0,0 ) of the ebl slot in question . the isl scaling factor and wavelegth dependent gradient are free parameters . the fit to the range 390 - 407 nm is shown in fig . the caii h line at 396.9 nm was not included because it is contaminated by the h@xmath21 line of ionised gas . our choice for another suitable slot with strong fraunhofer lines was 510 - 535 nm . by fitting the observed spectrum with an isl spectrum and ebl contributions of different amounts we obtain the following ebl values for these two slots : + @xmath22(400 nm ) = @xmath23 cgs @xmath24 or @xmath25 nw m@xmath26 sterad@xmath1 + @xmath22(520 nm ) @xmath27 cgs @xmath28 or @xmath29 nw m@xmath26 sterad@xmath1 the statistical error ( @xmath30 ) is small enough for putting to the ebl significant limits comparable with the ebl minimum value of @xmath31 cgs for @xmath32 nm derived by summing up deep galaxy counts ( madau and pozzetti 2000 ) . _ systematic errors . _ ( 1 ) blocking by the dark cloud is not complete for incident isotropic diffuse radiation like the ebl . for @xmath33 mag the blocking factor is 0.86 ( see mattila 1976 , table a2 ) . ( 2 ) the systematic error caused by our isl model is tested by observing the semi - transparent positions with @xmath34 mag for which the scattered starlight has the same spectrum but the ebl contribution is only @xmath35 of that for @xmath36 mag . they give @xmath37 0 ( within 1@xmath38 statistical errors ) and may thus indicate a need to increase the ebl upper boundary error bar by @xmath39 or @xmath3 0.5 cgs . with a 14% blocking factor correction and increasing the upper bound of the errors ( systematic ) by 0.5 cgs we end up with the following preliminary ebl estimates : + @xmath22(400 nm ) = @xmath40 cgs @xmath24 or @xmath41 nw m@xmath26 sterad@xmath1 + @xmath22(520 nm ) @xmath42 cgs @xmath28 or @xmath43 nw m@xmath26 sterad@xmath1 we show in fig . 6 a compilation from uv to nir ( 0.1 - 5 @xmath16 m ) of recent ebl measurements and upper limits ( colour symbols ) as well as lower limits from galaxy counts ( black symbols ) . upper limits from the tev gamma - ray absorption method are shown as dashed lines . the resulting values from the present paper are shown as red solid square with error bars at 400 nm and upper limit at 520 nm . the reanalysis by mattila ( 2003 ) of the bernstein et al . ( 2002 ) ebl values are shown as large red rectangles , indicating upper limits . in agreement with these , the reanalysis by bernstein ( 2007 ) resulted in values that were substantially increased from their original estimates . they are shown as the solid magenta squares with error bars . for references to the three upper limits at 400 - 520 nm see leinert et al . ( 1998 ) , and for the other points see mattila ( 2006 ) and dominguez et al . ( 2011 ) . for a new ebl estimate ( not shown in fig . 6 ) based on reanalysis of pioneer10/11 data see matsuoka et . al . ( 2011 ) and these proceedings . , _ apj _ , 666 , 663 , _ apj _ , 571 , 56 and _ apj _ , 571 , 85 , _ apj _ , 632 , 713 , _ mnras _ , 410 , 2556 , _ a&a _ , 402 , 433 , , 127 , 1 , _ mnras _ , 312 , l9 , _ apj _ , 736 , 119 , _ a&a _ , 47 , 77 , _ a&a _ , 82 , 373 and _ a&as _ , 39 , 53 , _ iaus _ , 139 , 257 , _ apj _ , 591 , 119 , _ mnras _ , 372 , 1253 , _ apj _ , 148 , 377	the extragalactic background light ( ebl ) at uv , optical and nir wavelengths consists of the integrated light of all unresolved galaxies along the line of sight plus any contributions by intergalactic matter including hypothetical decaying relic particles . the measurement of the ebl has turned out to be a tedious problem . this is because of the foreground components of the night sky brightness , much larger than the ebl itself : the zodiacal light ( zl ) , integrated starlight ( isl ) , diffuse galactic light ( dgl ) and , for ground - based observations , the airglow ( agl ) and the tropospheric scattered light . we have been developing a method for the ebl measurement which utilises the screening effect of a dark nebula on the ebl . a differential measurement in the direction of a high - latitude dark nebula and its surrounding area provides a signal that is due to two components only , i.e. the ebl and the diffusely scattered isl from the cloud . we present a progress report of this method where we are now utilising intermediate resolution spectroscopy with eso s vlt telescope . we detect and remove the scattered isl component by using its characteristic fraunhofer line spectral signature . in contrast to the isl , in the ebl spectrum all spectral lines are washed out . we present a high quality spectrum representing the difference between an opaque position within our target cloud and several clear off positions around the cloud . we derive a preliminary ebl value at 400 nm and an upper limit to the ebl at 520 nm . these values are in the same range as the ebl lower limits derived from galaxy counts . * unit : * we will use in this paper the abbreviation 1 cgs = @xmath0ergs@xmath1@xmath2sr@xmath1@xmath1
post - quantum theories have been proposed which provide explicit wave function collapse models to explain how the classical world emerges from the quantum domain , see e.g. @xcite . although the physics behind each collapse mechanism differs , they all operate as a spatial localization preventing massive objects to be in a quantum superposition of two or more positions . a possible approach to test them is to manipulate the motion of a mechanical oscillator through the radiation pressure . in this framework , it has been recently proposed to test collapse models by simply looking at the spectrum of the light driving the oscillator @xcite or through a spontaneous increase of the equilibrium temperature @xcite . alternatively , we can look for a method to push the mechanical oscillator down to the quantum regime where its spatial position is largely delocalized and a technique to record the decay of spatial quantum coherences . deviation from standard decoherence that occurs through the interaction with the environment @xcite might make it possible to confirm or rule out predictions from these hypothetic wave function collapse models . the proposals of refs @xcite have followed this approach . they consist of first preparing quantum light , entangling it with the mechanical oscillator position and subsequently observing the oscillator decoherence through the decay of quantum properties of light . while recent proposals have shown how to relax some of the constraints on the opto - mechanical coupling strength , they still need non - classical light to start with @xcite . in the resolved - sideband regime , techniques benefiting from an optomechanical squeezing interaction @xcite or based on conditioning @xcite have been put forward to create quantum optomechanical states while avoiding the initial preparation of non - classical light . + in this work , we use laser light to drive a mechanical oscillator in the pulsed regime where the light pulse duration is much shorter than the mechanical period . in this regime , the mechanical oscillator can be cooled and manipulated without sideband resolution @xcite as shown in an recent experiment @xcite . the basic principle for manipulation is that the kick imparted by the light is proportional to the photon number . in particular , we show that when the oscillator is driven by an intense laser pulse where the photon number is inherently and largely undefined , this results in an opto - mechanical entangled state in which the oscillator is delocalized over a large range of positions . we build up an entanglement witness that can be used to reveal the quantum nature of this opto - mechanical state . we also show how to disentangle the light and the oscillator while using the light to record the decay of the oscillator spatial coherences and ultimately , to probe hypothetic deviations from standard decoherence . we discuss the experimental feasibility of this test bench for wave function collapse models by studying the effects of various measurement inaccuracies and finite cooling efficiencies . consider the optical and mechanical modes of an optomechanical cavity with bosonic operators @xmath0 and @xmath1 respectively . the corresponding hamiltonian is given by @xmath2 where @xmath3 is the mechanical frequency and @xmath4 is the optomechanical coupling , l being the cavity length , m the effective mass of the oscillator and @xmath5 the cavity frequency . further consider the ideal case where the mechanical mode is initially prepared in its motional ground state @xmath6 when a @xmath7 photon fock state @xmath8 impinges upon the oscillator , they induce a displacement of the mechanical state whose amplitude is time dependent @xmath9 @xmath8 @xcite . the result of the interaction with a laser pulse @xmath10 ( where we assume that @xmath11 all along the paper ) , i.e. with a poissonian distribution of number states can be obtained directly from this expression . in particular , in the pulsed regime where the interaction time @xmath12 satisfies @xmath13 ( cf below for the exact conditions ) the optomechanical state reduces to @xmath14 a time @xmath15 after the interaction . the mechanical state involves coherent states @xmath16 with amplitudes that depend on the photon number . right after the interaction @xmath17 the photon number is entangled with the mechanical momentum . then the mechanical oscillator rotates in phase space . after a quarter of a mechanical period @xmath18 the photon number is entangled with the mechanical position before being entangled again with the mechanical momentum at @xmath19 and so on . before we show how to reveal this optomechanical entanglement , let us note that the mechanical position is delocalized over the range @xmath20 on average which can be made much larger than the mechanical zero - point fluctuation @xmath21 with an intense driving laser , even in the experimentally relevant regime where @xmath22 @xmath23 as the characteristic timescale of the wave - function collapse models decreases with @xmath24 some of them are expected to degrade the quantum properties of the state ( [ optomechanical_state ] ) on timescales that can be accessed experimentally even when dealing with small effective masses , as we show below . the question that we address in this paragraph is how to detect the quantum nature of the optomechanical state ( [ optomechanical_state ] ) . the entanglement witness that we propose is inspired by ref . @xcite . concretely , for two pairs of local observables @xmath25 and @xmath26 and a separable state @xmath27 , the following inequality holds @xmath28 \rangle \| + \| \langle [ b_1 , b_2 ] \rangle \|\right ) $ ] where @xmath29 stands for the variance . the idea is that the observables satisfy the heisenberg uncertainty relation locally whereas the pairs of observables @xmath30 and @xmath31 can only be classically correlated for a separable state . the aim is thus to find two couples of observables @xmath32 and @xmath33 such that the variances @xmath34 do not increase ( decrease ) during the evolution while the commutators @xmath35 \rangle\|$ ] and @xmath36 \rangle$ ] increase ( stay constant ) . in our case , the optomechanical interaction shifts the oscillator momentum by the photon number ( times the interaction strength ) suggesting @xmath37 and @xmath38 similarly , the light field acquires a phase proportional to the position of the oscillator . while the phase @xmath39 is not a physical observable , for a coherent state it can be indirectly accessed through homodyne detections @xmath40 therefore , we choose @xmath41 and @xmath42 this leads to the following inequality @xmath43 if the results of measurements do not fulfill this inequality , we can conclude that the light field and the mechanical oscillator are entangled . in particular , this inequality does not hold for the state ( [ optomechanical_state ] ) if @xmath44 this condition is obtained under the assumption @xmath45 it can be understood as a requirement on the laser intensity to significantly enlarge the mechanical zero point spread a condition that is necessary to correlate the photon number and the mechanical momentum . + interacts with the mechanical device , the photon number is entangled with the mechanical momentum . to detect this optomechanical entangled state , we need to look at the correlations between i ) the photon number and the mechanical momentum , ii ) the quadrature of the light and the mechanical position . the latter can be performed by using the fact that the phase shift in the reflected light is proportional to the position of the mirror . to record the mechanical decoherence , the mechanical oscillator is first disentangled from the light by measuring the mechanical position at a time which is a multiple of half the mechanical period ( see fig . [ fig1 ] ) . a feedback loop is then needed to correct the phase of the light depending on the result of the measurement of the mechanical position . the oscillator decoherence translates into a phase noise on the light that can be observed with an homodyne detection . deviation from environmental decoherence might make it possible to confirm or rule out predictions form explicit wave function collapse models.,scaledwidth=60.0% ] we first explain how to proceed with recording the decoherence of the mechanical oscillator : * a short pulse of duration @xmath12 generates the entangled optomechanical state ( [ optomechanical_state ] ) . * after a variable delay @xmath46 , another short pulse is used to measure the mechanical position , where no information is revealed about the photon number . * the phase of the light pulse in ( a ) is shifted using the measurement outcome of step ( b ) . * the phase quadrature of the outgoing light is measured . in the ideal case without decoherence , the conditional state of the light @xmath47 has a phase which depends on the result @xmath48 of the measurement of the oscillator position . correcting the phase of the light with a feedback loop yields a photonic state that is identical to the initial light state @xmath49 . things are different if we take the oscillator decoherence into account . these decoherence models , conventional or otherwise , operate as a decay of spatial coherences @xmath50 where @xmath51 and @xmath52 are position eigenstates . the exact expression of @xmath53 will differ according to the decoherence model being described @xcite , but all operate as a spatial localisation as a function of the distance @xmath54 , resulting in a phase noise on the light once the position of the mechanical device is measured @xcite . instead of @xmath55 , in the presence of decoherence , one obtains ( see @xcite , supplemental material ) @xmath56 where @xmath57 with @xmath58 indicating an average over half a period of the mechanical oscillator . the problem of identifying the oscillator decoherence is thus equivalent to characterizing a phase noise channel with light this characterization is more accurate when probed with large photon numbers . for example , by measuring the quadrature @xmath59 of the light field , we get @xmath60 and similarly @xmath61 where @xmath62 is defined by @xmath63 i.e. @xmath64 immediately after the optomechanical interaction ( k=0 ) , the mechanical oscillator has not yet undergone any decoherence and @xmath65 @xmath66 the phase of the light is well defined and the variance of @xmath55 is @xmath67 in the limit where @xmath68 the spatial coherences of the mechanical oscillator vanish @xmath69 @xmath70 . the light field is described by a mixture of coherent states with random phases and its spread in the phase space tends to @xmath71 focusing on @xmath72 the time taken to double the distribution of possible results , for example , is given by @xmath73 at the leading order in @xmath74 . we see that the use of intense laser pulses reduces the time it takes to observe the effect of collapse models . as particular examples , we compare conventional decoherence with explicit collapse models as proposed by ellis and co - workers @xcite on the one hand and diosi & penrose @xcite on the other hand . we show that with a combination of large @xmath74 and small thermal dissipation , we can find experimentally feasible parameters in which the standard decoherence time is longer than the timescale of these collapse models , hence opening the way to confirm or rule out their predictions . note that for standard decoherence processes where @xmath75 @xmath76 . in this case @xmath77 for the diosi & penrose model , @xmath78 decreases less rapidly with @xmath79 such that the ratio @xmath80 can be much larger than one . as this ratio can be accessed through the two first moments of @xmath59 - , this provides an unambiguous criteria to distinguish standard decoherence with the dioisi & penrose model . this is particularly relevant in the regime where @xmath81 is comparable to the parameter that is used to define the mass distribution in the diosi & penrose model . we now address the question of the feasibility in detail . first , our results have been derived under the assumption that the dynamics falls within the pulsed regime . this requires @xmath82 which provides an upper bound on the photon number given the optomechanical coupling , the pulse duration and the mechanical frequency placing an upper limit on the extent to which we can violate ( [ ent_witness ] ) . let us now consider various imperfections . first , let the measurement of the oscillator position be subject to a gaussian noise with a spread @xmath83 to reveal entanglement , one needs @xmath84 in the limit @xmath85 i.e. to resolve the zero point spread . the requirement is more demanding for recording decoherence . if the mechanical position is not precisely measured , the phase of the light can not be accurately corrected leading to phase noise . this effect has to be smaller than the phase noise induced by decoherence . since we are interested in the regime where decoherence doubles the variance of @xmath86 we need @xmath87 note that the mechanical position can be measured by homodyning a light pulse that has been sent into the mechanics , since the phase of the reflected light pulse depends on the mechanical position . it has been shown in ref . @xcite that for an input drive with duration @xmath88 the achievable precision @xmath89 depends on the photon number @xmath90 through the formula @xmath91 the limitation for @xmath90 and hence on the precision is given by the power @xmath92 that can be used before heating significantly the surrounding bath and ultimately by the power that can be homodyned before photo - detection saturates . similarly to the requirement on the precision of the measurement of the oscillator position , the measurement of the quadrature of the light field needs to be accurate . consider an imperfect quadrature measurement in which the phase of the local oscillator follows a gaussian distribution with a standard deviation @xmath93 this is equivalent to a phase noise on the light field and a perfect local oscillator . we found that the proposed witness can reveal optomechanical entanglement if @xmath94 to accurately record the decoherence of the mechanics , we find @xmath95 in the limit @xmath96 ( [ condition_violation ] ) , the constraint for observing entanglement is the most demanding . if the mechanical oscillator is not initially in its ground state , but in a thermal state with a mean occupation @xmath97 the variance of the light is unchanged but the correlations between the light and mechanics decreases . to detect entanglement , one needs @xmath98 in the limit where @xmath99 various cooling schemes have been proposed in the pulsed regime @xcite . for example , it has been shown in ref . @xcite that measuring the mechanical position with two pulses ( of duration @xmath100 and containing each @xmath90 photons ) delayed by a quarter of a mechanical period reduces the mechanical excitation to an effective thermal occupation @xmath101 for @xmath102 this results in @xmath103 i.e. ground state cooling . again , the limitation on @xmath104 is the power @xmath105 than can be homodyned and that can be undergone by the mechanics without increasing the temperature of the environment . to summarize , let us consider a given optomechanical system with fixed mechanical and cavity frequencies , effective mass and cavity length , i.e. @xmath106 and @xmath3 are given . we choose a pulse duration @xmath12 as large as possible to relax the constraint on the precision of the homodyne detection ( [ accuracy_homodyne_ent ] ) . in particular , for @xmath107 the maximum value is set by the cavity finesse . we then choose the largest possible @xmath74 so that the dynamics still holds in the pulsed regime ( [ condition_linear ] ) . we found that @xmath108 provides a significant violation of the inequality ( [ ent_witness ] ) . the detection of entanglement then gives a constraint on the initial effective occupation number of the optomechanical system through the formula ( [ cond_occupation ] ) . lastly , the formulas ( [ accuracy_position_deco ] ) and ( [ accuracy_homodyne_ent ] ) give the requirement on the measurement precision . the temperature of the environment and the mechanical quality factor are such that standard decoherence should operate on times scales longer than the collapse models we wish to test . for concreteness , we focus on a mechanical oscillator with an effective mass @xmath109 ng and a frequency @xmath110 s@xmath111 that is used as one of the mirrors of a perot - fabry cavity with length 0.5 cm , resulting in an optomechanical coupling @xmath112 s@xmath113 we consider a cavity finesse of @xmath114 which corresponds to the finesse of the perot - fabry cavity implemented with micromirrors in ref . @xcite . this leads to light pulses with a duration @xmath115 @xmath116s with a mean photon number of up to @xmath117 i.e. a power is 1 @xmath116w . to reveal entanglement , the thermal occupation number of the mechanics has to satisfy @xmath118 which can be achieved with a power @xmath119 nw and the phase of the local oscillator that is used for homodyning the light has to be set with an accuracy of @xmath120 @xmath121 this also takes a base temperature of @xmath122 mk and a mechanical qm factor of @xmath123 to record the decoherence of the mechanics , the mechanical position needs to be resolved with an accuracy @xmath124 which can be achieved with a power @xmath92 of 0.38 @xmath116w . furthermore , we found that the model by ellis and co - workers takes about @xmath125s to double the spread of results of @xmath126 , which would be testable with the proposed device with qm @xmath127 and @xmath128 mk . the model of diosi & penrose might also be testable despite the known ambiguity with respect to the mass distributions . under the assumptions that the mass is distributed over spheres corresponding to the size of the atomic nuclei , it takes about @xmath129s to double the spread of results of @xmath126 . this should be testable with qm @xmath130 and @xmath131 mk . we have shown how a simple laser light driving a mechanical oscillator in the pulsed regime results in an optomechanical entangled state . we have shown how to reveal the quantum nature of this state and how to use it to probe the oscillator decoherence . while we have focused on a realization with the trampoline resonator envisioned in ref . @xcite , various systems might be used to implement our proposal and ultimately , to test explicit wave function collapse models . we found for example that despite its small mass , the zipper cavities of ref . @xcite might be used to test the model of diosi & penrose model with a bath temperature of @xmath132 mk if their mechanical q factor is pushed to @xmath133 + we thank f. frowis and e. oudot for many discussions . this work was supported by the national swiss science foundation ( snsf ) , grant number pp00p2 - 150579 and the austrian science fund ( fwf ) , grant number j3462 and p24273-n16 . here , we clarify the constraint for the system to be in the pulsed regime . immediately after the interaction , the system is in the state @xmath134 and @xmath135 . @xmath136 refers to the displacement operator on the mirror ( light ) system respectively . performing a phase shift @xmath137 and taking the overlap of this state to the ideal case , we find @xmath138 \left [ e^{-2 i \alpha^2 \lambda a^\dag a } e^{i \lambda ( a^\dag a)^2 } d_{m}(a^\dag a \beta)d_{\ell}(\alpha ) \right]\|00 \rangle\|^2 \nonumber \\ & = & \|\langle 00 \| d_l^\dag ( \alpha ) e^{-2 i \alpha^2 \lambda a^\dag a + i \lambda ( a^\dag a)^2 } d_l ( \alpha)\|00\rangle\|^2 \nonumber \\ & = & \|\langle 00 \| e^ { + i \lambda [ ( a^\dag a)^2 + \sqrt{2 } \alpha \ { a^\dag a , x_{\ell } \ } + 2\alpha^2 x_{\ell } ^2 ] } \|00\rangle\|^2 \nonumber \\ & \approx&\|1 + i \lambda \alpha^2\|^2 \nonumber .\end{aligned}\ ] ] here we use the fact that @xmath139 , and requiring that @xmath140 yields the requirement that @xmath141 . the fact that the light pulse is not sufficiently short , can lead to a mirror with a different starting position in phase space , even right after the mirror - light interaction time . this can be seen from the displacement of the mirror state . @xmath142 \nonumber \end{aligned}\ ] ] to disentangle the light from the mechanics then , the mirror measurements should be adjusted by @xmath143 .	we show that the interaction of a pulsed laser light with a mechanical oscillator through the radiation pressure results in an opto - mechanical entangled state in which the photon number is correlated with the oscillator position . interestingly , the mechanical oscillator can be delocalized over a large range of positions when driven by an intense laser light . this provides a simple yet sensitive method to probe hypothetical post - quantum theories including an explicit wave function collapse model , like the diosi & penrose model . we propose an entanglement witness to reveal the quantum nature of this opto - mechanical state as well as an optical technique to record the decoherence of the mechanical oscillator . we also report on a detailed feasibility study giving the experimental challenges that need to be overcome in order to confirm or rule out predictions from explicit wave function collapse models .
in systems of reduced dimensionality , thermal and quantum fluctuations are the source of physical phenomena without analogues in ordinary 3d materials . since the original studies of mermin and wagner on the role of thermal fluctuations,@xcite it is known that in the isotropic spin-@xmath3 heisenberg model the divergent number of low - energy thermal excitations ( i.e. spin waves ) suppresses any long - range ordered phase for dimensions @xmath4 . due to the reduced dimensionality , the magnetic properties of spin arrays such as spin chains and ladders are further influenced by quantum effects that are masked in common 3d - materials . at @xmath5 , despite quantum corrections present in the antiferromagnetic case , magnetic order develops in the ground state of the 2d bipartite heisenberg model . due to the statistical analogy of a quantum @xmath6-dimensional antiferromagnet at zero temperature with a purely classical magnet at finite temperature in @xmath7 dimensions,@xcite quantum effects , particularly effective in the antiferromagnetic case , imply the absence of long - range order in quantum heisenberg antiferromagnets for @xmath8 . in real bulk materials , however , the same concept of one - dimensionality is ill - defined . for instance , it has often been shown that the effective dimensionality of certain materials can strongly change upon variations of temperature,@xcite magnetic field,@xcite or energy scale@xcite being probed . as a consequence , the interpretation of the experimental observations requires an interpolation between limits of different effective degrees of freedom . this is particularly true for the specific case of spin- antiferromagnetic ( af ) heisenberg chains realized in well - known model materials such as kcuf@xmath9,@xcite sr@xmath1cuo@xmath9,@xcite or copper pyrazine dinitrate.@xcite their high - temperature disordered phases , well described by the strictly one - dimensional luttinger liquid concept@xcite are , at low temperatures , replaced by a magnetically ordered state induced by weak interchain interactions . the need to interpolate between different dimensionalities suggested the introduction of the class of so - called _ quasi_-1d materials . the degree of one - dimensionality is usually measured in terms of the ratio between the nel temperature , marking the onset of the 3d magnetic order , and the intrachain exchange interaction.@xcite values of this ratio much below one signify the preservation of the 1d character of the system . nevertheless , 1d quantum fluctuations continue to be effective in the 3d domain too , i.e. , below the nel temperature . this is evident from both the strongly reduced saturation moment in the ordered state as well as from the deviations with respect to excitations predicted by the standard spin - wave theory.@xcite observations of the evolution of the properties of a physical system across the phase diagram , in regions characterized by different effective dimensionalities , has motivated a number of theoretical and experimental studies , dedicated to the phenomenon of dimensional crossover . the physics of an assembly of isolated spin- quantum chains upon increasing the interchain couplings is rather well understood.@xcite the impact of such a perturbation on chains experiencing a local transverse staggered field ( ltsf ) , however , is still an open question.@xcite a spin- quantum chain in a uniform field and a concomitant ltsf is usually modeled by the hamiltonian:@xcite @xmath10 where @xmath11 is the site - index along the chain direction , @xmath12 is the intrachain exchange coupling constant , @xmath13 is the bohr magneton , @xmath14 is the electronic gyromagnetic ratio , @xmath15 and @xmath16 are the directions parallel and perpendicular to the externally applied field @xmath17 , respectively . the locally induced transverse staggered field is @xmath18 , with @xmath19 as a constant . the interest in spin chains with an ltsf was first triggered by experimental studies on copper benzoate cu(c@xmath20h@xmath21coo)@xmath1@xmath223h@xmath1o,@xcite probing the compound s magnetism via susceptibility , nuclear magnetic resonance ( nmr ) and electron - spin resonance ( esr ) measurements . in zero magnetic field ( @xmath23 ) , copper benzoate was known to be just another realization of a spin- heisenberg chain , with an exchange constant @xmath24 k and the onset of a canted antiferromagnetic order at @xmath25 k.@xcite due to the relatively small intrachain exchange coupling , copper benzoate was considered as a favorable material for studying the field - dependent properties of the 1d spin- quantum heisenberg model . however , the field - dependent bulk properties were soon found to be incompatible with corresponding theoretical results.@xcite incommensurate soft modes at field - dependent reciprocal space positions of the excitation spectrum are expected for spin- antiferromagnets in an applied field . the first inelastic neutron scattering ( ins ) experiment , intended to verify these theoretical predictions , was performed on copper benzoate . @xcite field - driven incommensurate modes were indeed found at the expected positions in reciprocal space . due to an unexpected field- and orientation - dependent spin - gap @xmath26 , scaling as @xmath27 , these modes were not soft , however.@xcite this surprising ins observation was soon interpreted as the result of the particular character of the @xmath14-tensors and dzyaloshinskii - moriya ( dm ) interactions . along the chain direction , the non - diagonal tensor components change sign from one cu site to the next and the same is true for the dm @xmath28-vector . both can be mapped onto an effective ltsf and thus to eq . . formally it turns out that , in the absence of interchain perturbative terms , the effective low - energy theory of the ltsf hamiltonian is given by the quantum sine - gordon ( sg ) model.@xcite the sg model is one of the few non - linear problems which benefits from an exact solution . hence , the low - energy theory provides analytical expressions for the physical quantities in the temperature range @xmath29 when @xmath30 . + in the last 15 years the magnetic properties of a large number of both organic and inorganic compounds have successfully been described by the ltsf model . specifically , nmr and esr studies focused mostly on the joint detection of the temperature - dependent longitudinal ( @xmath31 ) and transverse ( @xmath32 ) local magnetization . examples are the cases of copper pyrimidine dinitrate @xcite and of bacu@xmath1ge@xmath1o@xmath2 @xcite . quite generally , any real material , unless directly excluded by symmetry arguments , will develop an arbitrarily small ltsf , captured in the @xmath33 term of eq . , when an external field @xmath17 is applied . this field - induced ltsf is expected to compete with that developing due to interchain interactions below the nel temperature ( @xmath34 ) . the resulting physics in the limit where both effects are of similar magnitude is currently not clear.@xcite one of the few experimental studies where both perturbations on the 1d spin- heisenberg model were strong enough to be experimentally visible is an ins experiment on cucl@xmath1 @xmath22 2(dimethylsulfoxide ) ( cdc).@xcite spin excitations measured at 40 mk showed that the competition leads to the opening of a gap at a non zero value of the staggered field @xmath33 , rather than in zero field , as the scaling relation @xmath35 would imply . besides the particular case of cdc , all other esr and nmr studies of materials modeled by eq . were usually limited to the range @xmath36 . by exploiting the high sensitivity of nuclear magnetic resonance to local magnetic fields , we present a @xmath0si nmr study of bacu@xmath1si@xmath1o@xmath2 . we demonstrate how the local magnetization in a chain , adequately modeled by eq . , develops above @xmath37 and how it evolves when @xmath38 . the latter evolution could be monitored thanks to the complete decoupling of the local field - induced magnetization from the order parameter which is related to the onset of a standard , temperature - driven second - order magnetic phase transition . in section [ sec : previousres ] we briefly summarize relevant results of previous research on bacu@xmath1si@xmath1o@xmath2 . our nmr data are presented in section [ sec : results ] , while a detailed discussion of the analysis is reported in section [ dataanaly ] . finally , section [ sec : summ ] offers a summary and the main conclusions of this work . a ) schematic view of the crystalline unit cell of bacu@xmath1si@xmath1o@xmath2 . red and black numbers identify the copper- and silicon atom sites , respectively . the small red spheres indicate the oxygen sites . zigzag cu@xmath39 spin chains run along the @xmath19 axis.,scaledwidth=50.0% ] bacu@xmath1si@xmath1o@xmath2 crystallizes in the orthorhombic space group @xmath40 ( @xmath41 ) with lattice constants @xmath42 , @xmath43 and @xmath44 .@xcite each unit cell contains eight cu@xmath39 ions , which determine the material s magnetic properties , and eight silicon atoms , each family being equivalent by local symmetry . both types of atoms , together with the respective closest - oxygen - atom configurations , are depicted in fig . [ fig : figure1 ] . the copper - ion spins form zigzag chains along the crystallographic @xmath19 axis . early zero - field ins studies of bacu@xmath1si@xmath1o@xmath2 at @xmath36 revealed a gapless spinon continuum . a fit to the lower boundary of the spectrum provided an intrachain exchange value @xmath45 mev.@xcite at the same time , elastic neutron - scattering measurements found evidence of a long - range antiferromagnetic order below @xmath46 k , with an ordered cu@xmath39 moment at saturation of 0.15 @xmath13 , collinear with the @xmath19 axis.@xcite these results qualify bacu@xmath1si@xmath1o@xmath2 to be among the best realizations of a spin-@xmath47 heisenberg chain system , characterized by a low ordering temperature and by a large intrachain exchange coupling . for a global overview of a material classification the reader is referred to table 1 in ref . . over the years , studies of bacu@xmath1si@xmath1o@xmath2 proceeded along two main directions , both related to the present work : _ i _ ) the study of the spin - reorientation transitions @xcite and , _ ii _ ) the investigation of the 1d - to-3d crossover.@xcite regarding the first topic , experiments employing neutron scattering,@xcite esr,@xcite and ultrasound techniques,@xcite were carried out in externally applied magnetic fields of varying strengths and orientations . these studies provided evidence of a number of phase transitions related to the realignment of spins ( and spin - flop transitions for an externally applied field @xmath48 ) at @xmath34.@xcite in spite of serious efforts , the cause for these transitions is not yet clear.@xcite in the following we show that nmr data provide direct evidence of the presence of an enhanced transverse susceptibility below @xmath37 , offering a simple explanation for the observed spin reorientations . we suggest that non - trivial phase diagrams below the magnetic ordering temperature may appear naturally in anisotropic quasi-1d antiferromagnets exhibiting a strong reduction of the ordered moment . as for the second topic , early studies have shown that bacu@xmath1si@xmath1o@xmath2 is also a model spin - chain compound for investigating the crossover from quantum - spin 1d dynamics to semi - classical 3d ( spin - wave ) dynamics.@xcite zero - field ins experiments probing the range below the nel temperature revealed how the presence of weak interchain interactions @xmath49 , produce an effective ltsf originating from neighboring chains , which leads to a confinement of massless spinon excitations at @xmath34 . therefore the situation is not equivalent to that of non interacting 1d chains in a staggered field . the additional @xmath50 interactions induce dispersive excitations with a momentum transverse to the chain direction . their energy vanishes at well defined positions in reciprocal space , coinciding with the bragg peaks of the ordered structure . for bacu@xmath1si@xmath1o@xmath2 in an applied magnetic field , intrinsic anisotropies lead to a field - induced ltsf causing a spin - gap to be present already above @xmath37 . the possibility of directly measuring the field - induced gap via ins is slim , but important consequences are expected for the local static magnetization . since the latter is accessible via nmr , we took advantage of this unique opportunity to study the 1d - to-3d crossover in a quantum - spin chain with an ltsf modeled by eq . . the nmr measurements were carried out on a @xmath51 mm@xmath52 single crystal of bacu@xmath1si@xmath1o@xmath2 . since the @xmath53 and @xmath19 lattice parameters are roughly the same , the correct identification of the two crystallographic directions is not trivial . field - dependent magnetization measurements on the same sample with @xmath48 could detect the two expected spin - flop transitions , hence confirming the correct identification of the crystal axes . the @xmath0si nmr lines were measured as a function of temperature at two different magnetic fields , with the external field being applied along the crystallographic @xmath53 or @xmath54 axis . spectra above the antiferromagnetic transition were recorded with a standard spin echo technique , while at lower temperatures the line intensities were established by a superposition of frequency sweeps as described in ref . . the nmr spectra reported in fig . [ fig : figure2]a show the height - normalized shapes vs. temperature , measured at 7 t for both crystal orientations . in the paramagnetic phase ( @xmath55 ) single lines are observed , which split into two in the ordered regime . while the results are qualitatively the same for both field directions , quantitative differences are obvious . with the field along the @xmath54 axis , the nmr line shifts towards higher frequencies ( by ca . 30 khz ) upon lowering the temperature . a much smaller shift is observed when @xmath56 and likewise the line splitting for @xmath57 is considerably reduced . the signal width , instead , is approximately the same in both cases and is essentially unaffected accross the entire temperature range in the paramagnetic regime . the monotonic decrease of the resonance frequency with increasing temperature in the paramagnetic phase ( see fig . [ fig : figure2]b ) is quite surprising . a spin-@xmath47 heisenberg chain is expected to display a broad maximum in the magnetic susceptibility @xmath58 at @xmath59.@xcite this maximum , also known as the bonner - fisher peak,@xcite should be located at @xmath60 k for bacu@xmath1si@xmath1o@xmath2 ( @xmath61 mev ) . this was indeed reported in previous works@xcite and confirmed by our low - field magnetization data ( see fig . [ fig : figurecomp]c ) . since the nmr line shift and the magnetic susceptibility are generally proportional , the absence of any maximum in fig . [ fig : figure2]b for @xmath55 is a striking feature , indicating a fundamental difference between the microscopically probed local field ( via nmr ) and that reflected in the macroscopic susceptibility . once the lineshape maxima in the 3d ordered regime ( @xmath57 ) were identified ( see fig . [ fig : figure2]b ) , the differences between the peak positions , @xmath62 , were evaluated . the corresponding values for the two different orientations are shown in fig . [ fig : figure2]c . we recall that @xmath62 is ultimately proportional to the order parameter of the phase transition and hence it can be used to monitor the transition . the resulting increase in frequency splitting between 10 and 5 k can be fitted by a power - law @xmath63 . although the chosen temperature range is too broad to really reflect a truly critical regime , the exponent @xmath64 is very close to @xmath65 , the value obtained from zero - field neutron diffraction data.@xcite figure [ fig : figure2]c also confirms that the monotonic increase of the order parameter for @xmath57 does not depend on field . indeed , for @xmath66 , practically the same ordered moment at saturation is found for @xmath67 t and 7 t. the postulated collinear antiferromagnetic order for @xmath57 is known to have a zero - field saturation moment of 0.15 @xmath13 and an easy axis which coincides with the @xmath19 direction.@xcite as shown in fig . [ fig : figure2 ] , the line positions reflect the nmr response to a magnetic field oriented along the @xmath53 or @xmath54 direction , respectively , i.e. , perpendicular to the easy axis @xmath19 . a standard collinear antiferromagnet with the field applied perpendicular to the easy axis exhibits a constant magnetization in the ordered phase . the nmr lines are thus supposed to split symmetrically with respect to their common relative shift,@xcite clearly at variance with our observation . in fact , the average positions of the two maxima , indicated in fig . [ fig : figure2]b by full black dots , contrary to expectations , are observed to decrease with decreasing temperature for @xmath57 . spin - wave corrections to the constant magnetic susceptibility below @xmath37 for antiferromagnets ordered collinearly along a direction perpendicular to the applied field show at most an increase of the longitudinal magnetization with decreasing temperature . this is due to zero - point spin fluctuations affecting the ordered moment.@xcite however , according to the data presented in fig . [ fig : figurecomp]c , this correction is modest in bacu@xmath1si@xmath1o@xmath2 . angular dependence of the nmr lines at 20 and 200 k. the orientation of the external field @xmath687.02 t varies within the crystalline @xmath69 plane.,scaledwidth=40.0% ] the @xmath70si nmr resonances of bacu@xmath1si@xmath1o@xmath2 depend strongly on sample orientation , a trend which is particularly conspicuous in the ordered phase , where the positions and shapes of the nmr lines are sensitive to even a small degree of misalignment . in order to avoid problems with data interpretation due to misorientation , a study to establish the orientation dependence of the nmr lines was carried out by mounting the sample on a two - axis goniometer , suitable for nmr experiments at cryogenic temperatures.@xcite the results for two temperatures above @xmath37 , 20 and 200 k , are reported in fig . [ fig : figure3 ] . once the @xmath53 and @xmath54 axes were identified , the sample was rotated such that the direction of the externally applied field was kept in the @xmath69 plane of the crystal lattice . as already mentioned in sec . [ intro ] , the zigzag geometry of the cu@xmath39 spin chains in bacu@xmath1si@xmath1o@xmath2 provokes electron anisotropies which may strongly affect the physics of the chain system in case of an externally applied magnetic field . two dominant contributions to the anisotropy originate either in off - diagonal components of the gyromagnetic tensor @xmath14 ( alternating in sign along the chain direction ) , and/or in spin - orbit effects in the cu - o - cu superexchange path along the chain . the latter is also known as the dzyaloshinskii - moriya ( dm ) interaction.@xcite to compare the ltsf model of eq . with experimental data , an estimate of these two contributions to anisotropy has to be made . a gap in the spin - wave excitation spectrum in the ordered phase was observed in zero - field ins measurements and attributed to two - ion anisotropy effects with an energy scale of @xmath71 mev,@xcite while an additional mode at a lower energy of @xmath72 0.17 mev was observed by esr.@xcite based on the local symmetry of the intrachain cu - o - cu bond , a dm @xmath28-vector lying almost in the @xmath69 plane , with unit vector components [ 0.86 , 0.51 , 0.07 ] was suggested.@xcite by making use of the crystal symmetry , we apply general space - group operations @xmath73 to a pair @xmath74 of copper sites interacting via oxygen superexchange along the @xmath19-axis . rotations @xmath75 and affine transformations @xmath76 are related to the symmetry operation @xmath77 . the original pair is transformed into a new set and the local environment transforms accordingly . the configuration of the various @xmath28 vectors in the unit cell can be established from the transformation rule:@xcite @xmath78 where @xmath28 transforms as an axial vector under the application of the rotation @xmath75 . with this rule the full pattern of alternating @xmath28 vectors , depicted as black arrows , halfway between the relevant cu sites in the upper panel of fig . [ fig : figure4 ] , can be derived . we note that the @xmath53 and @xmath54 components of the @xmath28 vector have alternating signs when moving along a given chain in the @xmath19 direction , or when moving between different chains in the @xmath53 direction . we obtain @xmath79 , @xmath80 and also @xmath81 , @xmath82 . the prime after the index site denotes copper ions belonging to the upper nearest - neighbor unit cell along the @xmath19 direction . ] the effect of the dm interactions between sites @xmath11 and @xmath83 , if not forbidden by crystal symmetry , can be taken into account via the following spin hamiltonian:@xcite upper panel : pattern of the field - induced ltsf in bacu@xmath1si@xmath1o@xmath2 . red ( blue ) lines correspond to the case of a longitudinal field applied along the @xmath54(@xmath53)-axis . the ltsf orientation of sites 3,4,7 and 8 is not affected by the change of field orientation . black arrows show the direction of the dm vector according to the symmetry rule in eq . ( see text ) . lower panel : local magnetization pattern of the order parameter in the magnetically ordered phase at @xmath84 and @xmath85= 0 . the same kind of structure is realized with a moderate external field in the @xmath69-plane.,scaledwidth=45.0% ] @xmath86 this contribution can be mapped onto a local transverse staggered field , @xmath87 , via a rotation in spin space.@xcite for small @xmath88 ratios , as is the case for bacu@xmath1si@xmath1o@xmath2 , the local transverse field at site @xmath11 can be approximated by : @xmath89 where @xmath90 is the uniform ( diagonal ) part of the local @xmath14-tensor valid at site @xmath11 and @xmath85 is the externally applied field . the term in eq . represents the second type of the ltsf components outlined above . in addition , strongly orientation - dependent , high - temperature magnetization data were interpreted as indicating a strong anisotropy of the local @xmath14-tensor.@xcite a direct measurement of the @xmath14-tensor components is in general possible via esr experiment . in the case of bacu@xmath1si@xmath1o@xmath2 , however , this is hampered by the broadening and the loss of intensity of the esr absorption in the paramagnetic phase , providing at best an estimation of @xmath14-factor @xmath91 and @xmath92.@xcite the evaluation of the local @xmath14-tensor is additionally complicated by the strong in - chain exchange interaction , leading to the exchange ( or motional ) narrowing of the resonance line.@xcite consequently , differences in the @xmath14-factor can not be resolved as long as the corresponding zeeman splittings are smaller than the exchange energy . since @xmath93 k , this is clearly the case here , even in very strong fields . as a working hypothesis we assume that the @xmath14-factor anisotropy is mostly determined by single - ion effects , which reflect the local configuration of oxygen atoms . figure [ fig : figure1 ] shows that each copper ion is located at the center of a tetrahedrally distorted cuo@xmath94 square . because of this distortion , the oxygen - to - oxygen distances of opposite o atoms differ by about 2% . if we neglect this detail , the local point group of the cuo@xmath94 unit is @xmath95 . an arbitrary @xmath14-tensor is then invariant under all the symmetry operations of the point group @xmath95 . in the local reference frame of a cuo@xmath94 unit , the tensor adopts a uniaxial form @xmath96@xmath97 , @xmath98 , @xmath99 ) , with two of the principal axes oriented along the two directions at 45 degrees from the square s diagonals and the third axis parallel to their cross product . the transformation matrix relating a cuo@xmath94 unit to the crystallographic unit cell may now be obtained . selecting the copper atom 1 in fig . [ fig : figure1 ] this transformation leads to : @xmath100 where the subscript @xmath11 of @xmath101 refers to the site index . since the eight copper sites are equivalent under the allowed symmetry operations , we can obtain the tensor @xmath96 for each of them . the matrix in eq . reveals that the components @xmath102 and @xmath103 are roughly equal ( in qualitative agreement with esr experiments ) , reflecting the coinciding high - temperature magnetization tails , measured with a field along the @xmath54 and @xmath19 direction , respectively . the qualitative behavior of the magnetization is also shown in fig . 1 of ref . . by considering the @xmath14-tensor for each site @xmath11 in the unit cell , and by using eq . for the dm contribution to the local field , the expected ltsf pattern for an external field applied along the @xmath53 or @xmath54 axis can be established . this pattern is the key for understanding the nmr results . in our case , whenever the external field lies in the @xmath69 plane , the ltsf @xmath104 is found to be parallel to the @xmath19 direction ( see fig . [ fig : figure4 ] ) . in order to exploit this favorable configuration , the field orientation @xmath48 has not been addressed in the present work . with the external field in the @xmath69 plane and forming an angle @xmath105 with the b - axis , we can calculate the ratio @xmath106 . for instance , by considering the copper sites 3 and 5 we obtain : @xmath107 the configurations of the ltsf , resulting from the orientation of an external field along the crystalline @xmath54 ( @xmath53 ) direction , are indicated by red ( blue ) arrows in the upper panel of fig . [ fig : figure4 ] . the terms @xmath108 in are to be inserted in eq . . these two patterns are found to be equivalent to the @xmath109 and @xmath110 irreducible representations of the magnetic - structure space , as previously established in ref . . they consist of the following linear combinations : @xmath111 with @xmath112 the component of the local magnetization at the @xmath11-site along the @xmath19-axis . note that the products @xmath113 and @xmath114 are symmetry invariants,@xcite i.e. , they are combinations of the irreducible representations ( ir ) of the magnetic structure which transform according to the trivial representation ( the 1d representation consisting of @xmath115 matrices containing the entry 1 ) of the little group of the propagation vector @xmath116 .. ] as mentioned above the local magnetization at a copper site @xmath11 , denoted as @xmath117 , is given by a uniform and a transverse component , such that @xmath118 , locally induced by an external uniform field @xmath119 and a staggered field @xmath87 . in the following we fix the convention that @xmath120 has a saturation value of . in order to recall the general results already known for @xmath121 and @xmath122 , and to present our new results , it is useful to introduce the following reduced units : @xmath123 a ) results of finite temperature dmrg calculations across an extended range of @xmath124 based on hamiltonian and valid in the small @xmath125 limit , in comparison with known analytical and tabulated results ( see text for details ) . b ) comparison of the calculated @xmath126 as obtained analytically or via dmrg at @xmath127 . c ) squid - magnetometry data measured at 1 t with the field applied along the @xmath53 and @xmath54 crystalline axes , respectively . solid lines are fits using eq . , dashed lines are predictions for an isotropic uniform chain with no ltsf.,scaledwidth=50.0% ] it has been shown @xcite that by using this notation and setting both @xmath128 and @xmath129 , the susceptibility @xmath130 has a peak at @xmath131 and a zero - temperature limit of @xmath132 . in fig . [ fig : figurecomp]a we reproduce the temperature dependence of @xmath130 , calculated and tabulated by a. klmper in ref . . the transverse field @xmath125 opens a gap in the excitation spectrum which , for @xmath133 , scales as:@xcite @xmath134 for small fields and @xmath135 analytic field - theoretical results for @xmath126 are available.@xcite in the chosen reduced units it reads : @xmath136 the high @xmath125 limit has been treated by previous density matrix renormalization group " ( dmrg ) calculations.@xcite no complete and unbiased numerical result is yet available for the susceptibility at small magnetic fields across an extended range of @xmath124 . in fig . [ fig : figurecomp]a , b we fill this gap with results of dmrg calculations@xcite for chains with 100 and 200 spin sites , respectively , and compare them with the analytical result of eq . , including or omitting the logarithmic correction . in our figure , the subscript @xmath77 stands for @xmath137 or @xmath138 , respectively . without the logarithmic term , the main difference between the analytical and the numerical result is , as expected , at low temperatures . in order to see how much a transverse field affects the longitudinal uniform magnetization , we followed i. affleck s approach @xcite and computed numerically the total derivative @xmath139 of the free energy of the system obtaining : @xmath140 with the parameters @xmath108 as given in eq . . two separate dmrg runs were employed to calculate the uniform and staggered susceptibilities independently . subsequently equation was used to obtain the value of @xmath141 . the possibility to compute the two quantities @xmath142 and @xmath143 separately , allowed us to make use of symmetries to reduce the hilbert space in the dmrg simulations . in order to compare our simulation results with the data for bacu@xmath1si@xmath1o@xmath2 , the temperature dependence of the magnetization @xmath144 was measured along the @xmath53- and the @xmath54-axis of the crystal . the data , measured with a squid magnetometer , are displayed in fig . [ fig : figurecomp]c . because of the small moment , particular care was taken in the choice of the sample - holder material that would cause at most a small magnetic background signal . with the model given by eq . we obtain good agreement with the experimental data down to approximately 20 k. the departure of the solid lines from the points is most likely due to approaching the onset of magnetic order . as previous authors@xcite we also tried a fit by imposing the value @xmath145 ( dashed lines ) . the resulting discrepancies are obvious . from the high - temperature tails of @xmath141(@xmath146 ) we extract values between 2.19 and 2 for @xmath147 and @xmath148 , respectively . the latter differ considerably , but are more realistic than the corresponding values between 2.5 and 2.2 quoted in refs . and . useful information can be extracted from the fit parameters @xmath108 . for the macroscopic uniform magnetization @xmath141 , the sign change of @xmath108 ( see eq . [ ltsf1 ] ) is irrelevant . thus in fig . [ fig : figurecomp]c , @xmath149 and @xmath150 are the corresponding parameters for fields @xmath85 along the @xmath53- or @xmath54-direction , respectively . first of all we consider the field - induced and angle - dependent spin - gap in eq . . we obtain @xmath151 mev and @xmath152 mev . these gaps , when expressed in @xmath153 units , are both of the order of 10 k , i.e. close to the temperature where the magnetic order sets in . this explains why an activated behavior of the spin - lattice relaxation time has not been observed in our previous nmr work.@xcite a proper estimate for the dm parameters @xmath154 can be obtained by solving eq . with @xmath155 from and the fitted @xmath156 . we obtain the values @xmath157 mev and @xmath158 mev . remarkably , the ratio @xmath159 is close to 1.68 , the value obtained from purely geometrical considerations.@xcite experimentally it turns out that the nmr response at low temperatures is dominated by the transverse magnetization @xmath160 , i.e. , the diverging susceptibility @xmath126 emerging from the dmrg calculation in fig . [ fig : figurecomp]b . the study of this quantity via nmr and its fate below @xmath37 is the main topic of the rest of this paper . having established the contributions to the local magnetic field at the cu sites , we now proceed to study their influence on the @xmath70si nmr - line data . in our case the local magnetization experienced by the silicon nuclei is dominated by the externally applied field , and the contribution due to the sample s magnetization is only of second order . for this reason the resonance frequency @xmath161 of the @xmath0si nucleus @xmath116 ( @xmath162 see fig . [ fig : figure1 ] for the notation ) can be written as:@xcite @xmath163 , \label{couplingsi}\ ] ] where @xmath164 is the @xmath0si nuclear gyromagnetic ratio , @xmath165 is the orbital shift tensor , @xmath166 is the dipolar tensor which couples the silicon nucleus @xmath116 to the copper atom @xmath11 , and @xmath167 is the relevant transferred hyperfine interaction . the dipolar sum in eq . can be calculated directly . this was done by fixing the cu - spin arrangement resulting from the ltsf configuration shown in fig . [ fig : figure4 ] and by including the contributions from the copper atoms within 50 from the considered silicon site . the sum of hyperfine interactions runs over the four nearest neighbor ( nn ) copper sites.@xcite given @xmath167 and @xmath166 for the silicon nucleus @xmath116 , the relevant tensors for the other silicon sites can be obtained by allowed symmetry transformations , meaning that tensors referring to the various silicon nuclei are not mutually independent . if a symmetry operation of the space group @xmath73 , applied to the silicon site @xmath168 , brings it to @xmath169 ( and consequently the copper site @xmath170 to @xmath171 ) , the hyperfine tensors are given by @xmath172 . for example : @xmath173 the four nn copper atoms surrounding the silicon atom located at site @xmath174 are @xmath175 ( see fig . [ fig : figure1 ] ) . contrary to dipolar interactions , the components of the transferred - hyperfine tensor are _ a priori _ unknown . for calculating directly the @xmath105-dependent component @xmath176 of the uniform local field at site @xmath174 or @xmath177 , parallel to @xmath178 , we define @xmath179 , @xmath180 , @xmath181 and @xmath182 and we denote @xmath183 and @xmath184 . by using the notation of eq . , we obtain : @xmath185 \cdot \frac{m}{g_{u } } \left ( \begin{array}{c } g_1^{11 } \sin{\theta } \\ g_1^{22 } \cos{\theta } \end{array } \right ) \nonumber \\ & = \frac{m}{2g_{u } } \left [ y_1 g_1^{11 } + y_5 g_1^{22 } + \left ( y_5 g_1^{22 } - y_1 g_1^{11 } \right ) \cos{2\theta } + \right . \nonumber \\ & \left . \pm \left ( y_2 g_1^{22 } + y_4 g_1^{11 } \right ) \sin{2\theta } \right ] , \label{uniformpart}\end{aligned}\ ] ] where the plus ( minus ) sign refers to @xmath174 ( 2 ) . in we take into account that the sample s uniform longitudinal magnetization and the externally applied field may not be collinear due to a possible @xmath14-tensor anisotropy . by definition , @xmath186 and @xmath187 . the reason for picking the si sites @xmath188 for describing the relevant nmr lineshapes is evident from the contribution of the transverse field @xmath189 to the resonance frequency . we get : @xmath190 . \label{stgloc12}\end{aligned}\ ] ] in the _ paramagnetic phase _ the following relations always hold by symmetry : @xmath191 by denoting @xmath192 , we obtain for the transverse local field in the paramagnetic phase : @xmath193 it turns out that by considering any other of the silicon sites , the only two possible _ orthogonal _ local fields are those given by . for an applied field along the @xmath53 or the @xmath54 axis these two local fields coincide . from eq . it follows that if @xmath194 ( field along @xmath54 ) , @xmath195 . the same is true if @xmath196 ( field along @xmath53 ) ( see fig . [ fig : figure4 ] ) . a single narrow line is thus expected for @xmath194 and @xmath196 , while two lines are expected in an intermediate angular range and at temperatures exceeding @xmath37 . this is indeed the case , as already shown in fig . [ fig : figure3 ] . we conclude this section with two principal results . the first concerns the prediction for the angular- , temperature- and field - dependent relative nmr line shift @xmath197 in the paramagnetic ( @xmath55 ) regime due to the transferred - hyperfine and orbital interactions : @xmath198 \right . \nonumber \\ & \left . + m^{3/5 } \left ( g_3 \sin{\theta } \pm g_6 \cos{\theta } \right ) \right . \nonumber \\ & \left . + \frac{1}{2 } \left [ \sigma_1 + \sigma_5 + \left ( \sigma_5 - \sigma_1 \right ) \cos{2\theta } \pm 2 \sigma_2 \sin{2\theta } \right ] \right \ } , \label{shift1}\end{aligned}\ ] ] where a symmetric orbital - shift tensor @xmath199 has been introduced . the consequences of eq . are discussed in the following section . the model behind eq . is ultimately independent of the exact geometry of the hyperfine couplings ; the qualitative result does not change even if , for instance , the parameters @xmath200 and @xmath201 were zero . next we discuss the situation in the _ ordered regime _ , below @xmath37 . here , due to the second - order phase transition , the symmetry of the system is spontaneously broken . the adopted order reflects one of the irreducible representations of the magnetic structure . its product with the applied field is , however , not necessarily an invariant upon symmetry transformations . consequently , in the ordered regime the lines are expected to split , even when the field is applied along the main crystallographic axes . from previous zero - field diffraction studies@xcite and following the conventions in ref . , it is known that the representation chosen by the spin system is @xmath202 , collinear with the @xmath19 axis ( see lower panel of fig . [ fig : figure4 ] ) . by calculating the local field employing eq . , the nmr lines split below @xmath37 according to : @xmath203 since in the @xmath202 representation @xmath195 is always valid , eq . suggests that the lines should coincide if @xmath204 . this is not the case , however , if all the possible silicon sites are considered . in the ordered phase , the local field at the sites @xmath205 is the same . also the sites @xmath206 experience the same field , but the latter differs slightly from the former . this explains the line splitting shown in fig . [ fig : figure2 ] for both field orientations with respect to the crystal axes @xmath53 and @xmath54 . the parameters @xmath207 @xmath208 @xmath209 are not directly accessible by experiment . nevertheless , eq . offers the possibility to average out the @xmath202 contribution to the nmr shift below @xmath37 , thus providing a _ * direct access * _ to the components @xmath210 and @xmath211 of the local magnetization in the ordered regime . as just explained at the end of the previous section , the average nmr line positions at @xmath212 ( shown as black dots in fig . [ fig : figure2]b ) are independent of the contribution of the @xmath202 representation and reflect the influence of the ltsf and the uniform magnetization . this holds true even if a dipolar term is added to eq . , since the average nmr line position is not affected by the expected symmetrical dipolar splitting below @xmath37 . with the external field in the @xmath69 plane , the ltsf is always collinear with the @xmath19 axis . for @xmath36 the transverse magnetization induced by the ltsf contributes to @xmath213 in the form of eq . . for @xmath34 both @xmath210 and @xmath211 are still present . since however @xmath210 ( in fig . [ fig : figurecomp]c ) is weakly temperature dependent , the strong variation of the average nmr shift for @xmath66 at @xmath34 ( see fig . [ fig : figure5]c ) is dominated by a contribution related to the @xmath214 representation . thus , below @xmath37 , nmr allows us to reveal the effects of the interaction between the representations @xmath215 and @xmath202 . we return to this issue after considering first the @xmath36 regime . the relative shift captured in eq . includes the anisotropic orbital - shift tensor . its components along the main crystal axes are usually determined via the so - called clogston - jaccarino plot , where the nmr line shift is plotted versus the corresponding susceptibility.@xcite this approach requires a sufficiently broad temperature range , in which the nmr shift mimics the sample s magnetization . because of the large cu - o exchange coupling in the bacu@xmath1si@xmath1o@xmath2 chains and the weak @xmath70si nmr signal at elevated temperatures , a reliable estimate of the orbital - shift components was not possible in this way . by assuming this tensor to be symmetric and temperature - independent , we released the parameters @xmath216 , @xmath217 and @xmath218 and the hyperfine couplings . in this way the whole data set could be fitted with a single set of parameters . the quantitative temperature dependences of @xmath210 and @xmath211 were established by using the calculations described in sec . [ sec : ltsf_physics ] , inserting the values of the ltsf as obtained from the fits in fig . [ fig : figurecomp]c . in fig . [ fig : figure5]a - d we display the result of the fits , as well as the individual contributions to the local magnetization , as a function of the angle @xmath105 and of temperature . in these figures the individual contributions to the total shift ( red curve ) of the nmr lines caused by the local longitudinal and transverse magnetization and by the orbital shift are highlighted as blue , black and green curves , respectively . for @xmath67 t , only the global fit is presented . also shown are the temperature independent contributions of the orbital shift ( broken lines in fig . [ fig : figure5]c and d ) . it may be seen that the temperature dependence of the shift due to the longitudinal magnetization is weak . the transverse component @xmath219 is small at 200 k but it grows significantly at low temperatures . the data were fitted in the range 20 to 230 k. the relative shift of the nmr lines ( at very low fields with respect to saturation ) scales linearly as a function of field . the fit parameters we obtain are @xmath220 t/@xmath13 , @xmath221 t/@xmath13 , @xmath222 t/@xmath13 and @xmath223 t/@xmath13 . since the fit parameters @xmath224 and @xmath225 are not lineary independent , we could fit only their combination @xmath226 t/@xmath13 . the computed dipolar tensor components , to be inserted in eq . , are of the order of 0.020.04 t/@xmath13 . the orbital shift values displayed in fig . [ fig : figure5 ] are of the order of 150 ppm . employing the same classification of representations as introduced in ref . , the antiferromagnetically - ordered phase in zero magnetic field is related to the @xmath202 representation and the corresponding order parameter . in the previous section we provided evidence for an enhanced transverse magnetic susceptibility even in the ordered regime . this enhancement is characteristic of quasi-1d chains in an ltsf and we argue that it is the reason for the unusual spin - reorientation transitions that are observed in bacu@xmath1si@xmath1o@xmath2 . the microscopic approach requires considering the effects of the 1d - to-3d dimensional crossover in specific features of the magnetic properties . in case of chains with no ltsf , this was done with a combined mean - field and random - phase approximation " approach.@xcite + here we tackle the problem with a ginzburg - landau ( gl ) expansion@xcite of the free energy @xmath227 close to @xmath37 . although this phenomenological approach neglects fluctuation effects , it has the advantage of retaining the exchange - energy contributions to the susceptibility of the ordered phase . the mean - field approach also includes interactions between different , possibly coexisting , order parameters . we start by constructing symmetry invariants of the little group of the @xmath116-vector.@xcite in bacu@xmath1si@xmath1o@xmath2 , even if exposed to an applied field , a commensurate antiferromagnetic structure with @xmath228 is realized , leading to a little point group which coincides with @xmath229 . we call @xmath230 the @xmath231-component of @xmath232 , with @xmath233 the @xmath234-ir ; clearly the product @xmath235 ( with @xmath236 as the number of equivalent copper sites in the unit cell ) . in terms of a gl free - energy expansion over all possible order parameters , a phase transition will occur whenever one of the coefficients @xmath237 of the quadratic term @xmath238 changes sign . since the @xmath239 representation is the one realized in the magnetically ordered regime in zero field , we can write that @xmath240 ( @xmath241 ) . all the irs @xmath242 of the magnetic structures in bacu@xmath1si@xmath1o@xmath2 are one - dimensional.@xcite it is therefore easy to construct invariant combinations of the @xmath230 since the representations of the powers of these terms , which have to transform according to the trivial representation , remain one dimensional . the expansion in eq . is based on the physics discussed in the previous sections of this paper . retained are the terms containing @xmath243 , related to the zero - field magnetic order , @xmath244 representing the ltsf , and @xmath17 the external magnetic field . we will limit our considerations here to the case of an ltsf pattern @xmath245 which is realized for @xmath66 . a more complete analysis will be published separately.@xcite the relevant expansion in powers of @xmath243 , @xmath244 and @xmath246 reads : @xmath247 with @xmath248 as the value of the free energy in the paramagnetic phase in zero field . the @xmath249 coefficient is positive above the nel temperature , reflecting the absence of a spontaneous symmetry breaking related with the @xmath250 order parameter . on the other hand we assume @xmath249 to be small in the vicinity of @xmath37 since , as indicated in fig . [ fig : figure4 ] , @xmath109 differs from the lowest energy - state configuration @xmath239 only by the mutual orientation of the spins in the neighboring chains , which are relatively weakly coupled . the temperature dependence of @xmath249 is assumed to be linear in the vicinity of @xmath37 : @xmath251 $ ] . the fourth order term for @xmath239 fixes the magnitude of the main order parameter below the transition ; as required , @xmath252 . the terms with prefactors @xmath253 and @xmath254 are crucial in our discussion , because they describe the exchange competition of the field - induced order @xmath255 and the spontaneous order @xmath256 . microscopically , these terms arise from the simple idea that both the main order parameter and the induced order parameter involve the same local spins , eventually along the same crystallographic direction . the @xmath253 coefficient is expected to be positive in order to enhance the energy cost for the coexistence of these two magnetic structures . finally the term related to @xmath254 defines the preferred mutual orientation of the two order parameters by means of the scalar product between the them . from the expansion in eq . alone it is not possible to predict whether a collinear ( @xmath257 ) or a transverse ( @xmath258 ) spin configuration is realized . the terms related to @xmath259 and @xmath260 describe interactions between the longitudinal magnetization and @xmath202 . the terms with prefactors @xmath261 describe the orientation of the zero - field order parameter . reported results of neutron diffraction@xcite and antiferromagnetic resonance@xcite imply that @xmath262 . the term @xmath263 is responsible for the fact that the magnetic structure @xmath109 is induced by an external field applied along the @xmath54 axis . the powers of @xmath85 completing the expansion in eq . provide a full description of the effects of the @xmath14-tensor anisotropy on the longitudinal magnetization . this may be seen by recalling that : @xmath264 minimizing over the components of @xmath250 for @xmath66 and assuming a zero - field collinear antiferromagnetism @xmath239 , we get @xmath265 and : @xmath266}. \label{formulal1a}\end{aligned}\ ] ] with a similar reasoning we obtain the longitudinal magnetization @xmath267 along the @xmath54-axis , in the former notation denoted as @xmath213 : @xmath268 the last two equations deserve some discussion . equation captures the temperature dependence of the transverse staggered magnetization . microscopically , the increase of @xmath109 upon cooling above @xmath37 ( @xmath269 ) is due to the divergent transverse susceptibility of a 1d spin- quantum heisenberg chain in an ltsf . this situation is modeled by the decrease of @xmath270 on cooling ( i.e. with @xmath271 ) . it also predicts a decrease of @xmath215 upon the growth of @xmath202 at @xmath34 , as observed in the nmr data . microscopically , the change of regime upon decreasing temperature , from a divergent transverse susceptibility ( characteristic of a spin chain ) to a progressive competition between the field - induced magnetization pattern and the zero - field order parameter , is argued to be a direct consequence of the dimensional crossover from 1d to 3d of a chain in an ltsf . the clear experimental identification of how 1d physics affects the static magnetization properties even below @xmath37 is the new result emerging from the present study . below we address the question of how these anomalous properties for @xmath57 can explain certain spin reorientation transitions observed in bacu@xmath1si@xmath1o@xmath2 . data representing relative @xmath0si nmr line shifts in the vicinity of the ordering temperature , compared with the classical predictions of eqs . and . although fully classical , the gl type approach grasps the competition between the spin structures described by @xmath109 and @xmath202 . the hyperfine parameters and the orbital shift are taken from the analysis for @xmath36 ( see text ) . the average nmr frequency is shown below the transition temperature . black , blue and green line show single components of the fits due to @xmath109 , @xmath267 and @xmath199 , respectively.,scaledwidth=50.0% ] by analyzing eq . we note that the first term is zero for an easy axis ( @xmath19 axis in our case ) orthogonal to the direction of the applied field ( along the @xmath54 direction ) . the second term , instead , provides corrections to the constant magnetization predicted by the standard mean - field theory below @xmath37 . microscopically it can be related to a semi - classical contribution of spin - waves.@xcite from the magnetization data in fig . [ fig : figurecomp]c it may be concluded that @xmath272 . next we single out a constant paramagnetic contribution , with prefactors @xmath273 and @xmath274 , which is related to the magnetization of the ideal spin- heisenberg chain at the nel temperature . we note that also in this case a contribution from the staggered transverse susceptibility affects the longitudinal magnetization data , @xmath275 . while @xmath109 scales as @xmath276 , the contribution to @xmath267 scales as @xmath277 , in full qualitative agreement with the microscopic approach . in magnetization measurements along either the @xmath53 or the @xmath54 axis , the contribution of the staggered magnetization matters , but it is not as outstanding as in nmr measurements , where both @xmath267 and @xmath278 are revealed . the enhanced transverse susceptibility accounts very well for the observed spin reorientations . for example , with the applied field along the @xmath54-axis , such a transition occurs at @xmath279 t.@xcite it is caused by a sudden change of the easy axis from the @xmath19- to the @xmath53-direction . below @xmath37 the quasi one - dimensionality extends itself in the form of a field - induced transverse susceptibility . it favours a field - induced spin alignment @xmath109 which competes with the zero - field order parameter @xmath239 . the higher the field , the larger is the energy cost to sustain this arrangement [ captured by the term @xmath253 of eq . ] . substituting eq . into yields an expression which depends on @xmath280 only . a spin reorientation is then expected as the result of competition of the field - dependent anisotropic corrections with the conventional anisotropy of the order parameter at a field : @xmath281 it can be shown that , for @xmath258 , eq . also accounts for two spin - reorientation transitions when @xmath48 . the inclusion of the staggered field pattern described by the representation @xmath110 [ see eq . ] could similarly account for the phase transition at @xmath56.@xcite based on formulas and , we now attempt a comparison with the experimental data in the temperature range @xmath282 k , where the microscopic 1d model does not properly describe our results . using the transferred hyperfine parameters determined in section [ sec : comparison1 ] , we compare the @xmath0si nmr line shift monitored for a field @xmath283 = 7.02 t oriented along the @xmath54-axis with the gl approach , postulating @xmath284^{\beta}$ ] and @xmath285 ( we arbitrarily set the zero - temperature limit of @xmath239 to unity ) . this is shown in fig . [ fig : landaufit ] . to obtain tentative estimates of the gl - model parameter , we first fitted @xmath267 of fig . [ fig : figurecomp]c in the vicinity of @xmath37 to eq . and obtained the parameters @xmath286 emu / mol cu , @xmath287 emu / mol cu and the ratio @xmath288 emu / mol cu . notice that the present numbers contain already a prefactor 0.15 in the numerator of eq . , corresponding to the experimentally reported zero - temperature limit of @xmath239 expressed in bohr magnetons . considering the decrease of @xmath109 with increasing temperature above @xmath37 we obtain @xmath289 0.051 k@xmath290 . next , with the fixed @xmath291 ratio we could fit the relative nmr peak positions , as shown in fig . [ fig : landaufit ] , and hence determine the parameters @xmath292 and @xmath293 . the fit shown in fig . [ fig : landaufit ] was obtained with @xmath294 emu / mol cu and @xmath295 emu / mol cu , which correspond to a value @xmath296 @xmath13 at 7.02 t. a detailed analysis of @xmath0si nmr data obtained by probing single - crystalline bacu@xmath1si@xmath1o@xmath2 revealed the influence of 1d physics into the regime of 3d magnetic order at temperatures below 10 k. in this way the problem of weakly interacting nearest neighbor chains , described in the non - interacting limit by the model in eq . , could be addressed . based on a classical ginzburg - landau analysis it is shown that in this type of compounds complicated ( @xmath17 , @xmath146 ) magnetic phase diagrams emerge . they are caused by the interaction of the transverse staggered local magnetization , originating from magnetic anisotropies in spin- heisenberg chains , with the effective magnetic field due to the weakly ordered spin moments on neighboring chains . we argue that the previously established spin - reorientation transitions in bacu@xmath1si@xmath1o@xmath2 reflect this situation and can , therefore , be understood in this framework . the authors thank k. pra ( epf lausanne ) and o. zaharko ( psi ) for useful discussions . we are thankful to m. zhitomirsky ( cea - grenoble ) for the enlightening comments concerning the ginzburg - landau approach . this work was financially supported in part by the schweizerische nationalfonds zur frderung der wissenschaftlichen forschung ( snf ) and the nccr research pool manep of snf . one of the authors ( v.g . ) thanks the russian foundation for basic research ( rfbr ) for the support of his studies . k. m. kojima , y. fudamoto , m. larkin , g. m. luke , j. merrin , b. nachumi , y. j. uemura , n. motoyama , h. eisaki , s. uchida , k. yamada , y. endoh , s. hosoya , b. j. sternlieb , and g. shirane , _ phys . lett . _ * 78 * , 1787 ( 1997 ) . c. broholm _ et al . _ in _ high magnetic fields _ ( lecture notes in physics vol . 595 ) , ed . c. berthier _ et al . _ ( springer , berlin , 2002 ) , pp 211234 . a. zheludev , t. masuda , i. tsukada , y. uchiyama , k. uchinokura , p. bni , s .- h . lee , _ phys . b _ * 62 * , 8921 ( 2000 ) ; a. zheludev , k. kakurai , t. masuda , k. uchinokura , and k. nakajima , _ phys . rev . lett . _ * 89 * 197205 ( 2002 ) ; a. zheludev , m. kenzelmann , s. raymond , e. ressouche , t. masuda , k. kakurai , s. maslov , i. tsukada , k. uchinokura , and a. wildes , _ phys . rev . lett . _ * 85 * 4799 ( 2000 ) .	heisenberg spin- chain materials are known to substantially alter their static and dynamic properties when experiencing an effective transverse staggered field originating from the varying local environment of the individual spins . we present a temperature- , angular- and field - dependent @xmath0si nmr study of the model compound bacu@xmath1si@xmath1o@xmath2 . the experimental data are interpreted in terms of the divergent low - temperature transverse susceptibility , predicted by theory for spin chains in coexisting longitudinal and transverse staggered fields . our analysis first employs a finite - temperature density matrix renormalization group " ( dmrg ) study of the relevant one - dimensional hamiltonian . next we compare our numerical with the presently known analytical results . with an analysis based on crystal symmetries we show how the anisotropic contribution to the sample magnetization is experimentally accessible even below the ordering temperature , in spite of its competition with the collinear order parameter of the antiferromagnetic phase . the modification of static and dynamic properties of the system due to the presence of a local transverse staggered field ( ltsf ) acting on the one - dimensional spin array are argued to cause the unusual spin reorientation transitions observed in bacu@xmath1si@xmath1o@xmath2 . on the basis of a ginzburg - landau type analysis , we discuss aspects of competing spin structures in the presence of magnetic order and the enhanced transverse susceptibility .
the uncertainty principle initially observed by heisenberg @xcite for the case of position and momentum , and further formulated by robertson @xcite for arbitrary pairs of observables is a central in quantum theory . it sets limits on our ability to predict the precise outcomes of two incompatible measurements on a quantum system , and at the same time provides the basis for new technologies such as quantum cryptography @xcite in quantum information . originally , the uncertainty relation is expressed in terms of the standard deviation @xmath0\rangle\|$ ] for two observables @xmath1 and @xmath2 @xcite . however , this uncertainty bound is state - dependent and also trivial for finite - valued observables @xcite . to remove this pitfall and to precisely capture its physical meanings , the original form has subsequently been recast to the entropic one @xcite , which reads @xmath3 where @xmath4 denotes the shannon entropy of the probability distribution of the outcomes when @xmath1 is measured , and likewise for @xmath5 . @xmath6 quantifies the complementarity of @xmath1 and @xmath2 , where @xmath7 for nondegenerate observables , with @xmath8 and @xmath9 being the eigenvectors of @xmath1 and @xmath2 . the above uncertainty limitation applies to the case that the observer can access only to the classical information . it may be violated through clever use of entanglement , which plays a central role in quantum information . this is the quantum - memory - assisted entropic uncertainty principle , which was previously conjectured by renes and boileau @xcite , and later being strictly proved by berta _ it states that if the observer can entangle the particle @xmath10 that he wishes to measure with another particle @xmath11 which serves as a quantum memory , then the uncertainty of this observer about any pair of observables can be dramatically reduced . particularly , if @xmath10 and @xmath11 are maximally entangled , then the observer is able to correctly predict the outcomes of whichever measurement is chosen . this new entropic uncertainty relation reads @xmath12 where @xmath13 is the conditional von neumann entropy of the postmeasurement state @xmath14 , and likewise for @xmath15 . compared with eq . ( 1 ) , there is an extra term @xmath16 ( the conditional von neumann entropy of @xmath17 ) appearing on the right - hand side of eq . ( 2 ) . in particular , if @xmath16 is negative , the berta _ _ bound ( bb ) of uncertainty in eq . ( 2 ) can be reduced comparing with previous uncertainty relation . it is also found that this negative value gives the lower bound of the one - way distillable entanglement between @xmath10 and @xmath11 @xcite . this new entropic uncertainty principle has been recently confirmed experimentally @xcite , and ignites interests of people on investigating its potential applications from various aspects @xcite . in this work , we will first relate it to quantum teleportation , and show that any two - qubit @xmath17 with negative @xmath16 gives nonclassical teleportation fidelity . it is known that teleportation is one fundamental protocol in quantum information processing @xcite . it is crucial , both theoretically and experimentally , to know whether the fidelity of teleportation is in classical regime or in quantum regime . our result relate this important problem to the entropic uncertainty principle . we will also investigate efficiency of this new entropic uncertainty relation on witnessing entanglement in a class of bosonic structured reservoirs . it is known that entanglement plays a key role in quantum information processing such as in teleportation and in condensed matter physics ; see , for example , refs . the point of departure for this practical application ( i.e. , entanglement witness ) is eq . ( 2 ) , from which one can note that if @xmath18 , then @xmath19 , and hence @xmath17 is entangled @xcite . experimentally , the value of @xmath20 can be estimated by conditional single - qubit tomography on @xmath11 @xcite , and this estimate is termed tomographic estimate ( te ) . there are other ways for estimating the uncertainty . the first is the measurement estimate ( me ) denoted by @xmath21 , which corresponds to the same measurements on both @xmath10 and @xmath11 , and are favored for its ease of implementation @xcite . this estimate provides an upper bound for the new uncertainty relation in that quantum measurements never decrease entropy . the second is the fano estimate ( fe ) obtained by using fano s inequality @xmath22 @xcite , where @xmath23 is the binary entropy function , with @xmath24 being the probability that the outcomes of @xmath25 on @xmath10 and @xmath25 on @xmath11 are different , and @xmath26 is the dimension of @xmath10 . for the two - qubit system ( i.e. , @xmath27 ) , the inequality @xmath28 is a signature of entanglement between @xmath10 and @xmath11 . in this section we relate the quantum - memory - assisted entropic uncertainty principle to quantum teleportation . we will show that any two - qubit state with negative conditional von neumann entropy , which thus lowers the upper bound of the uncertainty , is a manifestation of its usefulness for nonclassical teleportation . without loss of generality , we suppose the sender alice wants to teleport to the receiver bob a general one - qubit state , with a two - qubit state @xmath29 ( pure or mixed ) being used as the quantum channel . then if they adopt the standard teleportation scheme ( i.e. , alice performs the bell - basis measurement while bob is equipped to perform any unitary transformation ) , the maximal average fidelity achievable can be evaluated as @xcite @xmath30 where @xmath31 , with @xmath32 being the @xmath33 positive matrix with elements @xmath34 related to the bloch sphere representation of @xmath29 below @xmath35 where @xmath36 is the @xmath37 identity operator , @xmath38 is the vector of the pauli spin matrices , @xmath39 and @xmath40 are the local bloch vectors in @xmath41 with @xmath42 and @xmath43 . since the teleportation fidelity is locally unitary invariant , and we can always find unitary operators @xmath44 and @xmath45 which transform the state @xmath29 into @xmath46 with @xmath47 and @xmath48 , it suffices to restrict our concern to the representative class of quantum channels expressed in eq . ( 5 ) with less number of parameters , for which we always have @xmath49 @xmath50 gives @xmath51 and thus @xmath17 is competent for teleportation . the real numbers @xmath52 , @xmath53 , and @xmath54 ( @xmath55 ) in eq . ( 5 ) must satisfy certain constraints such that @xmath17 is a well defined density operator . particularly , for the special case of @xmath56 , physical @xmath17 are those with @xmath57 ( for simplicity , we just refer to @xmath58 in the following ) belongs to the tetrahedron @xmath59 ( see fig . 1 ) with vertices @xmath60 , @xmath61 , @xmath62 and @xmath63 , among which separable ones are confined to the octahedron @xmath64 with vertices @xmath65 , @xmath66 , and @xmath67 @xcite . the bell states sit at the four vertices of @xmath59 , while the werner states @xcite are those represented by the lines connecting the vertices of @xmath59 with the origin of @xmath68 . for the nonzero values of @xmath69 and/or @xmath70 , the vector @xmath58 which admits the positive semidefiniteness of @xmath17 also belongs to @xmath59 @xcite . but now some @xmath58 inside @xmath59 may not correspond to the physical states . moreover , while all the separable states are still confined to @xmath68 , there are also entangled ones with @xmath58 belongs to @xmath68 , which is different from the case of @xmath56 . since the octahedron @xmath68 is specified by latexmath:[$\sum_{k=1}^{3 } @xmath17 with @xmath58 belongs to the four small tetrahedra divided by @xmath68 ( i.e. , the regions inside @xmath59 and outside @xmath68 ) gives @xmath50 and thus is useful for quantum teleportation . physical @xmath17 confined to @xmath68 may also be entangled when the inner product @xmath72 , but they are useless for teleportation . now we begin to discuss the operational identification of @xmath17 useful for teleportation , and to what degree this identification can cover the whole set of physical @xmath17 useful for teleportation . when considering the standard protocol , horodecki _ et al . _ proved that any @xmath17 which violates the bell - clauser - horne - shimony - holt ( bell - chsh ) inequality is useful for teleportation @xcite . here , we will show that the negativity of the conditional von neumann entropy of @xmath17 is also a signature of its usefulness for teleportation . particularly , these @xmath17 can be witnessed by means of the new entropic uncertainty relation , and therefore is expected to have potential applications in experiments . to prove the above argument , we consider first @xmath17 associated with the vertices of @xmath68 . the positive semidefiniteness requires that the matrix elements @xmath73 and @xmath74 , which yield @xmath75 for physical states . for example , if @xmath76 the above requirements turn to @xmath77 and @xmath78 , which are satisfied only when @xmath75 . under the conditions of @xmath75 , we can write the explicit forms of @xmath17 at the vertices of @xmath68 , and further determine the constraints imposed on the parameters by positive semidefiniteness of @xmath17 . we use another property of the physical @xmath17 , which says that if a hermitian matrix is positive semidefinite then all of its principal minors must be non - negative . for @xmath17 with @xmath76 , we derive the second- and third - order leading principal minors as @xmath79 we see that @xmath80 , @xmath81 and @xmath82 must be satisfied for ensuring the positive semidefiniteness of @xmath17 . furthermore , the determinant of the @xmath83 minor formed by removing from @xmath17 its third row and third column ( i.e. , one of the third - order principal minor ) can be determined as @xmath84 , @xmath85 further gives @xmath86 . then under the conditions of @xmath87 and @xmath88 we obtain the eigenvalues of @xmath17 as @xmath89 , @xmath90 , and the eigenvalues of the reduced @xmath91 as @xmath92 . these give rise to the quantum conditional entropy @xmath93 . in fact , one can also determine @xmath86 by the argument that @xmath17 is positive semidefinite if @xmath94 for any projector @xmath95 . taking @xmath96 with @xmath97 and @xmath32 denoting transpose , we obtain @xmath98,\end{aligned}\ ] ] where @xmath99 , with @xmath100 and @xmath101 representing the real and imaginary parts of @xmath102 . one can check directly that only when @xmath86 can @xmath103 for any @xmath95 . by using the same methodology we obtain constraints imposed on the parameters of @xmath17 associated with the remaining five vertices of @xmath68 , they are : @xmath104 and @xmath105 for @xmath106 , @xmath107 for @xmath108 , @xmath87 and @xmath109 for @xmath110 . all of these correspond to physical @xmath17 with @xmath93 . on the other hand , physical states @xmath17 with @xmath58 belonging to @xmath68 can always be written as a convex combination of states with @xmath58 at the vertices of @xmath64 , so by using the concavity of the quantum conditional entropy @xcite , we obtain @xmath111 for any density matrix @xmath17 belongs to @xmath68 [ it is also possible for @xmath111 with @xmath17 lying beyond @xmath68 ] . this means that for any physical @xmath17 with negative conditional entropy , @xmath58 must belongs to the four tetrahedra separated by @xmath68 , which gives @xmath113 , and thus makes it useful for nonclassical teleportation . we would like to point out here that the negativity of the conditional von neumann entropy and the violation of the bell - chsh inequality @xcite identify different subsets of density matrices useful for teleportation ; namely , there are @xmath17 with @xmath19 but do not violate the bell - chsh inequality [ e.g. , @xmath17 of eq . ( 5 ) with @xmath114 and @xmath115 , while there are also @xmath17 which violate the bell - chsh inequality but with @xmath116 ( e.g. , the partial of the extended werner - like states @xcite ) . in fig . 1 we presented regions of the valid @xmath58 determined by @xmath19 with different @xmath69 and @xmath70 , from which one can see that for the bell - diagonal states ( i.e. , @xmath56 ) , they locate near the four vertices of @xmath59 and are symmetric with respect to the origin of @xmath68 . for the general case @xmath72 , however , the valid @xmath58 makes @xmath19 will shrink to small regions and their distribution are not symmetric with respect to the origin of @xmath68 . the geometric characterization of @xmath17 also allows us to determine fractions of different kinds of @xmath17 over the ensemble of physical @xmath17 . this can be estimated by calculating ratio of volumes of the three - dimensional spaces occupied by @xmath58 associated with different @xmath17 . here we obtained the corresponding volumes by performing monte carlo simulations , generating @xmath117 random @xmath58 uniformly distributed in the cube illustrated in fig . 1(a ) , and checking if they correspond to physical @xmath17 , if they give rise to @xmath51 and if they make @xmath19 . in this way we confirmed that @xmath118 of the bell - diagonal states @xmath17 give @xmath51 ( the volume of @xmath59 is twice that of @xmath68 ) , and about @xmath119 of the bell - diagonal @xmath17 make @xmath19 . therefore , about @xmath120 of the bell - diagonal @xmath17 useful for teleportation can be identified by negativity of the conditional entropy . using the same method , we have also performed simulation for @xmath69 and @xmath70 chosen in fig . 1(c ) and 1(d ) , and confirmed that for the former ( latter ) case about @xmath121 ( @xmath122 ) of the physical @xmath17 are useful for teleportation , among which about @xmath123 ( @xmath124 ) of them have negative quantum conditional entropy . entanglement is a precious resource for quantum computing , but it is fragile and can be easily destroyed by the environment . the measure of entanglement including the operational methods to distinguish it from the separable case , however , is a tricky problem . so it is of practical significance to find a straightforward witnessing method . here , we discuss efficiency of the new entropic uncertainty relation on witnessing entanglement in open quantum systems . we consider a system consists of two identical qubits which interact independently with their own reservoir , with the single `` qubit+reservoir '' hamiltonian given by @xcite @xmath125 where @xmath126 is the transition frequency of the qubit , and @xmath127 are the pauli raising and lowering operators . the index @xmath128 labels the reservoir field mode with frequency @xmath129 , with @xmath130 ( @xmath131 ) being the bosonic creation ( annihilation ) operator and @xmath132 being the coupling strength . when the reservoir is at zero temperature and there is no correlation between the qubit and the reservoir initially , the single - qubit reduced density matrix @xmath133 can then be determined as @xcite @xmath134 where @xmath135 in the standard basis @xmath136 , and the explicit time dependence of the single function @xmath137 contains the information on the reservoir spectral density and the coupling constants . after obtaining @xmath133 , the two - qubit density matrix @xmath138 can then be determined by the procedure presented in @xcite . here we suppose the two qubits are prepared initially in the extended werner - like ( ewl ) states @xcite @xmath139 where @xmath140 or @xmath141 , with @xmath142 and @xmath143 . the density matrix @xmath144 depends only on the chosen initial state @xmath145 and values of the function @xmath137 associated with the hamiltonian model of eq . ( 9 ) , regardless of the reservoir structure . thus in the following we ignore temporarily the explicit form of @xmath137 and consider only the dependence of the uncertainties on @xmath146 . the results obtained here thus apply to all cases where the single - qubit dynamics has the form of eq . ( 10 ) . to witness the smallest amount of entanglement @xcite , we choose the two observables as @xmath147 and @xmath148 in the following discussion . this choice of measurement operators gives the maximal complementarity between @xmath1 and @xmath2 ; that is , @xmath149 and thus @xmath150 . consider first the initial state @xmath151 , for which the nonzero elements of @xmath152 can be expressed in terms of @xmath153 ( the diagonal elements ) or @xmath154 ( the antidiagonal elements ) @xcite . then one can check directly that uncertainty of the te , bb as well as the concurrence ( a measure of entanglement ) @xcite are determined only by @xmath153 ( they are also independent of @xmath155 ) , while the other two estimates ( i.e. , me and fe ) are determined by @xmath154 . in fig . 2(a ) we plot the uncertainties versus @xmath156 ( the me and fe are plotted only with @xmath157 ) , while in fig . 3 we plot uncertainties of the me and fe versus the real and imaginary parts of @xmath146 , both for the initial state @xmath158 . clearly , when @xmath156 is larger than a critical value @xmath159 , the entanglement can always be witnessed by te , and the related entanglement region does not explicitly depend on the particular choice of the reservoir . @xmath159 depends on the chosen initial state @xmath151 , and for @xmath160 it is of about 0.9101 , which gives the entanglement region witnessed by it as @xmath161 $ ] . from fig . 3 one can see that the me and fe are determined , however , by both @xmath162 and @xmath163 , for there are @xmath164 and @xmath165 ( those with the same distance to the origin of the coordinates ) with @xmath166 but the uncertainties @xmath167 and @xmath168 ( here @xmath169 represents uncertainties estimated by me and fe ) . this means that , for this case , the entanglement regions witnessed by them are determined by the explicit structures of the reservoir and the coupling constant between the system and the reservoir . of course , if @xmath157 we still have the entanglement regions depending only on @xmath156 , and for the initial @xmath160 we obtained numerically @xmath170 , @xmath171 $ ] for the me , and @xmath172 , @xmath173 $ ] for the fe . for the initial @xmath174 , the density matrix @xmath175 is determined only by @xmath153 , therefore all the uncertainty estimates as well as the entanglement regions witnessed by them are also determined by @xmath153 , and thus are independent of the explicit structure of the reservoir . in fig . 2(b ) we give an exemplified plot of the @xmath156 dependence of different estimates for the initial state @xmath176 , for which we have @xmath177 , @xmath178 $ ] for the te , @xmath179 , @xmath180 $ ] for the me , and @xmath179 , @xmath181 $ ] for the fe . in the following we give some explicit examples of the structured reservoir to deepen our understanding of the above general arguments . * a. the sub - ohmic , ohmic and super - ohmic reservoirs * we consider first the structured reservoirs with spectral densities of the form @xcite @xmath182 with @xmath183 and @xmath184 being the dimensionless coupling constant and the cutoff frequency , which are related to the reservoir correlation time @xmath185 and the relaxation time @xmath186 ( over which the state of the system changes in the markovian limit of flat spectrum ) by @xmath187 and @xmath188 . depending on the value of @xmath189 , the reservoir is classified as sub - ohmic if @xmath190 , ohmic if @xmath191 , and super - ohmic if @xmath192 . for this kind of reservoir spectral densities , @xmath137 is determined by @xcite @xmath193 where the kernel function @xmath194 in the continuum limit the spectral density . in this work we take @xmath195 , 1 and 3 as three examples of the sub - ohmic , ohmic , and super - ohmic spectral densities . the kernel function can be integrated as @xmath196 ( @xmath197 ) for @xmath198 , and @xmath199 $ ] for @xmath195 , where @xmath200 denotes the factorial of @xmath189 , and @xmath201 . then @xmath137 can be solved numerically and the two - qubit density matrix can be derived by the procedure of @xcite . in fig . 4 we presented theoretical predictions of entanglement witness for the initial @xmath160 and @xmath202 in sub - ohmic , ohmic and super - ohmic reservoirs , with @xmath203 and @xmath204 . one can see that while the concurrence decays monotonically and disappears in the infinite time limit , the different uncertainty estimates become larger than 1 after finite timescales . thus there are critical @xmath205 after which the entanglement can not be witnessed by them . for the te , although the time interval @xmath205 is different for different spectral densities of the reservoirs and different system - reservoir coupling constants , the numerical results confirm that the entanglement regions witnessed by it agree with those predicted in the general arguments ( i.e. , @xmath161 $ ] for the initial @xmath160 , and @xmath178 $ ] for the initial @xmath202 ) . for the me and fe , the regions of entanglement being witnessed will shrink , particularly , for the initial @xmath160 , both the me and fe oscillate around 1 with increasing @xmath206 in the short - time region ( during which they are nearly overlapped ) , and the entanglement regions witnessed by them are discontinuous because here @xmath207 . for example , for parameters of fig . 4(a ) the discrete entanglement regions are @xmath208,~[0.8748,0.9491]$ ] and @xmath209 $ ] ; that is , some states with a small amount of entanglement can be witnessed , while some others with a relatively large amount of entanglement can not be witnessed . this reveals a counterintuitive fact ; that is , not the more the state is entangled , the easier it can be witnessed for some schemes . for the initial @xmath202 , the related entanglement regions are in accord with those predicted in the above general arguments , which are independent of the explicit time dependence of @xmath137 . we also examined effects of the cutoff frequency @xmath184 on entanglement witness ( for concise presentation in the paper , we did not plot them here ) , and found that for the sub - ohmic and ohmic spectral densities , @xmath205 decreases with increasing @xmath184 and their dependence on @xmath184 are weak ( @xmath210 for @xmath202 is only a slightly larger than that for @xmath160 ) . but for the super - ohmic spectral density , @xmath210 is strongly dependent on @xmath184 ; for example , for the parameters of figs . 4(e ) and 4(f ) , they increase dramatically from 8.125 and 9.55 for @xmath211 to 165.525 and 196.85 for @xmath212 . * b. the lorentzian reservoir * as the second example , we consider the structured reservoir with the lorentzian spectral density @xcite @xmath213 where @xmath214 denotes the spectral width of the reservoir and is related to the reservoir correlation time via @xmath215 , and @xmath216 is related to the relaxation time @xmath186 via @xmath217 . @xmath218 ( @xmath219 ) corresponds to the markovian ( non - markovian ) regime , and @xmath220 is the central frequency of the reservoir detuned from the transition frequency @xmath126 by an amount @xmath221 . the function @xmath137 can be derived analytically as @xcite @xmath222,\ ] ] with @xmath223^{1/2}$ ] . thus the two - qubit density matrix can also be derived analytically @xcite . to explore efficiency of entanglement witness for this kind of system - reservoir coupling , we plot in fig . 5 theoretical predictions of uncertainties in the non - markovian regime ( @xmath224 ) with the initial @xmath160 and @xmath202 . clearly , they oscillate with increasing @xmath225 , and the time regions during which the entanglement can be witnessed depend on @xmath221 and @xmath214 . but in the absence of detuning ( i.e. , @xmath226 ) , the numerical results show that the entanglement regions witnessed by different estimates are independent of @xmath214 , and are consistent with those predicted in the above general arguments , which can be understood from the fact that when @xmath226 we always have @xmath227 . introducing detuning will decrease the decay rate of entanglement , and as can be seen from figs . 5(c ) and 5(d ) , the time regions during which the entanglement can be witnessed become discontinuous , even for the te . but the entanglement regions witnessed by te remains the same ( see the green shaded regions in fig . 6 ) as that for @xmath226 . furthermore , for the initial @xmath160 , the entanglement regions witnessed by me and fe vary with the variation of @xmath214 ( their dependence on @xmath214 may be very weak for certain parameters , e.g. , @xmath228 and @xmath229 ) , which is caused by @xmath207 when @xmath230 . for the initial @xmath202 , they remain the same as those for @xmath226 , and do not depend on the parameter @xmath214 . in summary , we have studied relations between the quantum - memory - assisted entropic uncertainty principle , teleportation and entanglement witness . we proved geometrically that any two - qubit state with negative conditional von neumann entropy , which thus lowers down the upper bound of the entropic uncertainty relation , is useful for teleportation ( i.e. , @xmath51 ) . we have also examined efficiency of this new entropic uncertainty principle on witnessing entanglement in a general class of bosonic structured reservoirs , and found that the entanglement regions witnessed by te for the initial ewl state @xmath151 or that witnessed by all the three estimates te , me and fe for the initial ewl state @xmath174 are determined only by a function @xmath146 , which has no relation with the explicit form of its time dependence . these general arguments are corroborated by explicit examples of structured reservoirs with the sub - ohmic , ohmic , super - ohmic and lorentzian spectral densities . as a by - product , we also found that it is not a general result that the more the state is entangled , the easier it can be witnessed for certain chosen schemes . as the quantum - memory - assisted entropic uncertainty principle has been experimentally realized @xcite , and it is possible to simulate and control the markovian and non - markovian environments @xcite , we expect the results demonstrated in this work may be certified in future experiments with currently available technologies ; for example , by using two - level atoms confined in optical microcavities @xcite or simulated reservoirs @xcite . this work was supported by nsfc ( 11205121 , 10974247 , 11175248 ) , `` 973 '' program ( 2010cb922904 ) , nsf of shaanxi province ( 2010jm1011 ) , and the scientific research program of education department of shaanxi provincial government ( 12jk0986 ) . 50 w. heisenberg , z. phys . * 43 * , 172 ( 1927 ) . h. p. robertson , phys . rev . * 34 * , 163 ( 1929 ) . c. h. bennett and g. brassard , in _ proceedings of the ieee international conference on computers , systems and signal processing _ ( bangalore , india , 1984 ) , pp . 175 - 179 ; m. tomamichel , c. c. w. lim , n. gisin , and r. renner , nature commun . * 3 * , 634 ( 2012 ) . r. prevedel , d. r. hamel , r. colbeck , k. fisher , and k. j. resch , nature phys . * 7 * , 757 ( 2011 ) . i. biaynicki - birula and j. mycielski , commun . phys . * 44 * , 129 ( 1975 ) ; d. deutsch , phys . lett . * 50 * , 631 ( 1983 ) ; k. kraus , phys . d * 35 * , 3070 ( 1987 ) ; h. maassen and j. b. m. uffink , phys . . lett . * 60 * , 1103 ( 1988 ) . j. m. renes and j. c. boileau , phys . lett . * 103 * , 020402 ( 2009 ) . m. berta , m. christandl , r. colbeck , j. m. renes , and r. renner , nature phys . * 6 * , 659 ( 2010 ) . n. j. cerf and c. adami , phys . lett . * 79 * , 5194 ( 1997 ) ; i. devetak and a. winter , proc . a * 461 * , 207 ( 2005 ) . c. f. li , j. s. xu , x. y. xu , k. li , and g. c. guo , nature phys . * 7 * , 752 ( 2011 ) . m. tomamichel and r. renner , phys . lett . * 106 * , 110506 ( 2011 ) ; z. y. xu , w. l. yang , and m. feng , phys . rev . a * 86 * , 012113 ( 2012 ) ; p. j. coles , e - print arxiv:1203.3153 ; a. k. pati , a. r. usha devi , a. k. rajagopal , and sudha , e - print arxiv:1204.3803 . c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres , and w. k. wootters , phys . lett . * 70 * , 1895 ( 1993 ) . a. osterloh , l. amico , g. falci and r. fazio , nature * 416 * , 608 ( 2002 ) . j. cui , m. gu , l. c. kwek , m. f. santos , h. fan , and v. vedral , nature commun . * 3 * , 812 ( 2012 ) . m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _ ( cambridge university press , cambridge , uk , 2000 ) . r. horodecki , m. horodecki , and p. horodecki , phys . a * 222 * , 21 ( 1996 ) . r. horodecki and m. horodecki , phys . a * 54 * , 1838 ( 1996 ) . r. f. werner , phys . a * 40 * , 4277 ( 1989 ) ; b. bellomo , r. lo franco , and g. compagno , _ ibid . _ * 77 * , 032342 ( 2008 ) breuer and f. petruccione , _ the theory of open quantum systems _ ( oxford university press , oxford , 2002 ) . b. bellomo , r. lo franco , and g. compagno , phys . lett . * 99 * , 160502 ( 2007 ) . b. bellomo , r. lo franco , s. maniscalco , and g. compagno , phys . a * 78 * , 060302 ( 2008 ) . w. k. wootters , phys . lett . * 80 * , 2245 ( 1998 ) . a. j. leggett , s. chakravarty , a. t. dorsey , m. p. a. fisher , a. garg , and w. zwerger , rev . phys . * 59 * , 1 ( 1987 ) . q. j. tong , j. h. an , h. g. luo , and c. h. oh , j. phys . b * 43 * , 155501 ( 2010 ) . m. l. hu and h. fan , ann . * 327 * , 2343 ( 2012 ) . m. p. almeida , f. de melo , m. hor - meyll , a. salles , s. p. walborn , p. h. souto ribeiro , and l. davidovich , science * 316 * , 579 ( 2007 ) . j. s. xu , x. y. xu , c. f. li , c. j. zhang , x. b. zou , and g. c. guo , nature commun . * 1 * , 7 ( 2010 ) . b. h. liu , y. f. huang , c. f. li , g. c. guo , e. m. laine , h. p. breuer , and j. piilo , nature phys . * 7 * , 931 ( 2011 ) . k. j. vahala , nature * 424 * , 839 ( 2003 ) . i. buluta and f. nori , science * 326 * , 108 ( 2009 ) ; d. porras , f. marquardt , j. von delft , and j. i. cirac , phys . a * 78 * , 010101 ( 2008 ) .	we relate the principle of quantum - memory - assisted entropic uncertainty to quantum teleportation and show geometrically that any two - qubit state which lowers the upper bound of this uncertainty relation is useful for teleportation . we also explore the efficiency of this entropic uncertainty principle on witnessing entanglement in a general class of bosonic structured reservoirs . the entanglement regions witnessed by different estimates are determined , which may have no relation with the explicit form of the spectral density of the reservoir for certain special chosen sets of the initial states .
consider supercritical oriented percolation on @xmath6 . let @xmath7 denote the number of open paths from @xmath8 to @xmath9 . by concatenation of paths we get @xmath10 . in other words , the following superadditivity property holds : @xmath11 having in mind subadditive ergodic theorems , it seems then natural to think that , on the percolation event `` the cluster of the origin is infinite '' , the number @xmath0 of open paths with length @xmath1 starting from the origin should grow exponentially fast in @xmath1 . however , the possibility for edges to be closed implies that @xmath12 may be infinite , and therefore not integrable . this prevents from using subadditive techniques , at least in their simplest form . the growth rate of @xmath5 and related objects have already been studied . fukushima and yoshida @xcite proved that @xmath13 is almost surely strictly positive on the percolation event . lacoin @xcite proved that the straigtforward inequality @xmath14 is not always an equality . in spite of those works , to our knowledge , there was no proof of the convergence of @xmath5 in the literature . such a convergence has been obtained for a relaxed kind of percolation called @xmath15-percolation . let @xmath16 and let @xmath17 denotes the number of paths with length @xmath1 using at least @xmath18 open edges . the existence of the limit @xmath19 has been proved in comets popov vachkovskaia @xcite and in kesten sidoravicius @xcite by different methods . the present paper aims to prove that in supercritical oriented percolation , @xmath5 has an almost sure limit on the percolation event . the proof relies on essential hitting times which have been introduced in garet marchand @xcite in order to establish a shape theorem for the contact process in random environment . let us now define precisely the oriented percolation setting we work with . let @xmath3 be fixed , and let @xmath21 be the norm on @xmath22 defined by @xmath23 we consider the oriented graph whose set of sites is @xmath2 , where @xmath24 , and we put an oriented edge from @xmath25 to @xmath26 if and only if @xmath27 the set of these edges is denoted by @xmath28 . we say that @xmath29 is a _ path _ if and only if @xmath30 fix now a parameter @xmath31 $ ] , and open independently each edge with probability @xmath32 . more formally , consider the probability space @xmath33 , endowed with its borel @xmath34-algebra and the probability @xmath35 where @xmath36 stands for the bernoulli law of parameter @xmath32 . for a configuration @xmath37 , say that the edge @xmath38 is open if @xmath39 and closed otherwise . a path is said _ open _ in the configuration @xmath40 if all its edges are open in @xmath40 . for two sites @xmath41 in @xmath42 , we denote by @xmath43 the existence of an open path from @xmath44 to @xmath45 . by extension , we denote by @xmath46 the event that there exists an infinite open path starting from @xmath44 . there exists a critical probability @xmath47 such that : @xmath48 in the following , we assume @xmath49 , and we will mainly work under the following conditional probability : @xmath50 denote by @xmath0 the number of open paths of length @xmath1 emanating from @xmath51 . our main result is the following . [ theo - lebotheoun ] let @xmath49 . there exists a strictly positive constant @xmath52 such that , @xmath53-almost surely and in @xmath54 , @xmath55 we now recall some questions related to this convergence problem . first , note that @xmath56 . as noticed by darling @xcite , the sequence @xmath57 is a non - negative martingale , so there exists a non - negative random variable @xmath58 such that @xmath59\le 1.\ ] ] therefore , it is easy to see that @xmath60 so when @xmath61 , @xmath0 has the same growth rate as its expectation . in his paper @xcite , darling was seeking for conditions implying that @xmath61 . it seems that these questions have been forgotten for a while , but there is currently an increasing activity due to the links with random polymers see for example lacoin @xcite and yoshida @xcite . actually , it is not always the case that @xmath61 . let us summarize some known results : * @xmath62 . the random variable @xmath63 is @xmath53-almost surely constant ( see lacoin @xcite ) . note that a simple borel - cantelli argument ensures that @xmath64 . * @xmath65 a.s . if @xmath66 or @xmath67 ( see yoshida @xcite ) . * there exists @xmath68 $ ] such that : * * @xmath69 when @xmath70 and @xmath71 when @xmath72 . * * @xmath73 @xmath53-almost surely when @xmath74 and @xmath75 @xmath53-almost surely when @xmath76 . * * @xmath77 . * * @xmath78 if @xmath79 . + see lacoin @xcite sections 2.2 and 2.3 . * it is believed that @xmath80 and thus @xmath81 when @xmath82 . lacoin @xcite proved that the inequality is indeed strict for @xmath83-spread - out percolation for @xmath84 and @xmath83 large . in any case , it is clear that we need a proof of the existence of a limit for @xmath85 that would not require @xmath61 . our next result focuses on open paths with a prescribed slope . we first need to give a few more notations and results . oriented percolation is known as the analogue in discrete time for the contact process . usually , results are proved for one model , and it is commonly admitted that the proofs could easily be adapted to the other one . for the results concerning supercritical oriented percolation we use in this work , we will thus sometimes give the reference for the property concerning the contact process without any further explanation . we define @xmath86 as for the contact process , the growth of the sets @xmath87 is governed by a shape theorem when conditioned to survive : for every @xmath49 , there exists a norm @xmath88 on @xmath89 such that for every @xmath90 , @xmath53 almost surely , @xmath91^d \ ; \subset b_{\mu_p}(0,(1+\varepsilon)n),\ ] ] where @xmath92 . see , for the supercritical contact process , durrett @xcite or garet - marchand @xcite . for every set @xmath93 , we denote by @xmath94 the number of open paths starting from @xmath51 , with length @xmath1 and whose extremity lies in @xmath95 . [ theo - nombrean ] fix @xmath49 . there exists a concave function @xmath96,\ ] ] with the same symmetries as the grid @xmath97 , such that , for every set @xmath98 such that @xmath99 , @xmath53-almost surely , @xmath100 since @xmath101 is even and concave , the constant @xmath102 which appears in the statement of theorem [ theo - lebotheoun ] is indeed the value of the function @xmath101 at @xmath103 . by considering , in theorem [ theo - nombrean ] , the set @xmath104 for @xmath105 and for a small @xmath106 , we see that @xmath107 characterises the growth of the number of open paths with length @xmath1 and prescribed slope @xmath108 . using the very same technics of proof , one could for instance prove the following directional convergence result . if @xmath109 , denote by @xmath110 the number of open paths from @xmath51 to @xmath111 : [ coro - limdir1 ] fix @xmath49 and @xmath112 such that @xmath113 . + extract from the sequence @xmath114 the ( random ) subsequence , denoted @xmath115 , of indices @xmath116 such that @xmath117 . then @xmath53 almost surely , @xmath118 take now as a random environment a realization of oriented percolation on @xmath2 with parameter @xmath32 such that @xmath103 percolates . once this random setting is fixed , choose a random open path with length @xmath1 , uniformly among all open paths with length @xmath1 , and ask for the behavior of the extremity of this random path . more precisely , for every set @xmath98 with @xmath99 , the probability that the extremity of the random path stands in @xmath119 is @xmath120 then , theorem [ theo - nombrean ] can be rephrased as a quenched large deviations principle for the extremity of this random open path : [ theo - pgd ] fix @xmath49 . for every set @xmath98 such that @xmath99 , @xmath53-almost surely , @xmath121 here are a few open questions . * is the following statement true ? @xmath122 if the statement held , then the extremity of a random open path with length @xmath1 , uniformly chosen among open paths with length @xmath1 , would concentrate near @xmath123 . * is @xmath101 strictly concave ? this would imply the previous statement . * does @xmath101 vanish when @xmath108 tends to the boundary of @xmath124 ? the key ideas of our proofs are the following . first , in section 2 , we recall results for supercritical oriented percolation , and we build the essential hitting times . then , in section 3 , we fix a vector @xmath112 and we build an associated sequence of regenerating times @xmath125 ( see definition [ defi - sn ] ) . these random times satisfy @xmath126 and have good invariance and integrability properties with respect to @xmath53 . we can thus apply kingman s subbaditive ergodic theorem to obtain , in lemma [ lemm - ssadd ] , the existence of the following limit : @xmath127 section 4 is devoted to the proof of theorem [ theo - lebotheoun ] . the asymptotic behavior of @xmath128 should come from the `` direction '' @xmath129 in which open paths are more abundant , _ i.e. _ in the `` direction '' @xmath129 that maximizes @xmath130 . the key step to recover a full limit from the limit of a random subsequence is the continuity lemma [ lemm - limsup ] : using the coupled zone , we prove in essence that two points close in @xmath131 and reached from @xmath51 by open paths should have similar number of open paths arriving to them . finally , in section 5 , the same ideas are used to prove theorem [ theo - nombrean ] . the arguments are however more intricate . that is why we chose to present an independent proof of theorem [ theo - lebotheoun ] where to our opinion , each type of argument regenerating time , coupling appears in a simpler form . for @xmath132 , @xmath133 and any set @xmath134 , we denote by * @xmath0 the number of open paths from @xmath51 to @xmath135 , * @xmath136 the number of open paths from @xmath51 to @xmath135 that are the beginning of an infinite open path , * @xmath110 the number of open paths from @xmath51 to @xmath111 , * @xmath137 the number of open paths from @xmath51 to @xmath138 . we work with the oriented percolation model in dimension @xmath20 , as defined in the introduction . we set , for @xmath139 and @xmath109 , @xmath140 to simplify , we often write @xmath141 instead of @xmath142 . for instance , @xmath143 is the length of the longest open path starting from the origin , and the percolation event is equal to @xmath144 . first , finite open paths can not be too long ( see durrett @xcite ) : @xmath145 the set @xmath146 is called the coupled zone , and will play a central role in our proofs , by allowing to compare numbers of open paths with close extremities . as for the contact process , the growth of the sets @xmath87 and the coupled zones @xmath147 is governed by a shape theorem and related large deviations inequalities : [ prop - gd ] @xmath148 + fix @xmath49 . for every @xmath149 , there exist @xmath150 such that , @xmath151^d\;\\ \subset \;h_n + [ 0,1]^d \ ; \subset b_{\mu_p}(0,(1+\varepsilon)n)\end{array } \right ) \ge 1- ae^{-bn}.\ ] ] ( 0,-0.05 ) node @xmath152 ; ( 0,0 ) node[below]@xmath51 ; ( 0,0 ) ( 2.2,3 ) ; ( 0,0 ) ( 1.8,3 ) ; ( 0,0 ) ( -2.2,3 ) ; ( 0,0 ) ( -1.8,3 ) ; ( -2.4,2.5 ) ( 2.4,2.5 ) ; ( -2.4,0 ) ( 2.4,0 ) ; ( 1.1,2.45 ) node @xmath152 ; ( 1.1,2.5 ) node[above]@xmath111 ; ( -1.1,-0.05 ) node @xmath152 ; ( -1.1,0 ) node[below]@xmath153 ; ( -1.1,0)(-0.8,0.5)(-0.1,1)(0.5,1.5)(0.2,2)(1.1,2.5 ) ; ( 0,0)(-0.2,0.5)(0.3,1)(-0.2,1.5)(0.6,2)(1.1,2.5 ) ; if @xmath108 is in the coupled zone @xmath154 , and is reached by an open path starting from some point @xmath155 ( in blue ) , then @xmath156 ( in red ) . we now introduce the analogues , in the discrete setting of oriented percolation , of the essential hitting times used by garet marchand to study the supercritical contact process conditioned to survive in @xcite and @xcite ; we give their main properties in proposition [ prop - sigma ] . for a given @xmath109 , the essential hitting time will be a random time @xmath157 such that * @xmath53 almost surely , @xmath158 , * the associated random translation of vector @xmath159 leaves @xmath53 invariant . thus @xmath157 will be interpreted as a regenerating time of the oriented percolation conditioned to percolate . we define a set of oriented edges @xmath160 of @xmath97 in the following way : in @xmath161 , there is an oriented edge between two points @xmath162 and @xmath163 in @xmath97 if and only if @xmath164 . the oriented edge in @xmath28 from @xmath25 to @xmath26 can be identified with the couple @xmath165 . thus , we identify @xmath28 and @xmath166 . we also define , for @xmath167 , the translation @xmath168 on @xmath169 by : @xmath170 at some point , we will also need to look backwards in time . so , as set of sites , we replace @xmath42 by @xmath171 , and we introduce the following reversed time translation defined on @xmath172 by @xmath173 fix @xmath174 . we now recall the construction of the essential hitting times and the associated translations introduced in @xcite . fix @xmath109 . the essential hitting time @xmath157 is defined through a family of stopping times as follows : we set @xmath175 and we define recursively two increasing sequences of stopping times @xmath176 and @xmath177 with @xmath178 as follows : * assume that @xmath179 is defined . we set @xmath180 . + if @xmath181 , then @xmath182 is the first time after @xmath179 where @xmath108 is once again infected ; otherwise , @xmath183 . * assume that @xmath184 is defined , with @xmath185 . we set @xmath186 . + if @xmath187 , the time @xmath188 is the length of the oriented percolation cluster starting from @xmath189 ; otherwise , @xmath190 . we then set @xmath191 this quantity represents the number of steps before the success of this process : either we stop because we have just found an infinite @xmath192 , which corresponds to a time @xmath193 when @xmath108 is occupied and has infinite progeny , or we stop because we have just found an infinite @xmath194 , which says that after @xmath192 , site @xmath103 is never infected anymore . it is not difficult to see that @xmath195 and thus @xmath196 is @xmath197 almost surely finite . we define the essential hitting time @xmath157 by setting @xmath198 by construction @xmath199 on the event @xmath144 . note however that @xmath157 is not necessarily the first positive time when @xmath108 is occupied and has infinite progeny : for instance , such an event can occur between @xmath200 and @xmath201 , being ignored by the recursive construction . it can be checked that conditionally to the event @xmath202 , the process necessarily stops because of an infinite @xmath192 , and thus @xmath203 . at the same time , we define the operator @xmath204 on @xmath169 , which is a random translation , by : @xmath205 if @xmath206 is a sequence of points in @xmath97 , we also introduce the shortened notation @xmath207 for @xmath208 , we denote by @xmath209 the @xmath34-field generated by the maps @xmath210 . we denote by @xmath211 the @xmath34-field generated by the maps @xmath212 . [ prop - sigma ] fix @xmath49 and @xmath213 . 1 . suppose @xmath214 , @xmath215 . then for each @xmath216 , @xmath217 2 . the probability measure @xmath53 is invariant under @xmath218 . 3 . the random variables @xmath219 are independent under @xmath53 . 4 . suppose @xmath220 , @xmath221 , @xmath215 @xmath222 5 . for every @xmath216 , @xmath223 . there exists @xmath224 such that @xmath225 to prove a .- d . , it is sufficient to mimic the proofs of lemma 8 and corollary 9 in @xcite . the convergence e. has been proved for the contact process in @xcite , theorem 22 . the existence of exponential moments for @xmath34 has been proved for the contact process in @xcite , theorem 2 . the essential hitting times have good regenerating properties , but by construction ( see proposition [ prop - sigma ] e. ) , the vector @xmath226 lies close to the border of the percolation cone @xmath227 . we now need to build new regenerating points such that the set of directions of these points is dense inside the percolation cone . we define , for @xmath112 , a new regenerating time @xmath228 by setting @xmath229 and the associated translation : @xmath230 note that on @xmath144 , @xmath231 and @xmath232 ( with @xmath233 zeros ) . we can easily deduce from proposition [ prop - sigma ] the following properties of the time @xmath228 under @xmath53 : [ lemm - sigma ] fix @xmath49 , and @xmath112 . 1 . the probability measure @xmath53 is invariant under the translation @xmath234 . 2 . the random variables @xmath235 are independent and identically distributed under @xmath53 . 3 . the measure - preserving dynamical system @xmath236 is mixing . there exists @xmath224 such that @xmath237 we fix @xmath112 . we work under @xmath53 , and we set , for every @xmath132 , @xmath238 the points @xmath239 are the sequence of regenerating points associated to @xmath129 along which we are going to look for subadditivity properties . as , under @xmath53 , the random variables @xmath240 are independent and identically distributed with finite first moment ( see lemma [ lemm - sigma ] ) , the strong law of large numbers ensures that @xmath53-almost surely @xmath241 thus , for large @xmath1 , the point @xmath242 is not far from the line @xmath243 . to obtain directional limits along subsequences , we first apply kingman s subadditive ergodic theorem to @xmath244 for a fixed @xmath112 . [ lemm - ssadd ] fix @xmath49 and @xmath112 . there exists @xmath245 $ ] such that @xmath53-almost surely and in @xmath54 , @xmath246 fix @xmath112 . to avoid heavy notations , we omit all the dependence in @xmath129 . for instance @xmath247 and @xmath248 . note that by definition , @xmath53-almost surely , for every @xmath132 , @xmath249 and consequently , @xmath250 . for @xmath132 , we set @xmath251 let @xmath252 . note that @xmath253 . as @xmath254 counts the number of open paths from @xmath255 to @xmath256 , concatenation of paths ensures that @xmath257 which implies that @xmath258 as @xmath259 , @xmath260 the integrability of @xmath261 thus implies the integrability of every @xmath262 . so we can apply kingman s subadditive ergodic theorem . by property c. in lemma [ lemm - sigma ] , the dynamical system @xmath263 is mixing . particularly , it is ergodic , so the limit is deterministic : if we define @xmath264 we have @xmath53-almost surely and in @xmath54 : @xmath265 . + the limit of the lemma follows then directly from by setting @xmath266 finally @xmath267 . since @xmath268 @xmath53-a.s . and @xmath269 with positive probability , it follows that @xmath270 , and consequently @xmath271 . + as @xmath272 , we see that @xmath273 and that the convergence also holds in @xmath54 . we can now introduce a natural candidate for the limit in theorem [ theo - lebotheoun ] : @xmath274 indeed , at the logarithmic scale we are working with , we can expect that the dominant contribution to the number @xmath0 of open paths to level @xmath1 will be due to the number @xmath275 of open paths to level @xmath1 in the direction @xmath276 that optimizes the previous limit . note however that in our construction , @xmath129 has no real geometrical signification , but it is just a useful encoding : as said before , the asymptotic direction of the regenerating point @xmath242 in @xmath2 is @xmath277 to skip from the subsequences to the full limit , we approximate @xmath278 with a denumerable set of points : let @xmath279 [ lemm - dirdense ] for every @xmath49 , @xmath280 . note that the set @xmath281 is dense in @xmath278 . thus fix @xmath282 such that @xmath283 and consider @xmath284 then @xmath285 as @xmath1 goes to @xmath286 . finally , for @xmath112 , we denote by @xmath287 thus , for large @xmath1 , @xmath288 is the first point among the sequence of regenerating points associated to @xmath129 to be above level @xmath1 . by the renewal theory , @xmath53 almost surely , @xmath289 it is also not too far above level @xmath1 : [ lemm - sautemouton ] for every @xmath112 , there exist positive constants @xmath290 such that @xmath291 as we work in discrete time , @xmath292 . so @xmath293 as @xmath228 admits exponential moments thanks to lemma [ lemm - sigma ] , we can conclude with the markov inequality . fix @xmath49 . the proof of the almost sure convergence in theorem [ theo - lebotheoun ] is a direct consequence of the forthcoming lemmas [ lemm - liminf ] , [ lemm - limsup ] and [ lemm - pareil ] . the @xmath294 convergence follows from the remark that @xmath295 . remember that @xmath296 is defined in . take @xmath112 . note that @xmath298 is non - decreasing , and considering the increasing sequence @xmath299 , we see that , @xmath53 almost surely , for every integer @xmath1 such that @xmath300 , @xmath301 with and lemma [ lemm - ssadd ] , we deduce that @xmath53 almost surely , @xmath302 which completes the proof . fix @xmath149 and @xmath304 . we first approximate @xmath278 with a finite number of points : with lemma [ lemm - dirdense ] , we can find a finite set @xmath305 such that @xmath306 then , for @xmath1 large , we will control the number @xmath136 using these directions . we define @xmath307 as the first point in the sequence @xmath308 of regerating points associated to @xmath129 to be above level @xmath309 . using the notation introduced in , we set @xmath310 for a given @xmath311 , the law of large numbers says that @xmath312 so @xmath53 almost surely , for all @xmath1 large enough @xmath313 it follows then from the shape theorem , that @xmath53 almost surely , for all @xmath1 large enough @xmath314 the strategy is to prove that for @xmath1 large enough , for each @xmath315 , the @xmath1 first steps of an open path that goes from @xmath51 to @xmath111 and then to infinity are also the @xmath1 first steps of an open path which contributes to @xmath316 for any @xmath311 such that @xmath317 . to do so , we will use the coupled zone . note @xmath318 since @xmath319 preserves @xmath197 , we easily deduce from , proposition [ prop - gd ] and a borel cantelli argument that @xmath53 almost surely , @xmath320 holds for every @xmath1 large enough . fix @xmath323 such that @xmath324 . as holds , choose @xmath322 such that @xmath317 . since @xmath325 and @xmath326 , we know that @xmath327 . since @xmath328 , @xmath329 and @xmath320 holds , we have @xmath330 . note that @xmath331 , so @xmath332 . note also that @xmath324 implies that @xmath333 . by definition of the coupled zone , we have @xmath334 . going back to the initial orientation , it means that @xmath335 . so , if @xmath336 is a path from @xmath51 to @xmath111 , it is clear that @xmath336 is the restriction of a path that goes from @xmath51 to @xmath337 , and then to infinity . then , @xmath338 of paths from @xmath51 to infinity . green paths are the part above @xmath1 of paths from @xmath51 to infinity that meet some @xmath337 . to bound @xmath136 , we prove that each start of a red path is the start of a green path.,title="fig : " ] ( -165,5)@xmath51(-92,88)@xmath111(-92,130)@xmath339(-127,20)coupled zone of @xmath337(-131,163)asymptotic direction(-302,120)@xmath309(-292,85)@xmath1(-292,15)@xmath103 next , we use the directional limits given by lemma [ lemm - ssadd ] : @xmath53 almost surely , @xmath340 as @xmath341 , we obtain from the shape theorem that @xmath53 almost surely , for all @xmath1 large enough @xmath342 so , for @xmath1 large enough , we have @xmath53 almost surely , @xmath343 we complete the proof by letting @xmath106 go to @xmath103 . fix @xmath345 and define , for @xmath132 , the following event @xmath346 assume that @xmath347 occurs . consider a path @xmath348 from @xmath51 to @xmath135 and set @xmath349 : as @xmath350 , the event @xmath347 implies that @xmath351 . so @xmath352 contributes to @xmath353 and thus , on @xmath347 , @xmath354 the exponential estimate ensures that @xmath355 with the borel cantelli lemma , this leads to : @xmath356 by taking @xmath106 to @xmath103 , we obtain @xmath357 the proof for the inequality with @xmath358 instead of @xmath359 is identical . since @xmath360 , the reversed inequalities are obvious . recall that @xmath362 was defined in . our strategy is to prove that the identity @xmath363 defines a map on @xmath362 that is uniformly continuous on every compact subset of @xmath364 . we first refine the argument of lemma [ lemm - limsup ] implying the coupled zone : [ lemm - continuitemiam ] let @xmath365 . there exists @xmath366 such that the following holds . for every @xmath149 , for every @xmath367 , if @xmath368 then for any sequences of points @xmath369 and @xmath370 in @xmath131 , for any @xmath371 such that , @xmath53 almost surely , @xmath372 we have the following property : @xmath53 almost surely , for every @xmath1 large enough , if @xmath373 and @xmath374 , then @xmath375 . so @xmath381 almost surely , @xmath382 so @xmath381 almost surely , for every @xmath1 large enough , @xmath383 by the convergences for the @xmath384 , we know that @xmath53 almost surely , for every @xmath1 large enough , @xmath385^d \quad \forall k \in [ cn(1+{\varepsilon}-{\varepsilon}'),cn(1+{\varepsilon}+{\varepsilon } ' ) ] \\ & & \quad ( \tau\circ\theta^{\downarrow}_{(x , k)}\ge { \varepsilon } ' cn ) \rightarrow \forall m \ge { \varepsilon } ' cn \ ; b_{\mu_p}(0 , x,(1-\eta)m ) \subset \tilde k'_{m } \circ\theta^{\downarrow}_{(x , k)}\}.\end{aligned}\ ] ] with the large deviations for the coupled zone given in proposition [ prop - gd ] , there exist @xmath150 such that @xmath386 thus , the borel cantelli lemma ensures that @xmath387 assume then that @xmath388 . @xmath53 almost surely , for every @xmath1 large enough , we know that , and @xmath320 occur . assume that , for one of these large enough @xmath1 , @xmath373 and @xmath374 . note that @xmath389 so , on the event @xmath320 , as @xmath373 , we see that @xmath390 , so @xmath391 so , with , we see that @xmath392 . as @xmath393 , then @xmath394 , which gives an injection from the set of open paths from @xmath51 to @xmath395 into the set of open paths from @xmath51 to @xmath396 . for @xmath112 , we define @xmath307 as the first point in the sequence @xmath397 of regerating points associated to @xmath129 to be above level @xmath1 ( see definition ): @xmath398 the law of large numbers says that @xmath53 almost surely , @xmath399 the next lemma is a first step toward continuity . we define the following equivalence relation of the points in @xmath131 : @xmath409 lemma [ lemm - continuity ] ensures that if @xmath410 , then @xmath411 . we can thus define on the quotient set of directions @xmath362 , defined in , the following directional limit : @xmath363 lemma [ lemm - continuity ] ensures that the application @xmath412 is uniformly continuous on each @xmath413 . note that the @xmath414 given by lemma [ lemm - continuity ] gives an upper bound for its modulus of continuity . as , with lemma [ lemm - dirdense ] , @xmath415 is dense in the compact set @xmath416 , we can extend @xmath412 to any @xmath416 , and then to @xmath124 . fix @xmath417 , and consider @xmath418 . there exist @xmath419 such that @xmath420 let @xmath414 be given by lemma [ lemm - continuitemiam ] . let @xmath421 and @xmath149 . writing , for @xmath422 : @xmath423 we can choose @xmath424 such that : @xmath425 in particular , the uniform continuity of @xmath412 on @xmath426 ensures that @xmath427 by lemma [ lemm - dirdense ] , we can choose @xmath428 such that ( with the same notation ) : @xmath429 in particular , the uniform continuity of @xmath412 on @xmath430 and ensure that @xmath431 for every @xmath132 , set @xmath432 by the law of large numbers , @xmath433 and @xmath434 almost surely converge , respectively , to @xmath435 ans @xmath436 . since @xmath437 , complete convergence actually occurs ( see e.g. hsu - robbins @xcite ) , so , since the @xmath438 operators leave @xmath53 invariant , we get the following @xmath53 almost sure convergences : @xmath439 note that @xmath440 . in particular , @xmath441 using the @xmath294 convergence in the subadditive lemma [ lemm - ssadd ] , we then obtain @xmath442 for every @xmath132 , we now set @xmath443 we have the following @xmath53 almost sure convergence : @xmath444 by lemma [ lemm - continuitemiam ] we get , @xmath53 almost surely , for large enough @xmath1 : @xmath445 with the @xmath294 convergence in lemma [ lemm - ssadd ] we get that : @xmath446 with , and , we obtain @xmath447 letting @xmath106 tend to @xmath103 , we get the concavity of @xmath101 on @xmath413 . since @xmath101 is continuous , a density argument completes the proof . let @xmath450 with @xmath451 . there exists @xmath452 with @xmath453 . fix @xmath454 such that @xmath455 . by the continuity of @xmath412 , if we take @xmath106 small enough , we can also ensure that @xmath456 on @xmath457 . with lemma [ lemm - dirdense ] , we can find @xmath167 such that @xmath458 . we define @xmath307 as the first point in the sequence @xmath308 of regerating points associated to @xmath129 to be above level @xmath459 . using the notation introduced in , we set @xmath460 the law of large numbers says that @xmath461 note @xmath462 , \\ k \ge { \varepsilon}n /2 } } \ { \xi_k^0 \subset b_{\mu_p}(0 , ( 1+{\varepsilon } ) k ) \ } \circ \theta_{m}.\ ] ] since @xmath463 preserves @xmath197 , we easily deduce from , proposition [ prop - gd ] and a borel cantelli argument that @xmath53 almost surely , @xmath320 holds for @xmath1 large enough . now take @xmath1 large enough such that @xmath320 holds and , with , @xmath464 and @xmath465 , so that @xmath466 . then @xmath320 ensures that ( @xmath467 ) @xmath468 so @xmath469 , and then @xmath470 with and lemma [ lemm - ssadd ] , we deduce that @xmath53 almost surely , @xmath471 letting @xmath106 going to @xmath103 completes the proof . the proof is a refinement of that of lemma [ lemm - limsup ] . let @xmath474 . since @xmath475 is a compact subset of @xmath476 and @xmath477 is continuous on @xmath476 , one can find @xmath454 such that @xmath478 now take @xmath479 and @xmath480 as defined in the proof of lemma [ lemm - limsup ] and note @xmath481 now consider @xmath482 . since @xmath483 , for @xmath1 large enough , we can find @xmath322 such that @xmath484 . we have @xmath485 since @xmath486 , we get @xmath487 . now , following the proof of lemma [ lemm - limsup ] , for @xmath1 large enough , for each @xmath482 , the @xmath1 first steps of an open path that goes from @xmath51 to @xmath111 and then to infinity are also the @xmath1 first steps of an open path which contributes to @xmath316 for any @xmath488 , which gives @xmath489 as previously , we get @xmath490 we complete the proof by letting @xmath491 go to @xmath103 . assume that @xmath320 occurs . consider a path @xmath348 from @xmath51 to @xmath494 and set @xmath349 : as @xmath350 , the event @xmath320 implies that @xmath351 . looking backwards in time , we see that all these @xmath495 are in @xmath496 . so @xmath352 contributes to @xmath497 and thus , on @xmath320 , @xmath498 now , we first use lemma [ lemm - limsupa ] and take the @xmath359 , and then we use the continuity of @xmath412 and let @xmath106 go to @xmath103 : @xmath499 as @xmath500 , with lemma [ lemm - liminfa ] we obtain that @xmath501 this completes the proof . r. w. r. darling . the lyapunov exponent for products of infinite - dimensional random matrices . in _ lyapunov exponents ( oberwolfach , 1990 ) _ , volume 1486 of _ lecture notes in math . _ , pages 206215 . springer , berlin , 1991 .	we study the number @xmath0 of open paths of length @xmath1 in supercritical oriented percolation on @xmath2 , with @xmath3 . we prove that on the percolation event @xmath4 , @xmath5 almost surely converges to a positive deterministic constant . we also study the existence of directional limits . the proof relies on the introduction of adapted sequences of regenerating times , on subadditive arguments and on the properties of the coupled zone in supercritical oriented percolation .
let @xmath0 be a compact , connected , orientable surface of genus @xmath1 with @xmath2 boundary components . the mapping class group , @xmath13 , of @xmath0 is the group of isotopy classes of orientation preserving homeomorphisms of @xmath0 . the pure mapping class group , @xmath14 , is the subgroup of @xmath13 consisting of isotopy classes of homeomorphisms which preserve each boundary component of @xmath0 . the extended mapping class group , @xmath4 , of @xmath0 is the group of isotopy classes of all ( including orientation reversing ) homeomorphisms of @xmath0 . + assume that either @xmath5 and @xmath6 or @xmath7 and @xmath15 . the main results of the paper are the following two theorems : [ theorem1 ] a simplicial map , @xmath16 , is superinjective if and only if @xmath17 is induced by a homeomorphism of @xmath0 . [ theorem2 ] let @xmath10 be a finite index subgroup of @xmath4 and @xmath12 be an injective homomorphism , @xmath18 . then @xmath12 is induced by a homeomorphism of the surface @xmath0 ( i.e. for some @xmath19 , @xmath20 for all @xmath21 ) and @xmath12 has a unique extension to an automorphism of @xmath4 . these two theorems were proven for closed , connected , orientable surfaces of genus at least 3 by the author in @xcite . they were motivated by the following theorems of ivanov and the theorem of ivanov and mccarthy . * ( ivanov ) @xcite * [ theorema ] let @xmath0 be a compact , orientable surface possibly with nonempty boundary . suppose that the genus of @xmath0 is at least 2 . then , all automorphisms of @xmath22 are given by elements of @xmath4 . more precisely , if @xmath0 is not a closed surface of genus 2 , then @xmath23 . if @xmath0 is a closed surface of genus 2 , then @xmath24 . * ( ivanov ) @xcite * [ theoremb ] let @xmath0 be a compact , orientable surface possibly with nonempty boundary . suppose that the genus of @xmath0 is at least 2 and @xmath0 is not a closed surface of genus 2 . let @xmath25 be finite index subgroups of @xmath26 . then , all isomorphisms @xmath27 have the form @xmath28 . * ( ivanov , mccarthy ) @xcite * [ theoremc ] let @xmath0 and @xmath29 be compact , connected , orientable surfaces . suppose that the genus of @xmath0 is at least 2 , @xmath29 is not a closed surface of genus 2 , and the maxima of ranks of abelian subgroups of @xmath13 and @xmath30 differ by at most one . if @xmath31 is an injective homomorphism , then @xmath32 is induced by a homeomorphism @xmath33 , ( i.e. @xmath34)=[hgh^{-1}]$ ] , for every orientation preserving homeomorphism @xmath35 ) . for the surfaces that we consider in this paper , theorem [ theorem1 ] generalizes ivanov s theorem [ theorema ] , theorem [ theorem2 ] generalizes ivanov s theorem [ theoremb ] and ivanov and mccarthy s theorem [ theoremc ] ( in the case when the surfaces are the same ) . + in section 2 , we give some properties of superinjective simplicial maps of the complex of curves , @xmath3 . + in section 3 , we prove that a superinjective simplicial map @xmath36 , induces an injective simplicial map on the complex of arcs , @xmath37 , and by using this map we prove that @xmath17 is induced by a homeomorphism of @xmath0 . + in section 4 , we prove that if @xmath10 is a finite index subgroup of @xmath4 and @xmath38 is an injective homomorphism , then @xmath12 induces a superinjective simplicial map of @xmath3 , and by using this map we prove that @xmath12 is induced by a homeomorphism of @xmath0 . let @xmath0 be a compact , connected , orientable surface of genus @xmath1 with @xmath2 boundary components . we assume that either @xmath5 and @xmath39 or @xmath7 and @xmath8 . since our main results have already been proven for @xmath7 and @xmath40 in @xcite , we will assume that @xmath41 when @xmath7 throughout the paper . + a _ circle _ on @xmath0 is a properly embedded image of an embedding @xmath42 . a circle on @xmath0 is said to be _ nontrivial _ ( or _ essential _ ) if it does not bound a disk and it is not homotopic to a boundary component of @xmath0 . let @xmath43 be a collection of pairwise disjoint circles on @xmath0 . the surface obtained from @xmath0 by cutting along @xmath43 is denoted by @xmath44 . a nontrivial circle @xmath45 on @xmath0 is called _ ( k , m)-separating _ ( or a _ ( k , m ) circle _ ) if the surface @xmath46 is disconnected and one of its components is a genus @xmath47 surface with @xmath48 boundary components , where @xmath49 , @xmath50 . if @xmath51 is connected , then @xmath45 is called _ nonseparating_. let @xmath52 denote the set of isotopy classes of nontrivial circles on @xmath0 . the _ geometric intersection number _ @xmath53 of @xmath54 is the minimum number of points of @xmath55 where @xmath56 and @xmath57 . + the _ complex of curves _ , @xmath3 , on @xmath0 , introduced by w. harvey @xcite , is an abstract simplicial complex with vertex set @xmath52 such that a set of @xmath58 vertices @xmath59 forms an @xmath60 simplex if and only if @xmath61 have pairwise disjoint representatives . + * definition : * a simplicial map @xmath62 is called * superinjective * if the following condition holds : if @xmath63 are two vertices in @xmath3 such that @xmath64 , then @xmath65 . + in this section , we show some properties of superinjective simplicial maps of @xmath22 . the proofs of the lemmas 2.1 - 2.4 are similar to the proofs of the corresponding lemmas in the closed case which are given in @xcite . so , we will only state them here . [ injective ] a superinjective simplicial map , @xmath62 , is injective . [ conbyanedge ] let @xmath66 be two distinct vertices of @xmath3 , and let @xmath67 be a superinjective simplicial map . then , @xmath68 and @xmath69 are connected by an edge in @xmath3 if and only if @xmath70 and @xmath71 are connected by an edge in @xmath3 . let @xmath72 be a set of pairwise disjoint circles on @xmath0 . @xmath72 is called a _ pair of pants decomposition _ of @xmath0 , if @xmath73 is a disjoint union of genus zero surfaces with three boundary components , pairs of pants . a pair of pants of a pants decomposition is the image of one of these connected components under the quotient map @xmath74 together with the image of the boundary components of this component . the image of the boundary of this component is called the _ boundary of the pair of pants_. a pair of pants is called _ embedded _ if the restriction of @xmath75 to the corresponding component of @xmath73 is an embedding . an ordered set @xmath76 is called an _ ordered pair of pants decomposition _ of @xmath0 if @xmath77 is a pair of pants decomposition of @xmath0 . [ imageofpantsdecomp ] let @xmath16 be a superinjective simplicial map . let @xmath72 be a pair of pants decomposition of @xmath0 . then , @xmath17 maps the set of isotopy classes of elements of @xmath72 to the set of isotopy classes of elements of a pair of pants decomposition , @xmath78 , of @xmath0 . let @xmath72 be a pair of pants decomposition of @xmath0 . let @xmath45 and @xmath79 be two distinct elements in @xmath72 . then , @xmath45 is called _ adjacent _ to @xmath79 w.r.t . @xmath72 iff there exists a pair of pants in @xmath72 which has @xmath45 and @xmath79 on its boundary . + * remark : * let @xmath72 be a pair of pants decomposition of @xmath0 . let @xmath80 $ ] be the set of isotopy classes of elements of @xmath72 . let @xmath81 $ ] . we say that @xmath68 is adjacent to @xmath69 w.r.t . @xmath80 $ ] if the representatives of @xmath68 and @xmath69 in @xmath72 are adjacent w.r.t . @xmath72 . by lemma [ imageofpantsdecomp ] , @xmath17 gives a correspondence on the isotopy classes of elements of pair of pants decompositions of @xmath0 . @xmath82)$ ] is the set of isotopy classes of elements of a pair of pants decomposition which corresponds to @xmath72 , under this correspondence . [ adjacent ] let @xmath67 be a superinjective simplicial map . let @xmath72 be a pair of pants decomposition of @xmath0 . then , @xmath17 preserves the adjacency relation for two circles in @xmath72 , i.e. if @xmath83 are two circles which are adjacent w.r.t . @xmath72 and @xmath84=\alpha , [ b]=\beta$ ] , then @xmath85 are adjacent w.r.t . @xmath82)$ ] . let @xmath72 be a pair of pants decomposition of @xmath0 . a curve @xmath86 is called a _ 4-curve _ in @xmath72 , if there exist four distinct circles in @xmath72 , which are adjacent to @xmath87 w.r.t . @xmath72 . [ division ] let @xmath67 be a superinjective simplicial map . let @xmath45 be a @xmath88-separating circle on @xmath0 , where @xmath89 . let @xmath90 be the subsurfaces of @xmath0 s.t . @xmath91 has genus @xmath47 and has @xmath45 as its boundary , and @xmath92 . let @xmath93)$ ] . then @xmath94 is a @xmath88-separating circle and there exist subsurfaces @xmath95 of @xmath0 s.t . @xmath96 has genus @xmath47 and has @xmath94 as its boundary , @xmath97 , @xmath98 and @xmath99 . let @xmath45 be a @xmath88-separating circle on @xmath0 and @xmath89 . let @xmath91 be the subsurface of @xmath0 of genus @xmath47 which has @xmath45 as its boundary and @xmath92 . if @xmath100 , we choose a pair of pants decomposition @xmath101 of @xmath91 as shown in figure 1 , ( i ) and then complete @xmath102 to a pair of pants decomposition @xmath72 of @xmath0 in any way we like . if @xmath103 , we choose a pair of pants decomposition @xmath104 of @xmath91 and then complete @xmath102 to a pair of pants decomposition @xmath72 of @xmath0 such that each of @xmath105 is a 4-curve in @xmath72 for @xmath106 and @xmath107 are the boundary components of a pair of pants of @xmath108 . in figure 1 ( ii ) , we show how to choose @xmath108 when @xmath109 . in the other cases , when @xmath110 or @xmath111 , a similar pair of pants decomposition of @xmath91 can be chosen . + [ a1 ] = 1.85 in = 2.85 in let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath112 be the representative of @xmath113)$ ] which is in @xmath78 , for @xmath114 , and @xmath94 be the representative of @xmath115)$ ] which is in @xmath78 . let @xmath116 by using lemma 2.4 and following the proof of lemma 3.5 in [ 3 ] , we can see that @xmath94 is a @xmath88-separating circle , there exist subsurfaces @xmath95 of @xmath0 s.t . @xmath96 has genus @xmath47 and has @xmath94 as its boundary , @xmath117 and @xmath118 is a pair of pants decomposition of @xmath96 . let @xmath119 and @xmath120 . then @xmath121 are pair of pants decompositions of @xmath122 respectively as @xmath123 are pair of pants decompositions of @xmath124 respectively . + now , let @xmath68 be a vertex in @xmath125 . then , either @xmath126 $ ] or @xmath68 has a nonzero geometric intersection with an element of @xmath127 $ ] . in the first case , clearly @xmath128 since elements of @xmath127 $ ] correspond to elements of @xmath129 \subseteq \mathcal{c}(r_1')$ ] . in the second case , since @xmath17 preserves zero and nonzero geometric intersection ( since @xmath17 is superinjective ) and @xmath68 has zero geometric intersection with the elements of @xmath130 $ ] and @xmath84 $ ] , and nonzero intersection with an element of @xmath127 $ ] , @xmath70 has zero geometric intersection with elements of @xmath131 $ ] and @xmath132 $ ] , and nonzero intersection with an element of @xmath129 $ ] . then , @xmath133 . hence , @xmath134 . the proof of @xmath99 is similar . [ separating1 ] let @xmath67 be a superinjective simplicial map . then , @xmath17 sends the isotopy class of a nonseparating circle to the isotopy class of a nonseparating circle . let @xmath135 be a nonseparating circle . let s choose a separating ( 2 , 1 ) circle , @xmath45 on @xmath0 s.t . @xmath135 is in genus 2 subsurface , @xmath91 , bounded by @xmath45 . then , @xmath135 can be completed to a pants decomposition @xmath136 on @xmath91 , where @xmath137 as in figure 1 , ( i ) . then we can complete @xmath138 to a pair of pants decomposition , @xmath72 , on @xmath0 in any way we like . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath112 be the representative of @xmath113)$ ] which is in @xmath78 , for @xmath139 , and @xmath94 be the representative of @xmath115)$ ] which is in @xmath78 . let @xmath140 by lemma 2.5 , @xmath94 is a ( 2,1 ) separating circle bounding a subsurface , say @xmath96 , and @xmath118 is a pants decomposition on @xmath96 . then , by using the adjacency relation between the elements of @xmath141 it is easy to see that @xmath142)$ ] is a nonseparating circle . + for every 4-curve @xmath87 in a pants decomposition @xmath72 , there exist two pairs of pants @xmath143 and @xmath144 of @xmath72 having @xmath87 as one of their boundary components . let @xmath145 . the boundary of @xmath146 consists of four distinct curves , which are adjacent to @xmath87 w.r.t . @xmath72 . let @xmath67 be a superinjective simplicial map . then , @xmath17 sends the isotopy class of a @xmath147-separating circle to the isotopy class of a @xmath147-separating circle . let @xmath135 be a @xmath147-separating circle . a @xmath147-separating circle exists if the number of boundary components , @xmath2 , of @xmath0 , is at least 2 . so , we assume that @xmath6 . if @xmath148 , then it is easy to see that a circle is a @xmath147-separating circle iff it is a @xmath149 separating circle on @xmath0 . since @xmath135 is a @xmath147 separating circle , @xmath135 is a @xmath149 separating circle on @xmath0 . then , by lemma [ division ] , @xmath150)$ ] has a @xmath149 representative which is a @xmath147-separating circle on @xmath0 . + [ a0 ] = 3.5 in let @xmath45 be a @xmath149-separating circle on @xmath0 which is disjoint from @xmath135 . let @xmath91 be a subsurface of @xmath0 of genus @xmath1 having @xmath45 as its boundary and @xmath151 . let s choose a pair of pants decomposition @xmath108 of @xmath91 as in lemma [ division ] . if the @xmath152 we choose @xmath108 as in figure 1 ( i ) , if @xmath7 , we choose @xmath108 as in figure 1 , ( ii ) . + assume that @xmath153 for some @xmath154 and @xmath155 . + ( i ) if @xmath156 , then we choose a pants decomposition @xmath72 on @xmath0 s.t . @xmath157 where @xmath158 is a @xmath147 curve in @xmath72 . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath159 be the representatives of @xmath115 ) , \lambda([c ] ) , \lambda([c_1])$ ] in @xmath78 respectively . let @xmath160 . by lemma [ division ] and lemma 2.3 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath1 and @xmath161 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath162 and @xmath163 is a pants decomposition for @xmath164 . by using lemma 2.4 , we can see that @xmath165 is a genus @xmath1 surface having @xmath166 as its boundary components . since @xmath165 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , each of @xmath166 has to be a @xmath147 curve . + ( ii ) if @xmath167 , then we choose a pants decomposition @xmath72 on @xmath0 s.t . @xmath168 where , @xmath169 are 4-curves and @xmath170 are @xmath147 curves in @xmath72 , as shown in figure 2 ( for @xmath171 ) . in the other cases , a similar pair of pants decomposition of @xmath0 can be chosen . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath112 be the representative of @xmath113)$ ] in @xmath78 for @xmath172 , @xmath173 be the representatives of @xmath115 ) , \lambda([c])$ ] in @xmath78 respectively , @xmath174 be the representative of @xmath175)$ ] in @xmath78 for @xmath176 , @xmath177 be the representative of @xmath178)$ ] in @xmath78 for @xmath179 . let @xmath180 , @xmath181 . + by lemma [ division ] and lemma 2.3 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath1 and @xmath161 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath182 and @xmath183 are pants decompositions for @xmath95 respectively . by using lemma 2.4 , we can see that @xmath184 is a genus @xmath1 surface having @xmath185 as its boundary components . since @xmath186 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , each of @xmath187 has to be a @xmath147 curve . hence , @xmath188 is a @xmath147 curve . + assume that @xmath189 for some @xmath154 and @xmath190 . + ( i ) if @xmath191 , then @xmath192 is a pants decomposition on @xmath0 . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath173 be the representatives of @xmath115 ) , \lambda([c])$ ] in @xmath78 respectively . let @xmath193 . by lemma [ division ] and lemma 2.3 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath1 and @xmath161 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath182 and @xmath163 is a pants decomposition for @xmath164 . since @xmath94 is adjacent to @xmath188 in @xmath78 , there exists a pair of pants @xmath194 in @xmath164 having @xmath94 and @xmath188 on its boundary . by using lemma 2.4 , we can see that @xmath195 is a genus @xmath1 surface with two boundary components . one of the boundary components is @xmath188 . since @xmath195 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , @xmath188 has to be a @xmath147 curve . + [ ? ? ] = 3.33 in \(ii ) if @xmath156 , then we choose a pants decomposition @xmath72 on @xmath0 s.t . @xmath196 where @xmath197 are as in figure 3 . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath198 be the representatives of @xmath115 ) , \lambda([c ] ) , \lambda([c_1 ] ) , \lambda([b_1])$ ] in @xmath78 respectively . let @xmath199 . by lemma [ division ] and lemma 2.3 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath1 and @xmath161 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath97 and @xmath163 is a pants decomposition for @xmath164 . by using lemma 2.4 , we can see that @xmath165 is a genus @xmath1 surface having @xmath200 as its boundary components . since @xmath201 is adjacent to @xmath158 in @xmath72 , @xmath202 is adjacent to @xmath203 in @xmath78 , then there exists a pair of pants @xmath194 having @xmath202 and @xmath203 on its boundary . then , since @xmath204 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , each of @xmath166 has to be a @xmath147 curve . + ( iii ) if @xmath167 , then we choose a pair of pants decomposition @xmath72 on @xmath0 s.t . @xmath205 , @xmath206 where @xmath108 is a pair of pants decomposition as before , @xmath207 are 4-curves , @xmath201 is a 3-curve and @xmath170 are @xmath147 curves in @xmath72 as shown in figure 3 ( for @xmath208 ) . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath112 be the representative of @xmath113)$ ] in @xmath78 for @xmath172 , @xmath173 be the representatives of @xmath115 ) , \lambda([c])$ ] in @xmath78 respectively , @xmath174 be the representative of @xmath175)$ ] in @xmath78 for @xmath179 , @xmath177 be the representative of @xmath178)$ ] in @xmath78 for @xmath179 . let @xmath180 , @xmath209 . let @xmath95 be the subsurfaces of @xmath0 of genus @xmath1 and @xmath161 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath162 and @xmath163 is a pants decomposition for @xmath164 . by using lemma 2.4 , we can see that @xmath210 is a genus @xmath1 subsurface having @xmath211 as its boundary components . notice that @xmath202 is adjacent to @xmath212 in @xmath78 and @xmath212 live in @xmath213 . then , since @xmath201 is adjacent to @xmath158 w.r.t . @xmath72 , there is a pair of pants @xmath194 having @xmath202 and @xmath203 on its boundary . then , @xmath214 is a genus @xmath1 subsurface having @xmath215 as its boundary components . since @xmath216 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , each of @xmath185 has to be a @xmath147 curve . hence , @xmath188 is a @xmath147 curve . + let @xmath68 , @xmath69 be two distinct vertices in @xmath3 . we call @xmath217 to be a peripheral pair in @xmath3 if they have disjoint nonseparating representatives @xmath218 respectively such that @xmath218 and a boundary component of @xmath0 bound a pair of pants in @xmath0 . ] let @xmath67 be a superinjective simplicial map and @xmath219 be a peripheral pair in @xmath3 . then , @xmath220 is a peripheral pair in @xmath3 . let @xmath218 be disjoint nonseparating representatives of @xmath66 respectively such that @xmath218 and a boundary component of @xmath0 bound a pair of pants in @xmath0 . + we will first prove the lemma when @xmath7 . let @xmath45 be a @xmath221-separating circle which is disjoint from @xmath87 and @xmath222 . let @xmath91 be a subsurface of @xmath0 of genus @xmath223 having @xmath45 as its boundary and @xmath151 . let s choose a pair of pants decomposition @xmath108 of @xmath91 as in lemma 2.5 ( notice that @xmath108 was chosen depending on @xmath1 in lemma 2.5 ) . + [ ? ? ? ? ? ] = 3.45 in assume that @xmath224 . then @xmath225 is a pair of pants decomposition on @xmath0 ( see figure 4 , ( i ) ) . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath226 be the representatives of @xmath115 ) , \lambda([x ] ) , \lambda([y])$ ] in @xmath78 respectively . let @xmath118 be the set of elements in @xmath78 which corresponds to @xmath108 , and let @xmath227 . since @xmath7 , by lemma 2.5 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath223 and @xmath228 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath182 and @xmath229 is a pants decomposition for @xmath164 . since @xmath45 is adjacent to @xmath87 and @xmath222 w.r.t . @xmath72 in @xmath230 , @xmath94 has to be adjacent to @xmath231 and @xmath232 w.r.t . @xmath78 in @xmath164 . so , there exists a pair of pants @xmath194 in @xmath164 which has @xmath226 on its boundary . then , @xmath195 is a genus @xmath223 surface with two boundary components @xmath233 . then , since @xmath195 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , and @xmath0 is a genus @xmath1 surface with 1 boundary component , there has to be a pair of pants having @xmath234 and the boundary component of @xmath0 on its boundary . this proves the lemma when @xmath224 . + assume that @xmath148 . then @xmath235 ( see figure 4 , ( ii ) ) is a pair of pants decomposition on @xmath0 where @xmath236 is a nonseparating curve , @xmath237 are the boundary components of a pair of pants , and @xmath238 , [ z])$ ] is a peripheral pair . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath239 be the representatives of @xmath115 ) , \lambda([x ] ) , \lambda([y ] ) , \lambda([z])$ ] in @xmath78 respectively . let @xmath118 be the set of elements in @xmath78 which corresponds to @xmath108 , and let @xmath181 . + by lemma 2.5 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath223 and @xmath228 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath182 and @xmath240 is a pants decomposition for @xmath164 . since @xmath45 is adjacent to @xmath87 and @xmath236 w.r.t . @xmath72 in @xmath230 , @xmath94 has to be adjacent to @xmath231 and @xmath241 w.r.t . @xmath78 in @xmath164 . so , there exists a pair of pants @xmath194 in @xmath164 which has @xmath242 on its boundary . then , @xmath195 is a genus @xmath223 surface with two boundary components @xmath243 . then , since @xmath87 is adjacent to @xmath222 in @xmath230 , @xmath231 is adjacent to @xmath232 in @xmath164 . so , there exists a pair of pants @xmath244 having @xmath234 on its boundary . then , @xmath245 is a genus @xmath223 surface with three boundary components . let @xmath246 be the boundary component of @xmath244 which is different from @xmath247 . since @xmath248 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , and @xmath241 and @xmath232 are nonseparating circles by lemma 2.6 , @xmath246 is a boundary component of @xmath0 . this proves the lemma for @xmath148 . + assume that @xmath189 and @xmath190 . + ( i ) if @xmath249 , we choose a pants decomposition @xmath72 on @xmath0 s.t . @xmath250 where @xmath236 is a 4-curve , @xmath158 is a @xmath147 curve in @xmath72 , @xmath237 bound a pair of pants in @xmath72 and @xmath251 bound a pair of pants in @xmath72 . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath252 be the representatives of @xmath115 ) , \lambda([x ] ) , \lambda([y ] ) , \lambda([z ] ) , \lambda([c_1])$ ] in @xmath78 respectively . let @xmath118 be the set of elements in @xmath78 which corresponds to @xmath108 , and let @xmath253 . + by lemma 2.5 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath223 and @xmath228 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath182 and @xmath254 is a pants decomposition for @xmath164 . since @xmath45 is adjacent to @xmath87 and @xmath236 w.r.t . @xmath72 in @xmath230 , @xmath94 has to be adjacent to @xmath231 and @xmath241 w.r.t . @xmath78 in @xmath164 . so , there exists a pair of pants @xmath194 in @xmath164 which has @xmath242 on its boundary . then , @xmath255 is a genus @xmath223 surface with two boundary components @xmath243 . then , since @xmath236 is adjacent to @xmath222 and @xmath158 w.r.t . @xmath72 , @xmath241 is adjacent to @xmath232 and @xmath203 w.r.t . @xmath78 in @xmath164 . so , there exists a pair of pants @xmath244 in @xmath164 which has @xmath256 on its boundary . then , @xmath245 is a genus @xmath223 surface with three boundary components @xmath257 . since @xmath158 is a @xmath147 curve in @xmath72 , @xmath203 is a @xmath147 curve by lemma 2.7 . then , since @xmath245 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , there has to be a pair of pants containing @xmath231 , @xmath232 and a boundary component of @xmath0 as its boundary components . + ( ii ) if @xmath258 , we choose a pair of pants decomposition @xmath72 on @xmath0 such that @xmath259 where @xmath260 are 4-curves and @xmath261 are @xmath147 curves in @xmath72 and @xmath237 bound a pair of pants in @xmath72 . in figure 5 ( i ) , we show how to choose @xmath72 when @xmath262 . in the other cases , a similar pair of pants decomposition of @xmath0 can be chosen . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath263 be the representatives of @xmath115 ) , \lambda([x ] ) , \lambda([y ] ) , \lambda([z])$ ] in @xmath78 respectively , @xmath174 be the representative of @xmath175)$ ] in @xmath78 for @xmath179 and @xmath177 be the representative of @xmath178)$ ] in @xmath78 for @xmath264 . let @xmath118 be the set of elements in @xmath78 which corresponds to @xmath108 , and let @xmath265 . + by lemma [ division ] and lemma 2.3 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath223 and @xmath228 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath182 and @xmath163 is a pants decomposition for @xmath164 . we can see that @xmath266 is a genus @xmath223 surface having @xmath267 as its boundary components . since @xmath268 are @xmath147 curves in @xmath72 , @xmath269 are @xmath147 curves by lemma 2.7 . then , since @xmath266 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , there has to be a pair of pants containing @xmath231 and @xmath232 and a boundary component of @xmath0 as its boundary components . + [ ? ? ? ? ] assume that @xmath153 , and @xmath155 . let s choose a pants decomposition @xmath72 on @xmath0 s.t . @xmath270 where , @xmath271 are 4-curves , @xmath201 is a 3-curve and @xmath268 are @xmath147 curves in @xmath72 . in figure 6 ( i ) , we show how to choose @xmath72 when @xmath272 . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath239 be the representatives of @xmath115 ) , \lambda([x ] ) , \lambda([y ] ) , \lambda([z])$ ] in @xmath78 respectively , @xmath174 be the representative of @xmath175)$ ] in @xmath78 for @xmath179 , @xmath177 be the representative of @xmath178)$ ] in @xmath78 for @xmath179 . let @xmath118 be the set of elements in @xmath78 which corresponds to @xmath108 , and let @xmath227 . by lemma [ division ] and lemma 2.3 , there exist subsurfaces @xmath95 of @xmath0 of genus @xmath223 and @xmath228 respectively s.t . @xmath96 has @xmath94 as its boundary , @xmath182 and @xmath163 is a pants decomposition for @xmath164 . + ( i ) if @xmath273 , @xmath274 is a genus @xmath223 surface having @xmath275 as its boundary components . since @xmath201 is adjacent to @xmath158 w.r.t . @xmath72 , @xmath202 is adjacent to @xmath203 w.r.t . @xmath78 , so there exists a pair of pants @xmath276 in @xmath277 s.t . @xmath276 contains @xmath278 on its boundary . since @xmath158 is a @xmath147 curve in @xmath72 , @xmath203 is a @xmath147 curve by lemma 2.7 . then , since @xmath279 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , and @xmath231 and @xmath232 are nonseparating circles , there has to be a pair of pants containing @xmath231 and @xmath232 and a boundary component of @xmath0 as its boundary components . + ( ii ) if @xmath167 , we can see that @xmath280 is a genus @xmath223 surface having @xmath281 as its boundary components . since @xmath201 is adjacent to @xmath158 w.r.t . @xmath72 , @xmath202 is adjacent to @xmath203 w.r.t . @xmath78 , so there exists a pair of pants @xmath276 in @xmath282 s.t . @xmath276 contains @xmath278 on its boundary . since @xmath170 are @xmath147 curves in @xmath72 , @xmath283 are @xmath147 curves by lemma 2.7 then , since @xmath284 contains @xmath78 and @xmath78 is a pants decomposition of @xmath0 , and @xmath231 and @xmath232 are nonseparating circles , there has to be a pair of pants containing @xmath231 and @xmath232 and a boundary component of @xmath0 as its boundary components . this proves the lemma when @xmath7 . + when @xmath5 the proof is similar . we have @xmath6 in this case . instead of using a separating curve @xmath45 , we use pair of pants decompositions as given ( for special cases ) in figure 5 ( ii ) and figure 6 ( ii ) . by the proof of lemma 2.5 , there are pairwise disjoint representatives @xmath285 of @xmath286 ) , \lambda([a_2 ] ) , \lambda([z ] ) , \lambda([x])$ ] and a subsurface @xmath96 of genus 1 with two boundary components @xmath231 and @xmath241 such that @xmath231 , @xmath287 , @xmath288 bound a pair of pants in @xmath96 , and @xmath241 , @xmath287 , @xmath288 bound a pair of pants in @xmath96 . then , we follow the proof of the case when @xmath7 . + let @xmath289 be a sphere with @xmath47 holes and @xmath111 . a circle @xmath45 on @xmath289 is called an _ n - circle _ if @xmath45 bounds a disk with @xmath58 holes on @xmath289 where @xmath155 . if @xmath45 is a 2-circle on @xmath289 , then there exists up to isotopy a unique nontrivial embedded arc @xmath94 on the two - holed disk component of @xmath290 joining the two holes in this disk . if @xmath45 and @xmath79 are two 2-circles on @xmath289 such that the corresponding arcs @xmath94 , @xmath291 can be chosen to meet exactly at one common end point , and @xmath292 , \beta = [ b]$ ] , then @xmath293 is called a _ simple pair_. a _ pentagon _ in @xmath294 is an ordered 5-tuple @xmath295 , defined up to cyclic permutations , of vertices of @xmath294 such that @xmath296 for @xmath297 and @xmath298 otherwise , where @xmath299 . a vertex in @xmath294 is called an _ n - vertex _ if it has a representative which is an n - circle on @xmath289 . let @xmath300 be the interior of @xmath289 . there is a natural isomorphism @xmath301 which respects the above notions and the corresponding notions in @xcite . using this isomorphism , we can restate a theorem of korkmaz as follows : [ korkmaz ] * ( korkmaz ) @xcite * let @xmath289 be a sphere with @xmath58 holes and @xmath302 . let @xmath66 be two 2-vertices of @xmath303 . then @xmath304 is a simple pair iff there exist vertices @xmath305 of @xmath294 satisfying the following conditions : + ( i ) @xmath306 is a pentagon in @xmath294 , + ( ii ) @xmath307 and @xmath308 are 2-vertices , @xmath309 is a 3-vertex and @xmath310 and @xmath311 are k - vertices for @xmath312 , + ( iii ) @xmath313 , @xmath314 , @xmath315 , and @xmath316 are codimension - zero simplices . [ separating ] let @xmath67 be a superinjective simplicial map . then , @xmath17 sends the isotopy class of a @xmath317-separating circle to the isotopy class of a @xmath317-separating circle , where @xmath318 , @xmath50 . let @xmath292 $ ] where @xmath45 is a @xmath317-separating circle where @xmath319 , @xmath50 . + case 1 : assume the genus of @xmath0 is at least 3 . then @xmath45 separates a subsurface of genus at least 2 . so , it is enough to consider the cases when @xmath320 . if @xmath321 , then the lemma follows by lemma 2.5 . assume that @xmath322 . let @xmath91 be a subsurface of @xmath0 of genus @xmath47 with @xmath48 boundary components which has @xmath45 as one of its boundary components . let s choose a pair of pants decomposition @xmath323 of @xmath91 where @xmath324 are 4-curves and @xmath325 is a peripheral pair for @xmath326 as shown in figure 7 ( i ) ( for @xmath327 ) . then , we complete @xmath102 to a pair of pants decomposition @xmath72 of @xmath0 in any way we like . by lemma [ imageofpantsdecomp ] , we can choose a pair of pants decomposition , @xmath78 , of @xmath0 such that @xmath82 ) = [ p']$ ] . + let @xmath112 be the representative of @xmath113)$ ] which is in @xmath78 for @xmath328 and @xmath94 be the representative of @xmath115)$ ] which is in @xmath78 . let @xmath174 be the representative of @xmath175)$ ] which is in @xmath78 for @xmath329 . since @xmath105 is a 4-curve in @xmath72 , for @xmath328 , by lemma [ adjacent ] and lemma [ injective ] , @xmath112 is a 4-curve in @xmath78 , for @xmath328 . since @xmath325 is a peripheral pair , @xmath330 is a peripheral pair for @xmath326 by lemma 2.8 . then , there exist distinct pair of pants @xmath331 which has @xmath332 and a boundary component of @xmath0 on its boundary for @xmath326 . then , it is easy to see that @xmath333 is a genus @xmath47 subsurface with @xmath48 boundary components having @xmath94 on the boundary . all the other boundary components of this subsurface are boundary components of @xmath0 . hence , @xmath94 is a @xmath334 separating curve . + case 2 : assume the genus of @xmath0 is 2 and the number of boundary components is at least 2 . if @xmath45 separates a subsurface of genus 2 , then the proof is similar to the proof of case 1 . assume that @xmath45 separates a subsurface of genus 1 . then , since the number of boundary components is at least 2 , @xmath45 separates a subsurface of genus 1 with at least 2 boundary components . we will consider the following two cases : + ( i ) @xmath45 separates a subsurface @xmath335 of genus 1 with at least 3 boundary components . we will give the proof when @xmath45 is a @xmath336 circle . the proofs of the remaining cases are similar . let @xmath335 and @xmath337 be as shown in figure 7 ( ii ) . we complete @xmath338 to a pair of pants decomposition @xmath72 of @xmath0 . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath339 be the representatives of @xmath115 ) , \lambda([x ] ) , \lambda([y ] ) , \lambda([c])$ ] in @xmath78 respectively . since @xmath340 and @xmath341 are peripheral pairs , @xmath342 and @xmath343 are peripheral pairs by lemma 2.8 . then , there exist distinct pair of pants @xmath344 such that @xmath345 has @xmath233 and a boundary component of @xmath0 on its boundary , and @xmath346 has @xmath347 and a boundary component of @xmath0 on its boundary . there exists also a pair of pants @xmath348 which has @xmath349 on its boundary since @xmath231 is adjacent to @xmath94 and @xmath188 w.r.t . @xmath78 . then , it is easy to see that @xmath350 is a genus 1 subsurface with 3 boundary components having @xmath94 as one of its boundary components and all the other boundary components of this subsurface are boundary components of @xmath0 . hence , @xmath94 is a @xmath351 separating curve . + ( ii ) each connected component of @xmath51 is a genus 1 surface with 2 boundary components . let s choose a pair of pants decomposition @xmath352 on @xmath0 , where the curves are as in figure 7 ( iii ) . note that @xmath79 and @xmath45 have geometric intersection 2 , algebraic 0 . let @xmath78 be a pair of pants decomposition of @xmath0 such that @xmath82 ) = [ p']$ ] . let @xmath353 be the representatives of @xmath354 ) , \lambda([y ] ) , \lambda([z ] ) , \lambda([a_3 ] ) , \lambda([b])$ ] in @xmath78 respectively . since @xmath340 and @xmath355 are peripheral pairs , @xmath342 and @xmath356 are peripheral pairs by lemma 2.8 . then , there exist distinct pair of pants @xmath357 such that @xmath358 has @xmath233 and a boundary component of @xmath0 on its boundary , and @xmath359 has @xmath360 and a boundary component of @xmath0 on its boundary . then , since @xmath231 is adjacent to @xmath361 and @xmath291 w.r.t . @xmath78 , and @xmath232 is adjacent to @xmath241 and @xmath291 w.r.t . @xmath78 , there exist also distinct pair of pants @xmath362 and @xmath363 having @xmath364 and @xmath365 on their boundary respectively . then , it is easy to see that @xmath366 and there exists a homeomorphism @xmath367 , i.e. @xmath72 and @xmath78 are topologically equivalent . + let @xmath368 be as shown in figure 7 ( iii ) . let @xmath346 be the pair of pants in @xmath72 , with boundary components @xmath369 and a boundary component of @xmath0 . let @xmath289 be the subsurface of @xmath0 bounded by @xmath370 . @xmath289 is a sphere with four holes . let @xmath371 be the subsurface of @xmath0 bounded by @xmath372 and the boundary component of @xmath0 which is on the boundary of @xmath346 . @xmath371 is a sphere with five holes . + let @xmath373 and @xmath374 . let @xmath300 be the subsurface of @xmath0 bounded by @xmath375 . @xmath300 is a sphere with four holes . let @xmath376 be the subsurface of @xmath0 bounded by @xmath377 and the boundary component of @xmath0 which is on the boundary of @xmath359 . @xmath376 is a sphere with five holes . + since @xmath372 are essential circles in @xmath0 , the essential circles on @xmath371 are essential in @xmath0 . similarly , since @xmath377 are essential circles in @xmath0 , the essential circles on @xmath376 are essential in @xmath0 . furthermore , we can identify @xmath378 and @xmath379 with two subcomplexes of @xmath3 in such a way that the isotopy class of an essential circle in @xmath371 or in @xmath376 is identified with the isotopy class of that circle in @xmath0 . now , suppose that @xmath68 is a vertex in @xmath378 . then , with this identification , @xmath68 is a vertex in @xmath3 and @xmath68 has a representative in @xmath371 . then , @xmath380)=i(\alpha , [ t])=i(\alpha , [ w])= i(\alpha , [ y])=0 $ ] . then there are two possibilities : ( i ) @xmath381 $ ] or @xmath382 $ ] , ( ii ) @xmath383 ) \neq 0 $ ] or @xmath384 ) \neq 0 $ ] . since @xmath17 is injective , @xmath70 is not equal to any of @xmath385 , [ t ' ] , [ w ' ] , [ y']$ ] . since @xmath17 is superinjective , @xmath386)=i(\lambda(\alpha ) , [ t'])=i(\lambda(\alpha ) , [ w'])= i(\lambda(\alpha ) , [ y'])=0 $ ] . then , there are two possibilities : ( i ) @xmath387 $ ] or @xmath388 $ ] , ( ii ) @xmath389 ) \neq 0 $ ] or @xmath390 ) \neq 0 $ ] . then , a representative of @xmath70 can be chosen in @xmath376 . hence , @xmath17 maps the vertices of @xmath3 that have essential representatives in @xmath371 to the vertices of @xmath3 that have essential representatives in @xmath376 , ( i.e. @xmath17 maps @xmath391 to @xmath392 ) . similarly , @xmath17 maps @xmath393 to @xmath394 . + let @xmath395 be the isotopy classes of @xmath396 in @xmath371 respectively . it is easy to see that @xmath397 , [ a]\}$ ] is a simple pair in @xmath371 , @xmath398 , \gamma_3 , [ a])$ ] is a pentagon in @xmath378 , @xmath307 and @xmath399 are 2-vertices , @xmath309 is a 3-vertex , and @xmath397 , \gamma_3\}$ ] , @xmath397 , \gamma_2\}$ ] , @xmath400 , \gamma_3\}$ ] and @xmath401 are codimension - zero simplices of @xmath378 . + since @xmath17 is superinjective and @xmath402 are essential circles , we can see that @xmath403 , @xmath404 ) , \lambda(\gamma_3)$ ] , @xmath115))$ ] is a pentagon in @xmath379 . by the proof of case 2 ( i ) , we can see that @xmath405 is a 2-vertex in @xmath379 . using @xmath406 , it is easy to see that @xmath407 is a 2-vertex in @xmath379 . since @xmath408 is a peripheral pair in @xmath371 , by using lemma 2.8 and the existence of @xmath406 we can see that @xmath409 is a peripheral pair in @xmath376 and so @xmath410 is a 3-vertex in @xmath379 . since @xmath17 is an injective simplicial map @xmath411 ) , \lambda(\gamma_3)\}$ ] , @xmath411 ) , \lambda(\gamma_2)\}$ ] , @xmath412 ) , \lambda(\gamma_3)\}$ ] and @xmath413 are codimension - zero simplices of @xmath379 . then , by theorem [ korkmaz ] , @xmath411 ) , \lambda([a])\}$ ] is a simple pair in @xmath376 . since @xmath17 maps @xmath294 to @xmath414 , @xmath115)$ ] has a representative @xmath94 in @xmath300 such that @xmath415 ) , \lambda([a ] ) = \|b ' \cap a'\|$ ] . then , there exists a homeomorphism @xmath416 . hence , @xmath115)$ ] has a representative @xmath94 which is a @xmath417 circle . this proves the lemma for case 2 , ( ii ) . [ division2 ] let @xmath67 be a superinjective simplicial map . let @xmath418 be a @xmath317-separating circle on @xmath0 , where @xmath319 , @xmath419 separating @xmath0 into two subsurfaces @xmath90 . let @xmath420)$ ] . then @xmath421 is a @xmath317-separating circle , @xmath421 separates @xmath0 into two subsurfaces @xmath95 such that @xmath98 and @xmath99 . let @xmath418 be a @xmath317-separating circle where @xmath319 , @xmath422 . let @xmath90 be the distinct subsurfaces of @xmath0 of genus @xmath47 and @xmath423 respectively which come from the separation by @xmath418 . let @xmath424)$ ] . by lemma [ separating ] , @xmath421 is a @xmath317-separating circle . as we showed in the proof of lemma [ separating ] , there is a pair of pants decomposition @xmath108 of @xmath91 , and @xmath425 can be completed to a pair of pants decomposition @xmath72 of @xmath0 such that a set of curves , @xmath118 , corresponding ( via @xmath17 ) to the curves in @xmath108 , can be chosen such that @xmath118 is a pair of pants decomposition of a subsurface that has @xmath421 as a boundary component and all the other boundary components of this subsurface are boundary components of @xmath0 . let @xmath96 be this subsurface . let @xmath426 . a pairwise disjoint representative set , @xmath78 , of @xmath82)$ ] containing @xmath427 can be chosen . then , by lemma [ imageofpantsdecomp ] , @xmath78 is a pair of pants decomposition of @xmath0 . let @xmath428 and @xmath429 . then @xmath121 are pair of pants decompositions of @xmath122 respectively as @xmath430 are pair of pants decompositions of @xmath124 respectively . + now , let @xmath68 be a vertex in @xmath125 . then , either @xmath126 $ ] or @xmath68 has a nonzero geometric intersection with an element of @xmath127 $ ] . in the first case , clearly @xmath128 since elements of @xmath127 $ ] correspond to elements of @xmath129 \subseteq \mathcal{c}(r_1')$ ] . in the second case , since @xmath17 preserves zero and nonzero geometric intersection ( since @xmath17 is superinjective ) and @xmath68 has zero geometric intersection with the elements of @xmath130 $ ] and @xmath431 $ ] , and nonzero intersection with an element of @xmath127 $ ] , @xmath70 has zero geometric intersection with elements of @xmath131 $ ] and @xmath432 $ ] , and nonzero intersection with an element of @xmath129 $ ] . then , @xmath133 . hence , @xmath134 . the proof of @xmath99 is similar . [ top ] let @xmath67 be a superinjective simplicial map . then @xmath17 preserves topological equivalence of ordered pairs of pants decompositions of @xmath0 , ( i.e. for a given ordered pair of pants decomposition @xmath433 of @xmath0 , and a corresponding ordered pair of pants decomposition @xmath434 @xmath435 of @xmath0 , where @xmath436= \lambda([c_i])$ ] @xmath437 , there exists a homeomorphism @xmath438 such that @xmath439 @xmath440 ) . let @xmath72 be a pair of pants decomposition of @xmath0 and @xmath441 be a nonembedded pair of pants in @xmath72 . the boundary of @xmath441 consists of the circles @xmath218 where @xmath87 is a 1-separating circle on @xmath0 and @xmath222 is a nonseparating circle on @xmath0 . let @xmath91 be the subsurface of @xmath0 of genus @xmath223 with @xmath442 boundary components which has @xmath87 as one of its boundary component and let @xmath230 be the subsurface of @xmath0 of genus 1 which is bounded by @xmath87 . let @xmath108 be the set of elements of @xmath443 which are on @xmath91 and @xmath444 be the set of elements of @xmath443 which are on @xmath230 . then , @xmath445 are pair of pants decompositions of @xmath90 respectively . so , @xmath446 is a pair of pants decomposition of @xmath230 . by lemma [ division2 ] , there exists a 1-separating circle @xmath447)$ ] and subsurfaces @xmath95 , of @xmath0 , of genus @xmath223 and 1 respectively such that @xmath448 and @xmath449 . since @xmath127 \subseteq \mathcal{c}(r_1)$ ] , we have @xmath450 ) \subseteq \mathcal{c}(r_1')$ ] . since @xmath130 \subseteq \mathcal{c}(r_2)$ ] , we have @xmath451 ) \subseteq \mathcal{c}(r_2')$ ] . since @xmath17 preserves disjointness , we can see that a set , @xmath118 , of pairwise disjoint representatives of @xmath450)$ ] disjoint from @xmath231 can be chosen . by counting the number of curves in @xmath118 , we can see that @xmath118 is a pair of pants decomposition of @xmath96 . similarly , a set , @xmath163 , of pairwise disjoint representatives of @xmath451)$ ] disjoint from @xmath231 can be chosen . by counting the number of curves in @xmath163 , we can see that @xmath163 is a pair of pants decomposition of @xmath164 . since @xmath444 has one element , @xmath222 , @xmath163 has one element . let @xmath452 . since @xmath233 correspond to @xmath453 respectively and @xmath222 and @xmath232 give pair of pants decompositions on @xmath230 and @xmath164 ( which are both nonembedded pairs of pants ) and @xmath87 and @xmath231 are the boundaries of @xmath230 and @xmath164 , we see that @xmath17 sends " a nonembedded pair of pants to a nonembedded pair of pants . + let @xmath454 be an embedded pair of pants of @xmath72 . let @xmath455 be the boundary components of @xmath454 . there are three cases to consider : + ( i ) at least one of @xmath87 , @xmath222 or @xmath236 is a separating circle . + ( ii ) all of @xmath455 are nonseparating circles . + ( iii ) exactly one of @xmath455 is a boundary component of @xmath0 and the other two are nonseparating circles . + case ( i ) : w.l.o.g assume that @xmath87 is a @xmath317-separating circle for @xmath319 , @xmath50 . let @xmath90 be the distinct subsurfaces of @xmath0 of genus @xmath47 and @xmath423 respectively which comes from separation by @xmath87 . assume that @xmath456 are on @xmath230 . let @xmath447)$ ] . by lemma [ division2 ] , @xmath231 is a @xmath317 circle separating @xmath0 into subsurfaces , @xmath457 , of genus @xmath47 and @xmath423 respectively such that @xmath98 and @xmath99 . + ( a ) if @xmath222 and @xmath236 are nontrivial circles , then , since @xmath458 , @xmath459 , [ z]\ } ) \subseteq \mathcal{c}(r_2')$ ] . let @xmath460 ) , z ' \in \lambda([z])$ ] such that @xmath461 is pairwise disjoint . let @xmath78 be a set of pairwise disjoint representatives of @xmath82)$ ] which contains @xmath462 . @xmath78 is a pair of pants decomposition of @xmath0 . then , since @xmath87 is adjacent to @xmath222 and @xmath236 w.r.t . @xmath72 , @xmath231 is adjacent to @xmath232 and @xmath241 w.r.t . @xmath78 by lemma [ adjacent ] . then , since @xmath463 , and @xmath231 is a boundary component of @xmath164 , there is an embedded pair of pants in @xmath164 which has @xmath462 on its boundary . + ( b ) if each of @xmath222 and @xmath236 is a boundary component of @xmath0 , then @xmath87 is a ( 0,3)-separating circle . then , by lemma 2.7 , there exists @xmath447)$ ] s.t . @xmath231 is a ( 0,3)-separating circle . so , there is an embedded pair of pants which has @xmath231 and two boundary components of @xmath0 on its boundary . + ( c ) w.l.o.g . assume that @xmath222 is a boundary component of @xmath0 , and @xmath236 is a separating circle . then , @xmath236 is a @xmath464 circle . by using lemma 2.11 , it is easy to see that there exist @xmath465)$ ] and an embedded pair of pants which has @xmath243 and a boundary component of @xmath0 on its boundary . + hence , @xmath17 `` sends '' an embedded pair of pants to an embedded pair of pants in case ( i ) . + case ( ii ) : we can find a nonseparating circle @xmath246 and a @xmath466-separating circle @xmath418 on @xmath0 such that @xmath467 is pairwise disjoint and @xmath468 are on a genus 2 subsurface , @xmath91 , that @xmath418 bounds as shown in figure 8 . let @xmath469 . @xmath108 is a pair of pants decomposition of @xmath91 . we can complete @xmath425 to a pants decomposition @xmath72 of @xmath0 . + [ picture4 ] let @xmath230 be the subsurface of @xmath0 of genus @xmath470 which is not equal to @xmath91 and that comes from separation by @xmath418 . by lemma [ division2 ] , there exist a @xmath466-separating circle @xmath420)$ ] separating @xmath0 into two subsurfaces , @xmath95 , of genus 2 and @xmath470 respectively such that @xmath98 and @xmath99 . since @xmath471 , @xmath450 ) \subseteq \mathcal{c}(r_1')$ ] . we can choose a set , @xmath118 , of pairwise disjoint representatives of @xmath450)$ ] on @xmath96 . then , @xmath427 is a pair of pants decomposition of @xmath96 . we can choose a pairwise disjoint representative set , @xmath78 , of @xmath82)$ ] containing @xmath118 . @xmath78 is a pair of pants decomposition of @xmath0 . let @xmath472 be the representatives of @xmath354 ) , \lambda([y ] ) , \lambda([z ] ) , \lambda([w])$ ] respectively . then , since @xmath418 is adjacent to @xmath236 and @xmath246 w.r.t . @xmath72 , @xmath421 is adjacent to @xmath241 and @xmath473 w.r.t . @xmath78 by lemma [ adjacent ] . then , since @xmath474 and @xmath421 is the boundary of @xmath96 , there is an embedded pair of pants in @xmath96 which has @xmath475 on its boundary . since @xmath236 is a 4-curve in @xmath72 , @xmath241 is a 4-curve in @xmath78 . since @xmath236 is adjacent to @xmath218 w.r.t . @xmath72 , @xmath241 is adjacent to @xmath233 w.r.t . since @xmath241 is on the boundary of a pair of pants which has @xmath476 on its boundary , and @xmath241 is adjacent to @xmath233 , there is a pair of pants having @xmath462 on its boundary . so , @xmath17 `` sends '' an embedded pair of pants bounded by @xmath455 to an embedded pair of pants bounded by @xmath462 in this case too . + case ( iii ) : w.l.o.g assume that @xmath236 is a boundary component of @xmath0 and @xmath218 are nonseparating circles . then , @xmath477 , [ y])$ ] is a peripheral pair . then , by lemma 2.8 , @xmath478 ) , \lambda([y]))$ ] is a peripheral pair . let @xmath233 be disjoint representatives of @xmath354 ) , \lambda([y])$ ] respectively . then , there exists a pair of pants having @xmath233 and a boundary component of @xmath0 on its boundary . hence , in this case also @xmath17 `` sends '' an embedded pair of pants to an embedded pair of pants . + now , assume that @xmath479 is an ordered pair of pants decomposition of @xmath0 . let @xmath480)$ ] such that the elements of @xmath481 are pairwise disjoint . then , @xmath482 is an ordered pair of pants decomposition of @xmath0 . let @xmath483 be an ordered set containing the connected components of @xmath73 . by the arguments given above , there is a corresponding , `` image '' , ordered collection of pairs of pants @xmath484 . nonembedded pairs of pants correspond to nonembedded pairs of pants and embedded pairs of pants correspond to embedded pairs of pants . then , the proof of the lemma follows as in the proof of lemma 3.7 in [ 3 ] . + * remark : * let @xmath485 be an ordered set of vertices of @xmath3 having a pairwise disjoint representative set @xmath486 . then , @xmath486 can be completed to an ordered pair of pants decomposition , @xmath72 , of @xmath0 . we can choose an ordered pairwise disjoint representative set , @xmath78 , of @xmath82)$ ] by lemma [ imageofpantsdecomp ] . let @xmath487 be the elements of @xmath78 which correspond to the elements of @xmath486 . by lemma [ top ] , @xmath72 and @xmath78 are topologically equivalent as ordered pants decompositions . hence , the set @xmath486 and @xmath487 are topologically equivalent . so , @xmath17 gives a correspondence which preserves topological equivalence on a set which has pairwise disjoint representatives . + by using lemma 2.10 and following the proof of lemma 3.9 in [ 3 ] , we can prove the following lemma . this lemma will be used to see some more properties of superinjective simplicial maps . + [ intone ] let @xmath67 be a superinjective simplicial map . let @xmath68 , @xmath69 be two vertices of @xmath3 . if @xmath488 , then @xmath489 . an arc @xmath490 on @xmath0 is called _ properly embedded _ if @xmath491 and @xmath490 is transversal to @xmath492 . @xmath490 is called _ nontrivial _ ( or _ essential _ ) if @xmath490 can not be deformed into @xmath492 in such a way that the endpoints of @xmath490 stay in @xmath492 during the deformation . if @xmath45 and @xmath79 are two disjoint arcs connecting a boundary component of @xmath0 to itself , @xmath45 and @xmath79 are called _ linked _ if their end points alternate on the boundary component . otherwise , they are called _ unlinked_. the _ complex of arcs _ , @xmath37 , on @xmath0 is an abstract simplicial complex . its vertices are the isotopy classes of nontrivial properly embedded arcs @xmath490 in @xmath0 . a set of vertices forms a simplex if these vertices can be represented by pairwise disjoint arcs . + in this section , we assume that @xmath16 is a superinjective simplicial map . let @xmath493 be the set of vertices of @xmath37 . we prove that @xmath17 induces a map @xmath494 with certain properties . then we prove that @xmath495 extends to an injective simplicial map @xmath496 . + the proofs of lemma [ a ] - [ d ] are similar to the proofs of the corresponding lemmas given in @xcite . so , we do not prove these lemmas here . we only state them . = 6.6 in [ a ] let @xmath45 and @xmath79 be two disjoint arcs on @xmath0 connecting two distinct boundary components , @xmath497 , of @xmath0 . let @xmath335 be a regular neighborhood of @xmath498 in @xmath0 . then , @xmath499 where @xmath500 is a standard sphere with four holes and @xmath501 are arcs as shown in figure 9 , ( i ) . [ b ] let @xmath45 and @xmath79 be two disjoint arcs which are unlinked , connecting one boundary component @xmath502 of @xmath0 to itself . let @xmath335 be a regular neighborhood of @xmath503 on @xmath0 . then , @xmath504 where @xmath501 are the arcs drawn on a standard sphere with four holes , @xmath500 , as shown in figure 9 , ( ii ) . [ c ] let @xmath45 and @xmath79 be two disjoint arcs on @xmath0 such that @xmath45 connects one boundary component @xmath502 of @xmath0 to itself for some @xmath505 and @xmath79 connects the boundary components @xmath502 and @xmath506 of @xmath0 , where @xmath507 . let @xmath335 be a regular neighborhood of @xmath508 . then , @xmath509 where @xmath510 are the arcs drawn on a standard sphere with four holes , @xmath500 , as shown in figure 9 , ( iii ) . [ e ] let @xmath45 and @xmath79 be two disjoint arcs . suppose that @xmath45 connects @xmath502 to @xmath506 and @xmath79 connects @xmath502 to @xmath511 where @xmath502 , @xmath506 , @xmath511 are three distinct boundary components . let @xmath335 be a regular neighborhood of @xmath512 . then , @xmath513 where @xmath501 are the arcs drawn on a standard sphere with four holes , @xmath500 , as shown in figure 9 , ( iv ) . [ d ] let @xmath45 and @xmath79 be two disjoint , linked arcs connecting one boundary component @xmath502 of @xmath0 to itself for @xmath514 . let @xmath335 be a regular neighborhood of @xmath503 . then , @xmath515 where @xmath516 is a standard surface of genus one with two boundary components , and @xmath501 are as shown in figure 9 , ( v ) . by using the following lemmas , we see some more properties of @xmath17 . [ horver ] let @xmath68 and @xmath69 be two vertices in @xmath3 which have representatives with geometric intersection 2 and algebraic intersection 0 on @xmath0 . then , @xmath70 and @xmath71 have representatives with geometric intersection 2 and algebraic intersection 0 on @xmath0 . = 2.8 in let @xmath517 be representatives of @xmath66 with geometric intersection 2 and algebraic intersection 0 on @xmath0 . w.l.o.g assume that @xmath32 and @xmath518 do not intersect the boundary components of @xmath0 . let @xmath335 be a regular neighborhood of @xmath519 on @xmath520 . then , @xmath335 is a sphere with four boundary components . let @xmath521 be boundary components of @xmath335 such that there exists a homeomorphism @xmath522 @xmath523 where @xmath524 is a standard sphere with four holes having @xmath525 on its boundary and @xmath526 ( horizontal , vertical ) are two circles as indicated in figure 10 . since @xmath32 and @xmath518 have geometric intersection 2 and algebraic intersection 0 on @xmath0 , none of @xmath527 bound a disk on @xmath0 . if each of @xmath521 is an essential circle on @xmath0 , and @xmath0 is not a surface of genus two with two boundary components , then the proof of the lemma follows from the proof of lemma 4.6 in [ 3 ] , substituting lemma 2.12 above for the corresponding lemma in [ 3 ] . if each of @xmath521 is an essential circle on @xmath0 , and @xmath0 is a surface of genus two with two boundary components , then the proof of the lemma follows from the proof of lemma 2.10 . assume that exactly one of @xmath521 is not essential . assume that @xmath135 is not essential . then , since @xmath335 is a regular neighborhood in @xmath520 , and @xmath135 does not bound a disk on @xmath0 , @xmath135 and a boundary component , say @xmath528 , of @xmath0 bound an annulus , @xmath441 . let @xmath529 . then , @xmath289 is a regular neighborhood of @xmath519 . + let @xmath441= @xmath530 . any two elements in @xmath441 which are isotopic in @xmath0 bound an annulus on @xmath0 . let @xmath454 be a set consisting of a core from each annulus which is bounded by elements in @xmath441 , circles in @xmath441 which are not isotopic to any other circle in @xmath441 , and @xmath518 . we can extend @xmath454 to a pants decomposition @xmath72 of @xmath0 . since either @xmath5 and @xmath6 or @xmath7 and @xmath41 , there are at least four pairs of pants of @xmath72 . note that @xmath531 is a pair of pants decomposition of @xmath289 . each pair of pants of this pants decomposition of @xmath289 is contained in exactly one pair of pants in @xmath72 . it is easy to see that there is a pair of pants @xmath194 of @xmath72 such that interior of @xmath194 is disjoint from @xmath289 , @xmath194 has at least one of @xmath455 as one of its boundary components and all the boundary components of @xmath194 are essential circles in @xmath0 . we will give the argument for the case where @xmath194 has @xmath222 on its boundary . the other cases follow by similar arguments . + let @xmath244 be a regular neighborhood , in @xmath194 , of the boundary components of @xmath194 other than @xmath222 . let @xmath532 be the boundary components of @xmath244 which are in the interior of @xmath194 . then , @xmath533 bound an embedded pair of pants @xmath534 in @xmath194 . let @xmath535 . then , we can extend @xmath524 to @xmath536 and find a homeomorphism @xmath537 , where @xmath536 is as shown in figure 11 . + using lemma [ top ] , we can choose pairwise disjoint representatives @xmath462 , @xmath538 of @xmath354 ) , \lambda([y ] ) , \lambda([z ] ) , \lambda([v ] ) , \lambda([t ] ) , \lambda([w])$ ] respectively s.t . there exists a subsurface @xmath376 of @xmath0 and a homeomorphism @xmath539 @xmath540 @xmath541 for some boundary component @xmath502 of @xmath0 . since @xmath542 , [ v ] ) \neq 0 $ ] and @xmath17 is superinjective , we have , @xmath543 ) , \lambda([v ] ) ) \neq 0 $ ] . then , a representative @xmath544 of @xmath545)$ ] can be chosen such that @xmath544 is transverse to @xmath546 , @xmath544 does nt intersect any of @xmath547 , and @xmath543 ) , \lambda([v ] ) ) = \|h ' \cap v'\|$ ] . since @xmath543 ) , \lambda([v ] ) ) \neq 0 $ ] , @xmath544 intersects @xmath546 . hence , @xmath544 is in the sphere with four holes bounded by @xmath547 in @xmath376 . + = 2.8 in @xmath371 and @xmath376 are spheres with five holes in @xmath0 . since @xmath548 are essential circles in @xmath0 , the essential circles on @xmath371 are essential in @xmath0 . similarly , since @xmath549 are essential circles in @xmath0 , the essential circles on @xmath376 are essential in @xmath0 furthermore , we can identify @xmath378 and @xmath379 with two subcomplexes of @xmath3 in such a way that the isotopy class of an essential circle in @xmath371 or in @xmath376 is identified with the isotopy class of that circle in @xmath0 . now , suppose that @xmath68 is a vertex in @xmath378 . then , with this identification , @xmath68 is a vertex in @xmath3 and @xmath68 has a representative in @xmath371 . then , @xmath380)=i(\alpha , [ t])=i(\alpha , [ w])= i(\alpha , [ z])=0 $ ] . then there are two possibilities : ( i ) @xmath550 $ ] or @xmath551 $ ] , ( ii ) @xmath552 ) \neq 0 $ ] or @xmath553 ) \neq 0 $ ] . since @xmath17 is injective , @xmath70 is not equal to any of @xmath385 , [ t ' ] , [ w ' ] , [ z']$ ] . since @xmath17 is superinjective , @xmath386)=i(\lambda(\alpha ) , [ t'])=i(\lambda(\alpha ) , [ w'])= i(\lambda(\alpha ) , [ z'])=0 $ ] . then , there are two possibilities : ( i ) @xmath554 $ ] or @xmath555 $ ] , ( ii ) @xmath556 ) \neq 0 $ ] or @xmath557 ) \neq 0 $ ] . then , a representative of @xmath70 can be chosen in @xmath376 . hence , @xmath17 maps the vertices of @xmath3 that have essential representatives in @xmath371 to the vertices of @xmath3 that have essential representatives in @xmath376 , ( i.e. @xmath17 maps @xmath391 to @xmath392 ) . similarly , @xmath17 maps @xmath393 to @xmath394 . + it is easy to see that @xmath558 , [ v]\}$ ] is a simple pair in @xmath371 . then , by theorem [ korkmaz ] , there exist vertices @xmath395 of @xmath378 such that @xmath559 , \gamma_3 , [ v])$ ] is a pentagon in @xmath378 , @xmath307 and @xmath399 are 2-vertices , @xmath309 is a 3-vertex , and @xmath558 , \gamma_3\}$ ] , @xmath558 , \gamma_2\}$ ] , @xmath560 , \gamma_3\}$ ] and @xmath401 are codimension - zero simplices of @xmath378 . + since @xmath17 is superinjective and @xmath561 are essential circles , we can see that @xmath403 , @xmath545 ) , \lambda(\gamma_3)$ ] , @xmath562))$ ] is a pentagon in @xmath379 . by lemma [ top ] , @xmath405 and @xmath407 are 2-vertices , and @xmath410 is a 3-vertex in @xmath379 . since @xmath17 is an injective simplicial map @xmath563 ) , \lambda(\gamma_3)\}$ ] , @xmath563 ) , \lambda(\gamma_2)\}$ ] , @xmath564)$ ] , @xmath565 and @xmath413 are codimension - zero simplices of @xmath379 . then , by theorem [ korkmaz ] , @xmath563 ) , \lambda([v])\}$ ] is a simple pair in @xmath566 . since @xmath545)$ ] has a representative , @xmath544 , in @xmath300 , such that @xmath543 ) , \lambda([v ] ) = \|h ' \cap v'\|$ ] and @xmath563 ) , \lambda([v])\}$ ] is a simple pair in @xmath376 , there exists a homeomorphism @xmath567 . + the proof of the lemma in the remaining cases , when @xmath289 has more than one inessential boundary component is similar to the previous case . [ 3 ] let @xmath568 be curves which are either essential circles on @xmath0 or some boundary components of @xmath0 . let @xmath569 be essential circles on @xmath0 such that there exists a subsurface @xmath335 of @xmath0 and a homeomorphism @xmath570 @xmath571 @xmath572 where @xmath524 is a standard torus with two boundary components , @xmath573 , and @xmath574 , @xmath575 are circles as shown in figure 12 ( i ) . then , there exist @xmath576 , two simple closed curves , and @xmath577 ) , z ' \in \lambda([z ] ) , m ' \in \lambda([m ] ) , n ' \in \lambda([n ] ) , n ' \subseteq r$ ] and a homeomorphism @xmath578 , @xmath579 @xmath580 @xmath581 . if both @xmath135 and @xmath87 are essential circles , then the proof follows from lemma 4.7 in [ 3 ] , substituting lemma 2.12 above for the corresponding lemma in [ 3 ] . + since the genus of @xmath0 is at least 2 , both of @xmath135 and @xmath87 can not be boundary components of @xmath0 . w.l.o.g assume that @xmath87 is essential and @xmath135 is a boundary component of @xmath0 . we can complete @xmath530 to a pair of pants decomposition , @xmath72 , of @xmath0 . since @xmath582 gives a pair of pants decomposition on @xmath335 , by lemma [ top ] , there exists a subsurface @xmath583 which is homeomorphic to @xmath524 and there are pairwise disjoint representatives @xmath462 of @xmath354 ) , \lambda([y ] ) , \lambda([z])$ ] respectively and a homeomorphism @xmath584 such that @xmath585 for some @xmath586 . then by lemma [ intone ] , we have the following : + = 3.7 in @xmath587 , [ z])=1 \rightarrow i(\lambda([m ] ) , \lambda([z]))=1 $ ] , @xmath588 , [ z])=1 \rightarrow i(\lambda([n ] ) , \lambda([z]))=1 $ ] , @xmath587 , [ y])=1 \rightarrow i(\lambda([m ] ) , \lambda([y]))=1 $ ] , @xmath588 , [ y])=1 \rightarrow i(\lambda([n ] ) , \lambda([y]))=1 $ ] , @xmath587 , [ x])=0 \rightarrow i(\lambda([m ] ) , \lambda([x]))=0 $ ] , @xmath588 , [ x])=0 \rightarrow i(\lambda([n ] ) , \lambda([x]))=0 $ ] , @xmath587 , [ n])=0 \rightarrow i(\lambda([m ] ) , \lambda([n]))=0 $ ] . + let @xmath589 . there are representatives @xmath590 ) , n ' \in \lambda([n])$ ] such that @xmath591 , @xmath592 , @xmath593 with all intersections transverse . + since @xmath584 is a homeomorphism , we have @xmath594 , @xmath595 , @xmath596 of @xmath597 and @xmath598 of @xmath599 as shown in figure 12 ( ii ) so that each of them has transverse intersection one with @xmath600 and @xmath601 . let @xmath445 be the pair of pants with boundary components @xmath602 , and @xmath603 respectively . let @xmath604 be the annulus with boundary components @xmath605 , @xmath606 , @xmath607 respectively . by the classification of isotopy classes of families of properly embedded disjoint arcs in pairs of pants , @xmath608 , @xmath609 , @xmath610 and @xmath611 can be isotoped to the arcs @xmath612 respectively . let @xmath613 , @xmath614 be such isotopies . by a tapering argument , we can extend @xmath615 and @xmath616 and get @xmath617 and @xmath618 so that @xmath619 is @xmath620 on @xmath621 and @xmath622 is @xmath620 on @xmath621 for all @xmath623 . then , by gluing these extensions we get an isotopy @xmath624 on @xmath625 . + by the classification of isotopy classes of arcs ( relative to the boundary ) on an annulus , @xmath626 can be isotoped to @xmath627 for some @xmath628 . let s call this isotopy @xmath629 . let @xmath630 denote the extension by i d to @xmath524 . similarly , @xmath631 can be isotoped to @xmath632 for some @xmath633 . let s call this isotopy @xmath634 . let @xmath635 denote the extension by i d to @xmath524 . then , `` gluing '' the two isotopies @xmath636 and @xmath637 , we get a new isotopy , @xmath638 , on @xmath524 . then we have , @xmath639 . clearly , @xmath640 fixes @xmath525 . so , we get @xmath641 . let @xmath642 . then , because of the intersection information we also have that @xmath643 is isotopic to either @xmath644 or @xmath645 where @xmath645 is the curve that we get from @xmath644 by reflecting the picture in figure 12 ( i ) about the plane of the paper . let @xmath646 be this reflection . we have @xmath647 . if @xmath643 is isotopic to @xmath644 , we let @xmath648 , and we get @xmath649 . if @xmath643 is isotopic to @xmath645 , we let @xmath650 , and we get @xmath651 . this proves the lemma . + let @xmath490 be an essential properly embedded arc on @xmath0 . let @xmath441 be a boundary component of @xmath0 which has one end point of @xmath490 and @xmath454 be the boundary component of @xmath0 which has the other end point of @xmath490 . let @xmath335 be a regular neighborhood of @xmath652 in @xmath0 . by euler characteristic arguments , @xmath335 is a pair of pants . the boundary components of @xmath335 are called _ encoding circles of @xmath490 on @xmath0_. the set of isotopy classes of nontrivial encoding circles on @xmath0 is called the _ encoding simplex _ , @xmath653 , of @xmath490 ( and of @xmath654 $ ] ) . + an essential properly embedded arc @xmath490 on @xmath0 is called _ type 1 _ if it joins one boundary component @xmath511 of @xmath0 to itself . it is called _ type 1.1 _ if @xmath655 has a regular neighborhood @xmath335 in @xmath0 which has only one circle on its boundary which is inessential w.r.t . @xmath0 . if @xmath335 has two circles on its boundary which are inessential w.r.t . @xmath0 , then @xmath490 is called _ type 1.2_. we call @xmath490 to be _ type 2 _ , if it joins two different boundary components of @xmath0 to each other . an element @xmath654 \in \mathcal{v}(r)$ ] is called _ type 1.1 ( 1.2 , 2 ) _ if it has a type 1.1 ( 1.2 , 2 ) representative . @xmath490 is called _ nonseparating _ if its complement in @xmath0 is connected . + let @xmath656 be the boundary components of @xmath0 . we prove the following lemmas in order to show that @xmath17 induces a map @xmath657 with certain properties . + [ boundary ] let @xmath658 for some @xmath659 . then , there exists a unique @xmath660 for some @xmath661 such that if @xmath490 is a properly embedded essential arc on @xmath0 connecting @xmath511 to itself , then there exists a properly embedded arc @xmath662 on @xmath0 connecting @xmath663 to itself such that @xmath664 . assume that there are two boundary components , @xmath665 and @xmath666 such that each of them satisfies the hypothesis . let @xmath490 be a properly embedded , essential , nonseparating type 1 arc connecting @xmath511 to itself . then , there exist properly embedded arcs , @xmath667 , connecting @xmath665 to itself , and @xmath668 , connecting @xmath666 to itself , such that @xmath669 and @xmath670 . then , we have @xmath671 . note that a properly embedded essential arc @xmath490 is type 1.1 iff @xmath653 has exactly 2 elements . otherwise @xmath653 has 1 element . since @xmath490 is nonseparating type 1 , it is type 1.1 . then , since @xmath669 and @xmath670 and @xmath17 is injective , @xmath667 and @xmath668 are type 1.1 . we can choose a pairwise disjoint representative set @xmath672 of @xmath673 on @xmath0 . since @xmath674 , @xmath675 is a pairwise disjoint representative set for @xmath676 on @xmath0 . then , @xmath677 and @xmath665 bound a pair of pants , @xmath72 , on @xmath0 containing an arc , @xmath678 , isotopic to @xmath667 . similarly , @xmath679 bound a pair of pants , @xmath194 , on @xmath0 containing an arc , @xmath680 , isotopic to @xmath668 . let s cut @xmath0 along @xmath45 and @xmath79 . then , @xmath72 is the connected component of @xmath681 containing @xmath665 and @xmath194 is the connected component of @xmath681 containing @xmath666 . @xmath682 since @xmath666 is not in @xmath72 and @xmath666 is in @xmath194 . then @xmath72 and @xmath194 are distinct connected components meeting along @xmath45 and @xmath79 . hence , @xmath0 is @xmath683 , a torus with two holes which gives a contradiction since the genus of @xmath0 is at least 2 . so , only one boundary component of @xmath0 can satisfy the hypothesis . + since @xmath490 is nonseparating type 1 , @xmath653 contains two isotopy classes of nontrivial circles in @xmath0 . let @xmath78 be a pairwise disjoint representative set of @xmath684)$ ] . since the genus of @xmath0 is at least 2 , by the proof of lemma [ top ] , we can see that @xmath78 and a boundary component of @xmath0 bounds a unique pair of pants @xmath194 , in @xmath0 which has only one inessential boundary component . let @xmath685 be this inessential boundary component . let @xmath662 be an essential properly embedded arc connecting @xmath685 to itself in @xmath194 . then , we have @xmath686 . + = 2.65 in now , to see that @xmath685 is independent of the nonseparating type 1 arc @xmath490 connecting @xmath511 to itself , we prove the following claim : + claim 1 : if we start with two type 1 nonseparating arcs @xmath490 and @xmath662 starting and ending on @xmath511 , then @xmath687 . + proof : let @xmath490 and @xmath662 be nonseparating type 1 arcs connecting @xmath511 to itself . we can assume that @xmath490 and @xmath662 have minimal intersection . first , we prove that there is a sequence @xmath688 of essential properly embedded nonseparating type 1 arcs joining @xmath511 to itself so that each consecutive pair is disjoint , i.e. the isotopy classes of these arcs define a path in @xmath689 , between @xmath490 and @xmath662 . + if @xmath690 , then take @xmath691 , @xmath692 . assume that @xmath693 . we orient @xmath490 and @xmath662 arbitrarily . then , we define two arcs in the following way : start on the boundary component @xmath511 , on one side of the beginning point of @xmath662 and continue along @xmath662 without intersecting @xmath662 , till the last intersection point along @xmath490 . then we would like to follow @xmath490 , without intersecting @xmath662 , until we reach @xmath511 . so , if we are on the correct side of @xmath662 we do this ; if not , we change our starting side from the beginning and follow the construction . this gives us an arc , say @xmath667 . we define @xmath668 , another arc , by changing the orientation of @xmath662 and following the same construction . it is easy to see that @xmath694 are disjoint properly embedded arcs connecting @xmath511 to itself as @xmath490 and @xmath662 do . one can see that @xmath694 are essential arcs since @xmath695 intersect minimally . in figure 13 , we show the beginning and the end points of @xmath490 , the essential intersections of @xmath695 , and @xmath694 near the end point of @xmath490 on @xmath511 . + @xmath696 , @xmath697 since we eliminated at least one intersection with @xmath490 . we also have @xmath698 since we never intersected @xmath662 in the construction . notice that @xmath667 and @xmath668 are not oriented , and @xmath490 is oriented . it is easy to see from the construction that one of @xmath667 or @xmath668 has to be nonseparating type 1 arc , since @xmath662 is a nonseparating type 1 arc . + let @xmath699 and @xmath700 be nonseparating type 1 . by the construction , we get @xmath701 , @xmath702 . now , using @xmath490 and @xmath700 in place of @xmath490 and @xmath662 we can define a new nonseparating type 1 arc @xmath703 , with the properties @xmath704 . by an inductive argument , we get a sequence of nonseparating type 1 arcs such that every consecutive pair is disjoint , @xmath705 . this gives us a special path in @xmath37 between @xmath490 and @xmath662 . by using lemma [ b ] and lemma [ d ] , we can see a regular neighborhood of the union of each consecutive pair in the sequence and the boundary component of @xmath0 that the arcs are starting and ending at , and encoding circles of these consecutive arcs . then , by using the results of lemma [ horver ] and [ 3 ] , we can see that each pair of disjoint nonseparating type 1 arcs give us the same boundary component . hence , by using the sequence given above , we conclude that @xmath490 and @xmath662 give us the same boundary component . this proves claim 1 . + = 2.8 in let @xmath706 be a properly embedded nonseparating type 1 arc on @xmath0 connecting @xmath511 to itself . let @xmath707 . if @xmath490 is a properly embedded , nonseparating type 1 arc on @xmath0 connecting @xmath511 to itself , then by the arguments given above we have @xmath708 , and there exists a properly embedded arc @xmath662 on @xmath0 connecting @xmath663 to itself such that @xmath709 . + suppose that @xmath490 is a nontrivial properly embedded separating type 1 arc on @xmath0 connecting @xmath511 to itself . then it is easy to see that we can find an essential circle @xmath518 and a nonseparating type 1 arc @xmath246 , connecting @xmath511 to itself such that @xmath710 has a regular neighborhood , @xmath335 , which is a sphere with four boundary components , as shown in figure 14 . let @xmath711 be as shown in the figure . notice that @xmath87 and @xmath32 are encoding circles of @xmath246 . since @xmath246 is a nonseparating type 1 arc , @xmath87 is essential . then , by the proof of lemma [ horver ] , there exist essential simple closed curves @xmath447 ) , h ' \in \lambda([h ] ) , v ' \in \lambda([v ] ) , n ' \subseteq r$ ] , @xmath712 , where @xmath713 are boundary components of @xmath0 if @xmath456 are inessential circles respectively and @xmath577 ) , z ' \in \lambda([z])$ ] is @xmath456 are essential circles respectively and there exists a homeomorphism @xmath714 . + let @xmath715 be an arc connecting @xmath716 to itself in the pair of pants determined by @xmath717 and @xmath473 be an arc connecting @xmath716 to itself in the pair of pants determined by @xmath718 . we have @xmath719 . notice that since @xmath246 is a nonseparating type 1 arc , so is @xmath473 . since @xmath715 and @xmath473 connect @xmath716 to itself , and @xmath473 is a nonseparating type 1 arc , by using the previous arguments , we see that @xmath720 . so , the correspondence that we get on boundary components of @xmath0 using nonseparating type 1 arcs is the same as the one that we get by using separating type 1 arcs . hence , @xmath663 is the boundary component that we want . + we define a map @xmath721 using the correspondence which is given by lemma [ boundary ] . [ vertex ] let @xmath654 \in \mathcal{v}(r)$ ] . if @xmath490 connects @xmath511 to @xmath722 on @xmath0 where @xmath723 , then there exists a unique @xmath724 \in \mathcal{v}(r)$ ] such that @xmath662 connects @xmath725 to @xmath726 and @xmath686 . let @xmath654 \in \mathcal{v}(r)$ ] and let @xmath490 connect @xmath511 to @xmath722 on @xmath0 where @xmath723 . if @xmath727 , there exists a nontrivial properly embedded arc , @xmath662 , connecting @xmath725 to itself such that @xmath686 by lemma 3.8 . if @xmath728 , we can see the existence of @xmath662 as follows : let @xmath45 be a properly embedded nontrivial nonseparating arc which connects @xmath511 to itself and let @xmath79 be a properly embedded nontrivial nonseparating arc which connects @xmath722 to itself such that @xmath729 are pairwise disjoint and they are on a subsurface , @xmath335 , which is a sphere with four boundary components , as shown in figure 15 . let @xmath730 be as shown in the figure . + = 3.11 in since @xmath45 and @xmath79 are nonseparating , @xmath222 and @xmath236 are essential circles . by lemma [ horver ] , there exists a subsurface @xmath731 , representatives @xmath732 in @xmath545 ) , \lambda([v ] ) , \lambda([y]),$ ] @xmath733)$ ] respectively and two boundary components , @xmath734 of @xmath0 and a homeomorphism @xmath735 . then , by the proof of lemma [ boundary ] , we see that @xmath736 . let @xmath662 be a properly embedded arc connecting @xmath725 to @xmath726 in the pair of pants bounded by @xmath737 and @xmath544 . then , we have @xmath686 . + now , let @xmath738 be an essential properly embedded arc in @xmath0 such that @xmath738 connects @xmath725 to @xmath726 and @xmath739 . then , we have @xmath740 . let @xmath194 be a regular neighborhood of @xmath741 . since @xmath742 , there is a properly embedded arc @xmath667 isotopic to @xmath662 in @xmath194 . then , since both @xmath667 and @xmath738 connect the same boundary components in this pair of pants , they are isotopic . then , @xmath724 = [ e]$ ] . hence , @xmath724 $ ] is the unique isotopy class in @xmath0 such that @xmath662 connects @xmath725 to @xmath726 and @xmath686 . + @xmath17 induces a unique map @xmath657 such that if @xmath654 \in \mathcal{v}(r)$ ] then @xmath743)$ ] is the unique isotopy class corresponding to @xmath654 $ ] where the correspondence is given by lemma [ vertex ] . using the results of the following lemmas , we will prove that @xmath495 extends to an injective simplicial map on @xmath37 . + @xmath744 extends to a simplicial map @xmath745 . it is enough to prove that if two distinct isotopy classes of essential properly embedded arcs on @xmath0 have disjoint representatives , then their images under @xmath495 have disjoint representatives . let @xmath677 be two disjoint representatives of two distinct classes in @xmath493 . let @xmath746 be the boundary components of @xmath0 . we consider the following cases : + = 5.1 in case 1 : assume that there is no common boundary among the boundaries that @xmath45 and @xmath79 connect . let @xmath497 be the boundary components that @xmath45 starts and ends , and let @xmath747 be the boundary components that @xmath79 starts and ends where @xmath748 and @xmath749 and @xmath750 . then , since @xmath751 is disjoint from @xmath752 , we can find disjoint regular neighborhoods , @xmath753 , of @xmath754 and @xmath755 of @xmath756 on @xmath0 , which give us two disjoint pair of pants . then , by using lemma [ top ] , and the definition of @xmath495 , it is easy to see that the corresponding arcs ( images ) will have disjoint representatives . + case 2 : assume that @xmath677 are unlinked , connecting @xmath502 to itself for some @xmath514 . by lemma [ b ] , there is a homeomorphism @xmath584 such that @xmath757 where @xmath335 is a regular neighborhood of @xmath503 in @xmath0 and @xmath501 are as shown in figure 16 ( i ) . then , by using lemma [ horver ] and the definition of @xmath495 , we see that images have disjoint representatives . + case 3 : assume that @xmath45 connects one boundary component @xmath502 to itself for some @xmath514 , and @xmath79 connects @xmath502 to @xmath511 for some @xmath758 . then , by lemma [ c ] , there is a homeomorphism @xmath584 such that @xmath759 where @xmath335 is a regular neighborhood of @xmath760 in @xmath0 and @xmath501 are as shown in figure 16 ( ii ) . then , by using lemma [ horver ] and the definition of @xmath495 , we see that the images have disjoint representatives . + case 4 : assume that @xmath45 connects @xmath502 to @xmath506 and @xmath79 connects @xmath502 to @xmath511 where @xmath761 . then , by lemma [ e ] , there is a homeomorphism @xmath584 such that @xmath762 where @xmath335 is a regular neighborhood of @xmath763 in @xmath0 and @xmath501 are as shown in figure 16 ( iii ) . then , by using lemma [ horver ] and the definition of @xmath495 , as in the previous cases we see that the images have disjoint representatives . + = 4.2 in case 5 : assume that each of @xmath45 and @xmath79 connect two distinct boundary components , say @xmath497 , of @xmath0 . + we will first prove this case when @xmath0 has at least 3 boundary components . let @xmath511 be a boundary component different from @xmath502 and @xmath506 . let @xmath738 and @xmath764 be disjoint properly embedded arcs which are disjoint from @xmath677 such that @xmath738 connects @xmath502 to @xmath511 , @xmath764 connects @xmath506 to @xmath511 and @xmath765 are in a subsurface @xmath335 , which is a sphere with 4 holes , of @xmath0 as shown in figure 17 ( i ) . then , by applying the result of case 4 to each pair in @xmath766 , we can find disjoint representatives @xmath767 of @xmath768 ) , \lambda_([e])$ ] respectively , disjoint representatives @xmath769 of @xmath768 ) , \lambda_([l])$ ] respectively , disjoint representatives , @xmath770 of @xmath771 ) , \lambda_([l])$ ] respectively . then , by using lemma [ e ] , and lemma [ horver ] we can choose disjoint representatives @xmath772 of @xmath768 ) , \lambda_([e ] ) , \lambda_([l])$ ] respectively , a subsurface @xmath583 and a homeomorphism @xmath773 where @xmath774 are as shown in figure 17 ( i ) . since @xmath79 and @xmath738 are disjoint and @xmath79 connects @xmath502 to @xmath506 and @xmath738 connects @xmath502 to @xmath511 , by using case 4 , we can choose a representative @xmath201 of @xmath775)$ ] which is disjoint from @xmath776 . similarly , we can choose a representative @xmath777 of @xmath775)$ ] which is disjoint from @xmath778 . then , since @xmath776 and @xmath778 are disjoint , we can choose a representative @xmath779 of @xmath775)$ ] which is disjoint from @xmath780 . then , it is easy to see that @xmath404)$ ] has a representative which is disjoint from @xmath94 . + now , assume that @xmath0 has exactly two boundary components , @xmath497 . by lemma [ a ] , there is a homeomorphism @xmath781 @xmath782 where @xmath335 is a regular neighborhood of @xmath783 in @xmath0 and @xmath784 are as shown in figure 17 ( ii ) . since @xmath0 has exactly two boundary components , @xmath222 and @xmath236 are essential circles in @xmath0 . then , using lemma [ top ] , we can choose pairwise disjoint representatives @xmath713 of @xmath785 ) , \lambda([z])$ ] respectively , boundary components @xmath786 and a subsurface @xmath583 , and a homeomorphism @xmath787 . + @xmath335 and @xmath731 are spheres with four holes in @xmath0 . since @xmath456 are essential circles in @xmath0 , the essential circles on @xmath335 are essential in @xmath0 . similarly , since @xmath713 are essential circles in @xmath0 , the essential circles on @xmath731 are essential in @xmath0 . furthermore , we can identify @xmath788 and @xmath789 with two subcomplexes of @xmath3 in such a way that the isotopy class of an essential circle in @xmath335 or in @xmath731 is identified with the isotopy class of that circle in @xmath0 . now , suppose that @xmath68 is a vertex in @xmath788 . then , with this identification , @xmath68 is a vertex in @xmath3 and @xmath68 has a representative in @xmath335 . then as in the proof of lemma [ horver ] , we can see that @xmath17 maps @xmath790 to @xmath791 . then , since @xmath335 and @xmath731 has four boundary components , we can apply the arguments given in the proof of the first part to see that there are disjoint representatives of @xmath768)$ ] and @xmath775)$ ] in @xmath731 . this proves case 5 . + = 2.75 in case 6 : assume that @xmath677 are linked , connecting @xmath502 to itself for some @xmath792 . by lemma [ d ] , there is a homeomorphism @xmath793 where @xmath335 is a regular neighborhood of @xmath794 in @xmath0 and @xmath795 are as in figure [ figure18 ] . since @xmath796 and @xmath797 are the boundary components of regular neighborhoods of @xmath798 and @xmath799 on @xmath524 respectively , @xmath800 and @xmath801 are encoding circles for @xmath45 and @xmath79 on @xmath0 respectively . we have @xmath802 @xmath803 @xmath804 @xmath805 . since the genus of @xmath0 is at least 2 , @xmath806 is an essential circle on @xmath0 . by lemma [ 3 ] , there exists @xmath807 , @xmath808 , @xmath809 , @xmath583 and a homeomorphism @xmath810 , @xmath811 . since @xmath800 and @xmath812 are encoding circles for @xmath45 and @xmath79 on @xmath0 respectively , @xmath813 and @xmath814 are encoding circles for @xmath768 ) , \lambda_([b])$ ] on @xmath0 respectively . existence of @xmath406 shows that @xmath768 ) , \lambda_([b])$ ] have disjoint representatives . @xmath815 and @xmath816 are disjoint representatives for @xmath768 ) , \lambda_([b])$ ] respectively . + we have shown in all the cases that if two vertices have disjoint representatives , then @xmath495 sends them to two vertices which have disjoint representatives . hence , @xmath495 extends to a simplicial map @xmath817 . [ inj2 ] let @xmath67 be a superinjective simplicial map . then , @xmath818 is injective . it is enough to prove that @xmath495 is injective on the vertex set , @xmath493 . let @xmath654 , [ j ] \in \mathcal{v}(r)$ ] such that @xmath743)= \lambda_([j])=[k]$ ] . then , by the definition of @xmath495 , the type of [ i ] and [ j ] are the same . assume they are both type 1.1 . let @xmath819 , [ y]\}$ ] and @xmath820 , [ t]\}$ ] be the encoding simplices for @xmath654 $ ] and @xmath724 $ ] respectively . then , @xmath821 ) , \lambda([y])\}$ ] and @xmath822 ) , \lambda([t])\}$ ] are encoding simplices of @xmath823 $ ] . so , \{@xmath824 ) , \lambda ( [ y])\ } = \{\lambda ( [ z ] ) , \lambda ( [ t])\}$ ] . then , since @xmath17 is injective , we get @xmath819 , [ y]\ } = \{[z ] , [ t]\}$ ] . this implies @xmath654=[j]$ ] . the other cases can be proven similarly to the first case by using the injectivity of @xmath17 . + the following lemma can be proven similar to the proof of lemma 4.13 in @xcite , which uses the connectivity theorem for elementary moves of mosher , @xcite , appropriately restated for surfaces with boundaries . we will only state this lemma here . + [ fixing ] if an injection @xmath825 agrees with @xmath826 on a top dimensional simplex , where @xmath827 is induced by a homeomorphism @xmath828 , then @xmath629 agrees with @xmath827 on @xmath37 . notation : * a homeomorphism @xmath829 induces a map @xmath830 , where @xmath831 $ ] and @xmath832 induces a map @xmath833 in a similar way as @xmath17 induces @xmath495 . + * remark : * we have proven that @xmath17 is an injective simplicial map which preserves the geometric intersection 0 and 1 . if the number of boundary components of @xmath0 is at least 2 , using these properties and following n.v.ivanovs proof of his theorem 1.1 @xcite , it can be seen that @xmath495 agrees with a map , @xmath834 , induced by a homeomorphism @xmath828 on a top dimensional simplex in @xmath37 . then , by lemma [ fixing ] , it agrees with @xmath834 on @xmath37 . then , it is easy to see that @xmath17 agrees with a map , @xmath827 on @xmath3 . + if the number of boundary components of @xmath0 is exactly 1 , we cut @xmath0 along a nonseparating simple closed curve , @xmath135 on @xmath0 . let @xmath835 be this cut surface and let @xmath836 be the two boundary components of @xmath835 which comes from cutting @xmath0 along @xmath135 . then , considering how @xmath17 induced @xmath495 and using the techniques of @xcite , it is easy to see that @xmath17 induces a superinjective simplicial map @xmath837 and an injective simplicial map @xmath838 , where @xmath150 ) = [ d]$ ] . then , since @xmath835 and @xmath839 have 3 boundary components , by adapting the arguments given in the paragraph above and using lemma [ top ] , we see that @xmath840 agrees with a map @xmath841 on @xmath842 , where @xmath841 is induced by a homeomorphism @xmath843 such that @xmath844 . we can do this argument for any nonseparating simple closed curve on @xmath0 . then , to see that @xmath17 agrees with a map , @xmath834 , which is induced by a homeomorphism @xmath828 , on @xmath3 we use the following lemma . [ imp ] let @xmath845 be a superinjective simplicial map . assume that for any nonseparating simple closed curve @xmath135 on @xmath0 , @xmath840 agrees with a map , @xmath846 , which is induced by a homeomorphism @xmath847 where @xmath844 and @xmath150 ) = [ d]$ ] . then , @xmath17 agrees with a map @xmath848 which is induced by a homeomorphism @xmath828 . let @xmath135 be a nonseparating simple closed curve and @xmath849 be a simplicial map induced by a homeomorphism @xmath843 where @xmath844 and @xmath150 ) = [ d]$ ] such that @xmath840 agrees with @xmath841 on @xmath842 . let @xmath1 be a homeomorphism of @xmath0 which cuts to a homeomorphism @xmath850 which is isotopic to @xmath851 . then each homeomorphism of @xmath0 which cut to a homeomorphism @xmath850 which is isotopic to @xmath851 , is isotopic to an element in the set @xmath852 , [ 1 ] . it is easy to see that @xmath840 agrees with @xmath853 on @xmath842 for all @xmath854 . + let @xmath246 be a simple closed curve which is dual to @xmath135 ( i.e. @xmath246 intersects @xmath135 transversely once and there is no other intersection ) . let @xmath72 be a regular neighborhood of @xmath855 . then , @xmath72 is a genus one surface with one boundary component . let @xmath222 be the boundary component of @xmath72 . we have @xmath856 , [ y ] ) = 0 $ ] , @xmath857 , [ y ] ) = 0 $ ] , and @xmath856 , [ w ] ) = 1 $ ] . then , since @xmath17 is superinjective we have @xmath858 ) , \lambda([y ] ) ) = 0 $ ] , @xmath859 ) , \lambda([y ] ) ) = 0 $ ] and @xmath858 ) , \lambda([w ] ) ) = 1 $ ] . + let @xmath194 be the genus one subsurface with one boundary component of @xmath0 which has @xmath860 as its boundary . then , it is easy to see that @xmath861 , @xmath862 and @xmath863 is dual to @xmath864 , since @xmath246 is dual to @xmath135 . + since @xmath865 \in \mathcal{c}(r_c)$ ] , @xmath785)= g_\#([y ] ) = [ g(y)]$ ] . we also have @xmath866= \lambda([c])= g_\#([c])$ ] . since @xmath858 ) , \lambda([y ] ) ) = 0 $ ] , @xmath859 ) , \lambda([y ] ) ) = 0 $ ] and @xmath858 ) , \lambda([w ] ) ) = 1 $ ] , and @xmath867)$ ] , @xmath868)$ ] and @xmath864 and @xmath860 are disjoint , we can choose a simple closed curve @xmath869)$ ] such that @xmath473 is in @xmath870 and dual to @xmath864 . then , @xmath871 is dual to @xmath135 and @xmath871 is in @xmath872 . then , since both of @xmath246 and @xmath871 are dual to @xmath135 in @xmath872 , there exists @xmath873 such that @xmath874 ) = [ g^{-1}(w')]$ ] . then , @xmath875 ) = [ w']$ ] . since @xmath876 ) = [ w']$ ] , @xmath877 agrees with @xmath17 on @xmath878 $ ] . we can identify @xmath842 with a subcomplex , @xmath879 , of @xmath3 . let @xmath880 be the set of isotopy classes of simple closed curves which are dual to @xmath135 on @xmath0 . + claim 1 : @xmath881 agrees with @xmath17 on @xmath882\ } \cup l_c \cup d_c$ ] . + proof : it is clear that @xmath883)= \lambda ( [ c ] ) = [ d]$ ] . since @xmath841 agrees with @xmath840 on @xmath842 , @xmath881 agrees with @xmath17 on @xmath879 . + we have seen that @xmath877 agrees with @xmath17 on @xmath878 $ ] . let @xmath884 be a simple closed curve which is disjoint from @xmath246 and dual to @xmath135 . as we described before , there exists @xmath885 such that @xmath17 agrees with @xmath886 on @xmath887 $ ] . since @xmath246 and @xmath884 are disjoint , @xmath859 ) , \lambda([w_1 ] ) ) = 0 $ ] . if @xmath888 then @xmath889 ( since both @xmath246 and @xmath884 are dual to @xmath135 ) . then , since @xmath876)= ( gt_c^{m_c})([w])$ ] and @xmath890)= ( gt_c^{\tilde{m_c}})([w_1])$ ] , we would get @xmath859 ) , \lambda([w_1 ] ) ) \neq 0 $ ] , which gives a contradiction . therefore , @xmath891 . then , we see that @xmath881 agrees with @xmath17 on @xmath882\ } \cup l_c \cup \ { [ w ] \cup [ w_1 ] \}$ ] . + given any curve @xmath418 which is dual to @xmath135 , by using similar techniques as in lemma 3.8 , we can find a sequence of dual curves to @xmath135 , connecting @xmath864 to @xmath418 , such that each consecutive pair is disjoint , i.e. the isotopy classes of these curves define a path in @xmath892 , between @xmath864 and @xmath418 . then using the argument above and the sequence , we conclude that @xmath881 agrees with @xmath17 on @xmath880 . hence , @xmath881 agrees with @xmath17 on @xmath882\ } \cup l_c \cup d_c$ ] . this proves claim 1 . let @xmath893 . we have that @xmath894 agrees with @xmath17 on @xmath882\ } \cup l_c \cup d_c$ ] . + claim 2 : let @xmath518 be a nonseparating simple closed curve on @xmath0 . then , @xmath895 on @xmath3 . + proof : since the genus of @xmath0 is at least 2 , we can find a sequence of nonseparating simple closed curves connecting @xmath135 to @xmath518 such that each consecutive pair is disjoint . refining this sequence , we can get a sequence @xmath896 of nonseparating simple closed curves connecting @xmath135 to @xmath518 such that each consecutive pair in this sequence is simultaneously nonseparating . + = 2.3 in let s consider the first consecutive pair in the sequence , @xmath897 . since @xmath898 is simultaneously nonseparating , it can be completed to a set @xmath899 , ( shown in figure 19 , for @xmath900 ) , see @xcite , such that the isotopy classes of dehn twists about the elements of this set generate @xmath14 and all the curves in this set are ( i ) either disjoint from @xmath135 or dual to @xmath135 , and ( ii ) either disjoint from @xmath158 or dual to @xmath158 . then , since all the curves in @xmath899 are either disjoint from @xmath135 or dual to @xmath135 , by claim 1 we have that @xmath901 ) = \lambda ( [ x])$ ] for every @xmath902 . similarly , since all the curves in @xmath899 are either disjoint from @xmath158 or dual to @xmath158 , by claim 1 we have @xmath903 ) = \lambda ( [ x])$ ] for every @xmath904 . hence , @xmath905)= \lambda ( [ x ] ) = ( h_{c_1})_\ # ( [ x])$ ] for every @xmath904 . then , @xmath906 since @xmath907 . + by using our sequence , with an inductive argument we get that @xmath908 on @xmath3 and @xmath909 on @xmath882\ } \cup \{[v]\ } \cup l_c \cup d_c \cup l_v \cup d_v$ ] . in particular we see that , @xmath894 agrees with @xmath17 on any nonseparating curve @xmath518 and on @xmath910 . since every separating curve is in the link , @xmath911 , of some nonseparating curve @xmath912 , we see that @xmath894 agrees with @xmath17 on @xmath3 . this proves the lemma . + proof of theorem 1.1 follows from lemma 3.13 and the remarks made before this lemma . note that in @xcite , we referred to ivanov s cutting arguments for the result of the corresponding theorem . the proof of the cutting argument given above can be adapted easily to that case . we assume that @xmath913 where @xmath914 is the homomorphism defined by the action of homeomorphisms on the homology . a mapping class @xmath19 is called _ pseudo - anosov _ if @xmath52 is nonempty and if @xmath915 , for all @xmath68 in @xmath52 and any @xmath916 . @xmath1 is called _ reducible _ if there is a nonempty subset @xmath917 such that a set of disjoint representatives can be chosen for @xmath918 and @xmath919 . in this case , @xmath920 is called a _ reduction system _ for @xmath1 . each element of @xmath918 is called a _ reduction class _ for @xmath1 . a reduction class , @xmath68 , for @xmath1 , is called an _ essential reduction class _ for @xmath1 , if for each @xmath921 such that @xmath922 and for each integer @xmath923 , @xmath924 . the set , @xmath925 , of all essential reduction classes for @xmath1 is called the _ canonical reduction system _ for @xmath1 . the correspondence @xmath926 is canonical . in particular , it satisfies @xmath927 for all @xmath928 in @xmath4 . + the following two lemmas are well known facts . the isotopy class of a dehn twist about a circle @xmath45 , is denoted by @xmath929 , where @xmath84 = \alpha$ ] . [ reducible ] let k be a finite index subgroup of @xmath4 . let @xmath38 be an injective homomorphism . then there exists @xmath935 such that @xmath938 is a reducible element of infinite order for all @xmath939 . let @xmath944 . since @xmath945 is a finite index subgroup we can choose @xmath946 such that @xmath947 for all @xmath68 in @xmath52 . by lemma [ reducible ] @xmath938 is a reducible element of infinite order in @xmath4 . let @xmath43 be a realization of the canonical reduction system of @xmath948 . let @xmath135 be the number of components of @xmath43 and @xmath912 be the number of p - anosov components of @xmath949 . since @xmath950 . by theorem 5.9 @xcite , @xmath951 is a free abelian group of rank @xmath952 . by lemma [ rank=1 ] @xmath953 . then , either @xmath954 , @xmath955 or @xmath956 , @xmath957 . since there is at least one curve in the canonical reduction system we have @xmath954 , @xmath955 . hence , since @xmath958 , @xmath959 for some @xmath960 , @xmath961 , [ 1 ] , [ 5 ] . + * remark : * suppose that @xmath962 for some @xmath921 and @xmath963 and @xmath964 for some @xmath965 and @xmath966 . since @xmath967 , @xmath968 , @xmath969 . then , @xmath970 by lemma [ wellknown1 ] . therefore , by lemma [ correspondence ] , @xmath12 gives a correspondence between isotopy classes of circles and @xmath12 induces a map , @xmath971 , where @xmath972 . + in the following lemma we use a well known fact that @xmath973 for all @xmath68 in @xmath52 , @xmath974 , where @xmath975 if @xmath12 has an orientation preserving representative and @xmath976 if @xmath12 has an orientation reversing representative . [ identity ] let @xmath10 be a finite index subgroup of @xmath4 . let @xmath38 be an injective homomorphism . assume that there exists @xmath977 such that @xmath978 @xmath52 , @xmath979 such that @xmath980 . then , @xmath12 is the identity on @xmath10 . we use ivanov s trick to see that @xmath981 @xmath982 and @xmath983 @xmath984 @xmath985 , @xmath986 . then , we have @xmath987 @xmath940 , @xmath986 . hence , @xmath988 @xmath940 , @xmath986 by lemma [ wellknown1 ] . then , @xmath989 @xmath940 , @xmath986 . therefore , @xmath990 commutes with @xmath929 @xmath940 , @xmath986 . since @xmath14 is generated by dehn twists , @xmath991 @xmath986 . since the genus of @xmath0 is at least two and @xmath0 is not a closed surface of genus two , @xmath992 is trivial by 5.3 in [ 6 ] . so , @xmath993 @xmath986 . hence , @xmath994 . + by the remark after lemma [ correspondence ] , we have that @xmath1006 induces a map @xmath1007 , where @xmath10 is a finite index subgroup of @xmath4 . in the following lemma we prove that @xmath1008 is a superinjective simplicial map on @xmath3 . there exists @xmath1010 such that @xmath1011 and @xmath1012 . then we have the following : @xmath1013 @xmath1014 @xmath1015 ( by lemma [ wellknown2 ] ) @xmath1014 @xmath1016 ( since @xmath12 is injective on k ) @xmath1014 @xmath1017 where @xmath1018 ( by lemma [ correspondence ] ) @xmath1019 ( by lemma [ wellknown2 ] ) . by lemma [ intersection0 ] @xmath1008 is a superinjective simplicial map on @xmath3 . then , by theorem 1.1 , @xmath1008 is induced by a homeomorphism @xmath1021 , i.e. @xmath1022 for all @xmath68 in @xmath52 , where @xmath1023 $ ] . let @xmath1024 be the isomorphism defined by the rule @xmath1025 for all @xmath47 in @xmath4 . then for all @xmath68 in @xmath52 , we have the following : then , since @xmath1027 is injective , @xmath1028 by lemma [ identity ] . so , @xmath1029 . hence , @xmath12 is the restriction of an isomorphism which is conjugation by @xmath834 , ( i.e. @xmath12 is induced by @xmath32 ) . suppose that there exists an automorphism @xmath1030 such that @xmath1031 . let @xmath946 such that @xmath1032 for all @xmath68 in @xmath52 . since @xmath1033 and @xmath1034 , @xmath1035 for all @xmath68 in @xmath52 . then , by corollary [ i d ] , @xmath1036 . hence , the extension of @xmath12 is unique . harvey , boundary structure of the modular group , in : _ riemann surfaces and related topics : proc . of the 1978 stony brook conf . _ , ed . by i. kra and b. maskit , annals of math studies , 97 , _ princeton univ . press _ , 1981 , 245 - 251 .	let @xmath0 be a compact , connected , orientable surface of genus @xmath1 with @xmath2 boundary components . let @xmath3 be the complex of curves on @xmath0 and @xmath4 be the extended mapping class group of @xmath0 . suppose that either @xmath5 and @xmath6 or @xmath7 and @xmath8 . we prove that a simplicial map @xmath9 is superinjective if and only if it is induced by a homeomorphism of @xmath0 . as a corollary , we prove that if @xmath10 is a finite index subgroup of @xmath4 and @xmath11 is an injective homomorphism , then @xmath12 is induced by a homeomorphism of @xmath0 and @xmath12 has a unique extension to an automorphism of @xmath4 . this extends the author s previous results about closed connected orientable surfaces of genus at least 3 , to the surface @xmath0 . * superinjective simplicial maps of complexes of * + + + _ mathematics subject classification _ : 57m99 , 20f38 .
in the recent years quantum field theories have provided useful tools for addressing problems in algebraic topology . an example is the computation of the real cohomology of the spaces @xmath0 of imbeddings of the circle into @xmath1 for @xmath3 . the works of bar - natan @xcite and kontsevich @xcite , have shown how the perturbative expansion of chern - simons quantum field theory yields the so - called `` vassiliev invariants '' of the space of ordinary knots . the precise statement is that there exists a complex of ( trivalent ) feynman diagrams whose cohomology classes give rise to 0-cohomology classes of the space @xmath4 . chern - simons is a topological quantum field theory which can be defined in three dimensions only and produces feynman diagrams whose vertices have valence equal to three . there exist however other quantum field theories that , while sharing the same topological properties , are defined over manifolds of any dimension and produce feynman diagrams not necessarily trivalent . these topological quantum field theories are known as `` bf theories '' and their properties are studied in details in @xcite and @xcite . in @xcite it is shown how the perturbative expansion of bf theories can define a complex of feynman diagrams and a chain map into the de rham complex of @xmath0 , for any @xmath5 . moreover this map gives rise to an injective map in cohomology , when restricted to trivalent diagrams . the cohomology of this complex of feynman diagrams is called `` graph cohomology '' . a different approach to the problem of computing the cohomology of @xmath0 has been considered by v. vassiliev in @xcite : there exists in fact a spectral sequence converging to the cohomology of @xmath0 for every @xmath2 , whose @xmath6 term coincides with the graph cohomology ( see also @xcite ) . thus , vassiliev spectral sequence gives an `` upper bound '' to the cohomology of @xmath0 , since nontrivial cohomology classes of @xmath0 must come from some element in @xmath6 , while @xcite gives a `` lower bound '' , since at least those elements corresponding to trivalent diagrams give rise to nontrivial elements in cohomology . in general it is unclear whether the collapsing always happens at the @xmath6 term , as conjectured by kontsevich @xcite , or there are some `` graph cocycles '' which do not give rise to classes in @xmath7 . in this note we want to extend the results of @xcite in the following direction [ thm - main ] for every even @xmath2 , there exists a nontrivial class in@xmath8 associated to a nontrivalent graph cocycle . a similar question has been posed by r. bott @xcite regarding the possibility of constructing 1-dimensional cocycles on the space @xmath4 of ordinary knots . in this case , however , it is not even established the convergence of the spectral sequence . we also mention the work of d. sinha @xcite on the cohomology of the spaces of imbeddings of the real line into @xmath1 . using goodwillie s calculus , one can define a spectral sequence whose @xmath6-term , as in vassiliev s case , is isomorphic to ( a variant of ) the graph cohomology . recently p. lambrechts and i. volic have announced a proof of the collapse of sinha spectral sequence at the @xmath6 term . + here is the plan of the paper . we will first recall the definition of graph cohomology ( sections [ sec : grco ] and [ sec : grho ] ) , and the construction of the map from graph cohomology to the de rham cohomology of @xmath0 ( section [ sec : csi ] ) . then , in section [ sec : nontriv ] , we will prove theorem [ thm - main ] by considering a nontrivial graph cocycle consisting of nontrivalent diagrams and mapping it to @xmath7 . nontriviality of this class will follow from an explicit evaluation on the dual cycle . we first briefly recall some definitions given in @xcite . by @xmath9 and @xmath10 we mean the real vector spaces generated by decorated diagrams of order @xmath11 and degree @xmath12 , of odd and even type , respectively . the diagrams consist of an _ oriented circle _ and many _ edges _ joining vertices which may lie either on the circle ( _ external vertices _ ) or off the circle ( _ internal vertices _ ) . we require all the vertices to be at least trivalent . if we denote by @xmath13 the number of edges of a diagram , @xmath14 the number of its internal vertices and @xmath15 the number of its external vertices , then the order @xmath11 of a diagram is @xmath16 ( minus the euler characteristic ) while the degree @xmath12 is @xmath17 ( the deviation from being a trivalent diagram ) . we also define a _ chord diagram _ to be a decorated diagram whose vertices are all external . the decoration in @xmath9 is given by numbering the vertices ( up to even cyclic permutations ) , numbering the internal vertices ( up to even permutations ) and orienting the edges ( up to reversal ) . by convention , we number external vertices from 1 to @xmath14 and internal vertices from @xmath18 to @xmath19 . an extra decoration is needed on the edges connecting the same external vertex , namely an ordering of the two half - edges forming them . the decoration in @xmath10 is given by numbering the external vertices ( up to even cyclic permutations ) and numbering the edges ( up to even permutations ) . finally , both in @xmath9 and @xmath10 we quotient by the subspace generated by all diagrams containing two edges joining the same pair of vertices and diagrams containing edges whose end - points are the same internal vertex . the coboundary operators @xmath20 and @xmath21 are linear operators whose action on a diagram @xmath22 is given by the signed sum of all the diagrams obtained from @xmath22 by contracting , one at a time , all the regular edges and arcs of the diagrams . here by arc we mean a piece of the oriented circle between two consecutive vertices , and by regular edge we mean an edge with at least one internal end - point . the signs are as follows : if we contract an arc or edge connecting the vertex @xmath23 with the vertex @xmath24 , and oriented from @xmath23 to @xmath24 , then the sign is @xmath25 if @xmath26 or @xmath27 if @xmath28 . if we contract an edge labelled by @xmath29 , then the sign is @xmath30,where @xmath15 is the number of external vertices of the diagram . these complexes are called _ graph complexes _ , and the relevant cohomology groups @xmath31 and @xmath32 are know as _ graph cohomology_. when we write @xmath33 ( resp . @xmath34 ) , we mean either the odd or the even - type graph complex ( resp . graph cohomology ) . graph homology is given by the dual vector space @xmath35 and the boundary operator @xmath36 defined as the adjoint to @xmath37 . the homology groups are denoted by @xmath38 . notice that since there is a preferred basis of @xmath39 given by the diagrams , we can identify @xmath35 with @xmath39 , and consider @xmath40 as an operator on @xmath39 that decollapses the vertices of the diagrams in all possible ways . for instance , the boundary of a trivalent diagram is always zero , while the boundary of a diagram @xmath22 with all trivalent vertices except a quadrivalent external vertex is the sum of three trivalent diagrams as in figure [ de-4ev - d ] ( odd case ) and [ de-4ev - p ] ( even case ) . we claim that all classes of @xmath41 can be represented by chord diagrams only . more precisely , @xmath42 is isomorphic to the quotient of the space of chord diagrams @xmath43 by the subspace generated by the diagrams of figure [ 4t - rel - d ] and figure [ 1t - rel - d ] . similarly , @xmath44 is isomorphic to the quotient of the space of chord diagrams @xmath45 by the subspace generated by the diagrams of figure [ 4t - rel - p ] and figure [ 1t - rel - p ] . the proof of this fact in the odd case is given in @xcite , thm . the even case is completely analogous . the subspaces of figure [ 4t - rel - d ] and [ 4t - rel - p ] are know as the @xmath46 relations , while the subspaces of figure [ 1t - rel - d ] and [ 1t - rel - p ] are the @xmath47 relations . notice that the @xmath46 relations are the boundary of the sum ( with proper sign ) of two of the three nontrivalent diagrams considered in figure [ de-4ev - d ] and [ de-4ev - p ] . we now want to recall the definition and main properties of the _ configuration space integrals _ @xcite . first , for any compact smooth manifold @xmath48 , we define the open configuration space @xmath49 of @xmath50 points in @xmath48 to be the space of the @xmath50-uples @xmath51 such that @xmath52 for @xmath53 then , we consider the axerlod singer fulton macpherson compactification @xmath54 of @xmath49 , given by blowing - up all the diagonals @xcite . these @xmath54 are smooth compact manifolds with corners . the space @xmath55 is defined as the submanifold of @xmath56 where the @xmath57st point is fixed to be the point @xmath58 . if @xmath59 is an imbedding of @xmath60 into @xmath1 , then one can force some of the points in the configuration space to lay on @xmath59 , thus obtaining the configuration space @xmath61 of @xmath62 points on @xmath63 and @xmath64 points in @xmath1 . we denote by @xmath65 its compactification . putting all these spaces together , one gets a fiber bundle @xmath66 , whose fibers are compact . next , one considers the maps @xmath67 and observes that they extend smoothly to maps @xmath68 . let @xmath69 be a symmetric top form on @xmath70 which we assume to be concentrated around some fixed direction . here `` symmetric '' means that @xmath69 satisfies @xmath71 , where @xmath29 is the antipodal map on @xmath70 . the pull - back via @xmath72 of @xmath69 is a smooth @xmath73-form @xmath74 on @xmath75 called _ tautological form_. forms @xmath76 are defined as the pull - back of @xmath69 via the map @xmath77 where @xmath78 is a component of the compactified configuration space of @xmath62 points on @xmath60 , @xmath79 forgets the @xmath64 points outside the knot , @xmath80 is the projection on the @xmath23th point and @xmath81 is the normalized derivative @xmath82 . we are now ready to define the maps @xmath83 . we consider a diagram @xmath84 and number all its edges and vertices ( if they are not already numbered by their decoration ) . then to the edge between the vertices @xmath23 and @xmath24 we associate the tautological form @xmath74 , and take the wedge product over all the edges of @xmath22 . finally we integrate ( push - forward ) this product along the map @xmath66 , where @xmath62 is the number of internal vertices of @xmath22 and @xmath64 the number of external vertices . extending this map by linearity we obtain @xmath85 as shown in @xcite these are cochain maps for every @xmath5 and they induce injective maps in cohomology for @xmath86 . + one can wonder whether there exists an analogous construction for the dual theory , namely , whether one can define chain maps @xmath87 or , at least , maps in homology @xmath88 an answer can be given for the case @xmath86 by associating to every trivalent chord diagram @xmath89 a cycle of @xmath0 , denoted by @xmath90 , constructed as follows . consider first an `` imbedding '' @xmath91 of the diagram @xmath22 in @xmath1 , i.e. , an immersion of the oriented circle of @xmath22 into @xmath1 whose only singularities are transversal double points , and such that @xmath92 if and only if @xmath93 and @xmath94 are the end - point of a chord in @xmath22 . we assume that the indices @xmath23 and @xmath24 are the same indices of the chord in @xmath22 which has been contracted to form the singular point @xmath95 . next , for every pair of points @xmath96 ( say , @xmath97 ) such that @xmath92 and for every @xmath98 , we consider the following loop in @xmath1 : @xmath99$},\\ \mathbf{z}\,\delta\exp \left(1/[(t - t_i)^2-\epsilon^2]\right ) & \text{if $ t\in [ t_i-\epsilon , t_i+\epsilon]$ } , \end{cases}\ ] ] with @xmath100 . by adding all the loops @xmath101 to the immersion @xmath91 in correspondence with all the double points , we remove ( blow - up ) all the singularities and obtain a family of imbeddings which is parameterized by @xmath11 points on @xmath102 , where @xmath11 is the number of chords of @xmath22 . we look at this family of imbeddings as a @xmath103-cycle in @xmath0 . ( we remark that the main result of @xcite is that the cycle of imbeddings obtained from the chord diagram @xmath22 is dual to the de rham cocycle obtained from a trivalent graph cocycle containing @xmath22 , provided this graph cocycle exists ) . finally we extend these maps @xmath104 by linearity and we have : [ triv-4 t ] the maps @xmath105 defined above , descend to maps on @xmath41 . since @xmath41 is isomorphic to @xmath106 modulo the @xmath46 and @xmath47 relations , we have to check that the linear combination of the diagrams of figure [ 4t - rel - d ] , [ 1t - rel - d ] , [ 4t - rel - p ] and [ 1t - rel - p ] are sent to trivial cycles . consider in fact the four cycles of imbeddings corresponding to the diagrams of a @xmath46 relation . these cycles ( or better , their images ) are equal everywhere except near a small ball in @xmath1 , where they are described as follows : two strands of the imbedded circle meet in a double point @xmath107 ( which is then blown - up ) and a third strand is blown - up around one of the four ( half ) strands merging in @xmath107 . denote by @xmath108 the sum of these four cycles . now , the chain of imbeddings whose boundary is @xmath108 is constructed by considering a chain in @xmath0 equal to the cycles in @xmath108 , except that the third strand is lifted around @xmath107 along an @xmath109-sphere centered in @xmath107 , with 4 holes corresponding to the four ( half ) strands merging in @xmath107 ( see figure [ ch-4 t ] ) . notice that the orientation of the @xmath110-chain is automatically fixed by the orientation of the strand we are lifting and the fact that we have fixed , once for all , an orientation for @xmath1 . as for the @xmath47 relations , one can notice that when we collapse and blow - up a `` short chord '' ( i.e. , a chord whose vertices are consecutive vertices on the oriented circle ) we produce a cycle as in figure [ ch-1 t ] . if @xmath111 we have the difference of two ordinary knots which are clearly isotopic , namely there exist a 1-chain of imbeddings whose boundary is precisely the difference of these two knots . in higher dimensions , one can split the `` blow - up '' into two chains , @xmath112 and @xmath113 . suppose in fact that the two strands @xmath114 and @xmath115 meeting at the double point lie on a plane in @xmath1 ; then set @xmath112 to be given by those imbeddings lying on one side of the plane determined by @xmath114 and @xmath115 , and @xmath113 those lying on the other side , in such a way that the cycle of imbeddings which `` blows - up '' the double point is @xmath116 . therefore , the chains @xmath112 and @xmath113 play the role for @xmath5 of the two ordinary knots , and it is not difficult to see that there exists a higher dimensional analogue to the 1-chain connecting the two ordinary knots , namely that there exists a @xmath117-chain of embeddings whose boundaries are @xmath112 and @xmath118 . in the quest for nontrivalent graph cocycles , one easily sees that there are no nontrivial elements in @xmath119 , @xmath120 , @xmath121 , @xmath122 and @xmath123 , and that the first example of nontrivial nontrivalent graph cocycle is the generator of @xmath124 . this cocycle is represented in figure [ pla ] . theorem [ thm - main ] is proved by showing that the image of this graph cocycle through @xmath125 is a nontrivial cohomology class of @xmath0 . this will follow form the evaluation of this de rham cocycle on the dual cycle , which we are now going to construct . let us first consider the linear combination of diagrams of figure [ alp ] . we notice that it can be seen as the difference between two chains of diagrams @xmath126 ( the @xmath127 are the chains of diagrams in square brackets ) such that the boundary @xmath40 of each @xmath127 gives one of the @xmath46 relations of figure [ 4t - rel - p ] . it is also immediate to check that @xmath126 is indeed a cycle . next we associate a @xmath128-chain of @xmath0 to @xmath129 and @xmath130 as follows : since @xmath131 is a linear combination of trivalent diagrams , we can apply the map @xmath132 of equation . as shown in the proof of proposition [ triv-4 t ] the chain @xmath133 , resp . @xmath133 , is the boundary of a @xmath128-chain of @xmath0 which we denote by the symbol @xmath133 , resp . namely , we have : @xmath135 more specifically , what we have to do is to consider an `` imbbeding '' of the first diagram of @xmath129 , resp . @xmath130 , i.e. , an immersion of the oriented circle of this diagram into @xmath1 which is an imbedding except for a double and a triple point obtained by identifying the end - points of each chord . we get @xmath133 resp . @xmath134 , by blowning - up these intersections according to the constructions of section [ sec : csi ] . the difference @xmath136 of these two chains of imbeddings is the @xmath128-cycle of @xmath0 that we associate to @xmath137 , and it is represented in figure [ ch - alp ] . in other words , we have extended the map of equation to @xmath138 . let us define now @xmath139 so that @xmath140 is the configuration space integral associated to the diagram of figure [ pla ] . the theorem is proved by showing that the evaluation of @xmath141 on @xmath142 is different from zero . recall that the tautological forms @xmath74 are constructed using a symmetric top form @xmath69 on @xmath70 , concentrated around some fixed direction , say @xmath143 . this means that we are requiring @xmath69 to have support near @xmath144 and near @xmath145 . moreover we can always suppose that the images in @xmath1 of the chains @xmath133 and @xmath134 lay on a plane perpendicular to @xmath144 , except near the blow - ups . therefore the only contributions to the evaluation of @xmath146 and @xmath147 on @xmath148 , arise from those parts of the configuration space integral in which the tautological forms `` points toward the direction @xmath144 '' . this implies in particular that the evaluation of @xmath146 on @xmath148 is zero and that , in the evaluation of @xmath147 on @xmath148 , the only nontrivial contributions arise from that part of the configuration space integral @xmath147 in which @xmath149 is integrated near the blow - up of the double point and @xmath150 is integrated near the blow - up of the triple point . moreover , it not difficult to see that the evaluation of @xmath147 on @xmath148 can be explicitely computed as twice the degree of the map @xmath151 restricted to a compact subset @xmath152 of @xmath153 . this subset @xmath152 is given by the cartesian product of two 1-dimensional manifolds , two @xmath117-sphere and a @xmath73-manifold , imbedded in @xmath1 as follows : each of the 1-dimensional manifolds is linked to a @xmath117-sphere , and the @xmath73-manifold wraps around one of the spheres . for instance if we consider @xmath154 , we can easily see that the point is regular and that the number of elements in the counter - image is one . in other words , the evaluation of @xmath147 on @xmath148 is different from zero , and this concludes the proof of theorem [ thm - main ] . this work arises from a conversation with dylan thurston . i also thank alberto cattaneo , paolo cotta - ramusino and victor tourtchine for interesting discussions and useful comments .	we construct nontrivial cohomology classes of the space @xmath0 of imbeddings of the circle into @xmath1 , by means of feynman diagrams . more precisely , starting from a suitable linear combination of nontrivalent diagrams , we construct , for every even number @xmath2 , a de rham cohomology class on @xmath0 . we prove nontriviality of these classes by evaluation on the dual cycles .
in this paper we study the local minimizers of @xmath9 where @xmath10 is a bounded and smooth domain in @xmath11 , @xmath12 is the characteristic function of the set @xmath13 , and @xmath2 are positive constants such that @xmath14 the class of admissible functions @xmath15 consists of those functions @xmath16 , with @xmath17 , such that @xmath18 for a given boundary datum @xmath19 . this type of problems arises in jet flow models with two ideal fluids , see e.g. @xcite and @xcite page 126 , and has been studied in @xcite for @xmath8 . when the velocity @xmath20 of the planar flow depends on the gradient of the stream function @xmath5 in power law @xmath21 ( see @xcite ) , then the resulted problem for steady state admits a variational formulation with the functional . in higher dimensions , this models heat ( or electrostatic ) energy optimization under power fourier law , see @xcite . for admissible functions in @xmath22 the analogous problem has been studied in @xcite . however , the two - phase problem for general growth functionals has remained fundamentally open . towards this direction there are only some partial results available under the assumption of small lebesgue density on the negative phase , see @xcite . this is due to the lack of a monotonicity formula for @xmath23 . however , some weak form of monotonicity type formula is known for the modified alt - caffarelli - friedman functional , namely a discrete monotonicity formula in two spatial dimensions when @xmath24 is close to 2 , see @xcite . the aim of this paper is twofold and contributes into the regularity theory of the two - phase free boundary problems : first , we define a suitable notion of flatness for free boundary points which allows to partition the set @xmath25 into to disjoint subsets @xmath26 and @xmath27 . here @xmath28 is the set of flat free boundary points and @xmath27 the set of non - flat points . these sets are determined by the critical flatness constant @xmath29 , such that if the flatness at @xmath30 is less that @xmath29 then the free boundary must be regular in some vicinity of @xmath31 . consequently we can stratify the free boundary points and prove linear growth at the non - flat points of free boundary ( see section [ sec : main ] for precise definitions and statements ) . the advantage of this approach is that it avoids using the optimal regularity for @xmath5 everywhere and hence circumvents the obstacle imposed by the lack of monotonicity formula . however , our technique renders the local lipschitz continuity using a simple consequence of theorem a below . observe that the non - flat points @xmath32 are more interesting to study and it is vital to have linear growth at such points @xmath31 in order to classify the blow - up profiles . second , to study the flat points @xmath33 we apply the regularity theory developed for viscosity solutions of two - phase free boundary problems . to do so we prove that any local minimizer is also a viscosity solution . at flat points we get that the free boundary @xmath25 is very close to a plane in a suitable coordinate system . consequently , @xmath5 must be @xmath34monotone with @xmath35 small , which in turn implies that the free boundary is @xmath36 in some vicinity of @xmath31 . this approach , which is based on the fusion of variational and viscosity solutions , appears to be new and very useful . finally , from here we conclude the partial regularity of @xmath25 , that is @xmath25 is countably rectifiable and @xmath37 , where @xmath38 is the @xmath7-dimensional hausdorff measure . it is worthwhile to point out that our approach is new even for the classical case @xmath8 . in the forthcoming section [ sec : main ] we give the precise statements of the results that we prove . a detailed plan on the organization of the paper will be presented at the end of section [ sec : main ] . @xmath39 = + @xmath40 @xmath41 , + @xmath42 @xmath41 , + @xmath43 @xmath31 @xmath44 , @xmath45 , + @xmath46 @xmath47 , + @xmath48 mean value integral , + @xmath49 the volume of unit ball , + @xmath50 the positivity set of @xmath5 , + @xmath51 the negativity set of @xmath5 , + @xmath27 the set of non - flat free boundary points , see definition [ def - n - set ] , + @xmath28 @xmath52 , + @xmath53@xmath54 , + @xmath55 @xmath56 , @xmath57 the bernoulli constants . the existence of bounded minimizers of the functional in can be easily established using the semicontinuity of the @xmath58dirichlet energy and the weak convergence in @xmath59 , and can be found in @xcite . let now @xmath60 and @xmath61 be the slab of height @xmath62 in unit direction @xmath63 . let @xmath64 be the minimal height of the slab containing the free boundary in @xmath65 , i.e. @xmath66 put @xmath67 clearly @xmath68 is * non - decreasing * in @xmath69 . let @xmath5 be a local minimizer of . then , for any bounded subdomain @xmath70 there are positive constants @xmath29 and @xmath71 depending only on @xmath72 and @xmath73 such that , for any @xmath74 one of the following two alternatives holds : * if @xmath75 , for all @xmath76 , then @xmath77 for all @xmath78 , * if @xmath79 for some @xmath80 then the free boundary @xmath25 is @xmath36 in some neighbourhood of @xmath4 . we call @xmath81 * the critical flatness constant. the statement in theorem a leads to the following definition : [ def - n - set ] we say that @xmath82 is non - flat if @xmath83 for all @xmath84 such that @xmath85 . the set of all non - flat points is denoted by @xmath86 or @xmath27 for short . notice that if @xmath87 then @xmath88 , for some @xmath80 . so theorem a gives a partition of the free boundary of the form @xmath89 where @xmath90 is the set of flat free boundary points . let @xmath5 be as in theorem a. then , for any subdomain @xmath91 we have @xmath92 and @xmath93 in particular , @xmath94 . we remark that , as a consequence of theorem a , we also obtain local lipschitz continuity for the minimizers . let @xmath5 be as in theorem a. then for any subdomain @xmath70 there is a constant @xmath95 depending only on @xmath96 and @xmath71 such that @xmath97 . the methods and the techniques that we employ to prove theorems a and b pave the way to a number of new approaches . first , we fuse the variational methods with the viscosity theory . this is done by proving that any local minimizer @xmath16 is also a viscosity solution ( see section [ sec : visc ] , and in particular theorem [ th : viscosity ] ) . the key ingredient in the proof is the linear development of a nonnegative @xmath58harmonic function @xmath98 in @xmath13 near @xmath99 that vanishes continuously on @xmath100 , see lemma [ lemma : linear ] . there is a subtle point in the proof of the linear development lemma which amounts to the following claim : if @xmath60 and @xmath101 with @xmath102 then @xmath5 has linear growth near @xmath4 , i.e. there is a constant @xmath103 ( depending on @xmath4 ) such that @xmath104 near @xmath4 . indeed , by standard barrier argument we have that @xmath105 where @xmath106 . therefore @xmath107 has linear growth near @xmath4 . now the linear growth of @xmath108 near @xmath4 follows form lemma [ lem : dndooo ] . clearly the same claim is valid if @xmath109 and @xmath102 . we stress on the fact that lemma [ lemma : linear ] on linear development remains valid for solutions to a wider class of equations for which harnack s inequality and hopf s lemma are valid . second , we compare @xmath110 with the minimal height @xmath68 of the parallel slab of planes containing @xmath111 , for @xmath60 . more precisely , take @xmath112 , and fix @xmath113 , then @xmath114 or @xmath115 consequently , for given @xmath60 there are two alternatives : either for some @xmath116 we arrive at and this will mean that @xmath4 is a flat point of @xmath25 ( if @xmath113 is small ) or holds for sufficiently large @xmath117 . the latter implies linear growth at @xmath4 . note that the non - flat points are more interesting to study and having the linear growth at such points allows one to use compactness argument and blow - up @xmath5 in order to study the properties of the resulted configuration as done in the proofs of , and . note that if holds for @xmath118 then we have linear growth for @xmath5 near @xmath4 unto the level @xmath119 , see corollary [ cor : lin flat ] . altogether , this approach allows us to prove the main properties of the free boundary without using the full optimal regularity of @xmath5 and can be applied to a wide class of variational free boundary problems with two phases . a diagram showing the scheme of the proof is given below . as for the proof of the partial regularity result , i.e. @xmath120 we employ a non - degeneracy result obtained in proposition [ str - nd - hi ] for @xmath108 and some estimates for the radon measure @xmath121 given in lemma [ lem-7.11 ] . this is a standard approach but more involved because the linear growth is valid only at non - flat points of the free boundary . in section [ sec : tec ] we collect some material , mostly of technical nature , that we will use in the other sections . in particular , we prove the continuity of minimizers , by showing that @xmath122 if @xmath123 and @xmath124 if @xmath125 . we also recall the liouville s theorem and some basic properties of minimizers . finally we show that @xmath108 is non - degenerate , in the sense of proposition [ str - nd - hi ] , and a coherence lemma ( see lemma [ lem : dndooo ] ) . in section [ sec : visc ] we prove that any minimizer of the functional in is also a viscosity solution , according to definition [ def : visc ] . this will allow us to apply the regularity theory developed in @xcite for viscosity solutions and infer that the free boundary is @xmath36 regular near flat points . in section [ sec : eps ] we discuss and compare the notions of @xmath126-monotonicity of minimizers and of slab flatness of the free boundary . section [ lin vs flat ] is devoted to the proof of theorem a and section [ sec : thb ] contains the set up for the proof of theorem b. in section [ sec : blow ] we deal with the blow - up of minimizers proving some useful convergence and finish the proof of theorem b. then in section [ sec - ssuka ] we prove theorem c. the paper contains also an appendix , where we prove a result needed in section [ sec : visc ] . = [ diamond , draw , fill = red!20 , text width=7.5em , text badly centered , node distance=3 cm , inner sep=0pt ] = [ rectangle , draw , fill = blue!20 , text width=11em , text centered , rounded corners , minimum height=4em ] = [ draw , -latex ] = [ draw , ellipse , fill = red!20 , node distance=3 cm , minimum height=2em , text width=11em , text centered ] = [ draw , ellipse , fill = red!20 , node distance=3 cm , minimum height=2em ] in this section we prove some basic properties of minimizers . we first prove the continuity of minimizers of with any @xmath128hlder modulus of continuity , with @xmath129 , if @xmath130 and log - lipschitz modulus of continuity if @xmath123 . our method is a variation of @xcite and uses some standard inequalities for the functionals with @xmath58power growth . [ lemma : coherence ] let @xmath5 be a minimizer of . then for @xmath125 , we have that @xmath131 for any bounded subdomain @xmath132 , and consequently @xmath133 for any @xmath134 , * for @xmath123 , we have that @xmath135 , for any bounded subdomain @xmath136 , and thus @xmath5 is locally log - lipschitz continuous . in particular , @xmath137 for any @xmath138 and for any @xmath6 . fix @xmath139 and @xmath140 such that @xmath141 . let @xmath98 be the solution of @xmath142 comparing @xmath143 with @xmath144 in @xmath145 yields @xmath146 for some @xmath147 . on the other hand , the following estimate is true ( see @xcite page 100 ) @xmath148 for some tame constant @xmath149 depending on @xmath150 and @xmath24 . introduce the function @xmath151 defined as follows @xmath152 then from the basic inequalities @xmath153 that are valid for any @xmath6 ( see @xcite page 240 ) , we infer the estimate @xmath154 up to renaming @xmath155 . indeed , the case @xmath125 follows from the second inequality in . as for the remaining case @xmath123 we have by hlder s inequality @xmath156 and follows . furthermore , for any @xmath157 , we set @xmath158 then , from hlder s inequality we have @xmath159 we would also need the following estimate for a @xmath58harmonic function @xmath98 : there is @xmath160 such that for all balls @xmath141 , with @xmath139 , there exists a universal constant @xmath161 such that the following campanato type estimate is valid @xmath162 see @xcite theorem 6.4 for @xmath163 and @xcite theorem 5.1 for @xmath164 . denote @xmath165 , then , using , we obtain @xmath166 where , in order to get , we used campanato type estimate . from the triangle inequality for @xmath167 norm we have @xmath168 and so , combining this with , we obtain @xmath169\\\nonumber & \le & c\left\{\\|v({{\nabla}}u)-v({{\nabla}}v)\\|_{2 , r}+ \left(\frac rr\right)^{\frac{n+\alpha}2 } \\|v({{\nabla}}u)-(v({{\nabla}}u))_{x_0,r}\\|_{2 , r}\right\}\\\nonumber & \le & a\left(\frac rr\right)^{\frac{n+\alpha}2 } \\|v({{\nabla}}u)-v(({{\nabla}}u))_{x_0,r}\\|_{2 , r}+b r^{\frac{n}2},\\\nonumber\end{aligned}\ ] ] for some tame positive constants @xmath170 and @xmath171 . introduce @xmath172 then the former inequality can be rewritten as @xmath173 with some positive constants @xmath174 . applying lemma 2.1 from @xcite chapter 3 , we conclude that there exist @xmath175 and @xmath161 such that @xmath176 for all @xmath177 , and hence @xmath178 for some tame constant @xmath147 . this shows that @xmath179 is locally bmo . the log - lipschitz estimate for @xmath123 now follows from @xcite theorem 3 . the hlder continuity follows from sobolev s embedding and the john - nirenberg lemma . from lemma [ lemma : coherence ] it follows that for any @xmath70 there is a constant @xmath147 depending only on @xmath180 and @xmath181 such that if @xmath123 and @xmath182 , then @xmath183 see @xcite . this section is devoted to liouville s theorem , that we use in the proof of proposition [ prop : lin flat ] . we add the proof here . [ th : liouville ] let @xmath41 be a @xmath24-harmonic function in @xmath11 such that latexmath:[\[\label{pgrejdgkerh } some @xmath147 . then @xmath41 is a linear function in @xmath11 . for any @xmath44 , we introduce the scaled function @xmath185 hence @xmath186 is a @xmath24-harmonic function and @xmath187 thanks to . moreover , from the @xmath188-estimates for @xmath24-harmonic functions in @xmath189 ( for some @xmath190 ) , see @xcite , we have that @xmath191 and , moreover , @xmath192 for a positive constant @xmath193 , depending only on @xmath150 , @xmath24 and @xmath194 . hence , taking @xmath195 and @xmath196 in , we obtain that for any @xmath44 @xmath197 in particular , holds true for any @xmath198 . therefore , letting @xmath199 and sending @xmath200 in the formula above , we obtain that @xmath201 hence , @xmath41 is linear in @xmath189 . this completes the proof in view of the unique continuation theorem @xcite . [ prop : tec ] let @xmath203 be a local minimizer of . then * @xmath204 in the sense of distributions and @xmath205 in @xmath206 , * for any @xmath70 there is @xmath207 depending only on @xmath208 and @xmath181 such that if @xmath209 then @xmath210 where @xmath155 is a tame constant . p.1 follows from a standard comparison of @xmath5 and @xmath211 , where @xmath212 is a suitable smooth and compactly supported function . p.2 follows from @xcite . it is convenient to define @xmath213 with @xmath214 . as a consequence , the functional in can be rewritten in an equivalent form @xmath215 notice that the last term does not affect the minimization problem , and so if @xmath5 is a minimizer for @xmath202 , then it is also a minimizer for @xmath216 observe that if @xmath217 then the free boundary @xmath218 for the minimizer @xmath5 of @xmath202 coincides with @xmath47 . indeed , let @xmath219 , then we clearly have that if @xmath220 then there is @xmath44 such that @xmath221 in @xmath65 , and so @xmath5 is @xmath58superharmonic in @xmath65 . on the other hand , we have that @xmath222 , and so we get a contradiction with p.1 of proposition [ prop : tec ] . therefore @xmath223 . the functional @xmath224 preserves the minimizers under certain scaling . this property is a key ingredient in a number of arguments to follow . more precisely , let @xmath5 be a minimizer of , and take @xmath60 and @xmath44 such that @xmath225 . fixed @xmath157 , set also @xmath226 , for some constant @xmath227 . then one can readily verify that @xmath228^p\lambda{{\chi_{\{u_\rho>0\}}}}= \left[\frac{\rho}{s}\right]^p\frac1{\rho^n}\int_{b_\rho(x_0)}\|{{\nabla}}u\|^p+\lambda{{\chi_{\{u>0\}}}}.\end{aligned}\ ] ] in particular if we let @xmath229 then @xmath230 therefore if @xmath5 is minimizer of @xmath224 in @xmath231 then the scaled function @xmath232 is a minimizer of @xmath224 in @xmath189 . in this section we deal with a strong form of non - degeneracy for minimizers of . for @xmath8 , this result is contained in @xcite ( see in particular theorem 3.1 there ) . we use a modification of an argument from @xcite lemma 2.5 . [ str - nd - hi ] for any @xmath233 there exists a constant @xmath234 such that for any local minimizer of and for any small ball @xmath235 @xmath236 by scale invariance of the problem we take @xmath237 for simplicity and put @xmath238 since @xmath108 is @xmath58subharmonic ( recall p.1 in proposition [ prop : tec ] ) , then by @xcite theorem 3.9 @xmath239 introduce @xmath240 & \text{in}\ b_{\sqrt\kappa}\setminus b_{\kappa},\\ 0 & \text{in}\ b_{\kappa } , \end{array } \right.\ ] ] where @xmath241 and @xmath242 is chosen so that @xmath243 that is @xmath244 furthermore , by a direct computation we can see that @xmath245 and @xmath246 see @xcite . thus @xmath247 if @xmath248 is sufficiently small , say , @xmath249 it is clear that @xmath250 on @xmath251 , thanks to , hence by the minimality of @xmath5 ( recall also subsection [ rem : one phase ] ) @xmath252 now we observe that @xmath253 while @xmath254 therefore , from , we have that @xmath255 where to get the last line we also used the fact that @xmath98 is a @xmath58supersolution in @xmath256 ( recall ) and ) . moreover , by , we have that @xmath257 on @xmath258 , for some @xmath147 . thus @xmath259 on the other hand , from trace estimate , young s inequality and , we get @xmath260 where @xmath261 is the conjugate of @xmath24 and @xmath262 thereby , putting together and , we obtain @xmath263 which implies that @xmath264 in @xmath265 if @xmath266 is small enough . as a consequence of proposition [ str - nd - hi ] we have : [ cor - nondeg ] let @xmath5 be as in proposition [ str - nd - hi ] . let @xmath30 and @xmath44 such that @xmath267 . then @xmath268 where @xmath269 depends only on @xmath222 . the last technical estimate is very weak and of pointwise nature . it is used in the proof of theorem [ th : viscosity ] and serves a preliminary step towards the proof of theorem a. [ lem : dndooo ] let @xmath5 be a bounded local minimizer of . let @xmath60 and @xmath44 small such that @xmath225 . assume that @xmath270 ( resp . @xmath271 ) , for some constant @xmath95 depending on @xmath4 . then there exists a constant @xmath272 such that @xmath273 ( resp . @xmath274 ) . we will show only one of the claims , the other can be proved analogously . suppose that @xmath275 and we claim that @xmath276 where @xmath277 , for any @xmath278 . to prove this , we argue by contradiction and we suppose that fails . then there is a sequence of integers @xmath279 , with @xmath280 , such that @xmath281 observe that since @xmath5 is a bounded minimizer , then implies that @xmath282 as @xmath283 . also , notice that implies that @xmath284 now , we introduce the scaled functions @xmath285 , for any @xmath286 . then , from and , it follows that @xmath287 also , by ( used here with @xmath288 and @xmath289 ) we see that @xmath290 is a minimizer of the functional @xmath291^p\lambda{{\chi_{\{v_j>0\}}}}.\ ] ] furthermore , it is not difficult to see that implies that @xmath292 using this and caccioppoli s inequality , we infer that @xmath293 for some @xmath294 , implying that @xmath295 are uniformly bounded . so using lemma [ lemma : coherence ] we can extract a converging subsequence such that @xmath296 uniformly in @xmath297 and @xmath298 in @xmath299 for any @xmath300 . moreover , by , @xmath301^p\lambda{{\chi_{\{v_j>0\}}}}\to \int_{b_{\frac34}}\|{{\nabla}}v_0(x)\|^p , \quad { \mbox { as } } j\to+\infty.\ ] ] this , and give that @xmath302 which is in contradiction with the strong minimum principle . this shows and finishes the proof . in order to exploit the regularity theory of free boundary developed for the viscosity solutions in @xcite we shall prove that any @xmath59 minimizer of @xmath202 is also viscosity solution , as opposed to definition 2.4 in @xcite . for this , we recall that @xmath303 and @xmath304 . moreover , if the free boundary is @xmath305 smooth then @xmath306 is the flux balance across the free boundary , where @xmath307 and @xmath308 are the normal derivatives in the inward direction to @xmath309 and @xmath310 , respectively ( recall that @xmath311 is the bernoulli constant ) . [ def : visc ] let @xmath10 be a bounded domain of @xmath11 and let @xmath5 be a continuous function in @xmath10 . we say that @xmath5 is a viscosity solution in @xmath10 if * @xmath205 in @xmath312 and @xmath313 , * along the free boundary @xmath314 , @xmath5 satisfies the free boundary condition , in the sense that : * * if at @xmath315 there exists a ball @xmath316 such that @xmath317 and @xmath318 @xmath319 for some @xmath160 and @xmath320 , with equality along every non - tangential domain , then the free boundary condition is satisfied @xmath321 * * if at @xmath315 there exists a ball @xmath322 such that @xmath317 and @xmath323 @xmath324 for some @xmath325 and @xmath326 , with equality along every non - tangential domain , then @xmath327 the main result of this section is the following : [ th : viscosity ] let @xmath328 be a minimizer of . then @xmath5 is a viscosity solution in @xmath10 in the sense of definition [ def : visc ] . the proof of theorem [ th : viscosity ] , will follow from lemma [ lemma : linear ] below . it is a generalization of lemma 11.17 in @xcite to any @xmath24 ( see also the appendix in @xcite , where the authors deal with the one - phase problem in the half ball . ) we postpone the proof of lemma [ lemma : linear ] to appendix [ ap : lemma ] . [ lemma : linear ] let @xmath329 be a solution of @xmath205 in @xmath10 and @xmath330 . suppose that @xmath5 continuously vanishes on @xmath331 . then * if there exists a ball @xmath332 touching @xmath333 at @xmath4 , then either @xmath5 grows faster than any linear function at @xmath4 , or there exists a constant @xmath160 such that @xmath334 where @xmath63 is the unit normal to @xmath335 at @xmath4 , inward to @xmath10 . moreover , equality holds in in any non - tangential domain . * if there exists a ball @xmath336 touching @xmath333 at @xmath4 , then there exists a constant @xmath320 such that @xmath337 with equality in any non - tangential domain . with this , we are able to prove theorem [ th : viscosity ] . first we observe that i ) in definition [ def : visc ] is satisfied , thanks to p.1 in proposition [ prop : tec ] . to prove ii ) , we let @xmath338 , @xmath339 be a ball touching @xmath314 at @xmath4 and @xmath63 be the unit vector at @xmath4 pointing to the centre of @xmath171 . we want to show that and are satisfied for some @xmath160 and @xmath320 , with equality in every non - tangential domain . notice that @xmath340 is finite , thanks to lemma [ lemma : linear ] ( in particular , the statement b ) applied to @xmath107 ) . this follows from a standard barrier argument as one compares @xmath107 with @xmath341 where @xmath106 , @xmath69 is the radius and @xmath342 the centre of @xmath171 . thus @xmath343 is finite too , according to lemma [ lem : dndooo ] , that is @xmath344 recall that , using the notation in @xcite , the free boundary condition takes the form @xmath345 therefore it is enough to show that @xmath346 for this , we first consider the case @xmath347 , i.e. when @xmath107 is degenerate . we define the scaled function at @xmath4 @xmath348 since @xmath4 is a non - flat point of free boundary then it follows from that for any sequence @xmath349 as @xmath283 there is a subsequence @xmath350 such that @xmath351 converges to some @xmath352 . moreover , owing to lemma [ lemma : linear ] , in a non - tangential domain we have that @xmath353 without loss of generality , we may assume that @xmath354 . thus , after blowing - up , we have that @xmath355 in a cone @xmath356 for some @xmath357 . notice that @xmath358 in @xmath359 . also , @xmath360 in @xmath359 . then , by the unique continuation theorem ( see proposition 5.1 in @xcite ) we get that @xmath355 in @xmath11 . in turn , this implies that the free boundary condition is satisfied in the classical sense on the hyperplane @xmath361 . that is , @xmath362 on @xmath361 , and so @xmath363 on @xmath361 . hence is satisfied in the case @xmath347 . suppose now that @xmath326 , namely @xmath107 is non - degenerate . reasoning as above and blowing - up , we can prove that @xmath364 . it remains to show that @xmath365 . to do this , we set @xmath366 , that is @xmath367 is the free boundary of the blow - up @xmath352 . we take @xmath368 , @xmath369 , and we take the ball @xmath370 for some @xmath371 , see figure [ fig3 ] . either @xmath372 or @xmath373.,scaledwidth=70.0% ] there are three possibilities : * @xmath352 vanishes only on @xmath374 and @xmath358 in @xmath375 , * @xmath352 vanishes only on @xmath374 and @xmath372 in @xmath375 , * @xmath352 vanishes in @xmath376 . notice that * case 1 ) * can not occur , because it would imply that we deal with a one - phase problem in @xmath370 and the @xmath377density estimate for the zero set would be violated ( see theorem 4.4 in @xcite ) . consider now * case 2 ) * , and observe that on the hyperplane @xmath361 the free boundary condition is satisfied in classical sense : @xmath378 in @xmath370 . in particular , @xmath379 we claim that @xmath380 indeed , @xmath381 is @xmath24-harmonic in @xmath382 . moreover , yields that @xmath383 , therefore we have that @xmath384 pointwise in @xmath385 , and so follows . hence , from , and the unique continuation theorem @xcite we obtain that @xmath386 must be a linear function in @xmath387 . then , proposition 5.1 in @xcite implies that @xmath388 is a linear function in @xmath389 . thus the free boundary condition is satisfied in the classical sense on the plane @xmath361 including the origin , and this proves equality in in case 2 ) . now we deal with * case 3). we consider a cube @xmath390 centered at the origin such that @xmath391 , and we set @xmath392 . notice that @xmath393 in particular , @xmath394 on @xmath395 . according to the remark in subsection [ rem : one phase ] , @xmath352 is a minimizer in @xmath396 of the functional @xmath397 therefore @xmath352 is @xmath24-harmonic in @xmath396 . by maximum principle , @xmath352 can not achieve its maximum inside @xmath396 . this and imply that @xmath372 in @xmath396 , and so the free boundary coincides with @xmath361 . this concludes the proof of ii)-a ) in definition [ def : visc ] . similarly , one can also prove ii)-b ) . hence , @xmath5 is a viscosity solution , and the desired result follows . one of the main free boundary regularity theorems for viscosity solutions is formulated in terms of the @xmath34monotonicity of @xmath5 . more precisely , we have : we say that @xmath5 is @xmath398monotone if there are a unit vector @xmath399 and an angle @xmath400 with @xmath401 ( say ) and @xmath402 ( small ) such that , for every @xmath403 , @xmath404 we denote by @xmath405 the cone with axis @xmath399 and opening @xmath400 . we say that @xmath5 is @xmath398monotone in the cone @xmath406 if it is @xmath398monotone in any direction @xmath407 . one can interpret the @xmath34monotonicity of @xmath5 as closeness of the free boundary to a lipschitz graph with lipschitz constant sufficiently close to @xmath408 if we leave the free boundary in directions @xmath399 at distance @xmath266 and higher . the exact value of the lipschitz constant is given by @xmath409 . then the ellipticity propagates to the free boundary via harnack s inequality giving that @xmath314 is lipschitz . furthermore , lipschitz free boundaries are , in fact , @xmath36 regular . for @xmath8 this theory was founded by l. caffarelli , see @xcite . recently j. lewis and k. nystrm proved that this theory is valid for all @xmath6 , see @xcite . in fact , their argument does not require @xmath5 to be lipschitz . for viscosity solutions we replace the @xmath398monotonicity with the slab flatness measuring the thickness of @xmath410 in terms of the quantity @xmath411 introduced in . in other words , @xmath412 measures how close the free boundary is to a pair of parallel planes in a ball @xmath413 with @xmath414 clearly , planes are lipschitz graphs in the direction of the normal , therefore the slab flatness of @xmath314 is a particular case of @xmath34monotonicity of @xmath5 . hence , under @xmath415flatness of the free boundary we can reformulate the regularity theory `` flatness implies @xmath36 '' as follows : [ th1 ] suppose that @xmath416 with @xmath60 . then there exists @xmath417 such that if @xmath418 then @xmath419 is locally @xmath188 in the direction of @xmath63 , for some @xmath190 . we first discuss a preliminary result , that we will use for the proof of theorem a. [ prop : lin flat ] let @xmath5 be a local minimizer of @xmath202 and @xmath421 . for any @xmath278 , set @xmath422 if @xmath113 is fixed and @xmath423 for some @xmath116 , then @xmath424 for some positive constant @xmath71 , that is independent of @xmath4 and @xmath116 . otherwise if @xmath425 for some @xmath116 , then @xmath426 is a @xmath188 smooth surface , for some @xmath129 . we first deal with the case @xmath423 . in order to prove , we use a contradiction argument discussed in @xcite . hence , we suppose that fails , that is there exist integers @xmath427 , local minimizers @xmath428 and points @xmath429 such that @xmath430 and @xmath431 since @xmath428 is a local minimizer of @xmath202 in @xmath189 and @xmath432 , then @xmath428 is bounded ( see theorem 1 in @xcite ) . namely , there exists a positive constant @xmath193 , that is independent of @xmath433 , such that @xmath434 . therefore , from we have that @xmath435 , which implies that @xmath436 . hence , @xmath279 tends to @xmath437 when @xmath283 . we set @xmath438 using once more , we see that @xmath439 for any @xmath433 , we now define the function @xmath440 then , by construction , @xmath441 furthermore , from we have that @xmath442 which in turn implies that @xmath443 finally , since @xmath432 , we have that @xmath444 notice that @xmath290 is a minimizer ( according to its own boundary values ) of the scaled functional @xmath445 for @xmath446 and @xmath433 large . indeed , from and an easy computation , we get @xmath447 hence , by the change of variable @xmath448 and recalling , @xmath449 since @xmath428 is a minimizer for @xmath202 , the last formula implies that @xmath290 is a minimizer for @xmath450 . hence , from lemma [ lemma : coherence ] we obtain that for any @xmath300 and @xmath446 there exists a constant @xmath451 independent of @xmath433 such that @xmath452 for some @xmath190 . therefore , by a standard compactness argument , we have that , up to a subsequence , @xmath453 from , and we obtain that @xmath454 we claim that @xmath455 for this , notice that for any @xmath456 @xmath457 because @xmath290 is a minimizer for @xmath450 defined in . by taking @xmath458 in , we have that @xmath459 as @xmath283 . moreover , from we obtain @xmath460 as @xmath283 . thus , sending @xmath283 in and using these observations , we get @xmath461 for any @xmath456 . this implies . hence , from liouville s theorem ( see theorem [ th : liouville ] ) we deduce that @xmath98 must be a linear function in @xmath11 . without loss of generality we can take @xmath462 for some positive constant @xmath155 . on the other hand , implies that the following inequality holds true for the function @xmath290 : @xmath463 by the uniform convergence in , we have that for any @xmath402 there is @xmath464 such that @xmath465 whenever @xmath466 . since @xmath467 is @xmath81 thick in @xmath189 it follows that there is @xmath468 such that @xmath469 , for some @xmath470 , where @xmath471 is the unit direction of @xmath472 axis and @xmath473 . then we have that @xmath474 , which is a contradiction if @xmath126 is small . this finishes the proof of . if @xmath475 for some @xmath116 , we use theorem [ th : viscosity ] to obtain that @xmath5 is also a viscosity solution in the sense of definition [ def : visc ] . therefore , we can apply the regularity result in theorem [ th1 ] , thus obtaining the desired claim . this concludes the proof of proposition [ prop : lin flat ] . with the aid of proposition [ prop : lin flat ] we now complete the proof of theorem a. the argument in proposition [ prop : lin flat ] shows that either there are finitely many integers @xmath116 such that @xmath476 and @xmath477 or there are infinitely many @xmath116 such that and hold true . in the first case , there exists @xmath478 such that @xmath479 , and so @xmath480 is a @xmath188 smooth surface . in the second case , we have linear growth of @xmath5 at the free boundary point @xmath4 where the flatness does not improve . suppose now that we are given @xmath44 . then , either @xmath481 or @xmath482 . in the first case , we obtain that @xmath314 is a @xmath188-surface . in the second case we argue as follows : there exists @xmath278 such that @xmath483 hence , by the definition of @xmath484 given in , we have that @xmath485 this means that we are in the position to apply proposition [ prop : lin flat ] , that implies linear growth of @xmath5 at the level @xmath486 . a refinement of theorem a is given by the following : [ cor : lin flat ] let @xmath29 be the constant given in theorem a. then , if @xmath487 and @xmath488 , we have that @xmath489 where @xmath71 is the constant given by theorem a. here we introduce a functional that is a generalization to any @xmath6 of the one introduced by alt , caffarelli and friedman in the case @xmath8 , and we show that this functional is bounded at non - flat free boundary points , thanks to the linear growth ensured by theorem a. for this , we let @xmath490 , where @xmath491 and @xmath492 . we define the functional @xmath493 where @xmath60 and @xmath44 is such that @xmath225 . precisely , we show the following : let @xmath29 be fixed , @xmath70 be a subdomain and @xmath494 be such that @xmath495 . then there exist @xmath496 depending only on @xmath497 and @xmath181 such that @xmath498 since @xmath499 is nonnegative @xmath58subsolution ( recall p.1 in proposition [ prop : tec ] ) , we can apply caccioppoli s inequality , obtaining that @xmath500 from this and corollary [ cor : lin flat ] we have that @xmath501 for some @xmath502 . hence , using fubini s theorem , we have @xmath503 which implies the desired result . in @xcite we prove the converse statement in some sense . more precisely we show that if @xmath504 and @xmath123 is close to @xmath505 then @xmath506 is discrete monotone . in this section we introduce the set - up in order to prove theorem b. for this , we recall the notation introduced in section [ sec : main ] ( recall in particular definition [ def - n - set ] and formula ) . we first show that @xmath121 is radon measure . [ lem-7.11 ] let @xmath5 be a local minimizer of . then , the following statements hold true . @xmath121 is a radon measure and , for any @xmath507 and @xmath44 such that @xmath508 , there holds @xmath509 * for a given subdomain @xmath70 there is @xmath510 such that @xmath511 for all @xmath512 , where @xmath147 depends on @xmath513 , @xmath150 , @xmath24 , @xmath181 and @xmath71 ( given by theorem a ) . * for each @xmath33 there is @xmath514 such that @xmath515 for some @xmath147 that depends on @xmath513 , @xmath150 , @xmath24 , @xmath181 and @xmath31 . we first show . for this , we take for simplicity @xmath516 . observe that by p.1 in proposition [ prop : tec ] we have that @xmath517 in the sense of distributions . also , for any @xmath518 , @xmath519 therefore , integrating both sides of the last identity over the interval @xmath520 with respect to @xmath521 , we infer that @xmath522 this proves . to prove we argue towards a contradiction . so , for any @xmath280 , we let @xmath523 and @xmath524 such that @xmath525 we also introduce @xmath526 . since @xmath523 , it follows from theorem a that @xmath5 has uniform linear growth at @xmath527 . this property translates to the scalings of @xmath98 at @xmath527 giving uniform linear growth for the functions @xmath290 at the origin , i.e. @xmath528 where @xmath71 is the constant in theorem a. notice that @xmath290 is a minimizer of , so it is locally @xmath529 , for some @xmath190 , thanks to lemma [ lemma : coherence ] . hence @xmath530 is uniformly bounded in @xmath188 , and so is @xmath531 for any fixed @xmath502 , thanks to caccioppoli s inequality . therefore , we can extract a subsequence @xmath532 such that @xmath533 as @xmath534 and @xmath535 is a minimizer of @xmath202 in @xmath536 . moreover , by , @xmath537 with @xmath538 . as a consequence , @xmath539 vanishes identically in @xmath189 , by the minimum principle for the @xmath58harmonic functions . on the other hand , from corollary [ cor - nondeg ] we have that @xmath540 , and this gives a contradiction . thus the proof of is finished as well . the proof of the non - uniform estimate follows from a similar argument , by replacing @xmath71 with a constant @xmath541 depending on @xmath542 and @xmath543 . as a consequence of lemma [ lem-7.11 ] , we obtain the first part of theorem b. more precisely : [ coro : zero ] let @xmath544 be such that @xmath545 . then @xmath546 . it follows from and that for each @xmath547 there is @xmath514 such that @xmath548 thus @xmath549 is a besicovitch type covering of @xmath550 . applying besicovitch s covering lemma , we have that there is a subcovering @xmath551 of balls @xmath552 such that @xmath553 for some dimensional constant @xmath554 and @xmath555 where the balls @xmath556 in each @xmath557 are disjoint and @xmath557 are countable . now we take a small number @xmath558 , and we observe that if @xmath559 then holds for any @xmath560 . hence , without loss of generality , we take @xmath561 for any @xmath562 . therefore , using , @xmath563 where @xmath564 is the @xmath565 neighbourhood of @xmath550 . thus , choosing a finite covering of @xmath564 with balls @xmath566 , with @xmath567 , such that @xmath568 and @xmath569 and using , we have that @xmath570 and letting @xmath571 we arrive at the desired result . we end this section by the following density type estimate to be used in the final stage of the proof of theorem b. [ density est ] for any subdomain @xmath70 there is a positive constant @xmath572 depending on @xmath573 and @xmath574 such that @xmath575 notice that if @xmath576 then holds true with @xmath577 . so we focus on the case in which @xmath578 . we fix @xmath44 such that @xmath225 , and we take a function @xmath579 that is @xmath58harmonic in @xmath65 and such that @xmath580 on @xmath581 . then , reasoning as at the beginning of the proof of lemma [ lemma : coherence ] ( in particular , using , , and ) , we have that there exists a tame constant @xmath582 such that @xmath583 now we claim that there is a constant @xmath584 independent of @xmath69 such that @xmath585 notice that by comparison principle it follows that @xmath586 . we prove the first inequality in using a contradiction argument based on compactness , the second one can be proved analogously . suppose that , for any @xmath280 , there are @xmath587 and @xmath524 with @xmath588 such that @xmath589 now , define @xmath590 and @xmath591 , for any @xmath286 . we recall that implies that @xmath592 is a minimizer for @xmath202 in @xmath189 . so , it follows from p.1 in proposition [ prop : tec ] , caccioppoli s inequality and theorem a that @xmath593 where @xmath71 is the constant introduced in theorem a. also , we observe that @xmath594 in @xmath189 and that @xmath595 on @xmath596 . in particular , @xmath597 this and imply that @xmath598 , up to renaming @xmath599 ( recall that @xmath600 are @xmath24-subharmonic , thanks to p.1 in proposition [ prop : tec ] ) . moreover , from the local regularity theory for @xmath58harmonic functions we have that @xmath601 are uniformly @xmath36 in @xmath602 . consequently , we have that there is a subsequence ( still denoted by @xmath601 ) such that @xmath603 weakly in @xmath604 and uniformly in @xmath602 , as @xmath283 . in particular , by , @xmath605 as for the sequence @xmath592 , from and lemma [ lemma : coherence ] we infer that there is a subsequence ( still denoted by @xmath592 ) such that @xmath606 strongly in @xmath607 for any @xmath300 and @xmath608 uniformly in @xmath609 , as @xmath283 . furthermore , @xmath352 is a minimizer of @xmath202 and from the convergence of traces it follows that @xmath610 on @xmath596 . also , by corollary [ cor - nondeg ] we have that @xmath611 , and by proposition [ prop : tec ] we have that @xmath352 is @xmath58subharmonic in @xmath189 . altogether , we have obtained that @xmath612 but this is a contradiction to the comparison principle for @xmath24-harmonic functions . the second inequality of can be proven analogously . now we are ready to finish the proof of . from and we have @xmath613 for @xmath614 to be chosen later . observe that by standard gradient estimates @xmath615 up to renaming @xmath147 , where the last inequality follows from the maximum principle and theorem a. therefore , for any @xmath616 @xmath617 if we choose @xmath618 small enough . returning to we finally get that @xmath619 this finishes the proof of lemma [ density est ] . in this section we study the blow - up sequences of a minimizer of and prove a simple compactness result , that we use to conclude the proof of theorem b. for this , let @xmath5 be a minimizer of @xmath202 and @xmath60 . consider a sequence of balls @xmath620 , with @xmath621 . we call the sequence of functions defined by @xmath622 the blow - up sequence of @xmath5 with respect to @xmath620 . clearly @xmath623 is also a local minimizer . [ tech-2 ] let @xmath624 and @xmath623 be a blow - up sequence . then there is a blow - up limit @xmath625 with linear growth such that for a subsequence * @xmath626 in @xmath627 for any @xmath628 , * @xmath629 weakly in @xmath630 for any @xmath300 , * @xmath631 locally in hausdorff distance , * @xmath632 in @xmath633 . the first and second claims follow from lemma [ lemma : coherence ] and a customary compactness argument to show that the blow - up limit @xmath352 exists . we recall the definition of hausdorff distance : @xmath634 let @xmath635 be a ball not intersecting @xmath636 . if @xmath358 in @xmath637 then , by locally uniform convergence , @xmath638 in @xmath639 , thus implying that @xmath640 . as for the case @xmath641 in @xmath642 , it follows from proposition [ str - nd - hi ] that @xmath643 , for any small @xmath35 . thus , by the uniform convergence , we have that @xmath644 if @xmath116 is sufficiently large . from proposition [ str - nd - hi ] we conclude that @xmath645 in @xmath646 . in both cases we infer that @xmath647 does not intersect @xmath646 if @xmath116 is large enough . conversely , if @xmath642 does not intersect @xmath647 for any large @xmath116 , then either @xmath638 in @xmath648 or @xmath649 in @xmath642 . in the first case , @xmath623 is @xmath58harmonic in @xmath642 and hence so is @xmath352 . consequently , either @xmath358 in @xmath642 or @xmath650 in @xmath642 . thus @xmath642 does not intersect @xmath651 . in the second case , we have that @xmath641 , so that again @xmath651 does not intersect @xmath642 . reasoning as above and using a covering argument one can show that , for a fixed compact set @xmath652 , the quantity @xmath653 in the definition of @xmath654 , with @xmath655 and @xmath656 , can be chosen as small as we wish . the last statement follows from the non - degeneracy of @xmath108 given by corollary [ cor - nondeg ] , the convergence of @xmath631 in hausdorff distance and the fact that the @xmath150-dimensional hausdorff measure @xmath657 , since @xmath352 is also minimizer and corollary [ coro : zero ] applies . hence the proof of proposition [ tech-2 ] is concluded . [ multi - blow ] in view of proposition [ str - nd - hi ] we see that when we consider the blow - up of a minimizer , the limit can not vanish , no matter how many times we blow - up the minimizer @xmath5 at a non - flat point . we now finish the proof of theorem b. more precisely , we show that @xmath658 first observe that @xmath659 , see the discussion in section [ sec : eps ] . since the current boundary @xmath660{0.5pt}{4mm}\rule[-0.5mm]{4mm}{0.5pt}}\{u>0\}\cap b_r(0))$ ] is representable by integration , @xmath661 , we get from section 4.5.6 . on page 478 of @xcite that @xmath662 let us show that @xmath663 to see this , for @xmath278 , we define @xmath664 , where @xmath665 as @xmath666 . by the compactness properties obtained in proposition [ tech-2 ] , we have that @xmath667 , as @xmath666 , for some function @xmath352 and , for any test function @xmath212 , @xmath668 where was also used . hence we infer that @xmath669 is a function of bounded variation which is constant a.e . in @xmath670 . the positive lebesgue density property of @xmath671 obtained in lemma [ density est ] and translated to @xmath352 through compactness , and the strong maximum principle for @xmath58harmonic functions demand @xmath352 to be zero . this is in contradiction with the non - degeneracy of @xmath108 stated by proposition [ str - nd - hi ] ( notice that , by a compactness argument , the non - degeneracy property translates to @xmath352 ) . thus is proved . from and we obtain that @xmath672 . the proof of theorem b is then finished . with the aid of theorem a , in this section we complete the proof of theorem c. it is well - know that in order to prove the estimate it is enough to show that @xmath5 grows linearly away from the free boundary . for this , let @xmath673 and @xmath674 . notice that , if for all @xmath112 , @xmath675 , we have that @xmath676 , then it follows from theorem a that @xmath677 . therefore , suppose that there is @xmath80 such that @xmath678 but @xmath679 from and proposition [ prop : lin flat ] ( or corollary [ cor : lin flat ] ) it follows that @xmath680 denote @xmath681 and introduce @xmath682 then by it follows that @xmath535 is a minimizer in @xmath189 . furthermore , yields @xmath683 and by we see that @xmath684 is @xmath81 flat . therefore , we infer from the second part of theorem a that there are @xmath685 and @xmath160 depending on @xmath150 , @xmath24 , @xmath513 , @xmath29 and @xmath686 such that @xmath687 is @xmath36 regular . applying the boundary gradient estimates for @xmath58harmonic functions we finally obtain @xmath688 for some tame constant @xmath689 . recalling and we conclude that @xmath690 for some small universal constant @xmath691 . this completes the proof of theorem c. here we show lemma [ lemma : linear ] . now , we want to establish . for this , we first construct a function that can be used as a barrier to control @xmath5 from below in the ring @xmath697 . we consider the scaled @xmath58capacitary function @xmath698 , that is @xmath24-harmonic in @xmath697 , that vanishes on @xmath699 and that is equal to 1 on @xmath700 . observe that near the origin @xmath701 for some @xmath702 . using the harnack inequality we see that @xmath703 in @xmath704 , for some @xmath207 . thus , multiplying @xmath698 with a suitable constant @xmath705 we obtain that @xmath706 on @xmath700 . moreover , @xmath707 on @xmath699 . hence , by comparison principle , we get that @xmath708 from this and we obtain that @xmath709 near the origin . now we take @xmath478 the smallest positive integer such that @xmath710 and we define @xmath711 thanks to the set of numbers @xmath712 in the definition of @xmath713 is not empty . notice also that the sequence @xmath714 is increasing , and so we let @xmath715 . now we claim that equality in holds in any non - tangential domain . in what follows we denote by @xmath723 if the claim fails then there exist a sequence of points @xmath724 and @xmath725 such that @xmath726 . notice that implies that @xmath727 where @xmath728 . this implies that @xmath729 on some fixed portion of @xmath730 , for some @xmath207 . so , and the harnack inequality give that @xmath731 where @xmath732 has been introduced in and we can take @xmath733 . since @xmath623 are uniformly @xmath734 in @xmath735 and uniformly continuous in @xmath736 , we have that , up to a subsequence , @xmath623 converges uniformly to some @xmath737 in @xmath738 . therefore , by construction of @xmath170 , @xmath739 we recall and define functions @xmath740 as solutions to the following boundary value problem @xmath741 now from the definition of @xmath713 in we have that @xmath742 in @xmath743 , and so @xmath744 on @xmath745 . moreover , on @xmath746 we have that @xmath747 , thanks to . by comparison principle we get that @xmath748 in @xmath749 . by construction @xmath750 uniformly in @xmath751 and @xmath752 from the stability of @xmath188 norm in @xmath753 ( recall that we chose @xmath754 in ) we conclude @xmath755 this allows to estimate the hlder norm of @xmath756 near the flat portion of the boundary of @xmath757 . by hopf s lemma there is @xmath149 such that @xmath758 near the origin . combining we get that @xmath759 in @xmath760 . returning to @xmath5 we get that @xmath761 in @xmath762 . this is a contradiction with the definition of @xmath713 in . now we show part b ) . for this , we take a ball @xmath763 touching @xmath4 from outside @xmath10 . we construct the barrier as follows : we let @xmath764 to be a @xmath24-harmonic function in @xmath765 , such that @xmath766 on @xmath699 and @xmath767 on @xmath768 . then , from comparison principle we have that @xmath769 in @xmath770 . moreover , by hopf s lemma @xmath771 near the origin , for some @xmath772 . we take @xmath478 to be the smallest positive integer such that @xmath773 , and we define @xmath774 and , for any @xmath775 , @xmath776 since @xmath777 is a decreasing sequence , we can take @xmath778 . hence , @xmath779 , and , setting @xmath780 , from we deduce . : uniform lipschitz regularity for classes of minimizers in two phase free boundary problems in orlicz spaces with small density on the negative phase . h. poincar anal . non linaire _ * 31 * ( 2014 ) , no . 4 , 823850 . : a harnack inequality approach to the regularity of free boundaries . existence theory , compactness , and dependence on @xmath781 . _ ann . scuola norm . ( 4 ) _ * 15 * ( 1988 ) , no . 4 , 583602 . : _ the theory of jets in an ideal fluid_. translated from the russian by r. e. hunt . translation edited by e. e. jones and g. power . international series of monographs in pure and applied mathematics , vol . 93 pergamon press , oxford - new york - toronto , ont . 1966 viii+412 pp .	in this paper we study the two - phase bernoulli type free boundary problem arising from the minimization of the functional @xmath0 here @xmath1 is a bounded smooth domain and @xmath2 are positive constants such that @xmath3 . we prove the following dichotomy : if @xmath4 is a free boundary point then either the free boundary is smooth near @xmath4 or @xmath5 has linear growth at @xmath4 . furthermore , we show that for @xmath6 the free boundary has locally finite perimeter and the set of non - smooth points of free boundary is of zero @xmath7-dimensional hausdorff measure . our approach is new even for the classical case @xmath8 .
magnetars are extremely interesting objects both from the physical and astrophysical point of view . from one hand they are associated with sgr and axp pulsars considered as isolated neutron stars with unusual spectral properties . from the other hand they allow one to investigate different phenomena taking place in super strong magnetic field condition not available elsewhere . magnetic field strength of magnetar is believed to be @xmath0 g @xcite , i.e. @xmath1 , where @xmath2 g , @xmath3 is the electron mass , @xmath4 is the elementary charge . ] is the critical magnetic field . the spectra analysis of these objects is also providing evidence for the presence of electron - positron plasma in magnetar environment . it is well - known that strong magnetic field and/or plasma could influence essentially on different quantum processes @xcite . one of such phenomena is the radiation transfer in strongly magnetized plasma . this process is connected with the sgr and axp spectral formation . moreover it is the crucial ingredient of the models of sgr burst ( see e.g. @xcite ) where the creation of magnetically trapped high temperature ( @xmath5 mev ) plasma fireball is assumed ( see fig . [ fig : geom ] ) . it also defines the cooling rate of the outer crust of magnetar @xcite . the various studies indicate that electromagnetic processes such as compton scattering and photon splitting @xmath6 ( merging @xmath7 ) could play a crucial role in these models . in the present work the influence of these processes on radiation transfer is investigated in the presence of strong magnetic field and electron positron plasma , when the magnetic field strength @xmath8 is the maximal physical parameter , namely @xmath9 . here @xmath10 is the plasma temperature , @xmath11 is the chemical potential , @xmath12 and @xmath13 is the initial photon and electron ( or positron ) energies . in this case almost all electrons and positrons in plasma are on the ground landau level . the main goal of this talk is to demonstrate that the self - consistent accounting of strong magnetic field and dense plasma influence is necessary for the correct description of radiation transfer . the propagation of the electromagnetic radiation in any active medium is convenient to describe in terms of normal modes ( eigenmodes ) . in turn , the polarization and dispersion properties of normal modes are connected with eigenvectors and eigenvalues of polarization operator correspondingly . in the case of strongly magnetized plasma in the one loop approximation the eigenvalues of the polarization operator can be derived from the previously obtained results @xcite : @xmath14 - q^2\ , \lambda(b ) , \label{p1 } \\[2 mm ] { \cal p}^{(2)}(q ) & \simeq & -\frac{2 eb \alpha}{\pi}\left [ h\left(\frac{q^2_{\mprl}}{4m^2 } \right ) + { \cal j}({q_{\mbox{\tiny $ \\|$ } } } ) \right ] - q^2 \ , \lambda(b ) , \label{p2 } \\[2 mm ] \p^{(3)}(q ) & \simeq & - \frac{\alpha}{6 \pi}\ , \left [ q_{\mbox{\tiny $ \bot$}}^2 -\sqrt{q_{\mbox{\tiny $ \bot$}}^4 + \frac{(6n\omega)^2 q^2}{q_{\mprl}^2}}\ , \right ] - q^2\ , \lambda(b ) , \label{p3}\end{aligned}\ ] ] where @xmath15 , \quad n = \int \limits_{-\infty}^{+\infty } dp_z \ , \left [ f_{-}(e ) - f_{+}(e)\right ] , \ ] ] ( q _ ) = 2 q_^2 m^2 , e = , @xmath16^{-1}$ ] are the electron ( positron ) distribution functions , [ eq : h0 ] & & h(z)= - 1 , 0 z 1 , + [ 3 mm ] [ eq : h1 ] & & h(z ) = - - 1 + , z > 1 . here @xmath17 , the four - vectors with indices @xmath18 and @xmath19 belong to the euclidean \{1 , 2}-subspace and the minkowski \{0 , 3}-subspace correspondingly in the frame were the magnetic field is directed along @xmath20 ( third ) axis ; @xmath21 , @xmath22 , where the tensors @xmath23 , @xmath24 , with equation @xmath25 are introduced . @xmath26 and @xmath27 are the dimensionless field tensor and dual field tensor correspondingly . and neutral plasma vs. temperature : @xmath28 mev 1 , @xmath29 mev 2 and @xmath30 mev 3 . photon dispersion without plasma is depicted by dashed line . dotted line corresponds to the vacuum dispersion law , @xmath31 . the angle between the photon momentum and the magnetic field direction is @xmath32 . [ fig : dis1 ] ] the dispersion properties of the normal modes could be defined from the dispersion equations q^2 - ^()(q ) = 0 (= 1 , 2 , 3 ) . [ disper ] their analysis shows that 1 and 2 modes with polarization vectors _ ^(1)(q ) = , _ ^(2)(q ) = . [ epsilon ] are only physical ones in the case under consideration , just as it is in the pure magnetic field and @xmath33 polarizations in pure magnetic field @xcite and @xmath13- and @xmath34- modes in magnetized plasma @xcite . ] . however , it should be emphasized that this coincidence is approximate to within @xmath35 and @xmath36 accuracy . ) and degenerate plasma vs. chemical potential @xmath37 mev 1 , @xmath38 mev 2 and without plasma 3 . dotted line corresponds to the vacuum dispersion law , @xmath31 . the angle between the photon momentum and the magnetic field direction is @xmath32 . [ fig : dis0 ] ] notice , that in plasma only the eigenvalue @xmath39 is modified in comparison with pure magnetic field case . it means that the dispersion law of the mode 1 is the same one as in the magnetized vacuum , where its deviation from the vacuum law , @xmath31 , is negligibly small . from the other hand , the dispersion properties of the mode 2 essentially differ from the magnetized vacuum ones . in the fig . [ fig : dis1 ] [ fig : dis0 ] the photon dispersion in both strong magnetic field and magnetized plasma are depicted at various temperatures ( for the charge - symmetric plasma ) and chemical potential ( for the degenerate plasma ) . one can see that in the presence of the magnetized plasma there exist the kinematical region , where @xmath40 contrary to the case of pure magnetic field . it is connected with the appearance of the plasma frequency in the present of the real electrons and positrons which can be defined from equation @xmath41 this fact could lead to the modification of the kinematics of the different processes with photons . the analysis shows that the main channels of photon scattering and photon splitting / merging are * mode 1 ( extraordinary photon ) : @xmath42 & & \gamma_1 \to \gamma_2 \gamma_2 , \gamma_1 \gamma_2 \to \gamma_1 , \gamma_1 \gamma_1 \to \gamma_2 ; \nonumber\end{aligned}\ ] ] * mode 2 ( ordinary photon ) : @xmath43 & & \gamma_2 \gamma_2 \to \gamma_1 , \gamma_2 \gamma_1 \to \gamma_1 . \nonumber\end{aligned}\ ] ] it follows from eq . ( [ p2 ] ) that the eigenvalue of the polarization operator @xmath44 becomes large near the electron - positron pair production threshold . this suggests that the renormalization of the wave function for a photon of this polarization should be taken into account : _ ^(2)(q ) , z^-1_2 = 1 - . [ eq : eps ] in general case the propagation of photon modes through a magnetized plasma can be described by the following equations : @xmath45 @xmath46^{-1}.\ ] ] where @xmath47 are photon occupation numbers for extraordinary and odinary modes , @xmath48 is photon distribution function , @xmath49 are diffusion , absorption and scattering coefficients for different photon modes correspondingly ( @xmath50 ) which can be obtained by the angle averaging of the photon splitting / merging and photon scattering rates : @xmath51 , \\ s_{\lambda \lambda ' } & = & \int \frac{d\omega}{4\pi } w_{\lambda \to \lambda ' } ( \theta , r),\end{aligned}\ ] ] where @xmath52^{-1}. \label{eq : l}\end{aligned}\ ] ] in turn , the rates of the processes under consideration are given by the following formulas : @xmath53 w_{\lambda \to \lambda^ { ' } \lambda^ { '' } } & = & \frac{1 - ( 1/2)\delta_{\lambda ' \lambda '' } } { 32 \pi^2 \omega } \int \mid { \cal m_{\lambda \lambda^ { ' } \lambda^{''}}}\mid^2 z_{\lambda}z_{\lambda^{'}}z_{\lambda^ { '' } } \times \label{eq : wsp } \\ & \times & ( 1 + f_{\omega'})(1 + f_{\omega '' } ) \delta ( \omega_{\lambda}({\bf k } ) - \omega_{\lambda^{'}}({\bf k } - { \bf k^ { '' } } ) - \omega_{\lambda^{''}}({\bf k^ { '' } } ) ) \frac{d^3 k^{''}}{\omega_{\lambda^ { ' } } \omega_{\lambda^ { '' } } } , \nonumber\\[2 mm ] w_{\lambda \lambda^ { ' } \to \lambda^ { '' } } & = & \frac{1 } { 32 \pi^2 \omega } \int \mid { \cal m_{\lambda \lambda^ { ' } \lambda^{''}}}\mid^2 z_{\lambda}z_{\lambda^{'}}z_{\lambda^ { '' } } \times \label{eq : wmer } \\ & \times & f_{\omega'}(1 + f_{\omega '' } ) \delta ( \omega_{\lambda}({\bf k } ) + \omega_{\lambda^{'}}({\bf k ' } ) - \omega_{\lambda^{''}}({\bf k } + { \bf k^ { ' } } ) ) \frac{d^3 k^{'}}{\omega_{\lambda^ { ' } } \omega_{\lambda^{''}}}. \nonumber\end{aligned}\ ] ] where @xmath54 is the electron distribution function , @xmath55 and @xmath56 are the partial amplitudes of the photon scattering and photon splitting processes . using the expressions for @xmath57 ( [ eq : eps ] ) and taking account for photon dispersion properties in energy conservation law inside @xmath58-function in ( [ eq : wc])-([eq : wmer ] ) one could obtained the self - consistent result for the coefficients in transfer equations ( [ eq : te1 ] ) , ( [ eq : te2 ] ) . to calculate the corresponding amplitudes in the presence of strong magnetic field one should use the dirac equation solutions at the ground landau level . for the electron propagator it is relevant to use its asymptotic form @xcite . it is possible to present them in the covariant form . for compton scattering one has @xcite : @xmath59 { \cal m}_{1 2 } & = & -\frac{8 \pi \alpha m}{eb}\ , \frac{(q \lambda q')(q ' \tilde \lambda q ) } { \sqrt{q^2_{\mprp } q'^2_{\mprl } ( -q^2_{\mprl } ) } } , \label{eq : m12 } \\[2 mm ] { \cal m}_{2 1 } & = & \frac{8 \pi \alpha m}{eb}\ , \frac{(q \lambda q')(q \tilde \lambda q ) } { \sqrt{q^2_{\mprl } q'^2_{\mprp } ( -q^2_{\mprl } ) } } , \label{eq : m21 } \\ { \cal m}_{2 2 } & = & 16 i \pi \alpha m\ , \frac{\sqrt{q^2_{\mprl } q'^2_{\mprl}}\,\sqrt{(-q^2_{\mprl})}\ , \varkappa } { ( q \tilde \lambda q')^2 - \varkappa^2 ( q \tilde \varphi q')^2 } , \label{eq : m22}\end{aligned}\ ] ] where @xmath60 and @xmath61 . ( solid line ) and @xmath62 ( chain line ) . dotted line corresponds to the diffusion coefficient calculated using approximation ( [ eq : llow ] ) . ] the amplitudes of photon splitting are given by the following equations @xcite : @xmath63^{\mbox{\tiny $ 1/2 $ } } } { \cal g } ( q^{''}_{\mbox{\tiny $ \\|$ } } ) , \label{eq : m112 } \\[3 mm ] { \cal m}_{1 2 2 } & = & i 4\pi \left ( \frac{\alpha}{\pi}\right)^{\mbox{\tiny $ 3/2 $ } } \frac{(q ' q'')_{\mbox{\tiny $ \\|$ } } } { [ { q'^2_{\mbox{\tiny $ \\|$ } } } { q''^2_{\mbox{\tiny $ \\|$ } } } q^2_{\mprp}]^{\mbox{\tiny $ 1/2 $ } } } \label{eq : m122 } \\[2 mm ] & \times & \left \{(q q'')_{\mbox{\tiny $ \bot$ } } { \cal g}(q'_{\mbox{\tiny $ \\|$ } } ) + ( q q')_{\mbox{\tiny $ \bot$ } } { \cal g}(q''_{\mbox{\tiny $ \\|$ } } ) \right \ } , \nonumber \\[3 mm ] { \cal m}_{2 1 1 } & = & { \cal m}_{1 1 2 } ( q \leftrightarrow q '' ) , \quad \label{eq : m211}\end{aligned}\ ] ] where @xmath64 . the analysis of the last equations shows that all amplitudes much smaller than @xmath65 . therefore one could expect that mode 2 has the largest scattering absorption rate . it means that the radiation transfer in the magnetically trapped plasma may be described as s diffusion of the 1-mode photons whereas 2-mode photons are locked @xcite . ( solid line ) . dotted line corresponds ton the diffusion coefficient calculated using approximation ( [ eq : llow ] ) . ] in general case , the calculation of the reactions rates ( [ eq : wc])-([eq : wmer ] ) is rather complicated mathematical problem . however in some limiting cases it is possible to obtain the simple expression for the rates . for example , in low temperature ( @xmath66 ) and low energy ( @xmath67 ) limits the mean free path ( [ eq : l ] ) for the 1-mode photon in the charge symmetric plasma ( @xmath68 ) can be presented in the following form : @xmath69 where @xmath70 is the thompson cross section and the number of electron ( positron ) density in a strongly magnetized , charge - symmetric rarefied plasma can be estimated as @xmath71 . ] in formula ( [ eq : llow ] ) the fist term corresponds to the compton scattering process and the second one comes from the photon splitting contribution . it is the estimation for the photon mean free path that usually is used in the radiation transfer analysis in strongly magnetized plasma . moreover , the process of photon splitting / merging is not taking into account . we would like to show that even in low energy limit this approximation is not appropriate . we have made the numerical calculation of the coefficients in ( [ eq : te1 ] ) and ( [ eq : te2 ] ) in charge symmetric plasma . our results are represented in figures [ fig : d1]-[fig : s ] . ( solid line ) . dashed line corresponds to the diffusion coefficient calculated using approximation ( [ eq : llow ] ) . ] in figures [ fig : d1 ] , [ fig : d1low ] and [ fig : s ] one can see that diffusion coefficient calculated with taking account of photon dispersion and large radiative correction strongly deviate from the coefficient obtained by using approximation ( [ eq : llow ] ) . in addition , in fig . [ fig : r ] the ratio of the diffusion coefficient for only compton scattering to the diffusion coefficient with photon splitting is depicted . one can see that at low temperatures the additional absorption process of photon splitting leads to the significant decreasing of the diffusion coefficient . we have also analysed the problem of the radiation transfer in the cold degenerate plasma . in this case the main channel of photon splitting is @xmath72 . in the figures [ fig : w211 ] and [ fig : w21 ] the contributions of photon splitting and compton scattering to photon mean free path are depicted . one can see that in cold plasma the photon splitting contribution is negligibly small in comparison with compton scattering . it means that the only process of photon scattering on electrons defines the radiation transfer in cold plasma . at @xmath73 ( lower line ) , @xmath74 ( lower line ) , @xmath75 . ] at @xmath76 ( lower line ) , @xmath77 ( upper line ) , @xmath78 . ] we have investigated the influence of strongly magnetized plasma on the radiation transfer with taking into account of photon dispersion and large radiative corrections . we have studied the processes of compton scattering and photon splitting into two photons and calculated their contribution in transfer equation coefficient . the main conclusions of this work are * photon dispersion and radiative correction in strongly magnetized plasma could essentially influence on the radiation transfer process . * in charge symmetric plasma ( @xmath79 ) it is necessary to take into account the processes of photon splitting and photon merging . * in strongly degenerate plasma the influence of the photon splitting and photon merging processes on radiation transfer is negligibly small . we express our deep gratitude to the organizers of the seminar `` quarks-2008 '' for warm hospitality . this work was supported in part by the russian foundation for basic research under the grant no . 07 - 02 - 00285-a , and by the council on grants by the president of the russian federation for the support of young russian scientists and leading scientific schools of russian federation under the grant no . nsh-497.2008.2 and no . mk-732.2008.2 ( mc ) . 99 c. kouveliotou _ et al . _ , nature * 393 * ( 1998 ) 235 . c. kouveliotou _ et al . _ , astrophys . j. * 510 * ( 1999 ) l115 f. p. gavriil , v. m. kaspi , p. m. woods , nature * 419 * ( 2002 ) 142 . a. i. ibrahim , s. safi - harb , j. h. swank , w. parke and s. zane , astrophys . j. * 574 * ( 2002 ) l51 r.c . duncan , c. thompson , astrophys . j. * 392 * no . 1 . ( 1992 ) l9 c. thompson , r.c . duncan , mon . not . r. astron * 275 * ( 1995 ) 255 r.c . duncan , c. thompson , astrophys . j. * 473 * ( 1996 ) 322 r. c. duncan , fifth huntsville gamma - ray burst symposium , eds : r. marc kippen , robert s. mallozzi , gerald j. fishman . 526 ( american institute of physics , melville , new york ) p.830 [ astro - ph/0002442 ] . d. lai , rev . * 73 * ( 2001 ) 629 a. v. kuznetsov , n. v. mikheev , _ electroweak processes in external electromagnetic fields _ , springer - verlag , new york ( 2003 ) , p.1 . a. k. harding , d. lai , rept . * 69 * ( 2006 ) 2631 d.g . yakovlev , a.d . kaminker , o.y . gnedin , p. haensel , _ phys . rep . _ * 354 * , p1 ( 2001 ) . harding , m.g . baring , astrophys . j. * 507 * ( 1998 ) l55 m. g. baring and a. k. harding , astrophys . j. * 547 * ( 2001 ) 929 m.c . miller , astrophys . j. lett . * 448 * ( 1995 ) l29 t. bulik , m.c . miller , mon . not . r. astron * 288 * ( 1997 ) 596 a. k. harding , m. g. baring , p. l. gonthier , astro . * 120 * ( 1996 ) 111 a. k. harding , m. g. baring , p. l. gonthier , astrophys . j. * 476 * ( 1997 ) 246 h. herold , phys . rev . * d19 * ( 1979 ) 2868 d.b . melrose , a.j . parle , aust . j. phys . * 36 * ( 1983 ) 799 j.k . daugherty , a.k . harding , astrophys . j. * 309 * ( 1986 ) 362 p.l . gonthier , a.k . harding , m.g . baring et al . , astrophys . j. * 540 * ( 2000 ) 907 h. prez rojas , a. e. shabad , _ ann . _ * 121 * , 432 ( 1979 ) . h. prez rojas , a. e. shabad , _ ann . _ * 138 * , 1 ( 1982 ) . a.e . shabad , tr . nauk sssr * 192 * ( 1988 ) 5 s. l. adler , ann . ( n.y . ) * 67 * ( 1971 ) 599 d. a. rumyantsev , m. v. chistyakov , _ proceedings of 14th international seminar quark-2006 , st . petersburg , russia , 19 - 25 may , 2006 , _ [ hep - ph/0609192 ] d. a. rumyantsev , m. v. chistyakov , jetp * 101 * ( 2005 ) 635 y. e. lyubarsky , mon . not . * 332 * ( 2002 ) 199	the electromagnetic processes of compton scattering and photon splitting / merging are investigated in the presence of strongly magnetized electron - positron plasma . the influence of these processes on the radiation transfer in the astrophysical environment is studied . in particular , the contribution of the processes under consideration in coefficients of the transfer equation is calculated . we show the importance of photon splitting / merging contribution and taking into account of photon dispersion and wave function renormalization in strong magnetic field and plasma .
interference effects in quantum transport in semiconductor quantum dots ( qd ) have been among the highlights in electron transport studies.@xcite a simple example of coherent transport is resonant tunneling through a pair of qds independently coupled to left and right doped semiconductor leads that shows a conductance peak narrowing due to the interference between the tunneling electron pathways.@xcite in the presence of magnetic field , the tunneling current through a qd system exhibits aharonov - bohm ( ab ) oscillations @xcite as a function of magnetic flux through a surface enclosed by pathways.@xcite ab oscillations have been widely studied in systems where electron motion is constrained by the system geometry such as , e.g. , metal or semiconductor rings,@xcite carbon nanotubes @xcite or , more recently , graphene nanorings.@xcite at the same time , in _ open _ two - dimensional ( 2d ) electron systems , i.e. , when the electron motion in a 2d plane is unconstrained , oscillations of magnetoresistance were observed in the presence of a weak 1d @xcite or 2d @xcite periodic potential and in a system of antidots.@xcite in such structures , the oscillations are caused by geometric resonances occurring when the size of the electron s larmor orbit , which changes with magnetic field , is commensurate with the potential period.@xcite for magnetic fields corresponding to magnetoresistance maxima , electron trajectories run close to the potential energy minima , indicating that oscillations originate from electron orbital motion rather than its phase . here we show that ab oscillations can occur in an open 2d system where electron transport takes place via multiple pathways . we consider resonant tunneling through a square periodic array of qds sandwiched between two 2d electron gases ( 2degs ) in doped semiconductor quantum wells separated from the qd plane by tunneling barriers ( see the inset in fig . [ fig:1 ] ) . highly periodic square arrays of qds have been recently manufactured.@xcite tunneling current , e.g. , from left to right 2degs involves an electron traversing back and forth along closed pathways comprising electron trajectories within 2degs and tunneling between them through qd lattice sites ( we assume that direct interdot coupling is negligibly small ) . for each closed pathway , an in - plane magnetic field @xmath0 generates a flux @xmath1 , where @xmath2 is the area of the surface enclosed by such a pathway projected onto the plane normal to @xmath0 . we demonstrate that the array magnetoconductance exhibits a highly complex ab oscillation pattern originating from multiple pathways traversed by the tunneling electron ( see the inset in fig . [ fig:1 ] ) . for high mobility 2deg characterized by a mean free path @xmath3 that is much larger than the qd lattice constant @xmath4 , the conductance ab beats correspond to pathways of length @xmath5 which , hence , can be tracked by sweeping the magnetic field . remarkably , conductance oscillations are _ aperiodic _ unless the magnetic field slope relative to the qd lattice axes is a rational number . the lack of ab beat periodicity for a general field orientation implies the absence of pathway degeneracies caused by two or more pathways accommodating the same flux . ( color online ) normalized conductance vs. electron fermi momentum . ( a ) emergence of sidebands for zero - field conductance with decreasing 2deg scattering rate . ( b ) conductance lineshape change with varying in - plane magnetic field . inset : schematic view of a qd lattice placed between 2degs with several closed electron pathways . ] to obtain electron conductance through a qd array , we adopt the tunneling hamiltonian formalism.@xcite the hamiltonian of a square lattice of @xmath6 qds with in - plane coordinates @xmath7 separated by potential barriers from left and right 2deg planes has the form @xmath8 where @xmath9 , @xmath10 , and @xmath11 are , respectively , the creation and annihilation operators and energies for qd localized states , @xmath12 , @xmath13 , and @xmath14 are those for 2deg states ( @xmath15 ) , and @xmath16 is a transition matrix element between localized and 2deg states . we assume that direct tunneling between qds is weak and does not include interdot coupling in hamiltonian ( [ h ] ) . we restrict ourselves to the single - electron picture of transport and disregard electron interaction effects due to a low probability of qd double occupancy in a large array . we also assume that the magnetic field , to be included below , is sufficiently weak and/or the electron _ g_-factor is sufficiently small to cause significant zeeman splitting and suppress spin indices throughout . the zero - temperature conductance through a system of @xmath6 qds is given by @xcite @xmath17 where @xmath18 is the self - energy matrix of qd states due to coupling to electron states in the left and right 2degs , @xmath19 here the principal and singular parts of @xmath20 determine the energy matrix @xmath21 and the decay matrix @xmath22 , respectively , and the trace is taken over qd lattice sites . the transition matrix element can be presented as @xcite @xmath23 , where @xmath24 and @xmath7 are , respectively , the electron momentum and coordinate in 2degs , @xmath25 is the tunneling amplitude between qd and 2deg , and @xmath26 is the normalization area . we assumed that the barrier is sufficiently high so that electron tunneling between 2deg and the qd plane takes place along the shortest path and the dependence of @xmath25 on energy is weak.@xcite then the self - energy ( [ self - matrix ] ) takes the form @xmath27 , where @xmath28 is the electron green s function between the qd lattice sites projected onto 2deg planes . the coupling between the qd lattice states and the continuum of electronic states in 2degs gives rise to an _ in - plane _ quasimomentum @xmath29 that conserves across the system @xcite . the 2deg momentum space @xmath24 splits into bloch bands @xmath30 , where @xmath31 are reciprocal lattice vectors ( @xmath32 and @xmath33 are integers ) and @xmath29 lies in the first 2d brillouin zone ( @xmath34 ) . the energy spectrum of the qd lattice states can be obtained by performing a fourier transform of the self - energy matrix eq . ( [ self - matrix ] ) as @xmath35 , where @xmath36 here @xmath37 is the momentum space 2deg green s function of the band @xmath38 electron having a quasimomentum @xmath29 , @xmath39 is its dispersion ( @xmath40 is the electron mass ) , and @xmath41 is its scattering rate . in momentum space , the self - energy @xmath42 is a complex function of @xmath29 that determines the qd lattice band dispersion , @xmath43 , and its decay width , @xmath44 , due to the coupling to left and right 2degs . we now include an in - plane magnetic field @xmath0 tilted by an angle @xmath45 relative to the @xmath46-axis ( see the inset in fig . [ fig:1 ] ) through the vector potential @xmath47 . this leads to a momentum shift in the left and right 2degs , located at @xmath48 and @xmath49 , respectively , as @xmath50 , where @xmath51 . in the presence of the qd lattice , the momentum space in the left or right 2deg is now split as @xmath52 , where @xmath53 is a field - dependent band wave vector . here @xmath54 is the magnetic flux through the _ elementary area _ enclosed by pathways running between 2degs ( pathways i or ii in the inset of fig . [ fig:1 ] ) and @xmath55 is the flux quantum . the field dependence of @xmath56 in the 2deg electron dispersion @xmath57 translates to field - dependence of the qd lattice self - energy , @xmath58 , still given by eq . ( [ self - lsb ] ) but with @xmath38 replaced by @xmath59 . finally , the array conductance is obtained via a fourier transform of eq . ( [ cond ] ) as @xmath60 where the @xmath29 integral is taken over the 2d brillouin zone . the field dependence of @xmath61 comes from those of the qd lattice band dispersion @xmath62 and its width @xmath63 . changing the magnetic field magnitude may cause a 2deg electron to jump to another band , according to eq . ( [ band ] ) . since the qd lattice energy spectrum , @xmath64 , includes contributions from all 2deg bands , these field - induced interband transitions lead to oscillatory behavior of @xmath65 and @xmath63 which , in turn , gives rise to ab conductance oscillations . importantly , depending on the field orientation @xmath45 , interband transitions for the @xmath46 and @xmath66 components of @xmath67 can take place at _ different _ field values . for example , for @xmath0 oriented along the @xmath46 or @xmath66 axes ( @xmath68 or @xmath69 ) , only one component of @xmath67 can jump to the next value ( @xmath70 or @xmath71 ) with changing field magnitude [ see eq . ( [ band ] ) ] ; in real space , only the surfaces projection onto the @xmath72 or @xmath73 planes , respectively , would contribute to oscillations ( e.g. , either pathways ii or i in the inset of fig . [ fig:1 ] ) . however , for @xmath74 , interband transitions simultaneously take place for both components of @xmath67 ( @xmath70 and @xmath71 ) , i.e. , pathways i and ii would lead to similar oscillations . note that an interband transition involves contributions from many pathways , but , due to the square lattice symmetry , only two independent sets of oscillations in 2d momentum space , described by eq . ( [ band ] ) , are generated by all electron pathways . for a general field orientation , these two oscillation sets are incommensurate , implying that the resulting ab oscillation pattern is _ aperiodic_. the ab beats are periodic for field orientations that render commensurate interband transitions for both @xmath67 components , i.e. , for @xmath75 , where @xmath76 and @xmath77 are integers . in other words , the fluxes through projections of a surface , enclosed by an electron pathway , onto the @xmath72 and @xmath73 planes must be commensurate , which only takes place , given the lattice symmetry , if the magnetic field slope is a rational number . below we present numerical calculations for a symmetric configuration , i.e. , for a qd lattice at midpoint between similar quantum wells ( @xmath78 and @xmath79 , @xmath80 ) . the lattice constant was chosen to set @xmath81 , where @xmath82 is a geometric energy scale associated with the lattice , so that the transmission resonance @xmath83 occurs at a fermi momentum @xmath84 . the 2deg electron width @xmath85 due to elastic scattering was varied in the range from @xmath86 to @xmath87 , yielding @xmath88 in the range from 20 to 200 ; for the lattice period @xmath89 nm , this corresponds to a low - to - intermediate 2deg mobility in the range @xmath90@xmath91 @xmath92/vs . in fig . [ fig:1](a ) , zero - field per qd normalized conductance is shown for several values of @xmath93 . for low mobility 2deg with @xmath94 , the conductance shows a single peak centered at a qd resonance with weak shoulders on the left and right sides . with decreasing @xmath93 , these shoulders develop into sidebands while the main peak gets slightly shifted . these features are due to the appearance of new resonances in the integrand of eq . ( [ cond2 ] ) satisfying @xmath95 originating from qd lattice coupling to 2degs [ see eq . ( [ self - lsb ] ) ] . at the same time , with decreasing @xmath93 , the electron escape rate from the qd lattice to 2deg , @xmath96 , becomes a sharp function of @xmath97 . the combination of these two factors results in sharp features near the resonance and in the emergence of minor features away from it , the latter coming from neighboring bands . ( color online ) ab oscillations of magnetoconductance for the 2deg scattering rate @xmath93 in the range @xmath87@xmath86 . with increasing 2deg mobility , the oscillation pattern develops a fine structure associated with longer electron pathways . ] ( color online ) ab oscillations of magnetoconductance at @xmath98 for a magnetic field slope equal ( a ) 1 , ( b ) 1/2 , and ( c ) @xmath99 . the oscillation pattern in ( c ) is aperiodic . ] an in - plane magnetic field leads to significant decrease of the overall conductance and to a change of sidebands positions and widths [ see fig . [ fig:1](b ) ] . while the latter behavior reflects the field dependence of @xmath100 resonances , the amplitude drop comes from the change of interference between tunneling paths caused by ab flux . sweeping the magnetic field reveals pronounced ab oscillations of peak conductance with amplitude exceeding half of its zero - field value ( see fig . [ fig:2 ] ) . for @xmath101 , the largest period in ab oscillations pattern , in units of flux through elementary area @xmath102 , is provided by pathways enclosing surfaces that project area @xmath103 onto the @xmath72 plane , e.g. , pathways ii , iii , and v in fig . [ fig:1 ] inset , while strong half period beats come from pathways enclosing area @xmath104 when projected onto @xmath72 plane , e.g. , pathway iv . with decreasing @xmath93 , conductance oscillations develop _ fine structure _ due to the short - period beats coming from longer pathways . tilting the magnetic field reveals a dramatic increase of the fine structure complexity due to the reduction of pathways degeneracies , which are maximal for @xmath0 oriented along the lattice axes ( see fig . [ fig:3 ] ) . individual beats in the ab oscillation pattern correspond to specific pathways traversed by an electron , i.e. , a varying magnetic field enumerates the pathways by highlighting those with surfaces accommodating the integer ab flux ( in units of @xmath105 ) . note that the pathway degeneracies persist for any rational field slope ( i.e. , @xmath106 ) ; e.g. , for @xmath107 , pathways i and ii accommodate the same flux @xmath108 corresponding to the largest period in fig . [ fig:3](a ) . by changing the field slope , a substantially different oscillation pattern is generated that highlights a different set of pathways . for example , for @xmath109 [ see fig . [ fig:3](b ) ] , pathways i and ii are now distinct by producing beats with periods 2.24 and 1.12 , respectively ( in units of @xmath110 ) . note also that , for any rational slope , there are `` missing '' pathways that do not produce ab beats , e.g. , pathway i for @xmath68 , ii for @xmath111 , iii for @xmath74 , iv for @xmath112 , and v for @xmath113 . for a general field slope , however , the ab oscillation pattern has no periodic structure and there are no degenerate or missing pathways . an example of aperiodic beats for @xmath114 is shown in fig . [ fig:3](c ) . in summary , we have shown that the tunneling current through a periodic array of quantum dots sandwiched between 2d electron gases in quantum wells exhibits a highly complex pattern of aharonov - bohm oscillations originating from multiple pathways that the electron traverses in the course of transport . for high mobility samples , the ab beats corresponding to individual pathways are well resolved and could allow tracking the electron motion in the system by sweeping the magnetic field . we find that oscillation pattern is aperiodic unless the magnetic field slope relative to the lattice axes is a rational number . a. yacoby , m. heiblum , d. mahalu , and h. shtrikman , phys . lett . * 74 * , 4047 ( 1995 ) . r. schuster , e. buks , m. heiblum , d. mahalu , v. umansky , and h. shtrikman , nature ( london ) * 385 * , 417 ( 1997 ) . e. buks , r. schuster , m. heiblum , d. mahalu , and v. umansky , nature ( london ) * 391 * , 871 ( 1998 ) . t. v. shahbazyan and m. e. raikh , phys . rev . b 49 , 17123 ( 1994 ) . a. w. holleitner , c. r. decker , h. qin , k. eberl , and r. h. blick , phys lett . * 87 * , 256802 ( 2001 ) . m. sigrist , t. ihn , k. ensslin , d. loss , m. reinwald , and w. wegscheider , phys . lett . * 96 * , 036804 ( 2006 ) . t. hatano , t. kubo , y. tokura , s. amaha , s. teraoka , and s. tarucha , phys . * 106 * , 076801 ( 2011 ) . d. loss and e. v. sukhorukov , phys . lett . * 84 * , 1035 ( 2000 ) . j. knig and y. gefen , phys . lett . * 86 * , 3855 ( 2001 ) . y. gefen , y. imry , and m. ya . azbel , phys . lett . * 52 * , 129 ( 1984 ) .	we study resonant tunneling through a periodic square array of quantum dots sandwiched between modulation - doped quantum wells . if a magnetic field is applied parallel to the quantum dot plane , the tunneling current exhibits a highly complex aharonov - bohm oscillation pattern due to the interference of multiple pathways traversed by a tunneling electron . individual pathways associated with conductance beats can be enumerated by sweeping the magnetic field at various tilt angles . remarkably , aharonov - bohm oscillations are _ aperiodic _ unless the magnetic field slope relative to the quantum dot lattice axes is a rational number .
the analysis and characterization of circumstellar disks , which provide the raw material to form planets , are crucial for our understanding of disk dispersal and planet formation processes . observations and modeling of the dust content in protoplanetary disks are nowadays relatively abundant ( see testi et al . 2014 for a review ) . the gaseous component of circumstellar disks is instead less well characterized ( e.g. , dutrey et al . 2014 for a review ) even if major advances have been made in the past few years thanks to space observatories such as _ spitzer _ ( e.g. , pontoppidan et al . 2014 ) and _ herschel _ ( e.g. , dent et al . gas plays a pivotal role in the structure and evolution of disks representing the bulk of the disk mass ( e.g. , alexander et al . it is then necessary to extend our knowledge of the disk gas content in order to improve our understanding of the natal environments of planets . the sensitivity of the infrared spectrograph ( irs ) on the _ spitzer _ space telescope led to the discovery of gas lines that can be used to characterize the immediate surroundings of young stellar objects ( ysos ) . for instance , the rich water spectrum of ysos has been attributed to emission arising from warm ( @xmath6600 k ) layers of water vapor out to a few aus from the central star ( carr & najita 2008 , 2011 , salyk et al . 2008 , pontoppidan et al . 2010 , salyk et al . 2011 , banzatti et al . 2013 ) . among the wide forest of lines revealed by _ the most common one is the singly ionized neon , [ ne ii ] at 12.81@xmath0 m ( e.g. , pascucci et al . 2007 , lahuis et al 2007 , gdel et al . 2010 ) . different studies of this line show that objects with known jets / outflows have higher [ ne ii ] luminosity than objects without ( gdel et al . 2010 , baldovin - saavedra et al . moreover , follow up observations of this line with ground - based high - resolution spectrographs have shown that [ ne ii ] is predominantly tracing an outflow in high - accreting stars , while its origin is from photoevaporative disk winds driven by high - energy photons from the central star in lower accretors ( pascucci & sterzik 2009 , van boekel et al . 2009 , sacco et al . 2012 , baldovin - saavedra et al . 2012 ) . other gas lines that are commonly used to characterize objects undergoing accretion are hydrogen recombination lines . h i lines from the near - ultraviolet ( uv ) to the near - infrared ( ir ) have been extensively used as indicators of the rate at which the matter from the inner circumstellar disk is accreting onto the star ( e.g. , muzerolle et al . 2001 , calvet et al . 2004 , natta et al . 2006 , alcal et al . 2014 , edwards et al . 2013 among many others ) . recently , salyk et al . ( 2013 ) using nirspec at the keck ii telescope and crires at the very large telescope have introduced the h i pf@xmath7 line at 4.65@xmath0 m as accretion diagnostic . this is to date the longest wavelength hydrogen line proposed as accretion indicator . as early as 2007 , pascucci et al . ( 2007 ) reported also the detection of h i lines in the _ spitzer _ spectrum of rx1852.3 - 3700 . subsequently , ratzka et al . ( 2007 ) and najita et al . ( 2010 ) reported hydrogen line detections in both the low- and high - resolution _ spectrum of twhya . the analysis of these two objects did not constrain the origin of these lines and left several possibilities open : stellar chromosphere , hot disk atmosphere , accretion shock regions , or internal wind shock very close to the stellar surface ( hollenbach & gorti 2009 ) . more recently , carr & najita ( 2011 ) reported the detection of the ir hydrogen lines in 10 t tauri stars but did not discuss their origin . here , we improve upon studies of gas lines from circumstellar disks by analyzing hydrogen lines observed in the _ spitzer_/irs spectra of a large sample of objects . in particular we focus on the strongest of the humphreys series lines in the mid - infrared , the hu@xmath8 at 12.37@xmath0 m ( hereafter h i(7 - 6 ) ) and the h i(9 - 7 ) line at 11.32@xmath0 m . with the aim to investigate the origin of these mid - ir hydrogen lines , we compare their properties both with hydrogen lines observed at shorter wavelengths , mainly tracing accretion , and with the other most common gas line observed in the _ spitzer _ spectra ( [ ne ii ] ) which is instead an outflow or disk wind tracer depending on the mass accretion rate . the paper is organized as follows . in section 2 the source selection and data reduction are explained . in section 3 we describe the method used to account for the water contamination of the h i lines , and how we measure the h i and [ ne ii ] line fluxes and luminosities . in section 4 we discuss the origin of the h i mid - ir lines , highlighting the interesting presence of h i lines in a sample of debris disks . the summary of the results and an outlook are given in section 5 . using the _ spitzer _ heritage archive we downloaded and re - reduced all irs short - high ( sh ) spectra of disks - bearing young stars acquired in the standard staring mode and which have multiple on - source exposures ( n - cycle @xmath91 ) and sky observations . the latter two restrictions were applied to better identify and correct for bad pixels as summarized below . a list of the 114 objects identified in this way and with a signal - to - noise on the continuum greater than 5 is reported in table [ all_obj_log ] . this table also contains the source coordinates , the astronomical observation request number ( aor ) , the identification number ( i d ) of the _ spitzer _ program the object belongs to , and the exposure time per frame times the number of exposures . we also report in table [ tab_all_prop ] the sources properties , and in table [ flux_obj ] the mid - ir line fluxes , errors and continuum levels around the three line of interest . l c c c c r + hd 377 & 0h08m25.76s & + 6d37m00.3s & 13462272 & 148 & 120 @xmath10 3 + hd 12039 & 1h57m48.97s & -21d54m05.2s & 13461760 & 148 & 120 @xmath10 4 + hd 17925 & 2h52m32.13s & -12d46m11.0s & 9780480 & 148 & 30 @xmath10 5 + hd 19668 & 3h09m42.28s & -9d34m46.4s & 13462784 & 148 & 120 @xmath10 3 + lkha 326 & 3h30m44.01s & + 30d32m47.0s & 27063552 & 50641 & 30 @xmath10 12 + ic348 - 67 & 3h43m44.62s & + 32d08m17.9s & 22850816 & 40247 & 30 @xmath10 100 + ic348 - 31 & 3h44m18.16s & + 32d04m57.0s & 22849280 & 40247 & 30 @xmath10 16 + ic348 - 72 & 3h44m22.58s & + 32d01m53.7s & 22851840 & 40247 & 30 @xmath10 100 + ic348 - 68 & 3h44m28.51s & + 31d59m54.1s & 22851328 & 40247 & 30 @xmath10 100 + ic348 - 55 & 3h44m31.36s & + 32d00m14.7s & 22850304 & 40247 & 30 @xmath10 80 + ic348 - 2 & 3h44m35.36s & + 32d10m04.6s & 22847744 & 40247 & 30 @xmath10 4 + ic348 - 6 & 3h44m36.94s & + 32d06m45.4s & 22848256 & 40247 & 30 @xmath10 16 + ic348 - 37 & 3h44m37.99s & + 32d03m29.8s & 22849792 & 40247 & 30 @xmath10 30 + ic348 - 133 & 3h44m41.74s & + 32d12m02.4s & 22852352 & 40247 & 30 @xmath10 100 + ic348 - 21 & 3h44m56.15s & + 32d09m15.5s & 22848768 & 40247 & 30 @xmath10 30 + lkha 330 & 3h45m48.28s & + 32d24m11.9s & 27063040 & 50641 & 30 @xmath10 7 + hd 25457 & 4h02m36.74s & -0d16m08.1s & 9779712 & 148 & 6 @xmath10 4 + fm tau & 4h14m13.58s & + 28d12m49.3s & 24403968 & 30300 & 120 @xmath10 2 + fn tau & 4h14m14.59s & + 28d27m58.0s & 27062528 & 50641 & 30 @xmath10 6 + cw tau & 4h14m17.00s & + 28d10m57.8s & 27062272 & 50641 & 30 @xmath10 6 + fp tau & 4h14m47.30s & + 26d46m26.4s & 26321408 & 50498 & 30 @xmath10 8 + cx tau & 4h14m47.86s & + 26d48m11.0s & 23551744 & 40338 & 30 @xmath10 21 + cy tau & 4h17m33.72s & + 28d20m46.8s & 27061504 & 50641 & 30 @xmath10 14 + dd tau & 4h18m31.13s & + 28d16m28.9s & 26320384 & 50498 & 30 @xmath10 2 + bp tau & 4h19m15.84s & + 29d06m27.0s & 14548224 & 20363 & 30 @xmath10 12 + de tau & 4h21m55.67s & + 27d55m06.2s & 27060992 & 50641 & 30 @xmath10 10 + ft tau & 4h23m39.18s & + 24d56m14.3s & 27060224 & 50641 & 30 @xmath10 14 + 04216 + 2603 & 4h24m44.58s & + 26d10m14.2s & 27059968 & 50641 & 30 @xmath10 12 + ip tau & 4h24m57.08s & + 27d11m56.3s & 24404224 & 30300 & 120 @xmath10 5 + dg tau & 4h27m04.70s & + 26d06m15.8s & 14547968 & 20363 & 6 @xmath10 12 + dh tau & 4h29m41.57s & + 26d32m58.2s & 26320128 & 50498 & 30 @xmath10 10 + ux taua & 4h30m04.00s & + 18d13m49.4s & 18013952 & 30300 & 30 @xmath10 30 + dk tau & 4h30m44.28s & + 26d01m24.6s & 14548736 & 20363 & 30 @xmath10 8 + hk tau & 4h31m50.58s & + 24d24m17.8s & 27059200 & 50641 & 30 @xmath10 14 + fz tau & 4h32m31.77s & + 24d20m02.6s & 27058688 & 50641 & 30 @xmath10 6 + hn tau & 4h33m39.36s & + 17d51m52.3s & 27058176 & 50641 & 30 @xmath10 6 + dl tau & 4h33m39.08s & + 25d20m38.1s & 27058432 & 50641 & 30 @xmath10 6 + dm tau & 4h33m48.72s & + 18d10m10.0s & 18014720 & 30300 & 30 @xmath10 30 + aa tau & 4h34m55.44s & + 24d28m53.4s & 14551552 & 20363 & 30 @xmath10 12 + dn tau & 4h35m27.37s & + 24d14m58.9s & 24404480 & 30300 & 30 @xmath10 9 + do tau & 4h38m28.58s & + 26d10m49.4s & 14548480 & 20363 & 30 @xmath10 8 + lkca 15 & 4h39m17.80s & + 22d21m03.5s & 23551488 & 40338 & 30 @xmath10 21 + 04385 + 2550 & 4h41m38.82s & + 25d56m27.0s & 27057408 & 50641 & 30 @xmath10 6 + dp tau & 4h42m37.69s & + 25d15m37.3s & 27057152 & 50641 & 30 @xmath10 6 + dq tau & 4h46m53.06s & + 17d00m00.4s & 26321152 & 50498 & 30 @xmath10 2 + dr tau & 4h47m06.22s & + 16d58m42.9s & 27067136 & 50641 & 6 @xmath10 12 + ds tau & 4h47m48.60s & + 29d25m11.6s & 26320896 & 50498 & 30 @xmath10 4 + uy aur & 4h51m47.38s & + 30d47m13.9s & 14551040 & 20363 & 6 @xmath10 10 + st 34 & 4h54m23.68s & + 17d09m53.5s & 18016256 & 30300 & 30 @xmath10 40 + gm aur & 4h55m10.98s & + 30d21m59.5s & 18015488 & 30300 & 30 @xmath10 10 + su aur & 4h55m59.38s & + 30d34m01.5s & 27066880 & 50641 & 6 @xmath10 10 + hd 32297 & 5h02m27.44s & + 7d27m39.7s & 25677568 & 50150 & 30 @xmath10 5 + v836 tau & 5h03m06.60s & + 25d23m19.7s & 23552000 & 40338 & 30 @xmath10 21 + rw aur & 5h07m49.56s & + 30d24m05.0s & 14547200 & 20363 & 6 @xmath10 13 + hd 35850 & 5h27m04.76s & -11d54m03.5s & 9777920 & 148 & 30 @xmath10 6 + hd 37484 & 5h37m39.62s & -28d37m34.7s & 9780224 & 148 & 30 @xmath10 10 + ngc2068/ 2071-h9 & 5h46m11.86s & + 0d32m25.9s & 22853888 & 40247 & 30 @xmath10 100 + ngc2068/ 2071-s1 & 5h46m18.89s & -0d05m38.2s & 22854912 & 40247 & 30 @xmath10 50 + ngc2068/ 2071-h13 & 5h46m22.44s & -0d08m52.6s & 22854400 & 40247 & 30 @xmath10 80 + ao men & 6h18m28.24s & -72d02m41.7s & 5458688 & 148 & 120 @xmath10 2 + hd 61005 & 7h35m47.47s & -32d12m14.2s & 13462528 & 148 & 120 @xmath10 3 + hd 69830 & 8h18m23.95s & -12d37m55.8s & 12710656 & 41 & 30 @xmath10 2 + sx cha & 10h55m59.76s & -77d24m40.1s & 27066624 & 50641 & 30 @xmath10 7 + tw cha & 10h56m30.52s & -77d11m39.4s & 27066368 & 50641 & 30 @xmath10 28 + sz cha & 10h58m16.77s & -77d17m17.1s & 22846208 & 40247 & 30 @xmath10 4 + tw hya & 11h01m51.91s & -34d42m17.0s & 18017792 & 30300 & 30 @xmath10 2 + cs cha & 11h02m24.91s & -77d33m35.7s & 18021632 & 30300 & 30 @xmath10 50 + ced110-irs2 & 11h06m15.41s & -77d21m56.8s & 18022400 & 30300 & 30 @xmath10 10 + ced110-irs4 & 11h06m46.58s & -77d22m32.6s & 18019328 & 30300 & 30 @xmath10 36 + sz 18 & 11h07m19.15s & -76d03m04.8s & 18020864 & 30300 & 30 @xmath10 100 + vw cha & 11h08m01.47s & -77d42m28.7s & 27066112 & 50641 & 30 @xmath10 6 + sz 27 & 11h08m39.05s & -77d16m04.2s & 22847232 & 40247 & 30 @xmath10 100 + vz cha & 11h09m23.79s & -76d23m20.8s & 27065856 & 50641 & 30 @xmath10 8 + wx cha & 11h09m58.74s & -77d37m09.0s & 27065600 & 50641 & 30 @xmath10 12 + hen 3 - 600a & 11h10m27.53s & -37d31m53.6s & 18018560 & 30300 & 30 @xmath10 2 + xx cha & 11h11m39.66s & -76d20m15.3s & 27065344 & 50641 & 30 @xmath10 14 + rx j1111.7 - 7620 & 11h11m46.31s & -76d20m09.1s & 5451776 & 148 & 30 @xmath10 6 + chx22 & 11h12m42.69s & -77d22m23.1s & 22846720 & 40247 & 30 @xmath10 60 + 1rxs j121236.4 - 552037 & 12h12m35.77s & -55d20m27.4s & 5460992 & 148 & 120 @xmath10 6 + 1rxs j122233.4 - 533347 & 12h22m33.23s & -53d33m48.9s & 13463296 & 148 & 120 @xmath10 6 + twa 10 & 12h35m04.30s & -41d36m39.0s & 22934528 & 40338 & 30 @xmath10 10 + sz 50 & 13h00m55.38s & -77d10m22.1s & 27065088 & 50641 & 30 @xmath10 14 + 1rxs j130153.7 - 530446 & 13h01m50.69s & -53d04m58.1s & 13462016 & 148 & 120 @xmath10 6 + 1rxs j132207.2 - 693812 ( mpmus ) & 13h22m07.54s & -69d38m12.1s & 5451264 & 148 & 6 @xmath10 3 + 1rxs j133758.0 - 413448 & 13h37m57.30s & -41d34m42.0s & 5460224 & 148 & 120 @xmath10 6 + hd 119269 & 13h43m28.54s & -54d36m43.5s & 5461248 & 148 & 120 @xmath10 2 + hd 134319 & 15h05m49.91s & + 64d02m50.0s & 9779968 & 148 & 30 @xmath10 6 + v343nor & 15h38m57.56s & -57d42m27.3s & 5458944 & 148 & 120 @xmath10 2 + ht lup & 15h45m12.87s & -34d17m30.6s & 27064832 & 50641 & 6 @xmath10 10 + gq lup & 15h49m12.11s & -35d39m04.8s & 27064576 & 50641 & 30 @xmath10 8 + hd 141943 & 15h53m27.30s & -42d16m00.8s & 5462528 & 148 & 120 @xmath10 4 + hd 142361 & 15h54m59.86s & -23d47m18.2s & 5461504 & 148 & 120 @xmath10 12 + i m lup & 15h56m09.21s & -37d56m05.9s & 27064320 & 50641 & 30 @xmath10 6 + ru lup & 15h56m42.31s & -37d49m15.5s & 27064064 & 50641 & 30 @xmath10 5 + hd 143006 & 15h58m36.93s & -22d57m15.4s & 9777152 & 148 & 6 @xmath10 8 + rx j1600.6 - 2159 & 16h00m40.57s & -22d00m32.3s & 17145344 & 148 & 120 @xmath10 9 + ex lup & 16h03m05.49s & -40d18m25.3s & 27063808 & 50641 & 30 @xmath10 6 + rx j1612.6 - 1859 & 16h12m40.51s & -18d59m28.3s & 5457664 & 148 & 120 @xmath10 4 + rxjs j161410.6 - 230542 & 16h14m11.08s & -23d05m36.2s & 5453824 & 148 & 30 @xmath10 2 + 1rxs j161458.4 - 275013 & 16h14m59.17s & -27d50m23.0s & 17144832 & 148 & 120 @xmath10 9 + doar 21 & 16h26m03.03s & -24d23m36.4s & 24403456 & 30300 & 30 @xmath10 6 + doar 24e & 16h26m23.42s & -24d21m00.5s & 27062784 & 50641 & 30 @xmath10 5 + doar 25 & 16h26m23.69s & -24d43m14.0s & 24403712 & 30300 & 120 @xmath10 4 + sr 21 & 16h27m10.20s & -24d19m16.0s & 24403200 & 30300 & 6 @xmath10 10 + haro 1 - 16 & 16h31m33.46s & -24d27m37.4s & 27062016 & 50641 & 30 @xmath10 7 + rno 90 & 16h34m09.17s & -15d48m16.8s & 27061760 & 50641 & 6 @xmath10 9 + waoph6 & 16h48m45.63s & -14d16m36.0s & 27060736 & 50641 & 30 @xmath10 6 + ec 82 & 18h29m56.90s & + 1d14m46.5s & 27059712 & 50641 & 30 @xmath10 8 + rx j1842.9 - 3532 & 18h42m57.97s & -35d32m42.7s & 5451521 & 148 & 30 @xmath10 5 + rx j1852.3 - 3700 & 18h52m17.30s & -37d00m11.9s & 5452033 & 148 & 120 @xmath10 2 + as 353a & 19h20m30.98s & + 11d01m54.8s & 14547456 & 20363 & 30 @xmath10 8 + v1331 cyg & 21h01m09.20s & + 50d21m44.5s & 14547712 & 20363 & 30 @xmath10 8 + hd 202917 & 21h20m49.96s & -53d02m03.2s & 9778176 & 148 & 30 @xmath10 5 + hd 216803 & 22h56m24.05s & -31d33m56.1s & 9777664 & 148 & 30 @xmath10 5 + the irs sh module uses a cross - dispersed echelle design and covers the spectral range 9.9 - 19.6 @xmath0 m with a nominal resolving power of r@xmath6600 ( houck et al . the data reduction was performed starting from the _ droopres _ products following the procedure outlined in pascucci et al . ( 2013 ) . in brief , we computed the average of the on - source and sky exposures per nod position and rejected bad pixels using the noise statistics from multiple cycles . we then created background subtracted images and corrected for the additional known bad pixels reported in the irs bcd mask . the one dimensional spectrum was extracted from the background - subtracted pixel - corrected images using the full aperture extraction routine in smart ( higdon et al . we applied the same procedure to our sources and to the _ spitzer _ flux calibrator @xmath11 dra . from the reduction of 10 different exposures of @xmath11 dra and the known stellar model atmosphere we created two one - dimensional spectral response functions ( one at each nod position ) . finally , we divided the source spectra by the spectral response function . a compilation of four _ spitzer_/irs sh spectra analyzed in this paper is shown in figure [ ex_sh_spectra ] . the calibrated spectra of 54 sources belonging to the program ids 40247 , 20363 , 148 and 41 can be downloaded from the irsa spitzer science center at _ http://irsa.ipac.caltech.edu / data / spitzer / disks_sh_spectra_. the total sample includes objects classified according to their general disk properties . we use the generic term full disks ( f ) for those disks with strong excess emission above photospheric levels at infrared wavelengths , starting in the near - ir . transitional disks ( t ) are broadly defined as disks with reduced near- and mid - ir excess fluxes but excess emission comparable to full disks beyond @xmath620@xmath0 m , indicative of a depleted inner region . finally , debris disks ( d ) are defined in this paper as having pure photosphere emission out to about 10 but can have relatively small excess emission in the long portion of the mid - ir spectrum . this disk classification collected from the literature is based on the dust properties and the continuum spectral energy distributions , and it is mainly taken from furlan et al . ( 2006 , 2011 ) , espaillat et al . ( 2012 ) , silverstone et al . ( 2006 ) , chen et al . ( 2014 ) , as reported in table [ tab_all_prop ] . most of the debris disks included in our study belong to the recent catalog published by chen et al . full and transitional disks are primordial , gas - rich disks , while debris disks are second - generation disks , dominated by dust . in total our sample includes 68 full , 21 transitional , and 25 debris disks . in figure [ isto_stellar_prop ] we show the distribution in mass , luminosity , effective temperature and accretion luminosity ( see sect . [ accr_prop_sect ] ) for the total sample , and for the objects where the h i(7 - 6 ) line is detected . a k - s two sample test of the stellar properties shows the two dataset have most likely the same parent distribution . m , h i(9 - 7 ) at 11.32@xmath0 m and [ ne ii ] at 12.81@xmath0 m . [ ex_sh_spectra ] ] ] in the following , we summarize the main properties of our sample ( sect . [ sect_stellar_prop ] ) , compute accretion luminosities for each object in a self consistent manner ( sect . [ accr_prop_sect ] ) , and describe in detail how we calculate [ ne ii ] and h i fluxes taking into account any possible contamination from water lines ( sects . [ ne_par ] and [ hi_par ] ) . the last two sections ( sects . [ det_rates ] and [ cont_sect ] ) discuss detections and correlations between line fluxes and adjacent continuum emission . the properties of the 114 objects are summarized in table [ tab_all_prop ] . for each object we provide the stellar association it belongs to , the distance , spectral type ( spty ) , visual extinction ( a@xmath12 ) as well as the effective temperature ( t@xmath13 ) obtained using the t@xmath13-spty correlation of kenyon & hartmann ( 1995 ) and luhman et al . equivalent widths and stellar luminosities ( l@xmath14 ) have been collected from the literature using the references reported in table [ tab_all_prop ] , while the mass ( m@xmath14 ) , and radius ( r@xmath14 ) were derived from the hr diagram and the siess et al . ( 2000 ) evolutionary tracks . for multiple star systems the name we report in table [ tab_all_prop ] is that of the primary star , which is the one with the earliest spectral type . however , in the table caption we list the separation of any other stellar companion within 10@xmath15 of the primary star . given that the irs / sh slit width is 4.7@xmath16 , there are 15 objects with companions closer than 2.5@xmath15 that can contaminate the extracted irs spectra . however , mccabe et al . ( 2006 ) show that only in 25% of the binary systems the secondary flux becomes larger than the primary flux at wavelengths @xmath92 . based on this , we assign the spitzer spectrum always to the primary star in the 15 unresolved multiple systems . we emphasize that the sample collected is biased in several ways : i ) the number of objects with sky observations and multiple on - source exposures ( see sect . [ section : obs ] ) , ii ) we only consider objects where the continuum has been detected and the signal - to - noise ratio of continuum level is higher than 5 . hence , we do not attempt any statistical study based on this sample . to test whether mid - ir h i lines trace the accretion shock region and/or the accretion flow , we compare h i line luminosities with stellar accretion luminosities ( @xmath17 ) . one of the most direct methods to measure @xmath17 is through modeling of the uv excess continuum emission produced by the accretion shock ( e.g. gullbring et al . 1998 ) . however , only 17 objects in our sample have such measurements ( aa tau , cs cha , de tau , dk tau , dn tau , dm tau , dr tau , gm aur , hn tau , lkca 15 , rx 132207.2 - 693812 , tw hya , sz cha , sz 27 , lkha 330 , rx j1852.3 - 3700 and rx j1842.9 - 3532 , see ingleby et al . 2013 and manara et al . 2014 ) . to measure accretion luminosities in a consistent way for all the objects in our sample we will use the h@xmath8 line luminosity . even if this method allows an homogeneous estimate of the accretion luminosity , we stress that the hydrogen lines in general , and the h@xmath8 line in particular , have several contributing mechanisms : accretion , stellar winds , jets , chromosphere ( e.g. hartmann et al . 1990 , calvet & hartmann 1992 ) . while this line is known to be an imperfect tracer of accretion in young stars , rigliaco et al . ( 2012 ) , among others , have shown that @xmath17 obtained from the uv excess emission and via indirect methods ( e.g. , h@xmath8 line luminosity ) agree to within an order of magnitude . to measure the accretion luminosity we first compute the h@xmath8 luminosity from the h@xmath8 equivalent width , the _ r- _ and/or _ i-_band magnitude , and the extinction ( a@xmath12 ) for each object ( e.g. , hartigan et al . the data were dereddened using mathis s reddening law for r@xmath18 = 3.1 ( mathis 1990 ) . next , we convert the h@xmath8 luminosity into @xmath17 using the relationship found by rigliaco et al . ( although this method provides self - consistent values for @xmath17 we are aware that the not - simultaneous observation of h@xmath8 , photometry , and a@xmath12 might introduce additional uncertainties in our estimates of @xmath17 . in sect . [ hiasaccr ] we describe how we account for these sources of uncertainty . water is largely detected in protoplanetary disks ( e.g. , pontoppidan et al . 2010 , riviere - marichalar et al . 2012 ) , and its emission can contaminate other emission lines , as shown in the following sections . this is not the case for the [ ne ii ] line , that is quite isolated from other h@xmath19o transitions . for this water - isolated and spectrally unresolved gas line we compute the flux by fitting a gaussian profile to the observed line . we define the continuum level around the line as the mean value obtained sampling a 0.1@xmath0 m wide region shortward and longward of the transition , and the uncertainty on the measured flux is the rms in these regions . the line flux and the associated 1@xmath20 uncertainty are reported in table [ flux_obj ] for each detection . in the same table we provide 3@xmath20 upper limits to the flux when [ ne ii ] lines are not detected . line luminosities are then computed using the distances reported in table [ tab_all_prop ] . we have compared the measured [ ne ii ] line fluxes ( as reported in table [ flux_obj ] ) with those reported by other authors using either different data reduction techniques ( pascucci et al . 2007 , lahuis et al . 2007 , espaillat et al . 2013 ) , and/or including any possible unresolved contribution from other species ( h@xmath19o and oh , carr & najita 2011 ) . we find that the measured line fluxes and upper limits are in agreement within the errors ( see fig . [ neii_comparison ] ) . line fluxes measured following the data reduction chain explained in sect . [ section : obs ] and the line fluxes obtained from the literature using different data reduction techniques . [ neii_comparison ] ] the method used to compute the [ ne ii ] line fluxes and luminosities could be in principle applied to the h i lines . however , because the h i(7 - 6 ) and h i(9 - 7 ) transitions are very close to water lines we must subtract any possible contamination from water before measuring line fluxes . we proceed as follows . in sect . [ sect_water1 ] we identify the objects that lack detectable water emission ( hereafter , water - free disks ) and the objects where water is detected in their spectra ( water - contaminated disks ) . for the water - free disks we compute line fluxes as in sect . [ ne_par ] . for the water - contaminated disks we discuss a procedure to subtract water emission from the h i lines ( sect . [ sect_water2 ] ) and then we measure the residual h i fluxes after water subtraction . to search for water emission in the source spectra we used the two water line complexes at 15.17@xmath0 m and 17.22@xmath0 m identified by pontoppidan et al . these lines were selected because they are strong and more isolated than most other h@xmath19o lines . following pontoppidan et al . ( 2010 ) , we define as water contaminated those spectra in which both the h@xmath19o lines are detected above 3@xmath20 . we do not find any case where one of the two lines is detected above 3@xmath20 , and the other one is not detected . among the 114 disks in our sample , 33 are water - contaminated . the remaining objects have water - free spectra . we show in figure [ figwater ] a water - contaminated spectrum ( dr tau ) , and a water - free spectrum ( de tau ) in the region around the 15.17@xmath0 m and 17.22@xmath0 m water lines . for comparison we also show a synthetic water spectrum modeled assuming a gas temperature of 600 k and a water column density of 10@xmath21 @xmath22 ( see next section ) . line fluxes , together with the uncertainties and continuum level next to the lines are reported in table [ flux_obj ] . water emission lines have been detected towards many protoplanetary disks using the _ spitzer_/irs spectrograph ( pontoppidan et al . 2014 for a review ) . previous studies have shown that a simple model , which assumes a plane - parallel slab geometry and gas in local thermal equilibrium ( lte ) , can reproduce relatively well the fluxes of the observed h@xmath19o lines ( e.g. , carr & najita 2011 ) . however , we are aware that these assumptions ignore the non - lte effects and the complex geometric structure of disks ( see meijerink et al . we use this simple model as developed in pascucci et al . ( 2013 ) to subtract any water contamination from the h i lines . the specific lte model takes the molecular parameters from the hitran 2008 database ( rothman et al . 2009 ) and requires only 4 inputs : the gas temperature , the water column density , the radius of the emitting area , and the line width ( which we assume to be the thermal line width , i.e. , variable with temperature ) . we created a grid of synthetic spectra using five different gas temperatures ( t=[400 , 500 , 600 , 700 , 800 ] ) and water column densities ( n@xmath23=[10@xmath24 , 5@xmath1010@xmath24 , 10@xmath21 , 5@xmath1010@xmath21 , 10@xmath25 @xmath22 ] ) and assumed the same projected emitting area of radius 1 au . the chosen inputs span the range of parameters inferred for the warm disk atmosphere of t tauri stars ( carr & najita 2011 , salyk et al . 2011 ) . the water lines contaminating the hydrogen lines are centered at @xmath612.39@xmath0 m and @xmath611.31@xmath0 m for the h i(7 - 6 ) and h i(9 - 7 ) transition , respectively . in the following we will refer to these lines as `` h@xmath19o - contaminating '' . to subtract off the water contribution to the observed h i lines we have searched the synthetic spectra for _ isolated _ h@xmath19o lines with similar properties as the h@xmath19o - contaminating ones . the two water lines that we have identified will be called in the following `` h@xmath19o - isolated '' lines and have the following characteristics : they have similar lower energies ( e@xmath27 ) and einstein a - coefficients ( a@xmath28 ) ( within 15% at most ) as the h@xmath19o - contaminating lines ; h@xmath19o - isolated line is within 1 from the corresponding contaminating line , so that the continuum level does not substantially change . following these criteria we find that the h@xmath19o - isolated line at 13.13@xmath0 m is the best match for the h@xmath19o - contaminating line at 12.39@xmath0 m , while the h@xmath19o - isolated at 11.96@xmath0 m has very similar properties as the h@xmath19o - contaminating at 11.31@xmath0 m ( see table [ line_prop ] ) . we show in fig . [ fig_ratio ] the ratio between the h@xmath19o contaminating / isolated line fluxes for all the temperature and column density of the synthetic spectra . the ratios are quite constant among all the synthetic spectra , with h@xmath19o(12.39@xmath0m)/h@xmath19o(13.13@xmath0m)@xmath63.5 and h@xmath19o(11.32@xmath0m)/h@xmath19o(11.97@xmath0m)@xmath61.0 , and vary within a factor 1.5 at most over our grid of synthetic spectra . once we identified the best water proxies , we searched all the spectra for h@xmath19o - isolated lines . in the sh-_spitzer_/irs spectra none of the 83 water - free spectra shows these lines , as expected because the h@xmath19o lines become weaker at shorter wavelengths ( e.g. , salyk et al . in the 33 water - contaminated disks , we subtracted the water contribution from the h i line fluxes as follows : if the h@xmath19o - isolated line is detected above the 3@xmath20 level , we use it and its ratio to the nearby h@xmath19o - isolated line to compute the h@xmath19o - contaminating flux ( as h@xmath19o(12.39@xmath0m)=3.5@xmath10h@xmath19o(13.13@xmath0 m ) and h@xmath19o(11.32@xmath0m)=h@xmath19o(11.97@xmath0 m ) , respectively for h i(7 - 6 ) and h i(9 - 7 ) ) , and then subtract it from the measured hi line flux . if the h@xmath19o - isolated is not detected , we subtract its 3@xmath20 upper limit from the observed h i flux . after subtracting the water contamination from the observed line flux , we obtain the actual hydrogen line flux . the final line fluxes after water subtraction and errors are reported in table [ flux_obj ] . we also show in parenthesis the total flux observed before subtracting the water contribution in the objects where the h i lines are detected . in the case of h i non - detections we provide the 3@xmath20 upper limit to the flux . any other water contribution , if present , is within the 10% absolute flux uncertainty . as for the [ ne ii ] , we have compared our h i(7 - 6 ) line flux measurements with the few already reported in the literature ( pascucci et al . 2007 , carr & najita 2011 ) . we find that the reported values are in agreement within the errors and the absolute flux uncertainty . l c c c h@xmath19o - contaminating & 11.3241 & 1.576 & 2746.02 + h@xmath19o - contaminating & 12.3962 & 7.665 & 3211.21 + h@xmath19o - isolated & 11.9681 & 1.281 & 2629.33 + h@xmath19o - isolated & 13.1325 & 6.788 & 3244.60 + we detect the [ ne ii ] line at 12.81 in 49 objects in our sample . among the 67 objects where the [ ne ii ] line is not detected , 37 are full disks , 4 are transitional disks and 26 are debris disks . one debris disk has a [ ne ii ] line detection ( rigliaco et al . in preparation ) . after correcting for water contamination , we identify 46 objects with an h i(7 - 6 ) detection above 3@xmath20 , 11 of them also show the weaker h i(9 - 7 ) line . in 6 spectra we detect all three lines of interest : the [ ne ii ] , the h i(7 - 6 ) , and the h i(9 - 7 ) line . line detections are summarized in table [ tab_number ] . we note that all objects where water emission is detected are full disks and none of the transitional or debris disk spectra have detectable water emission . thus , expanding upon previous results ( e.g. , pontoppidan et al . 2010 ) , our analysis shows that the absence of water lines in the sh module is typical to more evolved transitional and debris disks , while full disks can either have a water rich spectrum or not . interestingly , we detect h i(7 - 6 ) in 8 young ( < 20 myrs ) debris disks . we will discuss these objects in sect . [ section_dd ] . l c \| c c c & 49 & 67 & 31 & 2 + h i(7 - 6 ) & 46 & 57 & 26 & 17 + h i(9 - 7 ) & 11 & 55 & 27 & 18 + + h i(7 - 6 ) & 24 & 50 & 46 & 4 + h i(7 - 6)+h i(9 - 7 ) & 11 & 55 & 27 & 18 + [ ne ii]+h i(7 - 6)+h i(9 - 7 ) & 6 & 50 & 50 & 0 + once we accounted for the water contribution and measured the h i(7 - 6 ) and h i(9 - 7 ) line luminosities , we investigated their behavior as a function of the continuum luminosity around the lines ( fig . [ h76_cont ] ) . as also seen in other ir lines ( h@xmath19s(2 ) , [ fe i ] at 24@xmath0 m , [ ne ii ] , e.g. , lahuis et al . 2007 ) , there is a correlation between the h i mid - ir lines and the continuum flux level next to the lines . this correlation could be influenced , but it is not exclusively driven , by the more favorable contrast between line and continuum luminosities where the continuum is lower ( as demonstrated by finding the same increasing trend between the line to continuum vs continuum flux ) . we have also investigated this correlation for different disk morphologies and find that debris disks have lower h i line luminosities and continuum luminosities than transitional and full disks . in order to understand whether these differences are statistically significant we have measured the mean and standard deviation of line and continuum luminosities for the three different disk morphologies ( shown as solid lines in fig . [ h76_cont ] ) . from this analysis we conclude that , even if transitional disks appear having lower h i and continuum luminosities than full disks , their difference is within 1@xmath20 uncertainty , and thus not statistically significant . on the other hand , debris disks show consistently lower luminosities , both in the continuum and in the lines . in the following we compare the [ ne ii ] line fluxes obtained from our reduction of the _ spitzer _ spectra with those of the h i lines to test if these two lines could trace the same region in / around the disk . in particular , we will test if [ ne ii ] and h i lines may trace the disk surface . as mentioned in sect . [ intro ] , [ ne ii ] lines have been studied both with _ spitzer _ and with ground - based spectrographs at higher spectral resolution . these studies have shown that the [ ne ii ] line mostly traces jets / outflows in high - accreting stars and the hot disk surface in lower accretors ( @xmath29@xmath30/yr , e.g. gdel et al . 2010 , sacco et al . unfortunately , similar ground - based followup studies have not yet been carried out for the mid - ir h i lines , hence a direct comparison of high - resolution spectra can not be done . the vast majority of the objects in our sample ( namely , 85% ) for which we have computed @xmath17 ( and @xmath29 ) can be classified as `` low - accretors '' according to the threshold value reported above ( gdel et al . hence , for most of our sources a detection of [ ne ii ] in the _ spitzer _ spectrum is most likely due to emission from the hot disk surface . the left panel of figure [ lh76_lne_comparison ] shows a log - log plot of the h i(7 - 6 ) line luminosity at 12.37@xmath0 m versus the [ ne ii ] line luminosity at 12.81@xmath0 m . the luminosities of the two lines , when they are both detected , appear positively correlated , with _ correlation coefficient of @xmath60.9 , and p - value@xmath610@xmath31 . the correlation becomes weaker when upper limits are considered . however , this correlation could be affected by the underlying correlation between the line luminosities and the continuum luminosities around the line ( see figure [ h76_cont ] and discussion in sect [ cont_sect ] ) . to account for this possible effect we plot in fig . [ lh76_lne_comparison ] ( right panel ) the equivalent widths of the two observed lines . the [ ne ii ] and h i(7 - 6 ) equivalent widths do not show any obvious correlation , and the _ pearson _ correlation coefficient drops to @xmath60.4 ( weak correlation ) and p - [email protected] . this result suggests that the positive relation between the h i(7 - 6 ) and [ ne ii ] line luminosity is caused by the correlation between the line and nearby continuum . to further test this finding , we compare the observed h i line luminosities with those predicted by models of disks irradiated by high - energy photons ( ercolano et al . 2008 , hollenbach & gorti 2009 , ercolano & owen 2010 ) . in an euv ( 13.6 [email protected] kev ) irradiated disk the predicted l@xmath33/ l@xmath34}$ ] ratios are as small as @xmath60.008 ( hollenbach & gorti 2009 ) because the radiative recombination rate of hydrogen is smaller than the electronic collisional excitation rate of [ ne ii ] . the observed line ratios are instead almost two orders of magnitude higher than the predicted ones ( see table [ flux_obj ] ) . a disk irradiated also by x - rays ( @xmath350.1kev ) still produces an h i(7 - 6 ) line luminosity more than one order of magnitude smaller than the ones observed ( ercolano et al . 2008 , ercolano & owen 2010 ) . the comparison between the observed line luminosity and different theoretical models shows that it is unlikely that the observed hi(7 - 6 ) lines trace the same hot disk atmosphere that produces the [ ne ii ] lines . in summary , both the lack of correlation between the [ ne ii ] and the h i(7 - 6 ) equivalent widths , and the comparison with model predictions suggest that the h i(7 - 6 ) line is not tracing the same disk region as the [ ne ii ] line , and that the h i lines are unlikely disk gas tracers . hydrogen recombination lines from the balmer to the paschen and brackett series ( in the uv , optical and near - ir wavelengths ) have been extensively studied over the years , and have been convincingly found to trace infalling material onto the star , as demonstrated by their broad line profiles consistent with gas at almost free - fall velocity onto the star ( e.g. , alcal et al . nevertheless , these line profiles are also known to have other contributing mechanisms : stellar wind , jet , chromosphere . recently , salyk et al . ( 2013 ) , using high - resolution spectroscopy extended the accretion tracers to the pfund series lines , including the pf@xmath7 hydrogen recombination line at 4.65@xmath0 m ( h i(7 - 5 ) ) . to test if the h i mid - ir lines can be used as accretion indicators , in the rest of this section we focus on the subsample of full and transitional disks for which an estimate of the h@xmath8 equivalent width is available in the literature ( see table [ tab_all_prop ] ) . we chose to use the h i(7 - 6 ) line ( the strongest among the h i lines observed with the _ spitzer _ sh module ) to maximize the number of detected sources . among the objects considered in this work , 73 have an estimate of the accretion luminosity from the h@xmath8 line ( 34 of those have an h i(7 - 6 ) detection ) . we show in fig . [ h76_lacc ] the h i(7 - 6 ) line luminosity as a function of the h@xmath8 line luminosity and the accretion luminosity measured as described in sect . [ accr_prop_sect ] . there is a linear correlation between logl@xmath36logl@xmath37 . to estimate the best fit regression line for the objects with detection in both quantities , and to account for the measured errors in both the x- and y - axes we employe the bayesian method as implemented in the idl routine _ linmix_err _ ( kelly 2007 ) . the error on the y - axes is driven by the uncertainty on the observed line fluxes . the uncertainty in our estimate of @xmath17 is given by the combination of several factors : the uncertainty on the h@xmath8 equivalent width , the non - simultaneous observation of h@xmath8 , photometry , and extinction , the error on the extinction value ( a@xmath12 ) and the errors on slope and intercept of the @xmath17-l@xmath38 relationship . following costigan et al . ( 2012 ) we assume a mean amplitude of variation in the derived accretion luminosities of @xmath60.37 dex , which includes h@xmath8 equivalent width uncertainties , time variability , and the error in the @xmath17-l@xmath38 relationship . we also consider an uncertainty on a@xmath12 of 1 mag ( e.g. edwards et al . 2013 ) , which gives a final uncertainty on @xmath17 @xmath60.5 dex . we find the following best fit regression line between the h i(7 - 6 ) line luminosity and the accretion luminosity : @xmath39 we have checked that the log@xmath40log@xmath17 correlation is not driven by the distance and/or stellar luminosity dependence . we have also ascertained that the relationship between the luminosity of the h i(7 - 6 ) line and @xmath17 measured through uv - excess emission for only 15 objects ( see sect . [ accr_prop_sect ] ) lies within the 95% confidence intervals . we note that the uncertainty on the intercept we find in eq . ( 1 ) is similar to the those found using different h i lines ( e.g. , alcal et al . 2014 ) , suggesting that the scatter around the fit is similar when using different h i lines . on the other hand , the slope found in eq . ( 1 ) is shallower than the ones found for the h i lines in the optical and near - ir wavelengths . this suggests that the mir lines are less sensitive to accretion luminosity variations . knowing @xmath17 and the stellar properties we can also compute the mass accretion rate ( @xmath29 ) using the classical relation from gullbring et al . ( 1998 ) : @xmath41 where r@xmath42 is the radius at which the accreting gaseous disk is truncated due to the stellar magnetosphere . in full disks , r@xmath42 has been found ranging between 3r@xmath14 and 10r@xmath14 ( johns - krull 2007 ) , and usually r@xmath42=5r@xmath14 is assumed . in our sample we do not only consider full disks , but also transitional disks where the geometry of the gas disk around the star could be different . we do not speculate on how that could change for more evolved disks , and we will use the r@xmath42=5r@xmath14 for all the different disk morphologies . for the subsample of objects where the pf@xmath7 line is detected ( salyk et al . 2013 ) and the h i(7 - 6 ) is detected and not water contaminated ( this work ) we find that their luminosities , as well as equivalent widths , are correlated with _ correlation coefficients @xmath60.7 and @xmath60.8 , respectively , and p - value @xmath60.007 and @xmath60.05 , respectively ( fig . [ lh76_lpfb_comparison ] ) . while salyk et al . 2013 report lower pf@xmath7 luminosities and high ews for 5 transitional disks with respect to full disks , we do not find any statistically significant difference in the h i(7 - 6 ) lines in our much larger samples of transitional and full disks ( see also sect . [ disk_tracers_sect ] ) . we also note that these two samples have not statistically different @xmath17 , as also reported in other studies where @xmath17are consistently derived ( keane et al . 2014 , fang et al . 2009 ) . the fact that the h i(7 - 6 ) line correlates both with the h@xmath8 and pf@xmath7 lines suggests that the humphreys @xmath8 line at 12.37@xmath0 m is likely an accretion indicator , as the hydrogen lines belonging to other series ( h@xmath8 ( e.g. , muzerolle et al . 1998 , pa@xmath7 ( e.g. , folha & emerson 2001 ) , br@xmath43 ( e.g. , natta et al . 2006 ) , pf@xmath7 ( salyk et al . 2013 ) ) . speculating on the correlation reported in eq . ( 1 ) , we can measure what the mass accretion rates would be if the h i(7 - 6 ) lines in the debris disks spectra were accretion related . we find that debris disks with detected h i lines span the lower range of mass accretion rate ( covering the range between @xmath1010@xmath42@xmath1010@xmath44 m@xmath5/yr ) , as would be expected by their more evolved stage and the lack of gas in their disks . hydrogen line ratios can be used as a diagnostic tool to constrain the physical conditions ( density and temperature ) of the emitting gas . among the sample of objects analyzed in this paper , in 46 the h i(7 - 6 ) transition is detected above 3@xmath20 , and 11 of those show as well a detection of the h i(9 - 7 ) transition . the detected h i(9 - 7)/h i(7 - 6 ) line ratios range between @xmath60.4 and @xmath61.1 . the upper limits cover a similar range . we can retrieve information on the physical properties of the emitting gas by comparing the observed ratios with those predicted by two different models : the case b ( hummer & storey 1987 , storey & hummer 1995 ) and the kwan and fischer ( kwan & fischer 2011 , kf ) models . in these models the h i line ratios ranges between @xmath60.30.5 in the optically thin regime , and become higher ( between @xmath61.02.0 ) when the lines are optically thick . the case b model for radiative ionization and recombination ( baker & menzel 1938 ) assumes that gas is optically thick to lyman series photons and optically thin to photons associated to all other transitions . this model has been largely used to retrieve the properties of gas accreting onto t tauri stars . however , results found considering individual objects and different hydrogen lines indicate very different physical properties of the accreting gas ( bary et al . 2008 , rigliaco et al . 2009 , and edwards et al . 2013 for a review ) . using the interactive online server ( hummer & storey 1987 , storey & hummer 1995 ) , we computed the predicted line ratios for a range of temperatures from 1000 k to 15,000 k and electron densities from 10@xmath4510@xmath46@xmath3 ( see fig . [ casebmodels ] ) . the predicted h i(9 - 7)/h i(7 - 6 ) line ratios are almost constant ( between @xmath60.3 - 0.4 ) for t@xmath475000 k over the electron density range considered here . as the temperature decreases under 5000 k , the predicted ratio increases from @xmath60.3 to @xmath60.7 until electron densities @xmath48@xmath3 , and then decreases to @xmath60.2 when @xmath49 becomes larger . thus , a large range of temperatures , down to t@xmath61000 k , is required to explain most of the observed ratios . still , ratios as high as 1.0 can not be explained by the case b models . in addition , as pointed out in edwards et al . ( 2013 ) , the necessity to maintain low line optical depths in the case b models puts some restrictions to the applicability of this model to regions with high densities such as accretion funnels in ysos . kwan & fischer ( 2011 ) predicted hydrogen line ratios using a different approach than the case b , in that they consider individual line optical depths . the inputs of these calculations are the gas temperature , the hydrogen density , the ionization rate ( which was not specified in the case b models ) and the velocity gradient with respect to the emission length scale ( see kwan & fischer 2011 and edwards et al . 2013 for more details ) . these parameters are chosen to replicate conditions in the region of winds and accretion flows in accreting ysos . we show in figure [ kfmodels ] the comparison between the kf model predictions for the h i(9 - 7)/h i(7 - 6 ) line ratio and the observations . the input gas temperatures span from 5000 k to 15,000 k , and given that the ( n@xmath50)/(n@xmath51 ) ratio is between 0.1 and 1.0 , the hydrogen densities ( n@xmath51 ) cover similar electron densities as those shown in fig . [ casebmodels ] for the case b models . the ratios suggested by the kf models are almost constant for all the temperatures until n@xmath52 @xmath3 . as the density increases , the line optical depth increases as well , first for the ( 7 - 6 ) transition , which explains the steep increase in line ratio . for ( n@xmath51)@xmath910@xmath2@xmath3 both lines have large optical depths which explains the plateau . while the general behavior is similar for all temperatures , the specific transition density is @xmath53 @xmath3 for t@xmath5410,000 k and @xmath55@xmath3 for lower temperatures . thus , unlike the case b models , the kf models can explain all the observed line ratios with a limited range of temperatures ( 5000 15,000 k ) and relatively narrow range of gas densities ( n@xmath56@xmath3 ) . according to martin ( 1996 ) , who modeled the temperature and density profiles for gas channeled in magnetospheric accretion funnels , the typical temperatures in these regions are @xmath66000 k , and hydrogen densities are as high as 10@xmath4510@xmath1@xmath3 . muzerolle et al . ( 2001 ) , considering the h@xmath8 and h@xmath7 lines , found that the temperature of magnetospherically accreting gas ranges between 6000 k 12000 k for a range of accretion rates between @xmath57 @xmath58 m@xmath59/yr . recently , edwards et al . ( 2013 ) examined near - ir hydrogen line ratios for a sample of t tauri stars . they find that the hydrogen density ( @xmath60 ) in the line formation region is constrained within 2@xmath1010@xmath12@xmath1010@xmath2@xmath3 , while the temperature is not well delineated . if the hydrogen mid - ir lines are tracing accretion at least in full and transitional disks , the observed h i(9 - 7)/h i(7 - 6 ) line ratios can not be explained by the gas properties predicted by the case b models . in fact , as we have seen is sect . [ casebmod ] and fig . [ casebmodels ] the observed line ratios require gas colder than 5000 k irrespective of the electron densities . such physical properties , and in particular such low temperatures , are unlikely in the magnetospheric accretion columns , in which we expect temperature of at least a factor of two higher and a much narrower range in density ( e.g. , muzerolle et al . 2001 ) . on the other hand , the kf models suggest gas properties that better match the scenario in which the mid - ir hydrogen lines are produced in the accretion funnels because they predict an hydrogen column density ( @xmath60 ) between @xmath61 @xmath3 , and a temperature range between 5,000 and 15,000 k ( see fig . [ kfmodels ] ) . the analysis of the h i(7 - 6)/pf@xmath7 line ratios for the subsample of objects where these two lines have been observed adds additional constraints . in all but one object the observed line ratio is consistent with the kf model predictions for @xmath60 between @xmath62 @xmath3 and t@xmath95,000 k ( see fig . [ kfmodels_pfb ] ) . these results not only suggest that in most cases h i(7 - 6 ) and pf@xmath7 might have a common origin , but that this origin is from hot gas with relatively high hydrogen column density ( as expected in the accretion funnels ) . the objects where both h i(7 - 6 ) and pf@xmath7 are detected , but the h i(7 - 6)/pf@xmath7 line ratio does not fit the kwan & fischer models are vw cha and tw hya , where the h i(7 - 6)/pf@xmath7 line ratios is @xmath60.9 . we note that similarly stringent constraints on the gas physical properties can not be placed when using shorter wavelengths hydrogen lines . in particular , we have analyzed model predictions ( both case b and kf models ) for the h i(7 - 6)/h@xmath8 and h i(7 - 6)/pa@xmath7 lines . we find that the observed ratios span a wider range of values , making the comparison with models inconclusive . this could be due to the different optical depths of the compared lines , which could affect the h@xmath8 line as well as other short wavelength transitions , or to time variability , which may be more pronounced at shorter wavelengths . in the case of the h i(7 - 6)/h@xmath8 line ratio we must also take into account that balmer series lines in general , and h@xmath8 line in particular , have quite complicated profiles , with a variety of structures that include blueshifted absorption and line asymmetries ( e.g. , basri & batalha 1990 ) . the analysis carried out so far has shown that the h i mid - ir lines are very unlikely disk tracers . moreover , the correlation between the h i(7 - 6 ) line and the accretion luminosity , and the h i line ratios support an origin of these lines in the accretion funnels for full and transitional disks . however , hollenbach & gorti ( 2009 ) discussed two other possibilities for the origin of h i mid - ir lines : an internal wind shock with high speed ( @xmath63 100 km / s ) and close to the wind origin ( @xmath641 au ) , or the stellar chromosphere ( see also pascucci et al . 2007 ) . an internal shock origin , even if plausible for full disks , would be very difficult to reconcile with the detection of h i in the spectra of transitional and debris disks . transition disks rarely show signs of outflows activity ( see for example the optical forbidden line profiles in hartigan et al . 1995 ) , and debris disks have already dispersed most of their gas disk mass ( e.g. pascucci et al . 2006 ) . while it is possible that mid - ir hydrogen lines trace different environments depending on the stellar / disk evolutionary stage , as shown for the [ ne ii ] line ( see sect . [ intro ] ) , a wind shock origin can not explain the detections in transitional and debris disks . the mean chromospheric electron density in t tauri stars is @xmath65@xmath3 ( giampapa et al . 1981 , or @xmath66@xmath3 if 0.1@xmath67(n@xmath50)/(n@xmath51)@xmath671 , see sect . [ kf_section ] ) . according to the kf model at these high densities the h i(7 - 6 ) and ( 9 - 7 ) transitions would be both optically thick , hence line ratios should be @xmath541 . only few sources in our sample sport such large line ratios . another way that we have employed to investigate the chromospheric origin of the h i lines is to consider the effect of the chromospheric emission on the measured accretion luminosity . manara et al . ( 2013 ) , quantified this effect by analyzing a sample of high dispersion spectra of non - accreting low - mass t tauri stars . measuring the h@xmath8 emission in non - accreting stars , and converting this value into an accretion luminosity , they find that the chromospheric activity can significantly contribute to the measured accretion luminosity when the latter is smaller that @xmath68l@xmath59 , and that this contribution is strongly dependent on the object s effective temperature ( hence spectral type ) . they derive a correlation between t@xmath69 and l@xmath70 ( the latter representing the chromospheric contribution to the accretion luminosity ) . unfortunately , their sample is limited to objects with spectral type in the range k7-m9.5 , and none of our debris disks falls in this range . in the following we will focus on the 42 objects with t@xmath714100 k and with an estimate of @xmath17 ( see table [ tab_all_prop ] ) . in fig . [ chrom_test ] we plot l@xmath70 obtained using the correlation provided by manara et al . ( 2013 , their eq . 2 ) versus @xmath17 as reported in table [ tab_all_prop ] . 13 of these objects also shown a detection in the h i(7 - 6 ) line ( green - filled symbols ) . there is a small number of targets , where the ratio is not more than 5 , for which the chromospheric contribution could substantially influence the estimate of the accretion luminosity . however , in the majority of objects @xmath17 is at least 5 times bigger than its chromospheric contribution , meaning that the latter is well within the estimated uncertainty on @xmath17 . moreover , the objects where the h i(7 - 6 ) line is detected ( identified as green - filled symbols in fig . [ chrom_test ] ) span all over the range of @xmath17 . /l@xmath70 ) , measured using the relation found by manara et al . ( 2013 ) , versus @xmath17 for all the objects with t@xmath714100 k. symbols are as labeled . filled symbols refer to objects where h i(7 - 6 ) is detected . [ chrom_test ] ] we can only test the chromospheric contribution over the accretion luminosity for @xmath640% of the total sample , where t@xmath714100 k. for this subsample the previous analysis suggests that the h i lines are not contaminated by the chromospheric contribution at a significant level . we could not extend this analysis to objects with higher temperatures ( earlier spectral types ) and to debris disks . _ it is possible that the h i mid - ir lines are tracing accretion in less evolved objects , while their contribution is only chromospheric in later evolutionary stages . _ the way to discriminate between accretion funnels , chromosphere and high - density wind is through ground - based high - resolution ( 10 - 20 km / s ) spectroscopy . in fact , for h i lines originating in accreting gas we will observe broad lines ( fwhm@xmath6100 km / s , muzerolle et al . 2001 ) close to the stellar velocity . a chromospheric component will be narrower ( @xmath640 km / s ) and centered at the stellar velocity ( e.g. herczeg et al . 2007 ) . an internal shock origin will instead produce h i profiles that are as broad as 100 km / s and can be also blueshifted by the same amount . as already mentioned in the previous sections , this paper reports the first detection of h i(7 - 6 ) and h i(9 - 7 ) gas lines in debris disks ( we also detect the [ ne ii ] gas line in the _ spitzer _ spectrum of hd 35850 , as will be discussed in rigliaco et al . , in preparation ) . we show in fig . [ spitz_dd ] a compilation of the 8 debris disks line profiles where the h i(7 - 6 ) line has been detected . in the following we will investigate the properties of all the debris disks we have re - reduced , either with detected h i(7 - 6 ) lines or not ( as reported in tables [ tab_all_prop ] and [ flux_obj ] ) . all these debris disks are observed within the feps _ survey ( meyer et al . 2006 ) , and we collect the quantities plotted in figure [ dd_prop_fig ] and reported in table [ dd_prop ] from the feps database . ] /l@xmath14 versus l@xmath36/l@xmath14 . bottom - left panel : chromospheric activity diagnostic r@xmath72 versus the h i line luminosity . bottom - right panel : h i(7 - 6 ) line luminosity versus age . [ dd_prop_fig ] ] l c c c c 1rxs j121236.4 - 552037 & 17 & 30.32 & & [ mam02],[feps ] + 1rxs j122233.4 - 533347 & 17 & 30.34 & & [ mam02],[feps ] + 1rxs j130153.7 - 530446 & 17 & 30.01 & & [ mam02],[feps ] + hd 119269 & 17 & 30.60 & & [ mam02],[feps ] + 1rxs j133758.0 - 413448 & 15 & 30.40 & & [ mam02],[feps ] + 1rxs j161458.0 - 275013 & 5 & 30.05 & & [ pre02],[feps ] + hd 142361 & 5 & 30.71 & -4.02 & [ pre02],[feps],[whi07 ] + rxj1600.6 - 2159 & 5 & 30.31 & & [ pre02],[feps ] + rxj1612.6 - 1859 & 5 & 30.65 & & [ pre02],[feps ] + hd 35850 & 12 & 30.34 & -4.22 & [ zuc01],[feps],[wri04 ] + v343nor & 12 & 30.33 & & [ zuc01],[feps ] + hd 19668 & 257 & & -4.38 & [ feps],[wri04 ] + hd 25457 & 112 & 29.74 & -4.39 & [ luh05],[feps],[wri04 ] + hd 37484 & 50 & 29.16 & -4.49 & [ feps],[feps],[whi07 ] + hd 12039 & 63 & 29.56 & -4.16 & [ feps],[feps],[whi07 ] + hd 202917 & 32 & 30.07 & -4.22 & [ feps],[feps],[wri04 ] + hd 141943 & 16 & 30.62 & -4.36 & [ feps],[feps],[hen96 ] + hd 377 & 87 & 29.52 & -4.36 & [ feps],[feps],[wri04 ] + hd 17925 & 55 & 29.07 & -4.31 & [ feps],[feps],[bal96 ] + hd 32297 & & & & + hd 61005 & 135 & & -4.26 & [ feps],[hen96 ] + hd 69830 & 1860 & & -4.95 & [ wri04],[wri04 ] + hd 134319 & 186 & & -4.35 & [ wri04],[wri04 ] + hd 216803 & 400 & 28.33 & -4.27 & [ feps],[feps],[hen96 ] + ao men & 12 & 30.16 & -3.75 & [ zuc01],[feps],[gre06 ] + we first investigate a chromospheric nature for the mid - ir hydrogen lines . as shown in figure [ dd_prop_fig ] ( top - left panel ) , all the debris disks where the h i(7 - 6 ) line is detected are younger and with higher x - ray luminosity . the latter is a chromospheric indicator of magnetic activity ( e.g. feigelson et al . 2007 ) , and its decay with age could let us conclude that the h i line in debris disks are chromospherical , and as the magnetic activity decreases ( with age ) they are not detected anymore in older disks . moreover , the x - ray emission of the debris disks with detected h i(7 - 6 ) lines almost reaches the saturation level for chromospheric emission ( l@xmath73/l@xmath74 , e.g. , patten & simon 1996 , figure [ dd_prop_fig ] , top - right panel ) . this saturation level indicates that the chromospheric radiation emitted from the star can not exceed a given fraction of the total stellar flux , and this limit is almost reached in all the observed debris disks , favoring again a chromospheric origin of the h i lines . another test to check whether or not h i lines in debris disks are from chromospheric activity is shown in the bottom - left panel of fig . [ dd_prop_fig ] , where we plot the stellar chromospherical activity indicator @xmath75 ( collected from white et al . this indicator measures the core emission in the ca ii h and k lines , and the larger is the value , the more chromospherically active is the star ( noyes et al . 1984 ) index may itself vary with time along the magnetic cycle of the star . ] . for the debris disks where this value has been measured there is no significant difference in terms of range covered by @xmath75 as a function of h i line luminosity , showing that stars with the same chromospheric activity ( similar @xmath75 value ) may or may not have detected h i line . on the other hand , the object with strongest h i line has the largest @xmath75 value , which would support chromospheric origin in this case . however , the sample is limited to only three objects , and statistical studies are not possible . we also speculate on the possibility that , as for full and transitional disks , the detection of the mid - ir h i lines in the _ spitzer _ spectra of debris disks might trace accretion at very low levels . in fact , fig . [ dd_prop_fig ] , bottom - right panel shows that the h i line is detected in the youngest systems from our sample , for which low level of accretion may be still ongoing . while multiplicity studies are not complete for debris disks , it is intriguing to note that the two debris disks with a confirmed companion close enough to contribute to the observed _ spitzer _ spectrum ( see table [ tab_all_prop ] in appendix a ) , have mid - ir hydrogen line detections . this raises yet another possibility that the h i(7 - 6 ) line observed in debris disks is due to an accreting or chromospherically active companion , instead of the debris disk object itself . however , this remains a speculation at this point given the low number of detections . in summary , even if a chromospheric origin appears the most likely based on the tests carried out so far , accretion can not be ruled out . only higher spectral resolution will be able to pin down the origin of mid - ir hydrogen lines . we have re - reduced archival _ spitzer_/irs spectra of 114 objects in different evolutionary stages , from full to transitional to debris disks , and we have investigated the origin of the h i mid - ir lines . our main findings are as follows . + 1 . after correcting for the disk water contamination , we detect h i(7 - 6 ) line in 26 full disks , 12 transitional and 8 debris disks . the fainter h i(9 - 7 ) line is detected in 6 full , 3 transitional and 2 debris disks . we find h i(7 - 6 ) luminosities similar to the [ ne ii ] 12.81 luminosities . these large luminosities exclude that the h i(7 - 6 ) line traces the same hot disk surface probed via the [ ne ii ] line . case b models can not reproduce the observed h i(9 - 7)/h i(7 - 6 ) ratios except for t<3000k . on the other hand , kf models reproduce the observed ratios for reasonable assumption of temperatures and hydrogen column densities ( n@xmath51 ) . using the kf models the n@xmath51 is fairly well constrained between @xmath76@xmath3 but the temperature is not . + 4 . for the subset of full and transitional disks we report a positive correlation between the h i(7 - 6 ) line luminosity and the accretion luminosity . this trend , together with the results shown in point 3 suggest an origin of these lines from gas accreting onto the star . + 5 . a pure chromospheric origin of the mid - ir h i lines , can only be tested for sources with t@xmath774100 k ( 40% of the total sample ) . for this subset of objects we find that the majority of them are not contaminated by chromospheric emission at a significant level . we report the first detection of h i lines in _ spitzer _ spectra of debris disks , which occurs only the in the youngest systems studied here ( < 20 myr ) . a chromospheric origin of these lines seems likely , but it is not possible to definitively rule out either a long - lasting accretion signature nature , or the presence of a non - resolved accreting companion . + ground - based high - resolution spectroscopy will enable testing this origin . extending accretion indicators at mid - ir wavelengths may enable measuring mass accretion rates in less evolved , still embedded class i objects ( where the star is still surrounded by a remnant infall envelope and accretion disk ) . still , care should be taken in converting the mid - ir h i line luminosities into mass accretion rate , because the envelope surrounding class i sources could absorb part of the emission . more interestingly and on a speculative base , if the mid - ir h i lines were confirmed as accretion indicators in debris disks , these lines may be used to extend mass accretion rate measurements to very low values . if the detected h i lines are indeed tracing accretion , the luminosities we measure for the 8 debris disks correspond to mass accretion rates @xmath78/yr . optical and uv accretion diagnostics are not sensitive to such low mass accretion rates for sun - like stars ( e.g. , manara et al . 2013 ) . finally , extending accretion indicators to the mid - ir will enable simultaneous observations of the [ ne ii ] line ( the only confirmed photoevaporative wind indicator ) and the h i line ( an accretion indicator ) , making possible a direct comparison of the two most efficient disk dispersal mechanisms , photoevaporation and viscous accretion ( alexander et al . 2014 ) . + the authors thank catherine espaillat and lynne hillenbrand for providing the optical spectra and physical parameters of a few objects , and john kwan for providing the mid - ir hydrogen line ratios . is supported by the nasa s astrophysics data analysis program research grant to i.p . ( i d : nnx11ag60 g ) . alcal , j.m . , natta , a. , manara , c. , et al . 2014 , a&a , 561 , 2a alexander , r. , pascucci , i. , andrews , s. et al . 2014 protostars and planets vi , university of arizona press ( 2014 ) , eds . h. beuther , r. klessen , c. dullemond , th . henning ( arxiv1311.1819 ) appenzeller , i. , krautter , j. & jankovics , i. 1983 , a&as , 53 , 291a baker , j. g. & menzel , d. h. 1938 , apj , 88 , 52b baldovin - saavedra , c. , audard , m. , gdel , m. et al . 2011 , a&a , 528a , 22b baldovin - saavedra , c. , audard , m. , carmona , a. et al . 2012 , a&a , 543a , 30b baliunas , s. , sokoloff , d. , soon , w. 1996 , apjl , 457 , 99 banzatti et al . 2013 , phd thesis : `` water vapor in protoplanetary disks '' bary , j. s. , matt , s. p. , skrutskie , m. f. et al . 2008 , apj , 687 , 376b basri , g. & batalha , c. 1990 , apj , 363 , 654b biller , b. , close , l. , masciadri , e. , et al . 2007 , apjs , 173 , 143b brown , j.m . , blake , g.a . , qi , c. et al . 2009 , apj , 704 , 496 calvet , n. & hartmann , l. 1992 , apj , 386 , 239c calvet , n. , dalessio , p. , hartmann , l. , et al . 2002 , apj , 568 , 1008 calvet , n. , muzerolle , j. , briceo , c. et al . 2004 , aj , 128 , 1294c carpenter , j. m. & stauffer , j. 2003 , feps memo no.4 http://feps.as.arizona.edu/index.html carr , j. & najita , j. 2008 , science , 319 , 1504c carr , j. & najita , j. 2011 , apj , 733 , 102 casali , m. m. & eiroa , c. 1996 , a&a , 306 , 427c chen , c.h . , mittal , t. , kuchner , m. et al . 2014 , apjs , 211 , 25 cieza , l. , schreiber , m. , romero , g. , et al . 2010 , apj , 712 , 925 cohen , m. & kuhi , l. v. 1979 , apjs , 41 , 743c costigan , g. , scholz , a. , stelzer , b. et al . 2012 , mnras , 427 , 1344c dent , w. r. f. , thi , w. f. , kamp , i. , et al . 2013 , pasp , 125 , 477d duchene , g. , monin , j. , bouvier , j. & menard , f. 1999 , a&a , 351 , 954d dutrey , a. , semenov , d. , chapillon , e. , et al . 2014 , protostars and planets vi , university of arizona press ( 2014 ) , eds . beuther , r. klessen , c. dullemond , th . henning ( arxiv1402.3503 ) edwards , s. , kwan , j. , fischer , w. et al . 2013 , apj , 778 , 148e ercolano , b. , drake , j. j. , raymond , j. c. & clarke , c. c. 2008 , apj , 688 , 398e ercolano , b. & owen , j. 2010 , mnras , 406 , 1553e espaillat , c. , ingleby , l. , hernndez , j. , et al . 2012 , apj , 747 , 103 espaillat , c. , ingleby , l. , furlan , e. , et al . 2013 , apj , 762 , 62e evans , t.m . , ireland , m.j . , kraus , a.l . , et al . 2012 apj , 744 , 120 fang , m. , van boekel , r. , wang , w. , et al . 2009 , a&a , 504 , 461f feigelson , e. ; townsley , l. ; gdel , m. & stassun , k. 2007 , prpl.conf , 313f fernandez , m. , ortiz , e. , eiroa , c. & miranda , l. f. 1995 , a&as , 114 , 439f folha , d. f. m. & emerson , j. p. 2001 , a&a , 365 , 90f furlan , e. , hartmann , l. , calvet , n. et al . 2006 , apjs , 165 , 568f furlan , e. , luhman , k. , espaillat , c. , et al . 2011 , apjs 195 , 3 giampapa , m. s. , calvet , n. , imhoff , c. l. & kuhi , l. v. 1981 , apj , 251 , 113 g gray , r. o. , corbally , c. j. , garrison , r. f. et al . 2006 , aj , 132 , 161 g gdel , m. , lahuis , f. , briggs , k. r. et al . 2010 , a&a , 519a , 113 g guenther , e. w. & emerson , j. p. 1997 , a&a , 321 , 803 g gullbring , e. , hartmann , l. , briceo , c. & calvet , n. 1998 , apj , 492 , 323 g johns - krull , c. 2007 , apj , 664 , 975j hartigan , p. , edwards , s. & ghandour , l. 1995 , apj , 452 , 736h hartmann , l. , avrett , e. h. , loeser , r. & calvet , n. 1990 , apj , 349 , 168h henry , t.j . , soderblom , d.r . , donahue r.a . , baliunas s.l . 1996 , aj , 111 , 439 herbig , g. h. & bell , k. 1988 , third catalog of emission - line stars of the orion population , lick observatory bulletin 1111 , herbst , w. 2008 , handbook of star forming regions vol ii , astronomical society of the pacific . herczeg , g. , najita , j. , hillenbrand , l. & pascucci , i. 2007 , apj , 670 , 509h hillendrand , l. , carpenter , j. , kim , s.j . , et al . 2008 , apj , 677 , 630 higdon , s. j. u. , devost , d. , higdon , j. l. , et al . 2004 , pasp , 116 , 975 hollenbach , d. , & gorti , u. 2009 , apj , 703 , 1203 houck , j. , roellig , t. l. , van cleve , j. , et al . 2004 , apjs , 154 , 18h hughes , j. & hartigan , p. 1992 , aj , 104 , 680h hummer , d. g. & storey , p. j. 1987 , mnras , 224 , 801h keane , j. t. , pascucci , i. , espaillat , c. , et al . 2014 , apj , 787 , 153k kelly , b. c. 2007 , apj , 665 , 1489k kenyon , s. j. & hartmann , lee 1995 , apjs , 101 , 117k kenyon , s. j. , brown , d. , tout , c. & berlind , p. 1998 , aj , 115 , 2491k kraus , a. , ireland , m. , martinache , f. & hillenbrand , l. 2011 , apj , 731 , 8k kraus , a. & ireland , m. 2012 , apj , 745 , 5k kwan , j. & fischer , w. 2011 , mnras , 411 , 2383k ingleby , l. , calvet , n. , herczeg , g. , et al . 2013 , apj , 767 , 112i lafrenire , d. , jayawardhana , r. , brandeker , a. , et al . 2008 , apj , 683 , 844l lafrenire , d. , jayawardhana , r. , van kerkwijk , m.h . , 2014 , apj , 785 , 47l leinert , c. , zinnecker , h. , weitzel , n. et al . 1993 , a&a , 278,m 129l levreault , r. 1988 , apj , 330 , 897l luhman , k. l. , stauffer , j.r . , muench , a. a. et al . 2003 , apj , 593 , 1093l luhman , k.l , stauffer , j.r . , & mamajek , e.e . , 2005 , apj , 628 , l69 lahuis , f. , van dishoeck , e. , blake , g. , et al . 2007 , apj , 665 , 492l mamajek , e. , meyer , m. , hinz , p. et al . 2004 , apj , 612 , 496 m mamajek , e. , meyer , m. & liebert , 2002 , aj , 124 , 1670 manara , c.f . , testi , l. , rigliaco , e. et al . 2013 , a&a , 551 , 107 manara , c.f . , testi , l. , natta , a. , et al . 2014 , a&a , 568a , 18 m manoj , p. , kim , k. h. , furlan , e. et al . 2011 , apjs , 193 , 11 m martin , s. c. 1996 , apj , 470 , 537 m mathis , j. s. 1990 , aspc , 12 , 63 m mccabe , c. , ghez , a. m. , prato , l. , et al . 2006 , apj , 636 , 932 m meijerink , r. , pontoppidan , k. m. , blake , g. a. , et al . 2009 , apj , 704 , 1471 m merin , b. , brown , j. , oliveira , i. et al . 2010 , apj , 718 , 1200 m metchev , s.r a. & hillenbrand , l. 2009 , apjs , 181 , 62 m meyer , m. , hillenbrand , l. , backman , d. , et al . 2006 , pasp , 118 , 1690 m muzerolle , j. , hartmann , l. & calvet , n. 1998 , aj , 116 , 455 m muzerolle , j. , calvet , n. & hartmann , l. 2001 , apj , 550 , 944 m muzerolle , j. , allen , l. , megeath , s. , et al . 2010 , apj , 708 , 1107 m najita , j. , carr , j. , strom , s. , et al . 2010 , apj , 712 , 274 natta , a. , testi , l. & randich , s. 2006 , a&a , 452 , 245n nguyen , d.c . , brandeker , a. , van kerkwijk , m.h . & jayawardhana , r. 2012 , apj , 745 , 119 noyes , r. w. , hartmann , l. w. , baliunas , s. l. , et al . 1984 , apj , 279 , 763n pascucci , i. , gorti , u. , hollenbach , d. , et al . 2006 , apj , 651 , 1177p pascucci , i. , hollenbach , d. , najita , j. , et al . 2007 , apj , 663 , 383 pascucci , i. & sterzik , m. 2009 , apj , 702 , 724p pascucci , i. , herczeg , g. , carr , j. s. & bruderer , s. 2013 , apj , 779 , 178p patten , b. & simon , t. 1996 , apjs , 106 , 489p petrov , p. p. , kurosawa , r. , romanova , m. m. , et al . 2014 , mnras , 442 , 3643p pinte , c. , padgett , d. , mnard , f. , et al . 2008 , a&a , 489 , 633 pontoppidan , k. , & dullemond , k. 2005 , a&a , 435 , 595 pontoppidan , k. , salyk , c. , blake , g. & k'aufl , h. 2010 , apjl , 722 , 173 pontoppidan , k. , salyk , c. , bergin , e. , et al . 2014 , protostars and planets vi , university of arizona press ( 2014 ) , eds . beuther , r. klessen , c. dullemond , th . ( arxiv1401.2423 ) prato , l. , greene , t. p. & simon , m. 2003 , apj , 584 , 853p preibisch , t. , brown , a. , bridges , t. , et al . , 2002 , aj , 124 , 404 ratzka , t. , khler , r. & leinert , ch . 2005 , a&a , 437 , 611 ratzka , t. , leinert , c. , henning , t. , et al . 2007 , a&a , 471 , 173r rebull , l. m. , padgett , d. l. , mccabe , c. et al . 2010 , apjs , 186 , 259r reipurth , b. & zinnecker , h. 1993 , a&a , 278 , 81r reipurth , b. , pedrosa , a. & lago , m. 1996 , a&as , 120 , 229r rigliaco , e. , natta , a. , randich , s. & sacco , g. 2009 , a&a , 495l , 13r rigliaco , e. , natta , a. , testi , l. , et al . 2012 , a&a , 548a , 56r riviere - marichalar , p. , mnard , f. , thi , w. f. , et al . 2012 , a&a , 538l , 3r rothman , l. s. , gordon , i. e. , barbe , a. , et al . 2009 , jqsrt , 110 , 533r sacco , g. g. , flaccomio , e. , pascucci , i. et al . 2012 , apj , 747 , 142s salyk , c. , pontoppidan , k. , geoffrey a. et al . 2008 , apj , 676l , 49s salyk , c. , pontoppidan , k. m. , blake , g. a. et al . 2011 , apj , 731 , 130s salyk , c. , herczeg , g. , brown , j. , et al . 2013 , apj , 769 , 21s siess , l. , dufour , e. & forestini , m. 2000 , a&a , 358 , 593s silverstone , m. d. , meyer , m. r. , mamajek , e. e. , et al . 2006 , apj , 639 , 1138s storey , p. j. & hummer , d. g. 1995 , mnras , 272 , 41s strom , k. m. , strom , s. e. & margulis , m. 1989 , esoc , 33 , 423s testi , l. , birnstiel , t. , ricci , l. , et al . 2014 , protostars and planets vi , university of arizona press ( 2014 ) , eds . beuther , r. klessen , c. dullemond , th . ( arxiv1402.1354 ) torres , c. a. o. , quast , g. r. , melo , c. h. f. , & sterzik , m. f. 2008 , handbook of star forming regions , volume ii : the southern sky asp monograph publications , vol . 5 . edited by bo reipurth , p.757 uchida , k. i. , calvet , n. , hartmann , l. , et al . , 2004 , apjs , 154 , 439 van boekel , r. , gdel , m. henning , t. , et al . 2009 , a&a , 497 , 137v walter , f. m. 1986 , apj , 306 , 573w walter , f. m. 1992 , aj , 104 , 758 white , r. & ghez , a. 2001 , apj , 556 , 265w white , r. j. & hillenbrand , l. 2005 , apj , 621l , 65w white , r. j. ; gabor , j. m. & hillenbrand , l. 2007 , aj , 133 , 2524w wilking , b. , meyer , m. , robinson , j. & greene , t. 2005 , aj , 130 , 1733w woitas , j. , leinert , ch . & khler , r. 2001 , a&a , 376 , 982w wright , j.t . , marcy , g.w . , butler , r.p . & vogt , s.s . 2004 , apjs , 152 , 261 zuckerman , b. , song , & webb 2001 , apj , 559 , 388 zuckerman , b. & song , i. 2004 , ara&a , 42 , 685z in figure [ all_hi_profile ] we show the h i(7 - 6 ) line for all the full and transitional disks where the line is detected . the h i(7 - 6 ) line profile for debris disks are shown in figure [ spitz_dd ] . in table [ tab_all_prop ] we report the source properties , and table [ flux_obj ] report the mid - ir line fluxes , errors and continuum levels around the three line of interest . l c c c c c c c c c c c c c c + ic348 - 2 & ic 348 & 315 & a2 & 3.8 & 8970 & 2.80 & 57.1 & 3.1 & & & [ her08],[esp12 ] + ic348 - 6 & ic 348 & 315 & g3 & 3.9 & 5830 & 2.40 & 16.6 & 4.0 & & & [ her08],[esp12 ] + ic348 - 31 & ic 348 & 315 & g6 & 8.6 & 5700 & 1.60 & 5.00 & 2.3 & & & [ her08],[esp12 ] + ic348 - 37 & ic 348 & 315 & k6 & 2.8 & 4200 & 0.90 & 1.30 & 2.2 & & & [ her08],[esp12 ] + ic348 - 55 & ic 348 & 315 & m0.5 & 8.5 & 3850 & 0.60 & 1.00 & 2.2 & & & [ her08],[esp12 ] + ic348 - 68 & ic 348 & 315 & m3.5 & 2.1 & 3740 & 0.30 & 0.50 & 2.0 & & & [ her08],[esp12 ] + 04216 + 2603 & taurus & 140 & m0 & & 3500 & 0.28 & 0.02 & 0.3 & & & [ fur06],[ken98 ] + 04385 + 2550 & taurus & 140 & m0 & 9.3 & 3850 & 0.57 & 3.67 & 4.0 & 20.0 & 0.50 & [ fur06],[ken98 ] + aa tau@xmath79 & taurus & 140 & k7 & 0.9 & 4060 & 0.73 & 1.38 & 2.3 & 37.1 & 0.07 & [ fur06],[coh79 ] + bp tau & taurus & 140 & k7 & 0.5 & 4060 & 0.73 & 1.62 & 2.4 & 40.1 & 0.07 & [ fur06],[coh79 ] + cw tau@xmath79 & taurus & 140 & k3 & 2.2 & 4730 & 1.15 & 0.76 & 1.3 & 134.6 & 0.30 & [ fur06],[whi01 ] + cx tau & taurus & 140 & m2.5 & 0.8 & 3490 & 0.36 & 0.62 & 2.0 & 19.8 & 0.003 & [ fur06],[coh79 ] + cy tau@xmath79 & taurus & 140 & m1 & 0.9 & 3705 & 0.47 & 0.71 & 1.9 & 69.5 & 0.06 & [ fur06],[coh79 ] + dd tau @xmath80 & taurus & 140 & m3 & 0.4 & 3415 & 0.29 & 0.13 & 1.0 & 181.8 & 0.05 & [ fur06],[whi01 ] , [ coh79 ] + de tau & taurus & 140 & m1 & 1.6 & 3705 & 0.46 & 1.47 & 2.8 & 54.0 & 0.23 & [ fur06],[whi01 ] , [ coh79 ] + dg tau & taurus & 140 & k6 & 1.4 & 4205 & 0.92 & 1.15 & 1.9 & 84.0 & 0.17 & [ fur06],[whi01 ] + dk tau@xmath80 & taurus & 140 & k7 & 1.4 & 4060 & 0.74 & 1.32 & 2.2 & 19.4 & 0.02 & [ fur06],[coh79],[str89 ] + dh tau@xmath79 & taurus & 140 & m0 & 1.0 & 3850 & 0.56 & 0.76 & 1.8 & 53.4 & 0.03 & [ fur06],[coh79 ] + dl tau@xmath79 & taurus & 140 & k7 & 2.0 & 4060 & 0.75 & 1.16 & 2.2 & 105.0 & 0.62 & [ fur06],[whi01 ] + dn tau@xmath79 & taurus & 140 & m0 & 0.4 & 3850 & 0.55 & 1.62 & 2.7 & 11.9 & 0.016 & [ fur06],[whi01 ] + do tau@xmath79 & taurus & 140 & m0 & 2.2 & 3850 & 0.56 & 1.05 & 2.2 & 100.0 & 0.19 & [ fur06],[whi01 ] + dp tau@xmath80 & taurus & 140 & m0.5 & 0.9 & 3850 & 0.56 & 0.85 & 1.9 & 85.4 & 0.05 & [ fur06],[coh79 ] + dq tau@xmath81 & taurus & 140 & m0 & 0.5 & 3850 & 0.56 & 0.93 & 2.0 & 112.9 & 0.04 & [ fur06],[coh79 ] + dr tau@xmath79 & taurus & 140 & k5 & 0.5 & 4350 & 1.16 & 0.96 & 1.9 & 106.0 & 0.49 & [ fur06],[whi01 ] + ds tau@xmath79 & taurus & 140 & k5 & 0.9 & 4350 & 1.10 & 0.68 & 1.3 & 58.7 & 0.16 & [ fur06],[whi01 ] + fm tau@xmath79 & taurus & 140 & m0 & 0.9 & 3850 & 0.57 & 0.52 & 1.5 & 70.9 & 0.02 & [ fur06],[coh79 ] + fn tau & taurus & 140 & m5 & 1.4 & 3125 & 0.22 & 0.79 & 2.8 & 24.7 & 0.007 & [ fur06],[coh79 ] + ft tau@xmath79 & taurus & 140 & m3 & 3.8 & 3415 & 0.31 & 0.16 & 1.0 & 254.3 & 0.49 & [ fur06],[coh79],[reb10 ] + fz tau@xmath79 & taurus & 140 & m0 & 2.7 & 3850 & 0.57 & 0.50 & 1.4 & 200.0 & 0.26 & [ fur06],[whi01 ] + hk tau@xmath82 & taurus & 140 & m0.5 & 3.0 & 3850 & 0.56 & 1.12 & 2.2 & 29.3 & 0.028 & [ fur06],[coh79 ] + hn tau@xmath80 & taurus & 140 & k5 & 1.2 & 4350 & 0.78 & 0.22 & 0.8 & 163.0 & 0.15 & [ fur06],[whi01 ] + ip tau & taurus & 140 & m0 & 0.5 & 3850 & 0.59 & 0.17 & 0.9 & 11.0 & 0.005 & [ fur06],[whi01 ] + rw aur@xmath80 & taurus & 140 & k3 & 0.4 & 4730 & 1.48 & 1.60 & 1.8 & 76.0 & 0.924 & [ fur06],[whi01 ] + su aur & taurus & 140 & g2 & 0.9 & 5860 & 1.85 & 8.90 & 2.7 & 3.5 & 0.14 & [ fur06],[coh79 ] + uy aur@xmath83 & taurus & 140 & k7 & 1.1 & 3850 & 0.60 & 0.91 & 2.1 & 72.8 & 1.02 & [ fur06],[coh79],[whi01 ] + as 353a@xmath82 & perseus & 150 & m3 & 2.1 & 3415 & 0.32 & 0.68 & 2.2 & 124.5 & 1.60 & [ pra03],[coh79 ] + lkha326 & perseus & 250 & m0 & 3.1 & 3800 & 0.49 & 0.05 & 0.5 & 52.7 & 0.10 & [ cas96],[coh79 ] + 2071-h13/ ngc2068 & ngc2068 & 400 & k1 & 2.9 & 5080 & 1.68 & 2.63 & 2.0 & 36.7 & 0.42 & [ esp12 ] + 2071-s1/ ngc2068 & ngc2068 & 400 & k4 & 4.1 & 4600 & 1.60 & 4.10 & 3.1 & 3.7 & 0.005 & [ esp12 ] + rx j1111.7 - 7620@xmath82 & cha & 160 & k1 & 1.5 & 5080 & 1.78 & 3.09 & 2.2 & 7.6 & 0.04 & [ sil06],[wal92 ] + sx cha@xmath80 & cha & 160 & m0.5 & 1.1 & 3850 & 0.57 & 0.61 & 1.6 & 33.2 & 0.02 & [ fur11],[gue97 ] + sz50 & cha & 160 & m3 & 2.1 & 3350 & 0.90 & 0.78 & 2.5 & 46.0 & 0.015 & [ fur11],[hug92 ] + twcha & cha & 160 & m0 & 1.2 & 3850 & 0.56 & 3.10 & 3.7 & 28.3 & 0.04 & [ fur11],[gue97 ] + vw cha@xmath80 & cha & 160 & k5 & 2.4 & 4350 & 1.14 & 4.02 & 3.3 & 71.7 & 0.81 & [ fur11],[rei96 ] + vz cha@xmath79 & cha & 160 & k6 & 0.5 & 4205 & 0.93 & 0.84 & 1.6 & 71.4 & 0.099 & [ fur11],[rei96 ] + wx cha@xmath80 & cha & 160 & k7 & 2.1 & 4060 & 0.77 & 0.77 & 1.8 & 65.5 & 0.07 & [ fur11],[app83 ] + xx cha@xmath80 & cha & 160 & m2 & 0.6 & 3560 & 0.37 & 0.18 & 1.1 & 133.5 & 0.03 & [ fur11],[app83 ] + hen-3 - 600a & twhya & 56 & m3 & 0.0 & 3415 & 0.26 & 0.06 & 0.6 & 21.8 & 0.007 & [ uch04 ] + twa10 & twhya & 100 & m2 & 0.0 & 5500 & 0.90 & 0.60 & 1.0 & & & [ zuc04 ] + 1rxs j132207.2 - 693812 & lcc & 86 & k1 & 0.6 & 5035 & 1.23 & 1.14 & 1.4 & 19.5 & 0.07 & [ sil06],[whi07 ] + ex lup@xmath79 & lupus & 190 & m0 & 0.0 & 3850 & 0.55 & 1.31 & 2.4 & 31.3 & 0.06 & [ lah07],[rei96 ] + gq lup@xmath79 & lupus & 190 & k7 & 0.4 & 4060 & 0.76 & 4.53 & 4.1 & 31.7 & 0.31 & [ lah07],[rei96 ] + ht lup@xmath82 & lupus & 190 & k3 & 0.0 & 4730 & 1.85 & 4.00 & 2.8 & 7.3 & 0.11 & [ lah07],[rei96],[woi01 ] + ru lup@xmath79 & lupus & 190 & k7 & 0.7 & 4060 & 0.74 & 1.28 & 2.2 & 136.0 & 0.69 & [ lah07],[rei96 ] + 1rxs j161411.0 - 230536@xmath82 & upper scorpius & 140 & k0 & 1.1 & 5250 & 1.62 & 2.95 & 2.0 & & & [ sil06],[mam04 ] + hd 143006 & upper scorpius & 169 & g7 & 0.6 & 5800 & 1.12 & 3.30 & 1.7 & 10.7 & 0.19 & [ sil06],[mam04 ] + doar 24e@xmath80 & @xmath84ophiucus & 160 & g7 & 6.1 & 5630 & 2.15 & 10.75 & 3.3 & 7.1 & 0.24 & [ fur11],[wil05 ] + f doar 25@xmath79 & @xmath84ophiucus & 160 & k5 & 2.9 & 4350 & 1.11 & 2.19 & 2.3 & 12.0 & 0.05 & [ cie10],[wil05 ] + sr21@xmath83 & @xmath84ophiucus & 125 & g2.5 & 9.0 & 5950 & 2.67 & 28.0 & 4.7 & & & [ pra03 ] + waoph6@xmath79 & @xmath84ophiucus & 160 & k6 & 1.0 & 4205 & 0.90 & 1.46 & 2.3 & 35.0 & 0.02 & [ lah07],[wal86 ] + ced110-irs4 & @xmath84ophiucus & 160 & k2 & & 4800 & 0.85 & 0.40 & 0.9 & & & [ pon05 ] + ced110-irs2 & @xmath84ophiucus & 160 & k3 & 3.3 & 4730 & 2.23 & 9.64 & 4.4 & & & [ fur11 ] + haro 1 - 16@xmath79 & @xmath84ophiucus & 160 & k3 & 1.2 & 4730 & 1.80 & 3.39 & 2.7 & 59.0 & 0.24 & [ fur06],[coh79 ] + rno 90@xmath79 & @xmath84ophiucus & 160 & g5 & 1.0 & 5770 & 1.28 & 2.74 & 1.6 & 76.0 & 0.47 & [ lah07],[lev88 ] + ec 82 & serpens & 415 & k7 & 1.0 & 4060 & 0.75 & 3.21 & 3.3 & 8.0 & 0.03 & [ lah07],[coh79 ] + rx j1842.9 - 3532 & cra & 136 & k2 & 1.1 & 4900 & 1.33 & 1.29 & 1.5 & 27.2 & 0.06 & [ sil06],[whi07 ] + rx j1852.3 - 3700 & cra & 136 & k3 & 1.0 & 4730 & 1.27 & 1.00 & 1.5 & 27.4 & 0.06 & [ sil06],[whi07 ] + v1331cyg & cyn & 550 & g8 & 0.0 & 5500 & 2.81 & 21.0 & 4.8 & 48.4 & 2.01 & [ coh79],[pet14 ] + + ic348 - 21 & ic 348 & 315 & k0 & 4.7 & 5250 & 1.60 & 3.80 & 2.4 & 6.4 & 0.08 & [ esp12 ] + ic348 - 67 & ic 348 & 315 & m0.75 & 2.0 & 3720 & 0.50 & 0.50 & 1.8 & 32.5 & 0.03 & [ esp12 ] + ic348 - 72 & ic 348 & 315 & m2.5 & 3.0 & 3580 & 0.40 & 0.70 & 2.1 & 2.0 & 0.0007 & [ esp12 ] + ic348 - 133 & ic 348 & 315 & m5 & 3.6 & 3200 & 0.20 & 0.20 & 1.5 & & & [ esp12 ] + dm tau & taurus & 140 & m1 & 0.0 & 3705 & 0.47 & 0.37 & 1.3 & 138.7 & 0.046 & [ fur06],[ken98 ] + fp tau & taurus & 140 & m5 & 0.3 & 3125 & 0.22 & 0.62 & 2.5 & 38.5 & 0.001 & [ fur06],[ken98 ] + gm aur & taurus & 140 & k3 & 1.2 & 4730 & 1.27 & 1.00 & 1.5 & 96.5 & 0.58 & [ fur06],[ken98 ] + lkca 15@xmath83 & taurus & 140 & k5 & 1.0 & 4300 & 1.06 & 0.74 & 1.5 & 13.0 & 0.027 & [ fur06],[ken98 ] + st 34@xmath82 & taurus & 140 & m3 & 0.2 & 3415 & 0.24 & 0.01 & 0.3 & 51.6 & 0.016 & [ lah07],[whi05 ] + ux tau a@xmath82 & taurus & 140 & k5 & 0.3 & 4350 & 1.17 & 0.91 & 1.9 & 9.5 & 0.03 & [ fur06],[her88 ] + v836tau & taurus & 140 & k7 & 0.9 & 4060 & 0.81 & 0.34 & 1.2 & 55.0 & 0.16 & [ fur06],[ken98],[ngu09 ] + lkha 330 & perseus & 250 & g3 & 1.8 & 5830 & 1.25 & 2.78 & 1.5 & 20.3 & 0.59 & [ bro09],[fer95 ] + 2071-h9/ ngc2068 & ngc2068 & 400 & k2 & 1.6 & 5250 & 1.41 & 2.03 & 1.7 & 1.2 & 0.004 & [ esp12 ] + chx 22 ( t54)@xmath83 & cha & 160 & g8 & 1.2 & 5520 & 1.69 & 4.63 & 2.2 & & & [ fur11],[wal92 ] , [ ngu09 ] + cs cha & cha & 160 & k6 & 0.8 & 4205 & 0.90 & 1.50 & 2.3 & 54.3 & 0.39 & [ fur11 ] , [ rei96 ] + sz cha@xmath82 & cha & 160 & k0 & 1.9 & 5250 & 1.40 & 1.90 & 1.7 & & & [ fur11 ] , [ rei96 ] + sz 18 ( t25 ) & cha & 160 & m2.5 & 1.6 & 3560 & 0.38 & 0.32 & 1.4 & 8.0 & 0.003 & [ fur11],[man11 ] + sz 27 ( t35 ) & cha & 160 & k8 & 3.5 & 4060 & 0.80 & 0.47 & 1.2 & 100.0 & 0.08 & [ fur11],[man11 ] + tw hya & twhya & 56 & m0 & 1.0 & 3850 & 0.57 & 0.64 & 1.7 & 194.0 & 0.09 & [ cal02],[rei96 ] + i m lup & lupus & 190 & m0 & 0.7 & 3850 & 0.56 & 3.52 & 3.9 & 4.7 & 0.02 & [ pin08],[rei96 ] + doar 21 & @xmath84ophiucus & 160 & k1 & 6.6 & 5080 & 2.20 & 27.3 & 6.7 & 1.5 & 0.12 & [ muz10],[mer10 ] + + 1rxs j121236.4 - 552037 & lcc & 108 & k0 & 0.2 & 5150 & 0.90 & 0.50 & 0.9 & & & [ car03 ] , [ che14 ] + 1rxs j122233.4 - 533347 & lcc & 124 & g0 & 0.3 & 5860 & 1.24 & 2.23 & 1.3 & & & [ car03 ] , [ che14 ] + 1rxs j130153.7 - 530446 & lcc & 108 & k2 & 0.2 & 4700 & 0.82 & 0.30 & 0.8 & & & [ car03 ] , [ che14 ] + hd119269 & lcc & 85 & g4 & 0.1 & 5700 & 1.07 & 1.31 & 1.2 & & & [ car03 ] , [ che14 ] + 1rxs j133758.0 - 413448 & ucl & 98 & k0 & 0.3 & 5250 & 0.90 & 0.68 & 0.9 & & & [ car03 ] , [ che14 ] + 1rxs j161458.4 - 27501 & upper scorpius & 114 & g5 & 0.6 & 5250 & 0.90 & 0.70 & 1.0 & & & [ sil06],[whi07 ] + hd 142361@xmath82 & upper scorpius & 101 & g3 & 0.5 & 5830 & 1.72 & 7.12 & 2.5 & & & [ car03 ] , [ che14 ] + rx j1600.6 - 2159 & upper scorpius & 160 & g9 & 0.8 & 5410 & 1.66 & 3.77 & 2.1 & & & [ sil06],[whi07 ] + rx j1612.6 - 1859@xmath82 & upper scorpius & 127 & k0 & 1.5 & 5250 & 1.50 & 2.40 & 1.8 & & & [ hil08],[whi07 ] + hd 35850 & @xmath7pictoris & 26.5 & f7 & 0.0 & 6000 & 1.20 & 1.78 & 1.2 & & & [ car03 ] , [ che14 ] + v343 nor@xmath82 & @xmath7pictoris & 38 & k0 & 0.0 & 5250 & 0.90 & 0.67 & 0.9 & & & [ car03 ] + hd19668 & ab dor & 40 & g8/k0 & 0.0 & 5500 & 0.90 & 0.60 & 0.8 & & & [ car03 ] , [ che14 ] + hd25457 & ab dor & 19 & f7 & 0.2 & 6000 & 1.20 & 2.50 & 1.5 & & & [ car03 ] , [ che14 ] + hd37484 & columba & 60 & f3 & 0.1 & 6740 & 1.30 & 3.90 & 1.5 & & & [ car03 ] , [ che14 ] + hd12039@xmath83 & tuc - hor & 42 & g4 & 0.2 & 6000 & 1.00 & 1.00 & 0.9 & & & [ car03 ] , [ che14 ] + hd202917 & tuc - hor & 46 & g7 & 0.1 & 5750 & 0.90 & 0.70 & 0.8 & & & [ car03 ] , [ che14 ] + hd 141943 & field & 67 & g0 & 0.1 & 6030 & 1.55 & 6.19 & 2.2 & & & [ car03 ] + hd377 & field & 39.6 & g2 & 0.2 & 6250 & 1.00 & 1.50 & 1.0 & & & [ car03 ] , [ che14 ] + hd17925 & field & 10 & k1 & 0.0 & 5080 & 0.92 & 0.55 & 0.9 & & & [ car03 ] , [ che14 ] + hd32297 & field & 112 & a0 & 0.2 & 8000 & 1.90 & 6.20 & 1.3 & & & [ car03 ] , [ che14 ] + hd61005 & field & 34 & g8 & 0.1 & 5759 & 0.90 & 0.60 & 0.8 & & & [ car03 ] , [ che14 ] + hd69830 & field & 13 & g8 & 0.0 & 5500 & 0.90 & 0.60 & 0.85 & & & [ car03 ] , [ che14 ] + hd134319 & field & 44 & g5 & 0.0 & 5770 & 1.00 & 0.72 & 0.8 & & & [ car03 ] , [ che14 ] + hd216803 & field & 8 & k4 & 0.0 & 4600 & 0.66 & 0.19 & 0.7 & & & [ car03 ] , [ che14 ] + aomen & field & 39 & k3 & 0.0 & 4435 & 0.70 & 0.26 & 0.9 & & & [ car03 ] + l \| c c c \| c c c \| c c c + + ic348 - 2 & 7.66e-15 & 1.81e-15 & 0.24 & 4.05e-15 & 7.45e-16 & 0.25 & @xmath676.46e-15 & & 0.33 + ic348 - 6 & @xmath672.23e-15 & & 0.078 & @xmath672.73e-15 & & 0.077 & @xmath671.18e-14 & & 0.11 + ic348 - 31 & @xmath672.68e-15 & & 0.076 & @xmath673.43e-15 & & 0.067 & @xmath673.62e-15 & & 0.067 + ic348 - 37 & @xmath672.46e-15 & & 0.055 & @xmath672.22e-15 & & 0.059 & @xmath672.92e-15 & & 0.081 + ic348 - 55 & 2.16e-15 & 7.28e-16 & 0.034 & @xmath671.94e-15 & & 0.037 & @xmath673.31e-15 & & 0.052 + ic348 - 68 & @xmath671.41e-15 & & 0.011 & 2.68e-15 & 4.97e-16 & 0.011 & @xmath673.51e-15 & & 0.012 + 04216 + 2603 & @xmath673.41e-15 & & 0.20 & @xmath674.77e-15 & & 0.20 & @xmath677.17e-15 & & 0.23 + 04385 + 2550@xmath79 & 4.33e-15 & 1.66e-15 & 0.35 & @xmath677.61e-15 & & 0.34 & @xmath671.08e-14 & & 0.38 + aa tau@xmath79 & 1.32e-14 & 2.14e-15 & 0.31 & @xmath677.61e-15 & & 0.32 & @xmath671.63e-14 & & 0.41 + bp tau@xmath79 & @xmath675.61e-15 & & 0.36 & @xmath676.70e-15 & & 0.38 & @xmath671.80e-14 & & 0.51 + cw tau@xmath79 & @xmath671.71e-14 & & 1.13 & 3.45e-14 ( 5.73e-14 ) & 4.20e-15 & 1.18 & @xmath673.73e-14 & & 1.33 + cx tau & @xmath672.38e-15 & & 0.17 & @xmath673.01e-15 & & 0.17 & @xmath675.55e-15 & & 0.20 + cy tau@xmath79 & 2.22e-15 & 9.68e-16 & 0.15 & @xmath674.93e-15 & & 0.16 & @xmath673.53e-15 & & 0.18 + dd tau@xmath79 & 2.02e-14 & 4.73e-15 & 0.93 & @xmath671.37e-14 & & 0.96 & @xmath675.02e-14 & & 1.14 + de tau & @xmath675.75e-15 & & 0.35 & 7.81e-15 & 1.40e-15 & 0.35 & @xmath671.34e-14 & & 0.49 + dg tau & 2.43e-13 & 1.13e-14 & 4.64 & 4.49e-14 & 6.49e-15 & 4.47 & @xmath675.44e-14 & & 4.52 + dk tau@xmath79 & @xmath679.78e-15 & & 0.78 & @xmath672.46e-14 & & 0.88 & @xmath671.54e-14 & & 1.24 + dh tau@xmath79 & @xmath673.99e-15 & & 0.12 & @xmath674.01e-15 & & 0.12 & @xmath671.04e-14 & & 0.16 + dl tau@xmath79 & 1.24e-14 & 2.15e-15 & 0.87 & 7.12e-15 ( 3.24e-14 ) & 2.09e-15 & 0.88 & 1.58e-14 ( 1.84e-14 ) & 2.52e-15 & 0.93 + dn tau@xmath79 & 3.83e-15 & 5.38e-16 & 0.28 & @xmath673.22e-15 & & 0.29 & @xmath672.92e-15 & & 0.31 + do tau@xmath79 & @xmath671.51e-14 & & 1.89 & @xmath671.75e-14 & & 1.94 & @xmath671.76e-14 & & 2.17 + dp tau@xmath79 & 1.16e-13 & 3.63e-15 & 0.63 & @xmath672.42e-14 & & 0.66 & @xmath672.44e-14 & & 0.78 + dq tau@xmath79 & 9.06e-15 & 3.33e-15 & 0.60 & @xmath671.14e-14 & & 0.61 & @xmath671.19e-14 & & 0.64 + dr tau@xmath79 & @xmath672.12e-14 & & 1.88 & 3.46e-14 ( 1.03e-13 ) & 6.93e-15 & 1.88 & 1.69e-14 ( 4.08e-14 ) & 8.65e-15 & 2.10 + ds tau@xmath79 & @xmath676.33e-15 & & 0.24 & 4.23e-15 ( 1.28e-14 ) & 1.10e-15 & 0.25 & @xmath676.32e-15 & & 0.36 + fm tau@xmath79 & @xmath672.26e-15 & & 0.25 & @xmath674.85e-15 & & 0.27 & @xmath676.71e-15 & & 0.34 + fn tau & 5.44e-15 & 1.23e-15 & 0.68 & @xmath675.71e-15 & & 0.67 & @xmath675.59e-15 & & 0.83 + ft tau@xmath79 & 5.97e-15 & 1.45e-15 & 0.25 & @xmath674.54e-15 & & 0.26 & @xmath671.07e-14 & & 0.31 + fz tau@xmath79 & @xmath671.75e-14 & & 0.98 & 2.46e-14 ( 8.65e-14 ) & 3.42e-15 & 1.02 & 2.23e-14 ( 3.17e-14 ) & 7.12e-15 & 1.15 + hk tau & 5.51e-15 & 1.08e-15 & 0.20 & 3.44e-15 & 7.21e-16 & 0.20 & @xmath674.42e-15 & & 0.24 + hn tau@xmath79 & 2.59e-14 & 4.81e-15 & 0.95 & @xmath671.66e-14 & & 1.01 & @xmath674.05e-14 & & 1.36 + ip tau & @xmath676.22e-16 & & 0.013 & @xmath678.16e-16 & & 0.014 & @xmath671.94e-15 & & 0.020 + rw aur@xmath79 & @xmath672.00e-14 & & 1.43 & 1.96e-14 ( 7.16e-14 ) & 5.92e-15 & 1.46 & @xmath673.23e-14 & & 1.76 + su aur & 1.78e-14 & 3.63e-15 & 2.14 & 3.27e-14 & 3.87e-15 & 2.17 & @xmath677.15e-14 & & 3.02 + uyaur & @xmath672.49e-14 & & 2.36 & @xmath673.39e-14 & & 2.38 & @xmath678.22e-14 & & 2.67 + as 353a & 1.11e-14 & 2.50e-15 & 1.11 & 1.96e-14 & 3.07e-15 & 1.13 & @xmath671.80e-14 & & 1.26 + lkha326 & 4.74e-15 & 1.23e-15 & 0.33 & @xmath678.34e-15 & & 0.33 & @xmath671.27e-14 & & 0.36 + 2071-h13/ ngc2068 & 1.01e-15 & 2.43e-16 & 0.024 & @xmath679.83e-16 & & 0.026 & @xmath671.71e-15 & & 0.038 + ngc20682071-s1 & @xmath671.97e-15 & & 0.046 & @xmath671.77e-15 & & 0.050 & @xmath672.23e-15 & & 0.061 + rx j1111.7 - 7620 & 9.97e-15 & 1.28e-15 & 0.22 & @xmath675.96e-15 & & 0.23 & @xmath679.55e-15 & & 0.27 + sx cha@xmath79 & @xmath677.08e-15 & & 0.49 & @xmath671.34e-14 & & 0.51 & @xmath671.42e-14 & & 0.65 + sz50 & @xmath673.20e-15 & & 0.20 & @xmath674.78e-15 & & 0.21 & @xmath677.16e-15 & & 0.22 + twcha & @xmath674.35e-15 & & 0.062 & @xmath674.01e-15 & & 0.055 & @xmath676.63e-15 & & 0.042 + vw cha@xmath79 & 3.13e-14 & 3.50e-15 & 0.73 & 3.74e-14 ( 5.98e-14 ) & 2.79e-15 & 0.76 & 1.91e-14 ( 2.64e-14 ) & 5.18e-15 & 0.95 + vz cha@xmath79 & @xmath675.92e-15 & & 0.37 & @xmath678.84e-15 & & 0.40 & @xmath678.61e-15 & & 0.46 + wx cha@xmath79 & @xmath675.08e-15 & & 0.28 & 1.29e-14 ( 2.62e-14 ) & 1.34e-15 & 0.30 & @xmath671.01e-14 & & 0.43 + xx cha@xmath79 & 4.57e-15 & 7.48e-16 & 0.13 & @xmath675.66e-15 & & 0.14 & @xmath676.46e-15 & & 0.16 + hen-3 - 600a & @xmath676.62e-15 & & 0.099 & @xmath674.78e-15 & & 0.095 & @xmath671.08e-14 & & 0.11 + twa10 & @xmath672.44e-15 & & 0.018 & @xmath672.09e-15 & & 0.019 & @xmath673.05e-15 & & 0.022 + rx 132207.2 - 693812 & @xmath672.59e-14 & & 0.75 & 3.46e-14 & 6.13e-15 & 0.74 & 3.17e-14 & 1.23e-14 & 0.84 + ex lup@xmath79 & @xmath671.63e-14 & & 3.13 & 1.71e-14 ( 5.34e-14 ) & 6.85e-15 & 3.21 & @xmath674.43e-14 & & 4.12 + gq lup@xmath79 & @xmath676.38e-15 & & 0.49 & 4.39e-15 ( 1.23e-14 ) & 2.29e-15 & 0.52 & @xmath679.91e-15 & & 0.65 + htlup & @xmath671.90e-14 & & 2.17 & @xmath671.96e-14 & & 2.21 & @xmath675.23e-14 & & 2.64 + ru lup@xmath79 & 4.54e-14 & 8.41e-15 & 4.29 & 3.61e-14 ( 7.74e-14 ) & 8.24e-15 & 4.36 & @xmath676.13e-14 & & 4.80 + rx j161411.0 - 230536 & 6.44e-15 & 1.36e-15 & 0.34 & @xmath675.69e-15 & & 0.35 & @xmath677.30e-15 & & 0.36 + hd143006 & @xmath671.08e-14 & & 0.69 & @xmath671.63e-14 & & 0.74 & @xmath673.68e-14 & & 1.12 + doar 24e@xmath79 & 1.38e-14 & 4.20e-15 & 2.54 & 1.83e-14 ( 3.96e-14 ) & 7.58e-15 & 2.58 & @xmath674.14e-14 & & 2.96 + doar 25@xmath79 & 2.87e-15 & 9.27e-16 & 0.21 & @xmath677.78e-15 & & 0.22 & @xmath673.84e-15 & & 0.24 + sr21 & @xmath672.84e-14 & & 0.79 & @xmath673.63e-14 & & 0.75 & @xmath676.01e-14 & & 0.91 + wa oph6@xmath79 & 8.55e-15 & 1.83e-15 & 0.82 & @xmath672.25e-14 & & 0.82 & @xmath671.65e-14 & & 0.92 + ced110irs2 & @xmath674.43e-15 & & 0.11 & 1.92e-15 & 6.24e-16 & 0.12 & @xmath676.97e-15 & & 0.15 + ced110-irs4 & 2.43e-14 & 8.57e-16 & 0.080 & @xmath671.84e-15 & & 0.070 & @xmath671.75e-15 & & 0.054 + haro 1 - 16@xmath79 & @xmath679.73e-15 & & 0.50 & 1.05e-14 ( 2.16e-14 ) & 3.42e-15 & 0.54 & @xmath672.43e-14 & & 1.05 + rno 90@xmath79 & 2.10e-14 & 1.08e-14 & 2.05 & 5.64e-14 ( 1.28e-13 ) & 6.22e-15 & 2.10 & 3.87e-14 ( 4.92e-14 ) & 9.83e-15 & 2.49 + ec 82 & 1.39e-14 & 3.05e-15 & 1.32 & @xmath671.48e-14 & & 1.45 & @xmath673.93e-14 & & 2.57 + rx j1842.9 - 3532 & 4.75e-15 & 7.21e-16 & 0.092 & 3.83e-15 & 7.96e-16 & 0.10 & @xmath675.25e-15 & & 0.19 + rx j1852.3 - 3700 & 8.66e-15 & 4.28e-16 & 0.035 & 4.62e-15 & 3.26e-16 & 0.037 & @xmath671.74e-15 & & 0.055 + v1331cyg & @xmath673.30e-15 & & 1.01 & 1.16e-14 & 6.24e-16 & 0.96 & @xmath675.64e-15 & & 0.95 + + ic348 - 21 & 3.12e-15 & 5.13e-16 & 0.088 & 3.71e-15 & 9.77e-16 & 0.081 & @xmath675.67e-15 & & 0.11 + ic348 - 67 & 2.40e-15 & 6.69e-16 & 0.018 & @xmath672.73e-15 & & 0.019 & @xmath673.43e-15 & & 0.023 + ic348 - 72 & @xmath672.93e-15 & & 0.011 & @xmath672.04e-15 & & 0.011 & @xmath672.12e-15 & & 0.012 + ic348 - 133 & @xmath672.90e-15 & & 0.017 & @xmath672.69e-15 & & 0.016 & @xmath674.85e-15 & & 0.023 + dm tau & 5.45e-15 & 4.59e-16 & 0.047 & @xmath672.13e-15 & & 0.048 & @xmath671.56e-15 & & 0.055 + fp tau & @xmath674.73e-15 & & 0.067 & @xmath673.76e-15 & & 0.070 & @xmath674.31e-15 & & 0.078 + gm aur & 1.02e-14 & 7.11e-16 & 0.097 & 1.89e-14 & 9.28e-16 & 0.10 & 1.20e-14 & 1.02e-15 & 0.17 + lkca 15 & 3.41e-15 & 5.54e-16 & 0.14 & 2.94e-15 & 6.71e-16 & 0.15 & @xmath677.89e-15 & & 0.27 + st 34 & 1.40e-15 & 5.66e-16 & 0.029 & 1.59e-15 & 4.68e-16 & 0.028 & @xmath672.12e-15 & & 0.029 + ux taua & 9.04e-15 & 9.85e-16 & 0.11 & 9.02e-15 & 8.53e-16 & 0.11 & @xmath674.98e-15 & & 0.14 + v836tau & @xmath672.05e-15 & & 0.10 & @xmath673.61e-15 & & 0.11 & @xmath673.33e-15 & & 0.17 + lkha 330 & @xmath674.99e-15 & & 0.49 & 8.84e-15 & 1.91e-15 & 0.50 & @xmath672.49e-14 & & 0.69 + 2071-h9/ ngc2068 & 1.21e-15 & 3.75e-16 & 0.023 & 2.44e-15 & 3.84e-16 & 0.024 & @xmath672.06e-15 & & 0.038 + chx 22 ( t54 ) & 5.48e-15 & 9.30e-16 & 0.032 & @xmath672.57e-15 & & 0.036 & @xmath671.04e-15 & & 0.049 + cs cha & 3.61e-14 & 6.12e-16 & 0.046 & 6.11e-15 & 2.53e-16 & 0.048 & 3.07e-15 & 7.43e-16 & 0.064 + sz cha & 1.48e-14 & 1.02e-15 & 0.16 & 6.10e-15 & 1.32e-15 & 0.17 & @xmath676.93e-15 & & 0.25 + sz 18 & 7.63e-16 & 2.33e-16 & 0.0075 & 8.68e-16 & 2.14e-16 & 0.0073 & @xmath672.63e-15 & & 0.0080 + sz 27 ( t35 ) & 6.32e-15 & 9.03e-16 & 0.021 & 3.03e-15 & 7.27e-16 & 0.022 & @xmath672.67e-15 & & 0.026 + tw hya & 5.01e-14 & 2.07e-15 & 0.61 & 7.93e-14 & 2.00e-15 & 0.62 & 3.34e-14 & 3.64e-15 & 0.81 + i m lup & 7.03e-15 & 1.54e-15 & 0.48 & @xmath674.58e-15 & & 0.49 & @xmath676.37e-15 & & 0.60 + doar 21 & 7.80e-14 & 2.58e-15 & 0.65 & @xmath671.32e-14 & & 0.63 & @xmath674.66e-14 & & 0.86 + + 1rxs j121236.4 - 55203 & @xmath676.09e-16 & & 0.014 & @xmath678.71e-16 & & 0.016 & @xmath678.11e-16 & & 0.019 + 1rxs j122233.4 - 533347 & @xmath674.75e-16 & & 0.020 & 8.56e-16 & 1.95e-16 & 0.021 & @xmath677.81e-16 & & 0.026 + 1rxs j130153.7 - 53044 & @xmath674.95e-16 & & 0.0084 & @xmath674.67e-16 & & 0.0092 & @xmath678.19e-16 & & 0.011 + hd119269 & @xmath671.15e-15 & & 0.024 & @xmath679.33e-16 & & 0.026 & @xmath671.49e-15 & & 0.031 + 1rxs j133758.0 - 413448 & @xmath676.79e-16 & & 0.019 & 5.44e-16 & 2.04e-16 & 0.021 & @xmath678.21e-16 & & 0.025 + 1rxs j161458.4 - 27501 & @xmath678.43e-16 & & 0.0088 & @xmath676.31e-16 & & 0.0098 & @xmath678.00e-16 & & 0.012 + hd 142361 & @xmath671.59e-15 & & 0.040 & 2.98e-15 & 1.47e-16 & 0.043 & 1.44e-15 & 2.60e-16 & 0.052 + rx 1600.0 - 2159 & @xmath677.96e-16 & & 0.012 & 1.20e-15 & 1.91e-16 & 0.013 & @xmath677.63e-16 & & 0.015 + rx 1612.2 - 1859 & @xmath676.28e-16 & & 0.028 & 9.32e-16 & 1.53e-16 & 0.031 & @xmath678.60e-16 & & 0.036 + hd 35850 & 3.38e-15 & 1.01e-15 & 0.26 & 9.02e-15 & 9.27e-16 & 0.27 & @xmath676.30e-15 & & 0.34 + v343nor & @xmath673.30e-15 & & 0.11 & 4.81e-15 & 8.09e-16 & 0.12 & @xmath675.64e-15 & & 0.15 + hd19668 & @xmath671.14e-15 & & 0.051 & @xmath671.62e-15 & & 0.056 & @xmath672.56e-15 & & 0.067 + hd25457 & @xmath671.86e-14 & & 0.53 & @xmath671.53e-14 & & 0.58 & @xmath671.86e-14 & & 0.68 + hd37484 & @xmath672.66e-15 & & 0.080 & @xmath672.16e-15 & & 0.086 & @xmath674.47e-15 & & 0.10 + hd12039 & @xmath677.55e-16 & & 0.065 & @xmath671.31e-15 & & 0.071 & @xmath671.43e-15 & & 0.082 + hd202917 & @xmath671.94e-15 & & 0.044 & @xmath673.19e-15 & & 0.048 & @xmath674.31e-15 & & 0.058 + hd 141943 & @xmath671.27e-15 & & 0.074 & 3.10e-15 & 4.13e-16 & 0.080 & 2.97e-15 & 4.91e-16 & 0.094 + hd377 & @xmath671.79e-15 & & 0.088 & @xmath671.31e-15 & & 0.095 & @xmath672.73e-15 & & 0.11 + hd17925 & @xmath673.42e-15 & & 0.61 & @xmath677.36e-15 & & 0.66 & @xmath678.26e-15 & & 0.78 + hd32297 & @xmath673.26e-15 & & 0.054 & @xmath674.28e-15 & & 0.053 & @xmath675.09e-15 & & 0.052 + hd61005 & @xmath671.49e-15 & & 0.066 & @xmath671.32e-15 & & 0.072 & @xmath672.40e-15 & & 0.084 + hd69830 & @xmath674.54e-15 & & 0.56 & @xmath674.75e-15 & & 0.62 & @xmath671.05e-14 & & 0.81 + hd134319 & @xmath672.21e-15 & & 0.047 & @xmath672.55e-15 & & 0.051 & @xmath672.73e-15 & & 0.060 + hd216803 & @xmath675.66e-15 & & 0.74 & @xmath677.99e-15 & & 0.81 & @xmath676.96e-15 & & 0.95 + aomen & @xmath671.04e-15 & & 0.051 & @xmath671.03e-15 & & 0.056 & @xmath672.08e-15 & & 0.065 +	in this paper we investigate the origin of the mid - infrared ( ir ) hydrogen recombination lines for a sample of 114 disks in different evolutionary stages ( full , transitional and debris disks ) collected from the _ spitzer _ archive . we focus on the two brighter h i lines observed in the _ spitzer _ spectra , the h i(7 - 6 ) at 12.37@xmath0 m and the h i(9 - 7 ) at 11.32@xmath0 m . we detect the h i(7 - 6 ) line in 46 objects , and the h i(9 - 7 ) in 11 . we compare these lines with the other most common gas line detected in _ spitzer _ spectra , the [ ne ii ] at 12.81@xmath0 m . we argue that it is unlikely that the h i emission originates from the photoevaporating upper surface layers of the disk , as has been found for the [ ne ii ] lines toward low - accreting stars . using the h i(9 - 7)/h i(7 - 6 ) line ratios we find these gas lines are likely probing gas with hydrogen column densities of 10@xmath1 - 10@xmath2 @xmath3 . the subsample of objects surrounded by full and transitional disks show a positive correlation between the accretion luminosity and the h i line luminosity . these two results suggest that the observed mid - ir h i lines trace gas accreting onto the star in the same way as other hydrogen recombination lines at shorter wavelengths . a pure chromospheric origin of these lines can be excluded for the vast majority of full and transitional disks . we report for the first time the detection of the h i(7 - 6 ) line in eight young ( < 20 myr ) debris disks . a pure chromospheric origin can not be ruled out in these objects . if the h i(7 - 6 ) line traces accretion in these older systems , as in the case of full and transitional disks , the strength of the emission implies accretion rates lower than 10@xmath4m@xmath5/yr . we discuss some advantages of extending accretion indicators to longer wavelengths , and the next steps required to pin down the origin of mid - infrared hydrogen lines .
ngc 1068 is a prototypical seyfert 2 galaxy . because of its proximity ( @xmath15 ) and brightness , it has been studied in nearly every possible detail . the polarimetric observation by @xcite , which reveals a seyfert 1 spectrum in scattered light , suggests that the nucleus and its associated broad - line region ( blr ) are obscured . this finding provides strong evidence for the unified theory in which viewing angles account for the differences between various active galactic nuclei @xcite the nuclear region of ngc 1068 harbors a variety of astrophysical phenomena . at the very center of the nucleus there is a bright compact ( @xmath1603 ) region commonly referred as `` the hot spot '' . within a few arcseconds from the nucleus , there are several bright and compact clouds that coincide with knots in the radio jets @xcite . the narrow - line region ( nlr ) is conical in shape toward the northeast ( fig . 1 ) , along a position angle of @xmath17 and with an opening angle of @xmath18 . beyond a 6 radius the surface brightness drops dramatically , and emission is dominated by two ring - like filaments at @xmath19 and 15 from the nucleus . high - spatial - resolution spectroscopy of ngc 1068 has been carried out in the optical @xcite as well as in the uv @xcite . from approximately 2 southwest of the nucleus to 4 northeast , emission lines exhibit multiple components @xcite : ( 1 ) major emission lines consist of narrow and broad lines ; ( 2 ) broad lines are approximately 2500 - 4000 wide , which may be linked to those that are found in polarized light and believed to be reflected light from the inner blr , and ( 3 ) narrow lines consist of a pair of red and blue components . the [ ] and [ ] line profiles suggest that the separation of these two components varies across the conical nlr . in addition to an overall biconical ionization configuration , there are compact knots whose optical spectra resemble kinematically the associated absorption line systems in quasars @xcite . these line - emitting knots have blueshifted radial velocities up to 3000 relative to the galaxy s systemic velocity , contributing mostly to the emission - line flux but not the continuum . between @xmath2025 and 45 northeast from the nucleus , uv line emission is redshifted relative to the systemic value , a pattern that is interpreted as the expansion of the plasma in the radio lobe @xcite . several important emission lines in the far - uv ( fuv ) region between 912 and 1150 are observable only with specially crafted uv instruments . during the astro-1 mission the hopkins ultraviolet telescope ( hut ) observed ngc 1068 with 18 and 30 apertures . the most striking features in the wavelengths below 1150 are the strong @xmath21977 and @xmath21991 lines . the line intensity ratios of i(@xmath211909)/i(@xmath21977 ) and i(@xmath211750)/i(@xmath21991 ) are temperature sensitive , and the derived temperature is @xmath22 000 k @xcite , higher than the values expected for a region producing and emission by photoionization , the line ratios in ngc 1068 are similar to those of the cygnus loop supernova remnant @xcite , suggesting a significant contribution from shock - heating mechanisms . astro-2 observations of ngc 1068 were obtained with a 12 aperture at three different positions . the results @xcite suggest that the emission lines observed with hut likely arise in the inner nuclear region imaged with hst . however , the poor angular resolution of the hut instrument ( @xmath23 ) does not allow a study of the nlr in terms of spatial details . the line - emitting mechanisms in the nlr , _ i.e. _ photoionization from the nucleus or shocks produced by jets , have long been under debate . @xcite and @xcite proposed that emission in the nlr may be entirely caused by shocks . velocity splitting over 1000 , reported by @xcite , in the vicinity of some of the bright emission - line knots provides evidence that fast shocks exist in the nlr of ngc 1068 . however , seyfert galaxies host powerful nuclear sources of ionizing radiation , and the situation can be more complex with shocks or nuclear photoionization prevailing in different environments @xcite . @xcite suggested that photoionizing shocks are important when a radio jet interacts with the interstellar medium , but not in the objects where sharp , straight - edged ionization cones are observed . @xcite argued that strong @xmath21977 and @xmath21991 emission may arise from fluorescence in photoionized gas if turbulent velocities exceed @xmath201000 . however , @xcite found that such a velocity would lead to extreme physical conditions that are inconsistent with the astro-2 data . more recent hst data @xcite found that the emission - line ratios are consistent with photoionization instead of shock heating mechanism . we have carried out observations with the _ far ultraviolet spectroscopic explorer _ ( fuse ) to study the spatial distribution of fuv emission lines . in this paper we present the results of both large - aperture and spatially resolved spectroscopic observations with fuse . for the first time , we are able to study the position dependence of several important diagnostic lines in the fuv band . fuse covers a wavelength range between 904 and 1188 , with a spectral resolution of @xmath24 @xcite . it is based on a rowland circle design and consists of four separate optical paths or channels . each channel consists of a mirror coated with aluminum plus lif or sic , a focal plane assembly , which includes the spectrograph apertures , a diffraction grating coated with aluminum plus lif or sic , and a portion of fuv detector ( two named a and b ) . in all , data are collected from eight traces , and the pairs of a and b are called channels . our original plan was to take spectra of the nuclear region of ngc 1068 at seven positions separated by @xmath20 1 . because of thermal instability of the optical structure in orbit , the four channels in the fuse spectrograph are not perfectly aligned , with orbit - dependent drifts possibly as large as 6 . therefore , a new observing strategy was developed to counter the drifting effect . the high - resolution aperture ( hirs : @xmath8 ) is used for spectroscopy on the finest spatial scales , to ensure that maximum resolution is maintainable even if the telescope imaging or pointing stability degrades below specifications . since the lif1 channel is mounted on the same optical system as the fuse fine error sensor , the spatial knowledge and stability is fully available for the data from this channel . the observations of ngc 1068 were carried out with two modes between 2001 november 28 and december 02 . the first set of data of 21951 seconds was taken with fixed pointing and the lwrs ( low - resolution ) square aperture of 30 . with such a large aperture , fluxes of the nuclear region are all collected ( fig . 2 ) , even with the thermally - induced mirror motions anticipated . the second set of observations was carried out in a scanning mode across the conical emission - line region , with a total exposure of 94434 seconds . a narrow hirs slit at a position angle of @xmath25 was used . because of orbital constraints , we were not able to place the slit completely perpendicular to the conical axis . the slit actually makes an angle of @xmath26 with respect to the conical axis . over years of fuse operation , a large amount of engineering data have been collected , and the drift pattern among the four channels has been well studied . after peak - up alignments of channels with a bright point source at the beginning of orbital nights , the relative drifts exhibit a dependence on orbital time ( fig . the lif2 channel exhibits relatively small drifts , within one arcsecond with respect to the lif1 channel . the sic channels exhibit drifts that can be represented by a constant drift rate during orbital nights . for example , the sic2 channel drifts relative to the lif1 channel at a rate of approximately 4 per 30 minutes . the starting drift at the orbital sunset is however variable , depending on the degree of occultation because the drift is generally in the reverse direction during orbital days . ngc 1068 itself is too diffuse to appear in the guide - star camera . for target acquisition , we used an offset guide star @xmath27 from the nucleus . we also used a fuv - bright star , feige 23 , which is approximately four degrees away from the source , for instrumental peak - up alignments at the beginning of each pointing . one orbit of peak - up operation on feige 23 was made to establish the drift pattern due to orbital motion . during the second orbit , one peak - up with the mdrs aperture ( @xmath28 ) and another with hirs were carried out on feige 23 . the instrument was then pointed to the southwest part of the nucleus of ngc 1068 , and we started observations in a scanning mode . as shown in table 1 , each of the five pointings consists of 7 - 9 orbits on the target . the observation window varied orbit by orbit , between @xmath29 and 2600 sec . the scanning speed in each individual orbit was set to complete 10 within the designated observation window . after accumulating a total exposure time of @xmath30 sec , fuse returned to feige 23 and carried out one peak - up with mdrs and another with hirs , then moved back to ngc 1068 for a new set of scans . after a total of five pointings , fuse returned to feige 23 for one last orbit of peak - up with mdrs , to acquire additional information on the drift pattern . since the channel alignments are not perfect except at the beginning of the first orbit , the orbit - time dependence of channel drifts derived from the engineering data is only useful in terms of the relative drift speed , not for the absolute timing . the drifts in the sic channels can be as large as 6 , which is near the size of the ionization cone . the drift direction is mainly along + x on the fuse aperture plate , i.e. , along the conical direction of ngc 1068 at the desired fuse aperture position angle of @xmath31 . therefore the observation procedure was designed so that each observation starts @xmath32 from the nucleus in the anti - cone ( southwest ) direction and sweeps across the cone towards the northeast direction at a constant angular velocity . 4 illustrates the time sequence in one orbit of observation : with a constant angular speed , the fuse hirs slit swept across the nuclear region of ngc 1068 . we identified the peak position of the count rate in the lif1 channel as the nuclear position . based on the angular speed illustrated in fig . 3 , seven time intervals were defined , and the spectral extraction in these segments yields the data a - g . the upper panel of fig . 4 marks the segments for the lif1 channel , whose spatial information is fully available . for the other three channels , their count rates peaked at different times as a result of drifting . we marked their respective peak positions . the sweeping speeds in these channels are the sum of two terms : that of the lif1 channel plus the relative drift speed of each channel ( fig . 3 ) with respect to lif1 . their time segments a - g are also illustrated in fig . the observation results in a series of segments that are dependent on the relative positions with respect to the nucleus . in each time sequence , one can determine the time when the observation slits coincide with the nucleus , as determined by the peak in the count rate . the lif1 slit positions are illustrated in fig . 1 . seven sets of data are extracted with a relative shift of 1 arcsec ( 125 at the nucleus ) , starting from 2 southwest from the nucleus ( in the opposite direction of the cone ) , to @xmath33 northeast of it . they were named as a , b , c ( the nucleus , see fig . 1 ) , d , e , f and g. the significant astigmatism of fuse yields a spatial resolution of @xmath32 at best , therefore it can not be used like a normal long - slit spectrograph to dissect the fine details along its slit . not all the segment data are extracted : if a time segment is incomplete ( `` g '' in fig . 4 , for example ) , they are not used . four pairs of spectra were derived from the fuse channels . each pair of `` a '' and `` b '' data is subject to the same normalization factor . we used the emission flux , which is common to all four channels , to normalize the spectra to the level in lif1a . we first calculated the continuum level between 1050 and 1070 for lif2b , sic1a , sic 2b and lif1a , then subtracted it from the spectrum to enable the measurement of emission - line fluxes . we then calculated the flux in two bins : a narrow component between 1041.1 and 1043.6 , and a broad component between 1027 and 1048 . the flux ratios with respect to their values in the lif1a spectrum were calculated , then their mean values were used as the normalization factors ( table 2 ) . for the first set of data , taken with lwrs , we corrected the `` worm effect '' in the lif1a , b channel , according to the procedures described in the _ fuse instrument and data handbook _ ( http://fuse.pha.jhu.edu/analysis/dhbook.html ) . fuse data reduction was carried out with pipeline version calfuse2.3 , with specific timing flags to extract time - dependent ( hence position - dependent ) segments of spectra in eight traces . the fuse spatial resolution is 15 , and the spacecraft jitter , as determined from the fuse engineering data , is approximately 06 . the total spatial resolution is therefore 16 . since the data were obtained with a 125 slit , there is overlap between positions . this can be seen by the fact that the sum of the fluxes in all positions is larger than that taken with one large 30 aperture . the normalized spectra in the four channels are binned to 0.1 , then merged , weighted by their signal - to - noise ratio . the merged spectra at the seven slit positions are plotted in fig . the wavelength bin between @xmath34 and 1087 falls into a gap between the lif channels , and only data in the sic2b channel exist . we retrieved archival hst / stis spectroscopic data to complement our fuse data . two sets of uv spectra are available : one taken at a position angle of @xmath35 and four adjacent slit positions of @xmath36 ( * ? ? ? * stis - a hereafter ) , and another at a position angle of @xmath37 taken with a slit of @xmath38 ( * ? ? ? * stis - b ) . both sets ( see fig 1 ) contain data taken with grating g140l and g230l , covering a wavelength range of @xmath39 to 3170 at a resolution of @xmath40 . these stis observations were made between 1988 and 2000 , therefore not simultaneous with the fuse data . for stis - a , we only used three slit positions as the other does not have g140l data . the fluxes in the three extracted spectra are summed up . since the fuse spatial resolution ( @xmath41 along its slit ) is considerably lower than that of stis , and since the position angles of these two sets of observations are nearly perpendicular , we must normalize the stis data to that of fuse . using the continuum level around 1180 in these spectra did not result in satisfactory matches as the line / continuum ratios are not constant at these spatial scales across the nucleus . we therefore use hst wpfc2 images retrieved from mast to determine the normalization factor between stis and fuse data . to extract the stis spectra , we first smoothed the 2-d images by 30 pixels ( 067 ) with a gaussian kernel along the spatial direction of the fuse scans , to match the spatial resolution of the corresponding fuse data . we then ran the standard stis pipeline task `` x1d '' with steps of one arcsecond ( 125 at the nucleus ) that correspond to the respective moving fuse slit positions , and a slit width of 125 . to normalize the stis fluxes , we used a hst / wfpc2 image of ngc 1068 , taken with filter f218w , and smoothed it along the same direction with the same kernel size as the stis 2-d image . we then measured the flux in each stis window . to normalize the fuse fluxes , we smoothed the wfpc2 image to the fuse resolution , namely 067 and 5 , respectively , along the direction perpendicular and parallel to that of the fuse slit ( fig . the flux ratios in respective extraction windows are used to derive the normalization factors ( table 2 ) . to the northwest direction from the source , there are several bright spots whose fluxes may be picked up because of fuse s low spatial resolution along its slit . we tested with several extraction windows that are sufficiently short not to include the fluxes from these bright spots in the smoothed wfpc2 image . since it is unlikely that these bright spots contribute to the redshifted uv emission lines , we used the flux values measured in a short window in calculating the scaling factor . any uncertainties introduced by this window selection only affect the scaling of ly@xmath14 and in position a and g. since the fuse slit positions are accurate to @xmath20 03 , we measured the fluxes in the neighboring regions , to make sure that the normalization factors are accurate to at least 20% . the height of stis extraction windows is the same as that of fuse ( @xmath9 ) , but their widths ( 01 - 02 ) are much smaller than fuse ( 20 ) . therefore considerable stis normalization factors were applied . the matching stis spectra are plotted in fig . we compared extracted spectra from the stis - a and stis - b datasets , and their line ratios are comparable within 25% . in the following section , we only use the data from stis - a , as they cover a spatial region six times as wide ( @xmath42 ) as that of stis - b . spectral analyses were carried out using iraf task _ specfit _ @xcite . for the fuse spectrum taken with a large aperture , we used a pair of narrow components and one broad component in the emission line profile . each narrow component was modeled as a doublet whose wavelength ratio is fixed by atomic data and whose line widths are identical . the emission feature is heavily absorbed by interstellar absorption at @xmath20 1037 , therefore we introduced several absorption components . the ly@xmath43emission is not prominent , and it is often overwhelmed by the broad emission . only in one or two positions is the ly@xmath43 emission visible . we therefore only modeled one component with a fixed line width . for and emission line features , one narrow component and one broad component were included in fitting . we fitted a power law with no extinction to the continuum . the fitted results are listed in table 3 . for the fuse spectra taken at positions a - g , we used one narrow component plus one broad component to model the emission , and one component for the and lines . to reduce the effect of channel drifts , the fuse spectra were fitted individually for each channel . the fitting results in table 4 are primarily derived from the lif1 data , which are not affected by channel drifts . for wavelengths below 995 , we used the fitting results from the scaled sic2a channel , which extends to 1005 and is therefore less susceptible than the sic1b channel to the detector edge effects . for 1085 , lif2a data were used . the fluxes of fitted components were normalized by the values in table 2 , and then tabulated in table 4 . the profiles of four major emission lines at seven different slit positions are plotted in fig . 7 , to show the changes in the narrow and broad components as a function of spatial location . the uv spectra suggest a low extinction level . we first fitted the continuum in the wavelength windows that are free of emission and absorption lines , and the results suggest a power - law index of @xmath44 ( @xmath45 ) and an extinction of @xmath46 . as a comparison , the hut spectrum between 912 and 1800 was fitted with a power - law continuum of @xmath47 and @xmath48 . if we adopt this extinction value , the power - law index is @xmath49 . the similar power indices suggest that one single power - law continuum can fit the far - uv spectrum of ngc 1068 without introducing significant reddening , as suggested from the uv / optical line ratios @xcite . many previous studies have gradually unraveled the kinematical complexities of gas in the nuclear region of ngc 1068 . the x - ray data @xcite suggest that emission lines are formed mainly in a photoionied plasma of a temperature around a few ev . the spatial resolution afforded by hst leads to a picture of decelerating jet @xcite : a biconical outflow from the nuclear region sweeps up denser , ambient clouds in the interstellar medium ( ism ) of ngc 1068 . other possibilities include overlapping , discrete ejection that gradually dissipates @xcite . in addition , a high - velocity radio jet impinges on some of the clouds . some gas expands perpendicularly to the axis of the jet , and the expanding radio lobe at the end of the jet also pushes on the ambient ism . near the nucleus , kinematic components span several thousand kilometers per second in velocity , and the continuum hot spot visible in hst images reflects a polarized view of the broad lines in the active nucleus . our spatially resolved fuse observations , while not at the resolution of hst , add information from major emission lines shortward of the hst bandpass at high spectral resolution . we plot the fluxes of major uv emission lines from our observations at seven slit positions in fig . 8 , and list the fitted line properties in table 4 . in this section , we discuss the four emission lines in the fuse spectral range , along with a comparison to emission lines observed in the stis spectra . the most prominent feature in the fuse spectra is the @xmath21@xmath211031,1037 emission line . in the data taken with lwrs , the two narrow components are of fwhm @xmath50 and a separation of @xmath10 . in the seven spectra taken with a narrow slit , we resolve this blend into one narrow component with fwhm of @xmath50 and one broad component . in fig . 9 we plot the profiles at the seven different slit positions . the high spectral resolution of fuse data enables us to compare with the results of optical fabry - perot spectroscopy @xcite , which reveal a narrow core of @xmath51 . according to the optical data , nearly 75% of the [ ] @xmath216583 flux is from components of @xmath52 wide . line widths at such a scale are consistent with that derived from hst uv spectroscopy . the narrow - line flux in the fuse spectra is highly concentrated ( 60% ) at the compact core , suggesting that the `` true '' nlr probably remains unresolved , at a sub - arcsecond scale . while the narrow o vi emission line is not the dominant component , its distribution is different from its broad counterpart , while only 40% of the broad line flux is from slit position c. it is surprising that the emission is dominated by a component that is broader than those seen in polarized light : in the lwrs spectrum , more than 3/4 of the flux is from a component with fwhm @xmath0 that is blue - shifted relative to the narrow component by @xmath53 . this broad component is present in all seven fuse slit spectra with considerable strength . it may arise from the reflected emission from the hidden blr , and/or may be the result of a significant velocity dispersion in the nlr . the hst foc data @xcite reveal that emission lines near the hot knots 2 northeast ( fuse slit positions d and e ) are split into two velocity systems separated by @xmath52 . the stis spectra discussed by @xcite show [ ] emission knots spanning such a broad velocity range in the immediate vicinity of the nucleus , but not at distances of several arcseconds . since the fuse slit collects emission from a block of regions spanning several arcseconds perpendicular to the conical axis , the total line emission from these regions may be blended into one broad component . the ratios of / may provide insight into the physical conditions of the line - emitting regions . the value is higher for the narrow components than their broad counterparts , implying a range of the ionization parameter @xmath54 in a typical photionization calculation . high values of @xmath55 are consistent with models that assume the same origin for the associated absorbers and blr @xcite , suggesting that the clouds that produce the broad emission components may be of the same origin as associated absorbers . the narrow line exhibits a systematic velocity shift from position a to g by approximately 220 . this gradient in the spatially resolved spectra explains why there are two narrow - line components in the integrated flux from the large - aperture fuse spectrum , where it is unresolved . @xcite reported a similar velocity pattern in their hst spectra . @xcite successfully modeled this as a biconical outflow in which radial velocity changes as a function of the distance to the central nucleus : the emission line knots show evidence for radial acceleration to a projected distance of 2 to the northeast direction , followed by deceleration up to 4 . the line widths also increase at @xmath56 from the nucleus , probably implying a larger dispersion in these regions . the broad component of exhibits a qualitatively similar kinematic pattern , but at a larger amplitude : its centroid shifts by @xmath52 ( fig . 10 ) across the same spatial region . 11 shows similar trends for the narrow and broad components of in the hst spectra which have not been explicitly noted in previous studies : the narrow component follows the kinematic pattern of the optical lines modeled by @xcite ; the broad component of shows behavior similar to that of in the fuse spectra . prior observations that noted this blue - shifted broad component invariably attributed it to the reflection of the blr . the blue shift and width ( at the position of the hot spot , and in integrated light ) are comparable to the broad polarized h@xmath43 line observed by @xcite . in a scattered blr picture , a blue shift is caused by the outflowing wind from the torus along our line of sight , and a line width is due to the intrinsic broad line width convolved with the thermal width of the hot reflecting wind . as a broad component is present in the fuse spectra at all seven positions , it is possible to assume that this is reflected light from the hidden blr . however , a large covering factor is needed to explain the observed fluxes . assuming a covering factor of 0.1 , the intrinsic flux of the broad emission in ngc 1068 would exceed that in ngc 4151 . a more reasonable explanation for the observed broad line widths is the large velocity dispersion between bright knots . at approximately 2 from the core ( position a and e ) , the fwhms of broad components are the broadest at 3200 . these maxima coincide with the widest spliting of velocity in bright knots @xcite . with an intrinsic dispersion of @xmath20800 and a separation in velocity of @xmath202500 between bright knots ( from stis results ) , data collected by the long fuse slit would exhibit a broad component of @xmath203200 , which is what observed at positions a and e. extended regions of hot , photoionized gas are seen in x - ray images of ngc 1068 @xcite that could be visible manifestations of this hot outflow . it is natural to assume that these high - velocity clouds may be related to the associated absorbers ( * ? ? ? * and references therein ) in agn , which are mostly blueshifted . as with the lower - velocity , lower ionization emission - line gas , the acceleration of this high ionization gas eventually is brought to a halt by an unknown deceleration mechanism , which might plausibly be interaction with the ambient ism of ngc 1068 . the evidence for deceleration at arcsecond scale may suggest that acceleration of the outflow materials may take place at sub - arcsecond scales . the flux ratio of @xmath211909 to 977 is extremely sensitive to temperature , and the value measured in n1068 suggests a high temperature that is consistent with shock - heating @xcite . the fuse data taken with a large aperture ( table 3 ) reveal that this emission line consists of a narrow and a broad component . the broad component is weak and hence can not be well separated in the data segments taken with a narrow fuse slit . the stis data reveal that @xmath211909 emission can be fitted with a narrow and a broad component of fwhm @xmath57 and 3500 , respectively . in principal , the ratio of @xmath211909 to 977 should be calculated only between the narrow components . as shown in fig . 8 , the flux of @xmath21977 is highly concentrated in position c and d. high temperatures implied by this line emission may therefore be associated with the compact core . in a large portion of the ionization cone , the line ratios ( @xmath211909/@xmath21977 ) are considerably higher , suggesting a lower temperature . however , the line ratios are considerably lower than that derived from the hut data , where broad @xmath211909 component was included in the calculation . the line ratio of @xmath211750 to @xmath21991 is also temperature dependent , and its value has also been used to derive a high temperature in the ionization cone . the fuse data also reveal a pair of components in the data taken with a large aperture . unlike @xmath21977 , the broad component of @xmath21991 , as shown in table 3 , is stronger than its narrow counterpart . the fuse data taken with a narrow slit shows that nearly a half of this broad component is in position c. in other positions , the @xmath21991 emission can only be fitted with one component of fwhm @xmath6 . in combination with the trend for @xmath21977 , we conclude that a bulk of flux in these temperature - sensitive emission lines is from the compact core of ngc 1068 , and their intensity is not directly tied to the physical conditions in the ionization cone . the distribution of @xmath211085 is different from other emission lines in the fuse spectra : as shown in fig . 8 , the flux varies smoothly across the nlr region , like that of ly@xmath14 , and other lines in the stis spectra . since emission is believed to be produced mainly by recombination and is insensitive to the gas temperature , this pattern of variation may simply reflect the distribution of nlr gas across the ionization cone . the ratio of emission @xmath211640 to @xmath211085 is believed to be a reddening indicator . as shown in fig . 12 , the ratio is nearly constant ( @xmath58 ) between slit positions b and f. this is consistent with the hut result of @xmath59 , suggesting an insignificant level of extinction ( @xmath60 ) . the low values at positions a and g may suggest that the data at these positions are not reliable . the i(1640)/i(1085 ) ratio is slightly lower than that expected from recombination , and this may suggest that the @xmath211085 may be slightly contaminated by other weak uv emission lines . @xmath211085 is a likely candidate , and the trend of decreasing ( 1640)/i(1085 ) ratio with distance from the nucleus supports this as there is a noticeable decrease in ionization state at larger distances from the nucleus @xcite . as with and , there appears to be a broad component that is considerably blueshifted . however , the relevant wavelength range is in a gap between the lif channels , and only sic2b data are available . because of the low s / n ratio , the reality of such a broad component is still questionable , as no such counterpart is found in the corresponding hst spectra of @xmath211640 . we have carried out high - spectral resolution , spatially resolved spectroscopy of the nuclear region of the seyfert 2 galaxy ngc 1068 . the high spectral resolution of fuse data enables us to study the line profiles of , , and as a function of position across the nuclear region . our observations using a long slit of @xmath8 at steps of @xmath61 along the axis of the emission - line cone resulted in a set of seven spatially resolved spectra running from @xmath11 southwest of the nucleus to @xmath12 northeast . the profiles exhibit considerable structure : a prominent broad component with fwhm @xmath0 @xmath1 , and a narrow component of @xmath51 @xmath62 . both components show a position - dependent velocity gradient along the conical axis with a velocity shift of @xmath63 @xmath1 in the narrow component and a shift of @xmath52 @xmath1 in the broad component . both patterns are consistent with radial outflow from the nuclear region . both the continuum and emission lines suggest low extinction : the uv continuum is flat to the shortest wavelengths in the fuse spectrum , and the best fit is a power law with virtually no reddening . the 1640/1085 ratio also is reflective of a consistently low extinction factor across the emission - line cone . the majority of the and emission arises in the compact core , suggesting that the region with extremely high temperature is very close to the nucleus and remains unresolved . the line emission at @xmath11 northeast of the nucleus is strong and broad , characterized by ( a ) a strong lya flux , but normal flux ; ( b ) a broad line ; and ( c ) a significantly enhanced flux , possibly the result of hot knots associated with shock heating . this work has been supported in part by nasa grant nag-8 - 1527 and nag-8 - 1133 . the fuse and hst data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) . allen , m. g. , dopita , m. a. , & tsvetanov , z. i. 1998 , , 493 , 571 antonucci , r. 1993 , araa , 31 , 473 antonucci , r. r. j. , & miller , j. s. 1985 , , 297 , 621 axon , d. j. , marconi , a. , capetti , a. , machetto , f. d. , schreier , e. , & robinson , a. 1998 , , 496 , l75 bicknell , g. v. , dopita , m. a. , tsvetanov , z. i. , & sutherland , r. s. 1998 , , 495 , 680 blair , w. p. , et al . 1991 , , 379 , l33 caganoff , s. , et al . 1991 , , 377 , l9 capetti , a. , axon , d. j. , & macchetto , f. d. 1997 , , 487 , 560 capetti , a. , axon , d. j. , machetto , f. d. , sparks , w. b. , & boksenberg , a. 1996 , , 469 , 554 cecil , g. , bland , j. , & tully , r. b. 1990 , , 355 , 70 cecil , g. , dopita , m. a. ; groves , b. , wilson , a. s. ; ferruit , p. , pcontal , e. , & binette , l. 2002 , , 568 , 627 crenshaw , d. m. , & kraemer , s. b. 2000@xmath64 , , 532 , l101 2000@xmath65 , , 532 , 247 2005 , , 625 , 680 crenshaw , d. m. , kraemer , s. b. , & george , i. m. 2003 , ara&a , 41 , 117 das , v. , crenshaw , d. m. , kraemer , s. b. , & deo , r. p. 2006 , , 132 , 620 dopita , m. a. , & sutherland , r. s. 1996 , , 102 , 161 dopita , m. a. , et al . 1997 , , 490 , 202 emsellem , e. , fathi , k. , wozniak , h. , ferruit , p. , mundell , c. g. , & schinnerer , e. 2006 , , 365 , 367 evans , i. n. , ford , h. c. , kinney , a. l. , antonucci , r. r. j. , armus , l. , & caganoff , s. 1991 , , 369 , l27 ferguson , j. w. , ferland , g. j. , & pradhan , a. k. 1995 , , 438 , l55 gerssen , j. , allington - smith , j. , miller , b. w. , turner , j. e. h. , & walker , a. 2006 , , 365 , 29 grimes , j. p. , kriss , g. a. , & espey , b. r. 1999 , , 526 , 130 groves , b. a. , cecil , g. , ferruit , p. , & dopita , m. a. 2004 , , 611 , 786 inglis , m. d. young , s. , hough , j. h. , gledhill , t. , axon , d. j. , bailey , j. a. , & ward , m. j. 1995 , , 275 , 398 kinkhabwala , a , et al . 2002 , , 575 , 732 kraemer , s. b. , & crenshaw , d. m. 2000@xmath64 , , 532 , 256 2000@xmath65 , , 544 , 763 kriss , g. a. 1994 , in _ astronomical data analysis software and systems iii _ , eds . d. r. crabtree , r. j. hanisch , & j. barnes , ( _ a . s. p. conf . series 61 _ , asp , san francisco ) , 437 kriss , g. a. , davidsen , a. f. , blair , w. p. , ferguson , h. c. , & long , k. s. 1992 , , 394 , l37 kriss , g. a. , blustin , a. , branduardi - raymont , g. , green , r. f. , hutchings , j. , & kaiser , m. e. 2003 , , 403 , 473 moos , h. w. , et al . 2000 , , 538 , l1 morganti , r. , oosterloo , t. , & tsvetanov , z. 1998 , , 115 , 915 morse , j. a. , raymond , j. c. , & wilson , a. s. 1996 , pasp , 108 , 426 sutherland , r. s. , bicknell , g. v. , & dopita , m. a. 1993 , , 414 , 510 unger , s. w. , lewis , j. r. , pedlar , a. , & axon , d. j. 1992 , , 258 , 371 wilson , a. s. , & ulvestad , j. s. 1983 , , 275 , 8 young , a. j. , wilson , a. s. , shopbell , p. l. 2001 , , 556 , 6 lccccl p1110202000 & 2001-nov-28 09:57:00 & 8 & 21951 & lwrs & mdrs + a1390201000 & 2001-nov-29 01:23:46 & 9 & 19541 & hirs & mdrs , hirs + a1390202000 & 2001-nov-29 16:34:46 & 9 & 19828 & hirs & mdrs , hirs + a1390203000 & 2001-nov-30 08:38:00 & 7 & 18023 & hirs & mdrs , hirs + a1390204000 & 2001-nov-30 20:53:34 & 9 & 17962 & hirs & mdrs , hirs + a1390205000 & 2001-dec-01 11:52:20 & 8 & 19080 & hirs & mdrs , hirs + cccccc a & 1.00 & 1.51 & 0.86 & 2.48 & 2.50 + b & 1.00 & 1.42 & 0.97 & 1.70 & 2.03 + c & 1.00 & 1.17 & 1.03 & 1.07 & 1.74 + d & 1.00 & 1.29 & 0.96 & 1.29 & 2.28 + e & 1.00 & 1.15 & 0.78 & 0.87 & 3.95 + f & 1.00 & 1.10 & 1.28 & 1.17 & 4.84 + g & 1.00 & 1.51 & 2.75 & 1.83 & 6.05 + llcccl + & 977.02 & @xmath66 & @xmath67 & @xmath68 + narrow & 990.98 & @xmath69 & @xmath70 & @xmath71 + broad & & @xmath72 & @xmath73 & @xmath74 + ly@xmath43 & 1025.72 & @xmath75 & @xmath76 & @xmath77 + narrow & 1037.63 & @xmath78 & @xmath79 & @xmath80 + broad & 1034.00 & @xmath81 & @xmath82 & @xmath83 + & 1085.15 & @xmath84 & @xmath85 & @xmath86 + ly@xmath14 narrow & 1215.67 & @xmath87 & @xmath88 & @xmath89 + ly@xmath14 broad & & @xmath90 & @xmath91 & @xmath92 + narrow & 1549.50 & @xmath93 & @xmath94 & @xmath95 + broad & & @xmath96 & @xmath97 & @xmath98 + & 1640.46 & @xmath99 & @xmath100 & @xmath101 + & 1750.00 & @xmath102 & @xmath103 & @xmath104 + ] narrow & 1908.73 & @xmath105 & @xmath106 & @xmath107 + ] broad & & @xmath108 & @xmath109 & @xmath110 + + & 977.02 & @xmath111 & @xmath112 & @xmath113 + narrow & 990.98 & @xmath114 & @xmath115 & @xmath116 + broad & & @xmath117 & @xmath118 & @xmath119 + ly@xmath43 & 1025.72 & @xmath120 & @xmath76 & @xmath121 + narrow & 1037.63 & @xmath122 & @xmath123 & @xmath124 + broad & 1034.00 & @xmath125 & @xmath126 & @xmath127 + & 1085.15 & @xmath128 & @xmath129 & @xmath130 + ly@xmath14 narrow & 1215.67 & @xmath131 & @xmath132 & @xmath133 + ly@xmath14 broad & & @xmath134 & @xmath135 & @xmath136 + narrow & 1549.50 & @xmath137 & @xmath138 & @xmath139 + broad & & @xmath140 & @xmath141 & @xmath142 + & 1640.46 & @xmath143 & @xmath144 & @xmath145 + & 1750.00 & @xmath146 & @xmath147 & @xmath148 + ] narrow & 1908.73 & @xmath149 & @xmath150 & @xmath151 + ] broad & & @xmath152 & @xmath153 & @xmath154 + + & 977.02 & @xmath155 & @xmath156 & @xmath157 + narrow & 990.98 & @xmath158 & @xmath159 & @xmath160 + broad & & @xmath161 & @xmath162 & @xmath163 + ly@xmath43 & 1025.72 & @xmath164 & @xmath165 & @xmath166 + narrow & 1037.63 & @xmath167 & @xmath168 & @xmath169 + broad & 1034.00 & @xmath170 & @xmath171 & @xmath172 + & 1085.15 & @xmath173 & @xmath174 & @xmath175 + ly@xmath14 narrow & 1215.67 & @xmath176 & @xmath177 & @xmath178 + ly@xmath14 broad & & @xmath179 & @xmath180 & @xmath181 + narrow & 1549.50 & @xmath182 & @xmath183 & @xmath184 + broad & & @xmath185 & @xmath186 & @xmath187 + & 1640.46 & @xmath188 & @xmath189 & @xmath190 + & 1750.00 & @xmath191 & @xmath192 & @xmath193 + ] narrow & 1908.73 & @xmath194 & @xmath195 & @xmath196 + ] broad & & @xmath197 & @xmath198 & @xmath199 + + & 977.02 & @xmath200 & @xmath201 & @xmath202 + narrow & 990.98 & @xmath203 & @xmath204 & @xmath205 + broad & & @xmath206 & @xmath207 & @xmath208 + ly@xmath43 & 1025.72 & @xmath209 & @xmath76 & @xmath210 + narrow & 1037.63 & @xmath211 & @xmath212 & @xmath213 + broad & 1034.00 & @xmath214 & @xmath215 & @xmath216 + & 1085.15 & @xmath217 & @xmath218 & @xmath219 + ly@xmath14 narrow & 1215.67 & @xmath220 & @xmath221 & @xmath222 + ly@xmath14 broad & & @xmath223 & @xmath224 & @xmath225 + narrow & 1549.50 & @xmath226 & @xmath227 & @xmath228 + broad & & @xmath229 & @xmath230 & @xmath231 + & 1640.46 & @xmath232 & @xmath233 & @xmath234 + & 1750.00 & @xmath235 & @xmath236 & @xmath237 + ] narrow & 1908.73 & @xmath238 & @xmath239 & @xmath240 + ] broad & & @xmath241 & @xmath242 & @xmath243 + + & 977.02 & @xmath244 & @xmath245 & @xmath246 + narrow & 990.98 & @xmath247 & @xmath248 & @xmath249 + broad & & @xmath250 & @xmath251 & @xmath252 + ly@xmath43 & 1025.72 & @xmath253 & @xmath76 & @xmath254 + narrow & 1037.63 & @xmath255 & @xmath256 & @xmath257 + broad & 1034.00 & @xmath258 & @xmath259 & @xmath260 + & 1085.15 & @xmath261 & @xmath262 & @xmath263 + ly@xmath14 narrow & 1215.67 & @xmath264 & @xmath265 & @xmath266 + ly@xmath14 broad & & @xmath267 & @xmath268 & @xmath269 + narrow & 1549.50 & @xmath270 & @xmath271 & @xmath272 + broad & & @xmath273 & @xmath274 & @xmath275 + & 1640.46 & @xmath276 & @xmath277 & @xmath278 + & 1750.00 & @xmath279 & @xmath280 & @xmath281 + ] narrow & 1908.73 & @xmath282 & @xmath283 & @xmath284 + ] broad & & @xmath285 & @xmath286 & @xmath287 + + & 977.02 & @xmath288 & @xmath289 & @xmath290 + narrow & 990.98 & @xmath291 & @xmath292 & @xmath293 + broad & & @xmath294 & @xmath295 & @xmath296 + ly@xmath43 & 1025.72 & @xmath297 & @xmath76 & @xmath298 + narrow & 1037.63 & @xmath299 & @xmath300 & @xmath301 + broad & 1034.00 & @xmath302 & @xmath303 & @xmath304 + & 1085.15 & @xmath305 & @xmath306 & @xmath307 + ly@xmath14 narrow & 1215.67 & @xmath308 & @xmath309 & @xmath310 + ly@xmath14 broad & & @xmath311 & @xmath312 & @xmath313 + narrow & 1549.50 & @xmath314 & @xmath315 & @xmath316 + broad & & @xmath317 & @xmath318 & @xmath319 + & 1640.46 & @xmath320 & @xmath321 & @xmath322 + & 1750.00 & @xmath323 & @xmath324 & @xmath325 + ] narrow & 1908.73 & @xmath326 & @xmath327 & @xmath328 + ] broad & & @xmath329 & @xmath330 & @xmath331 + + & 977.02 & @xmath332 & @xmath333 & @xmath334 + narrow & 990.98 & @xmath335 & @xmath336 & @xmath337 + broad & & @xmath338 & @xmath339 & @xmath340 + ly@xmath43 & 1025.72 & @xmath341 & @xmath76 & @xmath342 + narrow & 1037.63 & @xmath343 & @xmath344 & @xmath345 + broad & 1034.00 & @xmath346 & @xmath347 & @xmath348 + & 1085.15 & @xmath349 & @xmath350 & @xmath351 + ly@xmath14 narrow & 1215.67 & @xmath352 & @xmath353 & @xmath354 + ly@xmath14 broad & & @xmath355 & @xmath356 & @xmath357 + narrow & 1549.50 & @xmath358 & @xmath359 & @xmath360 + broad & & @xmath361 & @xmath362 & @xmath363 + & 1640.46 & @xmath364 & @xmath365 & @xmath366 + & 1750.00 & @xmath367 & @xmath368 & @xmath369 + ] narrow & 1908.73 & @xmath370 & @xmath371 & @xmath372 + ] broad & & @xmath373 & @xmath374 & @xmath375 +	we carry out high - resolution fuse spectroscopy of the nuclear region of ngc 1068 . the first set of spectra was obtained with a 30 square aperture that collects all emission from the narrow - line region . the data reveal a strong broad component of fwhm @xmath0 @xmath1 and two narrow @xmath2 components of @xmath3 . the @xmath4 and @xmath5 emission lines in this spectrum can be fitted with a narrow component of fwhm @xmath6 @xmath1 and a broad one of @xmath7 @xmath1 . another set of seven spatially resolved spectra were made using a long slit of @xmath8 , at steps of @xmath9 along the axis of the emission - line cone . we find that ( 1 ) major emission lines in the fuse wavelength range consist of a broad and a narrow component ; ( 2 ) there is a gradient in the velocity field for the narrow component of @xmath10 @xmath1 from @xmath11 southwest of the nucleus to @xmath12 northeast . a similar pattern is also observed with the broad component , with a gradient of @xmath13 . these are consistent with the hst / stis findings and suggest a biconical structure in which the velocity field is mainly radial outflow ; ( 3 ) a major portion of the and line flux is produced in the compact core . they are therefore not effective temperature diagnostics for the conical region ; and ( 4 ) the best - fitted uv continuum suggests virtually no reddening , and the 1085/1640 ratio suggests a consistently low extinction factor across the cone . at @xmath11 northeast of the nucleus there is a region characterized by ( a ) a strong ly@xmath14 flux , but normal flux ; ( b ) a broad line ; and ( c ) a significantly enhanced flux .
it is well known that energy spectra of particles from atmospheric family events which are observed by the emulsion chambers at mountain altitudes are essentially following power - like dependence . on the other hand the occurence of power - law distributions is a very common feature in nature and it is usually connected with such notions as criticality , fractals and chaotic dynamics and studied using generalized ( nonextensive ) maximum entropy formalism characterized by a nonextensivity parameter @xmath0 . such formalism leads in a natural way to power - laws in the frame of equilibrium processes @xcite . for non - equilibrium phenomena the sources of power - law distributions are self - organized criticality @xcite and stochastic multiplicative processes @xcite . in the former nonequilibrium systems are continuously driven by their own internal dynamic to a critical state with power - laws omnipresent . in the later power - law is generated by the presence of underlying replication of events . all these approaches can be unified in terms of generalized nonextensive statistics mentioned above ( widely known as tsallis statistics ) @xcite . + we shall look therefore at the development of cascades from this point of view . to be more specific , we shall use the notion of lvy flights as representatives of power - laws emerging from generalized statistics @xcite . ( 7.28,7.11 ) fig.1 distributions @xmath1 for multiplicative process , as given by eq . ( [ eq : multi ] ) ( histograms ) , for different generation numbers @xmath2 and @xmath3 are fitted by lvy distribution ( eq . ( [ eq : levy ] ) ) ( solid lines ) with parameters @xmath4 and @xmath5 , respectively . let us start with numerical ( monte - carlo ) consideration of the following model of multiplicative process : @xmath6 where @xmath7 is random number taken from the exponential distribution ( to account for the essentially exponential scaled energy dependence of the single elementary interaction ) : @xmath8 as was shown in @xcite , every exponential distribution with a fluctuating parameter results in the lvy distribution ( cf . 1 ) , which in our case has the following form @xmath9^{-\alpha}. \label{eq : levy}\ ] ] ( 7.28,7.11 ) fig.2 dependence of the parameter @xmath10 of the levy distribution ( [ eq : levy ] ) on the generation number @xmath11 . solid line represents fit by the formula ( [ eq : par ] ) with @xmath12 . asymptotically @xmath13 ( dotted line ) . in our present case parameter @xmath10 depends on the number of generations @xmath11 and on the parameter @xmath14 which defines exponential distribution of the variable @xmath7 ( in the limit of large @xmath11 dependence on @xmath14 effectively vanishes ) . as is illustrated in fig . 2 , dependence @xmath15 can be described by simple formula @xmath16 where @xmath17 is generation independent parameter which should be fitted ( in our case @xmath12 ) . if we use the canonical form of lvy distribution as emerging from tsallis statistics @xcite , @xmath18 ^{\frac{1}{1-q } } , \label{eq : levt}\ ] ] where mean @xmath19 is given by the parameter @xmath20 and @xmath0 is mentioned before nonextensive parameter such that @xmath21 @xcite then @xmath22 notice that for @xmath23 parameter @xmath24 ( or , respectively , @xmath25 ) , i.e. , we are recovering the initial exponential distribution . in the limit @xmath26 parameter @xmath0 approaches value @xmath27 , which is limiting value available for @xmath0 emerging from the normalization condition imposed on the probability distribution @xmath28 . in this limit @xmath29 . it is interesting to note that for @xmath30\cdot [ 1/(\alpha - 2 ) ] > > 1 $ ] , i.e. , for @xmath31 sufficiently large one gets power - like behaviour of @xmath28 : @xmath32 actually such situation is reached reasonably fast , because @xmath33 and @xmath10 tends to its limiting value @xmath34 rather quickly with increasing number of generations @xmath11 . in practice the equilibrium distribution @xmath35 is reached ( for @xmath36 ) already for @xmath37 . there exists a number of experimental data from the emulsion chambers exposed at mountain altitudes , which are relevant for our approach and which we shall compare with now @xcite . although , as we have already mentioned , the energy spectra of particles from atmospheric family events they represent are roughly expressed by the power - law distributions , so far there was no analysis showing wheather it is true and to what extend . our work is the first attempt in this direction as we show that they all can be described by the lvy type spectra . ( 7.28,7.11 ) fig.3 integral energy spectra for gamma quanta ( circles ) and hadrons ( squares ) in families registered at mt . kambala are fitted by lvy distributions ( solid lines ) . integral spectra @xmath38 from kambala experiment ( at @xmath39 ) @xcite are shown in fig . 3 where they are fitted using parameters @xmath40 ( for gamma families ) and @xmath41 ( for hadronic families ) . notice that hadronic component is little `` younger '' ( higher @xmath10 means smaller @xmath11 , cf . 2 ) than the electromagnetic component . families observed deeper in the atmosphere ( what means higher @xmath11 ) have smaller @xmath10 , as expected , cf . 4 , where @xmath38 from pamir experiment ( at @xmath42 ) @xcite are presented . here the corresponding parameters @xmath43 are equal to @xmath44 and @xmath45 for gamma and hadronic components , respectively . it is evident from figs . 3 and 4 that the observed gamma - hadron families are already reaching their quasi - equilibrium states . it should be noticed that most of the families registered come not from a single nuclear interaction but they usually contain particles which are decascadents of several ( @xmath46 at @xmath47 tev ) genetically connected nuclear interactions @xcite . our analysis indicates that the average number of cascade generations leading to the observed distributions is about @xmath48 . ( 7.28,7.11 ) fig.4 integral energy spectra for gamma quanta ( circles ) and hadrons ( squares ) in families registered at mt . pamir are fitted by lvy distributions ( solid lines ) . we have provided description of cascade processes encountered in cosmic ray emulsion experiments at high altitudes , which is based on the notion of lvy distrubutions ( which , in turn , originate from the nonextensive statistics described by parameter @xmath0 @xcite ) . in this way we have found that the observed distributions are to a high accuracy power - like , but at the same time we are able to account for the small deviations from the exact power - like behaviour . they are , in our approach , directly connected with the finite number @xmath11 of the cascade generations in the way provided by formulas ( [ eq : par ] ) or ( [ eq : q ] ) . the results presented before should be confronted with the result of pure nuclear origin , which is the case of the so called `` halo '' events with strong concentration of particles in the central region . fig . 5 shows such a case exposing integral energy distribution of shower cores ( recorded at distance @xmath49 cm from the energy - weighted center ) obtained in a `` halo '' event p06 registered at chacaltaya @xcite . in this case we observe almost exponential distribution ( @xmath50 in this case , what translates to @xmath51 ) with no influence of multiplicative processes . finally , it should be noticed that there exists a number of special cases more - or - less directly connected with our results : + @xmath52 energy transport equation @xcite for cascade processes gives @xmath53 as number of particles with energy @xmath54 at depth @xmath55 . notice that for the age parameter @xmath56 one approaches equilibrium spectrum @xmath57 with the slope @xmath13 . + @xmath58 one observes strong intermittent behaviour in the families @xcite , which results from the fluctuations in the atmospheric cascades alone and is not sensitive to details of the elementary interactions . + @xmath59 one observes multifractal behaviour in the cascades with fractal dimension @xmath60 at each cascade stage @xmath11 where @xmath61 @xcite . for each successive emission the distribution should be more inhomoheneous , i.e. , @xmath62 . + @xmath63 self - organized criticality @xcite to which all cascades considered here bear close resemblance because of the zipf s law @xcite , which predicts power - law behaviour of the type seen in eq.([eq : inf ] ) ( with @xmath64 ) . + acknowledgements : the partial support of polish committee for scientific research ( grants 2p03b 011 18 and 621/e-78/spub / cern / p-03/dz4/99 ) is acknowledged . + see , for example , g.wilk and z.wodarczyk , _ nucl . * b * ( proc . suppl . ) * a75 * ( 1999 ) 191 and references therein . j. phys . _ * 29 * no.1 ( 1999 ) ( also at the url : http://sbf.if.usp.br/ www_pages / journals/ bjp / vol29/num1/ index.htm ) for review .	we argue that atmospheric cascades can be regarded as example of the self - organized criticality and studied by using lvy flights and nonextensive approach . it allows us to understand the scale - invariant energy fluctuations inside cascades in a natural way .

Subsets and Splits

No saved queries yet

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