Datasets:
ccdv
/
arxiv-summarization

Dataset card Data Studio Files Community
7
article
stringlengths
0
745k
abstract
stringlengths
0
92.7k
let @xmath0 be a vector bundle and @xmath13 be a very ample line bundle on a variety @xmath14 . we say @xmath0 is an ulrich bundle for @xmath15 , if @xmath16 for each @xmath17 and @xmath18 . this notion was introduced in @xcite , where various other characterizations are given , i.e. , an ulrich bundle implies a linear maximal cohen - macaulay module . ulrich asked in @xcite whether every local cohen - macaulay ring admits a linear mcm - module . correspondingly , it was conjectured in @xcite that whether there exists an ulrich bundle on any variety for any embedding , and what is the smallest possible rank for such ulrich bundles . this conjecture is a wide open problem , and we know a few scattered results at present : @xcite for hypersurfaces and complete intersections , @xcite for curves and del pezzo surfaces , @xcite for abelian surfaces , @xcite for geometrically ruled surfaces of invariant @xmath19 , @xcite for sufficiently general k3 surfaces , @xcite for bielliptic surfaces , @xcite and @xcite for non - special surfaces with @xmath20 ( i.e. enriques surfaces ) . in this paper , we will work on an algebraically closed field @xmath21 of characteristic @xmath22 and @xmath7 will denote the projective space over @xmath21 of dimension @xmath6 . we consider the ulrich bundles on @xmath7 for the veronese embedding @xmath23 . in @xcite , beauville proved that for any positive integers @xmath6 and @xmath24 , there is an ulrich bundle @xmath0 of rank @xmath25 for @xmath2 , using the quotient map @xmath26 of degree @xmath25 . so the existence of ulrich bundles for @xmath2 has been solved completely . but as @xmath6 increases , @xmath25 grows very fast . so we want to know what is the smallest possible rank for such bundles . in section 3 , we will prove that there is a rank @xmath27 ulrich bundle @xmath0 for @xmath28 if and only if @xmath29 or @xmath30 . in @xcite , eisenbud , schreyer and weyman proved that if @xmath0 is an ulrich bundle for @xmath2 , then @xmath31 has natural cohomology for any integer @xmath32 . in section 4 , we will generalize this and prove the following result : * * theorem [ pro - space - lin - cohomology].**if @xmath0 be an ulrich bundle for @xmath28 and @xmath33 , then @xmath34 in particular , @xmath35 has natural cohomology . thus all the @xmath36 are determined by the formula @xmath37 based on this , we prove that every ulrich bundle has a resolution in terms of @xmath6 of the trivial bundle , which is the generalization of corollary 4.3 in @xcite : * * theorem [ pro - space - lin - sequence].**if @xmath0 is an ulrich bundle of rank @xmath1 for @xmath2 , then we have an exact sequence @xmath38 where @xmath39 as a corollary , we can give a necessary and sufficient condition for ulrich bundles if @xmath8 , and reclassify all ulrich bundles for @xmath40 and @xmath41 . in section 5 , we will construct some ulrich bundles and prove some properties using the exact sequence , i.e. , new examples of rank @xmath9 for @xmath10 and of rank @xmath11 for @xmath12 . for convenience , we recall two formulas to calculate the cohomology of bundles on @xmath7 . ( bott formula ) the values of @xmath42 are given by @xmath43 see okonek , schneider and spindler s book @xcite , chapter 1 section 1.1 . [ pro - space - eisenbud ] let @xmath0 be a rank @xmath1 vector bundle on @xmath7 . then @xmath0 is an ulrich bundle for @xmath28 if and only if @xmath44 thus @xmath45 has natural cohomology and all @xmath46 are determined by @xmath47 . see eisenbud , schreyer and weyman s paper @xcite , theorem 5.1 . here , we say a vector bundle @xmath48 on @xmath7 has natural cohomology if for each @xmath49 , at most one of the cohomology groups @xmath50 for @xmath51 . for any real number @xmath52 and positive integers @xmath6 and @xmath53 , we use the formal operation @xmath54 for example , if we assume that @xmath0 is an ulrich bundle for @xmath55 , we have the diagrammatic drawing of the cohomologies as above . since @xmath56 , we can get @xmath57 for @xmath58 and @xmath59 . in general , we have the following useful tip [ pro - space - eisenbud - cor-2 ] let @xmath0 be an ulrich bundle for @xmath28 . if @xmath60 , then @xmath61 for @xmath62 and @xmath63 . it s an immediate consequence of theorem [ pro - space - eisenbud ] . in this section , we construct all rank 1 ulrich bundles and give a criteria of rank 2 ulrich bundles . [ pro - space - rank 1 ] there is a rank 1 ulrich bundle @xmath0 for @xmath28 if and only if @xmath64 , @xmath65 or @xmath66 , @xmath67 . let @xmath68 be a rank 1 ulrich bundle for @xmath28 . if @xmath66 , by definition we have @xmath69 , which implies that @xmath70 . if @xmath71 , there are two ways to calculate the hilbert polynomial @xmath72 comparing the coefficients of @xmath73 and @xmath74 , we obtain the equations @xmath75 if we define @xmath76-\sum_{i=1}^{n}lni - nlnd\;\;(d\geq1),\ ] ] then the derivative of @xmath77 is @xmath78= \sum_{i=1}^{n}\frac{1}{2}[\frac{1}{d-1+\frac{2i}{n+1}}+\frac{1}{d-1+\frac{2(n+1-i)}{n+1}}-\frac{2}{d } ] > 0.\ ] ] so we have @xmath79 and @xmath80 for all @xmath81 , which implies that @xmath82 if and only if @xmath64 . solving the equations ( 1 ) , we obtain @xmath83 and @xmath84 . the converse is trivial by bott formula . coskun and genc have proved the case @xmath85 in @xcite proposition 3.1 . besides , in @xcite beauville claimed that there is a rank 1 ulrich bundle for @xmath15 with @xmath86 if and only if @xmath87 for some @xmath88 , which is different from proposition [ pro - space - rank 1 ] . [ pro - space - rank 2 ] the rank 2 vector bundle @xmath0 is an ulrich bundle for @xmath28 , if and only if it satisfies the following two conditions : 1 . @xmath89 ; 2 . if @xmath90 is even , then @xmath91 ; if @xmath92 is odd , then @xmath91 and @xmath93 . let @xmath0 be a rank 2 ulrich bundle for @xmath28 . if we assume that the topological equivalence class of @xmath0 is @xmath94 , then @xmath95 . there are two ways to calculate the hilbert polynomial @xmath96 comparing the coefficients of @xmath73 , we obtain @xmath89 . condition 2 ) is trivial by the definition of ulrich bundles . now we prove the converse . since @xmath97 , we have @xmath98 , which implies that @xmath99 by condition 1 ) . so we get @xmath100 by serre duality . actually , if we compare the coefficients of @xmath101 for all @xmath32 , we can calculate all chern classes of @xmath0 of any rank @xmath1 . for example , if @xmath102 , we have @xmath103 $ ] . as a corollary , when @xmath104 or @xmath105 for any positive integer @xmath21 , there is no ulrich bundles of rank 2 for @xmath28 , since @xmath106 . eisenbud , schreyer and weyman give a more effective criteria to find the triple @xmath107 such that there is no ulrich bundles of rank @xmath1 for @xmath28 , see @xcite corollary 5.3 . for convenience , we denote the sheaf of germs of holomorphic @xmath108-forms on @xmath7 by @xmath109 . if we dualize the euler sequence and takes the @xmath108th exterior power ( see hirzebruch @xcite , p. 55 ) , we get the following exact sequence of vector bundles @xmath110 now we use the notation @xmath111 . when we tensor the above sequence by @xmath31 , we have a series of exact sequences @xmath112 these sequences give us a way to calculate the cohomology of @xmath35 as follow . [ pro - space - lin - cohomology ] if @xmath0 be an ulrich bundle for @xmath28 and @xmath33 , then @xmath34 in particular , @xmath35 has natural cohomology . thus all the @xmath36 are determined by the formula @xmath37 to prove the first part of this thoerem , we discuss the value of @xmath32 . ( case 1 ) if @xmath113 , where @xmath114 is an integer with @xmath115 . since @xmath116 by theorem [ pro - space - eisenbud ] , we discuss the value of @xmath117 . for any @xmath118 , we have @xmath119 , which implies @xmath120 by the exact sequence @xmath121 . moreover , since @xmath122 by corollary [ pro - space - eisenbud - cor-2 ] , we can repeat the above step and conclude that @xmath123 for any @xmath124 , we have @xmath125 by corollary [ pro - space - eisenbud - cor-2 ] , which implies @xmath126 by the exact sequence @xmath127 . moreover , with the same reason , we have @xmath128 , so we can also repeat the above step and conclude that @xmath129 hence , at this case we have @xmath130 for @xmath131 . ( case 2 ) if @xmath132 . as in case 1 , since @xmath133 , we have @xmath134 . then for any @xmath135 , we can get @xmath136 . hence , at this case we have @xmath130 for @xmath137 . ( case 3 ) if @xmath138 . as in case 1 , since @xmath139 , we have @xmath140 . then for any @xmath141 , we can get @xmath142 . hence , at this case we have @xmath130 for @xmath143 . ( case 4 ) if @xmath144 , where @xmath114 is an integer with @xmath115 . since @xmath145 , we have @xmath146 by the exact sequence @xmath121 . now we only need to calculate @xmath147 . if @xmath148 , then @xmath149 . it turns into case 1 or case 2 , which we have done . if @xmath64 , then @xmath150 . for @xmath151 , we have @xmath152 and @xmath153 for any @xmath117 . for @xmath154 , we have @xmath155 and @xmath156 for any @xmath117 . if @xmath157 , using the same method we have @xmath158 if and only if @xmath159 hence , at all of these cases we have @xmath130 for @xmath131 . now we prove the last part of the theorem by induction . by the exact sequence @xmath121 , we have @xmath160 by theorem [ pro - space - eisenbud ] and induction , we can get @xmath161 so we have done . [ pro - space - lin - cohomology - remark ] if we want to find a sufficient and necessary condition as in theorem [ pro - space - eisenbud ] , we need to determinate all quintuple @xmath162 such that @xmath163 , which is very difficult . but for @xmath64 , we have proved that @xmath164 if @xmath165 and @xmath166 . [ pro - space - lin - sequence ] if @xmath0 is an ulrich bundle of rank @xmath1 for @xmath2 , then we have an exact sequence @xmath167 where @xmath39 by beilinson s theorem ( @xcite , chapter 2 theorem 3.1.4 ) , for any vector bundle @xmath48 on @xmath7 there is a spectral sequence @xmath168 let @xmath169 . by theorem [ pro - space - lin - cohomology ] , we have @xmath170 for @xmath171 and @xmath172 . by the properties of spectral sequence , we have @xmath173 for @xmath174 and @xmath175 , which gives the exact sequence ( [ pro - space - lin - sequence - eq ] ) . at this time , we have @xmath176 so we have done . [ pro - space - lin - sequence - cor ] a rank @xmath1 vector bundle @xmath0 is an ulrich bundle for @xmath40 if and only if @xmath177 . if @xmath177 , then @xmath178 for any @xmath17 and @xmath179 by bott formula . if @xmath0 is an ulrich bundles for @xmath40 , we have an exact sequence @xmath180 by theorem [ pro - space - lin - sequence ] and remark [ pro - space - lin - cohomology - remark ] . [ pro - space - lin - sequence - cor1 ] a rank @xmath1 vector bundle @xmath0 is an ulrich bundle for @xmath41 if and only if @xmath181 . if @xmath181 , then @xmath182 by bott formula . if @xmath0 is an ulrich bundles for @xmath41 , we have an exact sequence @xmath183 by theorem [ pro - space - lin - sequence ] . [ pro - space - lin - sequence - cor2 ] a rank @xmath1 vector bundle @xmath0 is an ulrich bundle for @xmath184 if and only if @xmath0 satisfies the following two conditions : 1 . there is an exact sequence @xmath185 2 . the induced map @xmath186 is injective ( or surjective ) . in particular , if @xmath102 , the condition 2 ) is trivial . if @xmath0 is an ulrich bundles for @xmath184 , we have the condition 1 ) by theorem [ pro - space - lin - sequence ] . the exact sequence implies @xmath187 for all @xmath17 and @xmath188 . condition 2 ) is equivalent to @xmath189 ( or @xmath190 ) . if @xmath102 , the exact sequence implies that @xmath191 . by proposition [ pro - space - rank 2 ] , the condition 2 ) is trivial . [ pro - space - lin - sequence - cor3 ] a rank @xmath1 vector bundle @xmath0 is an ulrich bundle for @xmath192 if and only if @xmath0 satisfies the following two conditions : 1 . there is an exact sequence @xmath193 2 . the induced map @xmath194 is injective and @xmath195 is surjective for @xmath196 . in particular , if @xmath102 , the condition 2 ) is replaced by that @xmath197 is injective . let @xmath48 be the cokernel of @xmath198 . the exact sequence in condition 1 ) is equivalent to two short exact sequence @xmath199 and @xmath200 so the sequences imply that @xmath201 for @xmath202 and all @xmath17 . the injection of @xmath203 is equivalent to @xmath204 , which is equivalent to @xmath205 . the surjection of @xmath206 is equivalent to @xmath207 . if @xmath102 , the exact sequence implies that @xmath208 . by proposition [ pro - space - rank 2 ] , to prove that the cokernel @xmath0 in condition 1 ) is an ulrich bundle , we only need to prove @xmath189 , which is is equivalent to the injection of @xmath209 . for @xmath85 , coskun and genc have found this exact sequence in @xcite . the results in corollary [ pro - space - lin - sequence - cor ] and corollary [ pro - space - lin - sequence - cor1 ] are well known , and there are other proofs using grothendieck s theorem and the splitting criterion of horrocks . by proposition [ pro - space - rank 1 ] , corollary [ pro - space - lin - sequence - cor ] and corollary [ pro - space - lin - sequence - cor1 ] , if we want to find a non - trivial ulrich bundle @xmath0 of rank @xmath1 for @xmath28 , then we must have @xmath210 and @xmath71 . in this section , we will give some interesting examples for @xmath102 or @xmath11 . consider the case @xmath85 and @xmath102 . by corollary [ pro - space - lin - sequence - cor2 ] , finding a rank 2 ulrich bundle for @xmath211 is equivalent to finding a morphism @xmath212 such that the cekernel of @xmath198 is locally free . for example , if we assume that @xmath213 are the homogeneous coordinates of @xmath214 , we can take @xmath215 for any point @xmath216 we have @xmath217 , then the cekernel @xmath218 of @xmath198 must be locally free . hence , @xmath219 is an ulrich bundle of rank @xmath9 for @xmath211 . this example is given by eisenbud , schreyer and weyman in @xcite proposition 5.9 . there are two other proofs for the existence of rank 2 ulrich bundles for @xmath211 given by beauville and casnati , in @xcite proposition 4 and @xcite theorem 1.1 . on one hand , by beauville s result , there exists an ulrich bundle of rank @xmath220 for any @xmath221 . on another hand , by serre construction , there exists a rank 2 ulrich bundle @xmath0 for @xmath221 fitting into the exact sequence @xmath222 where @xmath223 is a general set of @xmath224 points . in @xcite theorem 5.2 and theorem 5.4 , coskun and genc proved that there is a unique ulrich bundle of rank @xmath225 for @xmath226 . as an application of theorem [ pro - space - lin - sequence ] , we give another elementary proof for the case @xmath227 . we have found an ulrich bundle @xmath0 in the above example . let @xmath228 be another ulrich bundle of rank @xmath9 for @xmath226 , which is determinated by the morphism @xmath229 . in general , we can assume that @xmath230 since for any point @xmath216 we have @xmath231 , the determinant of matrix @xmath232 is non zero . hence , @xmath233 is an automorphism of @xmath234 and we have the commutative graph @xmath235 & \o(-2 ) \ar[r]^{\alpha}\ar[d]_{id}^{\sim } & \o^{3}(-1)\ar[r]\ar[d]_{a}^{\sim } & \e(-2 ) \ar[r]\ar[d ] & 0 \\ 0 \ar[r ] & \o(-2 ) \ar[r]^{\alpha ' } & \o^{3}(-1)\ar[r ] & \e'(-2)\ar[r ] & 0 } \ ] ] where @xmath236 . so we have @xmath237 . consider the case @xmath238 and @xmath239 . by corollary [ pro - space - lin - sequence - cor3 ] , finding a rank 2 ulrich bundle for @xmath10 is equivalent to finding two morphisms @xmath240 and @xmath241 , such that @xmath242 and the cekernel of @xmath243 is locally free and @xmath244 is injective . for example , if we assume that @xmath245 are the homogeneous coordinates of @xmath246 , we can take @xmath247 since @xmath248 and for any point @xmath249 the rank of @xmath250 is 3 , we have @xmath242 and the cekernel of @xmath243 is locally free . let @xmath48 be the cokernel of @xmath198 , then @xmath251 by definition . since @xmath252 by bott formula , we can get that @xmath244 is injective . hence , the cekernel @xmath218 of @xmath243 is locally free , and @xmath253 is an ulrich bundle of rank @xmath9 for @xmath10 . in @xcite proposition 5.11 , eisenbud , schreyer and weyman give another example for this case . let @xmath48 be the null correlation bundle on @xmath246 , which is defined by the exact sequence @xmath254 then @xmath255 is an rank 2 ulrich bundle for @xmath10 . moreover , they proved the uniqueness of rank 2 ulrich bundles for @xmath10 . similarly , as an application of theorem [ pro - space - lin - sequence ] , we give another proof for this uniqueness . we have found an ulrich bundle @xmath0 in the above example . let @xmath228 be another ulrich bundle of rank @xmath9 for @xmath256 , which is determinated by the morphism @xmath257 and @xmath258 . if we assume that @xmath259 is the cokernel of @xmath260 , as the proof of proposition [ pro - space - lin - n = d = r=2 ] , we have @xmath261 . without loss of generality , we can assume that @xmath262 . now we have the commutative graph @xmath263 & 0 \ar[d ] & \\ & & \o\ar[r]^{\sim}\ar[d ] & \o \ar[d]^{s ' } & \\ 0 \ar[r ] & \omega^2(2 ) \ar[r]\ar[d]^{\sim } & \o^{6}\ar[r]\ar[d]^{\pi } & \omega(2 ) \ar[r]\ar[d ] & 0 \\ 0 \ar[r ] & \mathcal{t}_{p^3}(-2 ) \ar[r ] & \o^{5}\ar[r]\ar[d ] & \e'(-1)\ar[r]\ar[d ] & 0 \\ & & 0 & 0 & } \ ] ] which implies that @xmath228 is determinated by a section @xmath264 . since @xmath265 is locally free , @xmath266 is base point free . if we consider the following short exact sequence @xmath267 then @xmath268 . let @xmath269 be the homogeneous coordinates of @xmath246 , we can assume that @xmath270 where @xmath271 . similarly , if @xmath0 is determinated by a section @xmath272 , we also have @xmath273 with @xmath274 . let @xmath275 . we can get the commutative graph @xmath235 & \o \ar[r]^{s}\ar[d]_{id}^{\sim } & \omega(2)\ar[r]\ar[d]_{b}^{\sim } & \e(-1 ) \ar[r]\ar[d ] & 0 \\ 0 \ar[r ] & \o \ar[r]^{s ' } & \omega(2)\ar[r ] & \e'(-1)\ar[r ] & 0 } \ ] ] so we have @xmath237 . consider the case @xmath276 and @xmath277 . by corollary [ pro - space - lin - sequence - cor2 ] , finding a rank 3 ulrich bundle for @xmath12 is equivalent to finding a morphism @xmath278 such that the cekernel of @xmath198 is locally free and @xmath279 is surjective . by serre duality , the surjection of @xmath280 is equivalent to the injection of @xmath281 . if we assume that @xmath282 are the homogeneous coordinates of @xmath214 , and @xmath283 we have the relation of equivalence @xmath284 @xmath285 let @xmath286 then the @xmath287 matrix of coefficients of the equations ( 3 ) of @xmath288 is @xmath289 so @xmath290 is injective if and only if @xmath291 . hence , in order to find an ulrich bundle of rank 2 for @xmath12 , it s enough to find 6 matrixes @xmath292 such that @xmath291 . for example , if we assume that @xmath293 then @xmath294 . let @xmath218 be the cekernel of @xmath198 , we have that @xmath295 is an ulrich bundle of rank @xmath11 for @xmath12 . in @xcite , coskun and genc conjectured that for all odd integer @xmath24 there exists a rank 3 ulrich bundle for @xmath184 . with the aid of computer technology , people know that it is correct when @xmath24 is very small , but no concrete example is given . here , we give the first example for the case @xmath277 . * * acknowledgement.**i am heavily indebted to my supervisor professor jinxing cai for his patient guidance and helping me out of the difficulties during the hard time of this paper . i also thank zhaonan li , dong zhang and xufeng guo for their assistance of computer technology .
we assume that @xmath0 is a rank @xmath1 ulrich bundle for @xmath2 . the main result of this paper is that @xmath3 has natural cohomology for any integers @xmath4 and @xmath5 , and every ulrich bundle @xmath0 has a resolution in terms of @xmath6 of the trivial bundle over @xmath7 . as a corollary , we can give a necessary and sufficient condition for ulrich bundles if @xmath8 , which can be used to find some new examples , i.e. , rank @xmath9 bundles for @xmath10 and rank @xmath11 bundles for @xmath12 .
at first glance , the study of the very first stars in the universe might appear to be a rather academic and even quixotic endeavor . after all , we have never directly seen such stars . since we believe that heavy elements are synthesized almost exclusively in stars , the very first `` population iii '' stars were , by definition , made of zero metal gas . intensive searches for metal poor stars in the halo galaxy , however , have only turned up stars with metallicities @xmath0 ( beers 2000 ) . furthermore , the formation of a zero metal star is in some sense a rather singular event . once that star goes supernova and pollutes its environment with metals , the stars that subsequently form in its vicinity will no longer be population iii ( with zero metallicity ) . population iii may also be suicidal in the sense that uv radiation from the first stars could destroy the molecular hydrogen that allows primordial gas to cool and form these stars in the first place . there has therefore been considerable speculation , e.g. , cayrel ( 1986 ) and haiman , rees , & loeb ( 1997 ) , that the population iii ( pop . iii ) phase in the evolution of our universe was very brief indeed . nonetheless , while the pop . iii phase may have been a brief one , it seems to have been a pervasive one . even in the least evolved regions of the universe that we can probe today , the high redshift lyman @xmath1 clouds , we still find evidence for a non - zero metallicity @xmath2 ( cowie & songaila 1998 ) . more importantly , the epoch of the pop . iii stars probably represents the first substantial input of energy , photons , and metals into the universe since the big bang and marks the end of the `` dark ages '' ( e.g. , loeb 1999 , rees 1999 ) that started when the cosmic background light redshifted out of the visible range ( at @xmath3 ) . because structure formation depends critically on the ability of baryons to cool , and this in turn depends critically on the metallicity and ionization state of the baryonic gas , the `` feedback '' effects of the first stars on the intergalactic medium ( igm ) in fact play a subtle but key role in determining the subsequent evolution of the universe . the significant interest in primordial star formation at this conference is therefore not an accident . large - scale cosmological simulations ( e.g. , ostriker & gnedin 1996 ; abel et al . 1998 ; fuller & couchman 2000 ) are just now becoming good enough to resolve the @xmath4 scales of the first objects in the universe that can cool and collapse ( tegmark et al . 1997 ) . understanding the star formation that will occur in these objects and accurately incorporating its effects into cosmological codes therefore represents the next major challenge in our quest to integrate forwards from the big bang . the interest in population iii , however , is not purely a theoretical one . studies of cosmic microwave background ( cmb ) anisotropies are now providing us with information on the physical conditions at @xmath5 when the universe became transparent to cmb radiation , while observations of high - redshift quasars and galaxies tell us about the universe at redshifts @xmath6 below . current observations , however , give us little information on the `` dark ages , '' the crucial epoch in between . reaching into this epoch is thus our next observational challenge and is the primary motivation , for example , of the _ next generation space telescope _ ( ngst ) which will provide unprecedented sensitivity at near - infrared wavelengths ( loeb 1998 ) . the study of the first stars is thus timely , providing a theoretical framework for the interpretation of what ngst might discover , less than a decade from now . even if ngst does not directly image the first stars , it will probe the epoch of the reionization of the igm ( e.g. , barkana , this proceedings ) . uv photons from the first stars , perhaps together with an early population of quasars , may have contributed significantly to this reionization ( e.g. , see ciardi et al . 2000 ; miralda - escud , haehnelt , & rees 2000 ) . if the reionization occurred early enough , cmb fluctuations on scales @xmath7 will be damped by electron scattering ( e.g , . haiman & loeb 1997 ) and this effect could be detected by other next generation instruments like map and planck . the energy input from the first stars may also have left a small but measurable imprint on the cmb on very small scales ( e.g. , see the contributions here by sugiyama et al . and bruscoli et al . ) . in sum , the implications of pop.iii star formation might be testable in the not too distant future ! before that day arrives , however , a skeptical observer might still question whether concrete progress can actually be made in understanding primordial star formation . in the case of present - day star formation , we can not predict the initial mass function from first principles despite the wealth of observational data available . how could we hope to do something like this for unseen primordial stars ? a scan of the considerable early literature on primordial star formation ( e.g. , yoneyama 1972 ; hutchins 1976 ; silk 1977 , 1983 ; carlberg 1981 ; kashlinsky & rees 1983 ; palla , salpeter , & stahler 1983 ; carr , bond , & arnett 1984 ; couchman & rees 1986 ; uehara et al . 1996 , haiman , thoul , & loeb 1996 ; omukai & nishi 1998 ) would tend to support this conclusion . the range of mass estimates for the first stars spans spans six ( ! ) decades , from @xmath8 to @xmath9 there are reasons for hope , however . first , the physics of the first stars is considerably simpler than that of present - day star formation ( larson 1998 ; loeb 1998 ) . the present - day interstellar medium is an exceedingly complex environment , but primordial gas initially has no metals , no dust grains and no cosmic rays to complicate the gas cooling function . since we think we know the primordial abundances well and the number of relevant species is small , the gas chemistry and cooling function are relatively simple and have been extensively studied ( e.g. , see galli & palla 1998 ) . also , because there are no stars yet , the only relevant external radiation is the cosmic background whose behavior we also think is well - understood . additionally , if one believes that galactic strength magnetic fields resulted from dynamo action perhaps enhanced by compression ( e.g. , kulsrud 1997 ) , then magnetic fields at early times are likely to be dynamically insignificant . finally , before the first supernovae went off , the early igm must also have been a rather quiescent place , with no sources to sustain turbulent motion . the remaining consideration in understanding how primordial gas collapses to form stars is knowledge of the typical initial conditions : the initial gas density and temperature profile , the gas angular momentum distribution , the density and velocity distribution of the dark matter halos containing the gas , and the underlying cosmology . estimating these initial conditions for a specific cosmological scenario is no longer a problem given the current state of simulations . in sum , at least during the initial phases of primordial star formation , we have a well - posed problem where the relevant physics is in hand . secondly , computers can now follow the inherently three dimensional process of primordial gas fragmentation and collapse . this is critical because it is not immediately obvious when gas fragmentation in primordial clouds halts . to appreciate the difficulties , note that since gas can cool and increase its density arbitrarily ( at least until an opacity limit sets in ) , the jeans mass for a collapsing gas cloud , i.e. , the scale below which fragmentation halts , can become extremely small . such behavior is indeed seen in one dimensional simulations of isothermal filament collapse . one might therefore predict primordial stars to have very low masses . however , if the initial density perturbations in the cloud are not very large or the cloud has a very strong central density concentration , the cloud can collapse into a single object before the perturbations have time to grow and fragment the cloud ( e.g. , tohline 1980 ) . thus , depending on exact initial conditions , one could also predict that the first objects to turn around and cool will collapse directly into very massive stars or black holes ( the so - called `` vmos '' or very massive objects ) . furthermore , one can not straightforwardly apply the intuition on fragmentation developed in the more extensive studies of present - day star formation . in the present - day case , gas cooling is very efficient and one typically takes the collapsing gas to be isothermal . molecular hydrogen is a very poor coolant , however , and the timescale for zero - metal gas to cool can often be comparable to or longer than the dynamical timescale for the gas to collapse . this has profound consequences , as we show next . the results shown are the thesis work of volker bromm . ( see bromm , coppi , & larson 1999 for a more extended discussion of the calculation . ) our goal here is to follow the gravitational fragmentation of a cloud to see if there is indeed a characteristic mass scale at which fragmentation stops and gravitational collapse proceeds unhindered . this `` clump '' mass scale and the overall spectrum of runaway clump masses that we find , of course , can not be directly translated into a stellar mass scale or imf , but it is an important first step . in the case of present - day star formation , at least , there is increasing evidence that the two may in fact be closely related . our calculational approach is intermediate to that of the other two primordial gas collapse calculations shown at this meeting . the first of these ( see contribution by nakamura & umemura ) uses a high resolution 2-d mesh code to follow the evolution of an idealized primordial gas cloud for many different initial conditions and perturbations . the second ( see contribution by abel et al . ) is a full 3-d calculation that starts from a large scale cosmological simulation and uses the adaptive mesh refinement ( amr ) technique to zoom in on the evolution of the first object to undergo collapse in their simulation volume . only a few realizations of the cosmological initial conditions have been explored . in an attempt to increase the number of realizations we can explore , we instead follow the cosmological evolution of a tophat density perturbation with parameters close to those expected for the first objects that can collapse ( @xmath10 e.g. , see tegmark et al . 1997 ) . we use a 3-d particle code based on the treesph code of hernquist & katz ( 1990 ) that incorporates the full primordial chemistry of galli & palla ( 1998 ) , including the effects of hd cooling . the sph technique does not handle shocks as well as mesh techniques , but strong shocks are not important in the regime considered here and sph is simple and flexible . for example , it easy to turn a group of gas particles into a single `` sink '' particle without having to worry about mesh artifact / resampling issues . this is useful for allowing a simulation to continue beyond the runaway collapse of the first clump ( which ordinarily would halt the calculation because of the courant limit ) . it is also easy to increase our spatial resolution in a desired region ( i.e. , perform a `` poor man s '' version of amr ) by tagging the particles that enter that region , and then restarting the calculation with each of the tagged particles replaced by many lower mass particles , e.g. , see fig . 7 . 1 - 4 show results one of our typical top - hat collapse/ fragmentation calculations . at @xmath11 we endow a spherical , uniform density halo of total mass @xmath12 ( baryonic plus dark matter ) with a hubble expansion such that virialization occurs at @xmath13 the dark matter is perturbed with a @xmath14 power spectrum expected from cdm on small scales . the baryons are uniformly distributed and have a mass fraction @xmath15 both halo components are initially in solid body rotation about the @xmath16 axis , with angular momentum corresponding to a cosmological spin parameter @xmath17 these are typical parameters for the first objects ( @xmath18 density fluctuation ) that turn around and are massive enough to cool in a hubble time ( e.g. , see tegmark et al . the dark matter initially plays a key role as the baryons fall into the potential wells of the growing small - scale dark matter perturbations ( fig 1 ) . eventually , the dark matter undergoes violent relaxation and starts to lose its substructure . the baryons sink into the center of the overall dark matter potential well and start to fragment ( fig . 2 ) . in fig . 3 , we plot the properties of the gas particles at @xmath19 note the `` pile up '' of particles at density @xmath20 and temperature @xmath21 k , corresponding to a jeans mass @xmath22 the pile up reflects the fact that gas undergoing collapse `` loiters '' at these values ( see the time history in fig . 4 , and discussion below ) . we have carried out many other runs , varying quantities like the total angular momentum of the cloud , the slope of the dark matter perturbation spectrum , the degree to which the mass is centrally concentrated ( the standard top - hat assumes a uniform density distribution , which is optimal in terms of producing many fragments but may not always be realistic ) , the baryon mass fraction ( @xmath23 ) , and the total mass and turnaround redshift of the cloud . we find two main results . first , in terms of the morphology of the collapsed gas and the overall `` efficiency '' of fragmentation ( the fraction of gas that ends up in clumps ) , we find that varying the initial conditions of the cloud _ does _ make a significant difference , e.g. , compare the gas morphology in fig . 2 with that in fig . 6 . similar dependences , e.g. , on the cloud s angular momentum and degree of central mass concentration , are in fact found in gas simulations of present - day star formation ( e.g. , tsuribe & inutsuka 1999 ) . note that this dependence on initial conditions means it is _ not _ possible to make statements about the overall efficiency of primordial star formation without first carrying out a comprehensive survey of the relevant conditions . second , despite the differences in gas morphology , we always find find roughly the _ same _ initial clump masses . here , initial clump mass is defined as the amount of gas that is gravitationally bound and infalling when the center of a clump starts it runaway collapse , i.e. , it does not include any further gas that may eventually accrete onto the clump . the reason for this perhaps surprising second conclusion can be found in fig . if we plot the temperatures and densities of our gas particles when the first clumps start to collapse , we always find an excess of particles with temperatures @xmath24 200 k and hydrogen densities @xmath25 @xmath26 these two numbers are not accidental and are set molecular hydrogen physics which does _ not _ depend on the initial conditions . specifically , a temperature @xmath27 k is the minimum one attainable via h@xmath28 cooling because of the molecular energy levels . the corresponding critical density , beyond which the h@xmath28 rotational levels are populated according to lte , is then @xmath29 @xmath30 . at the transition from nlte to lte , the cooling rate changes from being proportional to @xmath31 to merely linear in @xmath32 i.e. , the cooling time required for the gas to lose a significant fraction of its energy now becomes independent of density . due to this inefficient cooling , the gas ` loiters ' and passes through a phase of quasi - hydrostatic , slow contraction before undergoing runaway collapse ( see fig . this loitering appears to be crucial as it allows pressure waves to damp out density anisotropies and inhibits further fragmentation . although our results are still somewhat preliminary , we have carried out higher resolutions runs ( e.g. , fig . 7,8 ) to follow the collapse of a clump to much higher densities , and we indeed see no evidence for sub - fragmentation . abel et al . have reached the same conclusion in the even higher resolution runs that they have carried out . although we can not guarantee that some of our clumps will not break up into a few objects , e.g. , a binary system , it seems unlikely they will break up into hundreds or thousands of subclumps . in other words , to astrophysical accuracy , the jeans mass @xmath33 that follows from the typical density and temperature values in fig . 3 really is the characteristic clump mass scale for collapsing primordial gas . the fact that three groups at this conference arrived at the same conclusion using rather different codes and initial conditions tells us that a robust explanation , like the physics of molecular hydrogen , must lie behind it . although we are still far from solving the primordial star formation problem , the results presented at this meeting indicate we have made substantial progress . unless our understanding of primordial gas cooling is very wrong ( note that our simulations included hd cooling which some have speculated to be important ) or the typical physical conditions during the early dark ages are very different from current expectations , it appears inescapable that the first typical objects to collapse will fragment into clumps of initial mass @xmath22 it also appears likely that these clumps will not fragment much further as their interiors collapse to form a star or a few stars . therefore , typical primordial protostars are likely to look quite different from those around us today , and in particular , will be surrounded by much more massive @xmath34 envelopes . as noted at this meeting , it is not at all obvious how much of this mass actually makes it onto the final star . however , given that all the scales are so much larger and resemble those we find around present - day massive stars in the process of being born , it is difficult to see how one can make ordinary solar mass stars from such a gas configuration , i.e. , primordial star formation is probably strongly biased towards massive ( @xmath35 ) and possibly very massive ( @xmath36 ) stars . this would explain why we see no zero - metal stars today and has important consequences that have not been fully explored yet , e.g. , massive primordial stars produce many more ionizing uv photons per unit mass than low mass ones ( see bromm , kudritski , & loeb 2000 for a detailed calculation of the spectrum from a massive zero metal star).also , such massive stars could be good progenitors for hypernovae and gamma - ray bursts , or the seeds for massive black hole formation . there remain important questions that do not require a major leap in computing power to answer . first , at this meeting it became clear we need to deide what are typical , realistic initial conditions , e.g , nakamura & umemura pointed out that it is possible to fragment down to @xmath37 if one can start out with dense enough filaments ( @xmath38 we agree since this would skip the `` loitering '' phase of the collapse , but we do not see how such filaments arise in a realistic scenario . quantifying the relevant initial conditions for primordial gas collapse will let us determine the efficiency of clump formation , which in turn gives us an upper limit on the primordial star formation efficiency and provides a first indication as to the importance of the first stars . secondly , we can consider what happens to gas collapse and fragmentation in the presence of trace metals and uv background radiation from a previous generation of stars . population iii star formation is often considered to be a short - lived event because the pristine conditions required are wiped out once the first stars produce uv light and the first supernovae produce metals . however , as metallicity does not build up instantly , there may well be an extended window of time when star formation either proceeds in the massive clump / star mode described here ( if h2 is present ) or not at all ( if h2 is destroyed by uv radation ) . our preliminary calculations indicate that metal cooling does not become important until the gas metallicity reaches @xmath39 which coincidentally is the range of the lowest observed metallicities and also the range where abundance anomalies begin to appear in metal poor stars . ( these anomalies are often interpreted as increased scatter due to enrichment by individual supernova events , but they could also reflect atypical progenitor stars , e.g. , that had much hotter interiors than stars today . ) finally , it should be possible to push the spherically symmetric protostar calculation of omukai and nishi ( 1998 ) through to the accretion phase ( along the lines of masunaga , miyama , and inutskuka 1999 in the present - day star formation case ) . this will enable a first cut at understanding the feedback of the primordial protostar s radiation on its envelope . the feedback is likely to be strong , and abel conjectured at this meeting , for example , that all accretion may stop once the protostar reaches @xmath40 and produces enough ionizing radiation to destroy the envelope s molecular hydrogen , thereby removing its primary means of cooling . without a real calculation , however , it is not clear what the outcome will be . if the envelope gas simply becomes adiabatic , accretion can still occur if a sufficient central mass concentration has already been established ( bondi 1952 ) . abel , t. , anninos , p. , norman , m. l. , & zhang , y. 1998 , , 508 , 518 beers , t. c. 2000 , in `` the first stars '' , eso astrophysics symposia , ed . a. weiss , t. abel , & v. hill ( berlin : springer ) , 3 bromm , v. , coppi , p. s. , & larson , r. b. 1999 , , 527 , l5 bromm , v. , kudritski , r.p . , & loeb , a. , astro - ph/0007248 carlberg , r. g. 1981 , mnras , 197 , 1021 carr , b.j . , bond , j.r . , & arnett , w.d . , 1984 , , 280 , 825 cayrel , r. , 1986 , a&a , 168 , 81 ciardi , b. , ferrara , a. , governato , f. , & jenkins , a. 2000 , mnras , 314 , 611 couchman , h. m. p. , & rees , m. j. 1986 , mnras , 221 , 53 cowie , l. l. , & songaila , a. 1998 , nature , 394 , 44 fuller , t. m. , & couchman , h.m.p . 2000 , , submitted ( astro - ph/0003079 ) galli , d. , & palla , f. 1998 , a&a , 335 , 403 haiman , z. , thoul , a. a. , & loeb , a. 1996 , , 464 , 523 haiman , z. , rees , m.j . , & loeb , a. , 1997 , apj , 476 , 458 haiman , z. , & loeb , a. 1997 , , 483 , 21 hutchins , j. b. 1976 , , 205 , 103 kashlinsky , a. , & rees , m. j. 1983 , mnras , 205 , 955 kulsrud , r. m. 1997 , in `` critical dialogues in cosmology '' , ed . n. turok ( singapore : world scientific ) , 328 larson , r. b. 1998 , mnras , 301 , 569 loeb , a. 1998 , in asp conf . 133 , science with the next generation space telescope , ed . e. smith & a. koratkar ( san francisco : asp ) , 73 loeb , a. 1999 , astro - ph/9907187 miralda - escud , j. , haehnelt , m. , & rees , m. j. 2000 , , 530 , 1 omukai , k. , & nishi , r. 1998 , , 508 , 141 ostriker , j. p. , & gnedin , n. y. 1996 , , 472 , l63 palla , f. , salpeter , e. e. , & stahler , s. w. 1983 , , 271 , 632 rees , m. j. 1999 , aip conf . 470 , 13 silk , j. 1977 , , 211 , 638 silk , j. 1983 , mnras , 205 , 705 tegmark , m. , silk , j. , rees , m. j. , blanchard , a. , abel , t. , & palla , f. 1997 , , 474 , 1 tohline , j.e . , 1980 , apj , 242 , 209 tsuribe , t. & inutsuka , s. , 1999 , apj , 526 , 307 uehara , h. , susa , h. , nishi , r. , yamada , m. , & nakamura , t. 1996 , , 473 , l95 yoneyama , t. 1972 , pasj , 24 , 87
we briefly review the motivations for studying the formation of the first `` population iii '' stars and present recent results from our numerical simulations in this area . we discuss the new questions raised as a result of the simulations presented by us and others at this meeting . # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
useful discussions with immanuel bloch , pranjal bordia , andrew daley , markus heyl , henrik lueschen , evert van nieuwenburg , ulrich schneider , and lukas m. sieberer are gratefully acknowledged . we thank the erc for financial support through the uquam synergy grant . mhf acknowledges additional support from the swiss friends of the weizmann institute of science . 30ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop ( ) , @noop ( ) , @noop * * , ( ) link:\doibase 10.1103/physrevlett.114.117401 [ * * , ( ) ] @noop ( ) , @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevx.5.041047 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , classics in mathematics ( , ) link:\doibase 10.1103/physreva.82.063605 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) , @noop * * , ( ) @noop _ _ ( , ) link:\doibase 10.1103/physrevx.5.031032 [ * * , ( ) ] link:\doibase 10.1103/physrevx.5.031033 [ * * , ( ) ] in this section we provide a simple explanation for the observed stretched exponential relaxation of the imbalance . as a first step we derive an equation of motion that describes the spatial variations in the relaxation of the imbalance . as a starting point we take equation ( 4 ) of the main text for the local densities @xmath37 . deep in the localized phase we can take @xmath108 to connect only nearest neighbors , i.e. @xmath109 . we now replace the local density @xmath37 that has a dominant period two modulation on the lattice by the variable @xmath110 defined through _ = n_0 + e^ix _ f(x _ ) . physically , @xmath111 is the local amplitude of the period-2 charge density wave . the local normalized imbalance is @xmath112 . because @xmath111 is slowly varying in space , it facilitates a continuum limit of the rate equation = -(x)f-14(x)_x^2 f+14_x(x ) _ x f , where @xmath113 . neglecting the gradients of the smooth function @xmath114 we find @xmath115 $ ] . the relaxation of the imbalance then follows from ( t)=_0l_0^l dx e^-(x ) t. we have seen in the main text that in the strongly localized state the local relaxation rate @xmath58 depends on the energy offsets @xmath62 between neighboring sites as @xmath116 ^ 2 $ ] . after coarse graining we expect the energy offsets to follow a gaussian distribution p()=1e^-^22 ^2 , where @xmath117 is the variance of @xmath118 , the offset between two adjacent sites , and @xmath50 is the on - site disorder strength . note that due to the coarse graining , we expect @xmath119 , the value corresponding to the variance of the probability distribution of the local energy differences . putting all this together we can convert the integral over @xmath61 to one over @xmath62 to obtain ( t)=_0de^-^22_0 ^ 2h^2 - 4j^2 ^ 2 t . finally , a saddle - point evaluation of this integral yields ( t)~e^-=e^- , where @xmath120 is the scaled time defined in the main text and @xmath121 . for the numerical result of @xmath56 , we would require @xmath122 which corresponds to coarse graining over order four lattice sites . to emphasize how the type of disorder influences the form of the decay of conserved quantities when coupled to a bath , we compare the non - interacting case of dephasing for the random disorder to the quasi - periodic disorder of the aubry - andr model @xcite , @xmath123 where @xmath124 with @xmath125 and @xmath126 a random phase . as shown in fig . [ fig : comp ] , this type of disorder not only localizes more strongly for @xmath127 , but also changes the exponent @xmath128 and time scale @xmath129 to @xmath130 and @xmath131 . and @xmath28 . ( a ) the decay of the charge density wave as a function of @xmath132 , and ( b ) @xmath133 vs. @xmath134 in log - log scale to emphasize the different stretched exponential form . ] to lowest order in @xmath97 , we can trivially invert eq . ( 13 ) of the main text to express the jump operators in terms of the operators for the local integrals of motion . first , @xmath135 which we use to find the jump operators @xmath136 we therefore obtain the classical rate equations @xmath137 \nonumber\\ & & + \gamma \sum_{\beta\mu}\left[(w_{\alpha\beta}^{\mu\mu}+w^{\alpha\beta}_{\mu\mu})\tilde{n}_{\mu}\tilde{n}_{\beta}(1-\tilde{n}_{\alpha})-(w_{\beta\alpha}^{\mu\mu}+w^{\beta\alpha}_{\mu\mu})\tilde{n}_{\mu}\tilde{n}_{\alpha}(1-\tilde{n}_{\beta})\right]\nonumber\\ & & + \gamma \sum_{\beta\mu\nu}\left[(w_{\alpha\beta}^{\mu\nu}+w_{\beta\alpha}^{\mu\nu})(1-\tilde{n}_{\alpha})(1-\tilde{n}_{\beta } ) \tilde{n}_{\mu}\tilde{n}_{\nu } - 2w^{\alpha\beta}_{\mu\nu}(1-\tilde{n}_{\mu})(1-\tilde{n}_{\nu } ) \tilde{n}_{\alpha}\tilde{n}_{\beta}\right ] . \label{eq : full - eom - intdeph}\end{aligned}\ ] ] with @xmath138 and @xmath139 note that the first term of the right - hand side of eq . is identical to the rate equation ( 13 ) for the non - interacting case , which already leads to diffusion and a loss of the imbalance . for @xmath140 , the next - order contributions come from the second line which describes an ` assisted hopping ' , leading to a correction @xmath141 , such that the dissipation induced hopping rates are of the form @xmath89 , with @xmath90 a numerical constant of order 1 . indeed , the functional form stays essentially the same all the way up to @xmath142 , with the same parameters , as shown in fig . [ fig : betas ] . and ( b ) @xmath143 . ( c ) the stretched exponential @xmath128 for the decay of the imbalance @xmath144 $ ] for @xmath127 and @xmath145 disorder realizations for various @xmath97 . the inset shows the corresponding time scale @xmath129 . ] 1ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( )
coupling a many - body - localized system to a dissipative bath necessarily leads to delocalization . here , we investigate the nature of the ensuing relaxation dynamics and the information it holds on the many - body - localized state . we formulate the relevant lindblad equation in terms of the local integrals of motion of the underlying localized hamiltonian . this allows to map the quantum evolution deep in the localized state to tractable classical rate equations . we consider two different types of dissipation relevant to systems of ultra - cold atoms : dephasing due to inelastic scattering on the lattice lasers and particle loss . our approach allows us to characterize their different effects in the limiting cases of weak and strong interactions . _ introduction _ a closed quantum system can fail to reach thermal equilibrium when evolving under its own intrinsic dynamics if it is ( many - body ) localized by strong quenched disorder @xcite . direct evidence for such breaking of ergodicity in an interacting many - body system has recently been observed in a system of ultra - cold atomic fermions @xcite , as well as a chain of trapped ions @xcite . however , even these highly controlled implementations of many - body - localized ( mbl ) systems are affected by at least small coupling to the environment , for example through inelastic scattering on the lasers , that ultimately destroys the localization . therefore , in order to study the phenomenon of many - body localization in a realistic setting it is important to address the effect of dissipation on such systems and understand the interplay with intrinsic localization properties of the many - body system . here , we develop a general framework to investigate the dynamics of a mbl system weakly coupled to a dissipative markovian bath . our approach is based on the local integrals of motion of the closed localized system . using our general framework we discuss how observables relax in the presence of different types of dissipative baths . we note that earlier studies discussed the effects of a coupling to a thermal bath on the spectral properties of a mbl system @xcite . a particular example of dynamical relaxation of observables in a mbl system coupled to a bath has also been addressed recently in a numerical study @xcite , where a slow stretched exponential decay of these observables was found in a wide regime . we explain the origin of this behavior and address how the relaxation properties depend on the interactions and on the type of bath affecting the system . _ general scheme _ for concreteness , we exemplify our general approach using a one - dimensional model of spinless lattice fermions with random on - site energies and nearest - neighbor interactions , @xmath0 here , @xmath1 creates a fermion on site @xmath2 , @xmath3 , and the on - site energies @xmath4 $ ] are independent random variables . as a model of a closed system , this hamiltonian has been extensively studied and ( for @xmath5 ) is expected to be localized for @xmath6 @xcite . the time evolution of the density matrix of the system coupled to a markovian bath is given by the lindblad equation @xmath7 = -i[h , \rho ] + \gamma \sum_{i}\big(l^{{\phantom{\dag}}}_{i}\rho l_{i}^\dag - \frac12\{l_{i}^\dag l_{i}^{{\phantom{\dag } } } , \rho\}\big ) , \label{eq : lindblad}\ ] ] where the first term governs the unitary time evolution while the second term describes the coupling of the system to the environment after tracing out the bath degrees of freedom . @xmath8 are called jump operators , the system operators directly coupled to the bath . within a stochastic quantum - trajectory picture the unitary evolution of an ensemble of pure states is interrupted at random times by non - unitary quantum measurements ( or jumps ) generated by the operators @xmath8 @xcite . a full solution of the lindblad dynamics is a daunting quantum problem . however , we are able to greatly simplify it in the ( fully ) mbl phase , where the hamiltonian can be represented in terms of a complete set of quasi - local integrals of motion ( lioms ) @xcite . the corresponding operators @xmath9 are in one - to - one correspondence and have a finite overlap with the truly local physical operators @xmath10 appearing in the original hamiltonian . we utilize the fact that the unitary evolution deep in the mbl state is trivial in the liom basis . for physical initial conditions ( i.e. local in the physical basis @xmath10 ) the off - diagonal elements of the density matrix in this basis dephase rapidly . hence , after a microscopic time scale ( of order @xmath11 ) the density matrix becomes effectively diagonal . within the quantum - trajectory picture , the jump operators @xmath8 act after random time intervals set by @xmath12 and project to a new state @xmath13 with off - diagonal elements in the liom basis . however , for sufficiently weak coupling ( @xmath14 ) , the density matrix has enough time to dephase between successive quantum jumps as local operators @xmath8 do not create resonances deep in the localized phase . we can thus view the density matrix as a classical ( joint ) probability distribution @xmath15 at all times . the time evolution of the probability distribution is given by a classical master equation , where the quantum jumps generate the transition rates between classical configurations . hence , the change of the probability distribution of some configuration @xmath16 is just given by the transitions in and out of it , @xmath17 , \label{eq : classicalrate}\ ] ] where @xmath16 and @xmath18 represent classical configurations @xmath19 and @xmath20 is the transition rate from @xmath18 to @xmath16 . these rates are obtained by writing the jump operators in terms of the lioms . note that for the problem of spinless fermions the classical master equation then describes a symmetric exclusion process @xcite . for most of the conditions we consider , namely high temperature and/or weak interactions , a simple mean - field solution of the classical problem is justified and , as we show , gives an excellent approximation of the full quantum dynamics . motivated by the recent experiment of ref . @xcite we initialize the system in a state with a period - two density wave and track the relaxation of the density imbalance @xmath21 with @xmath22 ( @xmath23 ) the occupation of even ( odd ) sites . without coupling to a bath , @xmath24 approaches a non - vanishing value in the mbl phase , while it decays completely on a short timescale @xmath25 in the ergodic phase . we then consider two types of dissipative mechanisms that would ensue due to inelastic scattering of the atoms on the laser fields in the above mentioned experiment . the first mechanism is recoil and dephasing of the atom wavefunction implemented by a local measurement of the on - site density , @xmath26 @xcite . the second one is particle loss , which is implemented by the jump operators @xmath27 . _ non - interacting system _ having described the general approach we now demonstrate it for the case of a non - interacting system [ @xmath28 in eq . ] . the lioms in this case are simply the occupation numbers of the localized single - particle states @xmath29 , where @xmath30 creates a fermion in the localized state @xmath31 . the indices @xmath32 are in one - to - one correspondences with the site indices @xmath2 , so that @xmath33 are the single - particle wave functions localized near sites @xmath31 . we follow the procedure outlined above considering first the effect of a dephasing bath , i.e. , @xmath34 . these jump operators are written in terms of the `` quasi - particle '' operators @xmath35 as @xmath36 . this representation clarifies that the jump operators induce transitions between the single - particle localized states . the unitary evolution on the other hand dephases the initial state leading to a diagonal density matrix in the basis @xmath37 . hence , the equation of motion for @xmath37 generated by eq . ( [ eq : lindblad ] ) is a classical rate equation with transition rates set by the above jump operators , @xmath38 with @xmath39 . the random hopping between sites relaxes density modulations that may have existed in the initial state . for a disorder strength @xmath40 under dephasing , i.e. , @xmath34 , for a system of size @xmath41 for @xmath42 disorder realizations . the shaded regions denote standard errors.,scaledwidth=45.0% ] figure [ fig : anderson - decay ] shows a comparison of the result of the effective theory ( dashed lines ) with the exact solution available for the non - interacting problem ( solid lines ) . generally , eq . can be used to generate an infinite hierarchy of coupled equations for the @xmath43-point correlation functions . however , for quadratic hamiltonians and the jump operators of interest here , the equations for the two - point functions @xmath44 are closed . for the dephasing bath they yield @xmath45 with the transition rates @xmath46 . from eq . , we see that the off - diagonal terms are indeed averaged to zero because of the rapid oscillations with frequency of order @xmath47 , the disorder strength . the residual diagonal components then follow eq . ( [ eq : eom - ni - diag ] ) . in a strongly disordered system the transition rates are dominated by the spatial overlaps of nearest - neighbor states @xmath48 . hence , we expect the long - time behavior of @xmath24 taken for different parameter values @xmath49 and @xmath50 to collapse to a single universal curve in the strong - disordered regime when plotted as a function of the rescaled time @xmath51 . this collapse is demonstrated in fig . [ fig : anderson - scaling ] , showing the decay of the imbalance for different disorder strengths @xmath52 . we have plotted @xmath53 in this figure in order to better identify the form of the decay function . the result appears like a stretched exponential decay @xmath54 , with @xmath55 and @xmath56 , that persists for about two orders of magnitude in time . this is broadly consistent with the results reported in a recent numerical study @xcite . such a stretched exponential decay is known from systems exhibiting a distribution of time scales and can be understood as a superposition of exponential decays @xcite . in our system this could be a result of the imbalance decaying with different rates at distinct parts of the system because of fluctuations in the disorder strength . assuming the imbalance has slow spatial variations throughout the decay the equation of motion for the local imbalance has a well - defined continuum limit ( see si ) given by @xmath57 here , @xmath58 is a local , effective decay rate . deep in the localized phase , @xmath58 is dominated by short - range hopping rates , @xmath59 ^ 2 $ ] , where @xmath60 is the energy difference between nearby sites at location @xmath61 . after coarse graining we expect @xmath62 to have a gaussian distribution with variance @xmath63 . first solving for the local imbalance @xmath64 $ ] with @xmath65 the initial imbalance , we can obtain the global imbalance from the integral ( t)_0 dx e^-(x ) t = _ 0de^-^22 ^ 2 - 4j^2 ^ 2 t . [ eq : integratedi ] the saddle - point approximation to this integral gives a stretched exponential decay @xmath66 , where @xmath67 is the scaled time defined above and @xmath68 . the numerical value of @xmath69 thus corresponds to a variance for @xmath62 of @xmath70 . the small change in exponent , which was fit to 0.38 in fig . [ fig : anderson - scaling ] for @xmath41 , shows only weak dependence on system size ( see inset fig . [ fig : anderson - scaling ] ) and may be due to longer - range hopping or corrections from the gradient terms . ( solid , dashed , and dash - dotted lines , 150 disorder realizations each ) under dephasing with @xmath71 . the dashed black line denotes @xmath72 $ ] with @xmath55 and @xmath69 . the inset shows the size dependence using the diagonal rate equations.,scaledwidth=45.0% ] next , we consider dissipation in the form of single - particle loss . such a bath is implemented by the jump operators @xmath73 . in this case , the classical effective model leads to a trivial rate equation _ = -_. the exact equations of motion for the two - point functions are also simpler than in the dephasing bath , @xmath74 again , on a timescale @xmath75 off - diagonal terms die off and the effective model and exact dynamics coincide . since the decay due to single - particle loss in the non - interacting system does not induce hopping between different localized orbitals and is symmetric with respect to even and odd sites , it does not lead to decay of the ( normalized ) imbalance . only the total particle number relaxes with a trivial exponential decay . _ interacting system _ in the full mbl phase , where all eigenstates are localized , the hamiltonian @xmath76 only contains products of the quasi - local integrals of motion @xmath77 @xcite . in order to write the lindblad equation as a classical rate equation , we have to express the fermion operators @xmath78 in terms of the liom operators @xmath79 , which for the interacting case become non - trivial . we first consider the weak - coupling limit , @xmath80 . for a perturbative expansion of the lioms we express the hamiltonian eq . using the operators @xmath81 that diagonalize the problem without interactions @xmath82 where @xmath83 contains no diagonal terms @xmath84 and @xmath85 now , by requiring @xmath86=o(u^2/h^2)$ ] , we obtain @xmath87 for the validity of this expansion we must have no resonant terms in the above sum . this requirement is identical to the mbl criterion , formulated by @xcite , of no incoherent decay of a particle from state @xmath32 . inverting eq . we find the equations of motion for the lioms using eq . . for the case of dephasing , already the non - interacting system exhibits diffusion and the interaction effects are only sub - leading . specifically in the regime @xmath88 the dissipation induced hopping rates behave as @xmath89 , with @xmath90 a numerical constant of order 1 ( see si ) . , 100 disorder realizations ) for a system of size @xmath41 , @xmath91 , and @xmath40 . the shaded regions denote standard errors.,scaledwidth=45.0% ] we solve the effective classical rate equations obtained for the dephasing process and compare with a simulation of the quantum lindblad equation using the time - evolving block decimation ( tebd ) scheme for matrix product operators @xcite . fig . [ fig : interacting - dephasing ] shows excellent agreement between the effective model and simulations of a 20-site chain . the particle loss in the interacting case is governed by the rate equation @xmath92 with @xmath93 . in the interacting system , particle loss additionally induces particle hopping ( see inset fig . [ fig : ind - hopping ] ) , which in turn leads to diffusion . thus , the dynamics for particle loss in the interacting case differs clearly from the non - interacting case . only when the system becomes so dilute as to effectively be non - interacting , the imbalance stabilizes , as shown in fig . [ fig : ind - hopping ] . finally , we consider the strong - coupling limit @xmath94 . for simplicity , we study the dynamics of the initial state @xmath95 , the perfect density wave . in the limit @xmath96 this initial state is a stationary state of the dynamics if we include only the dephasing bath , hence the decay rate of the imbalance must vanish in this limit . when @xmath97 is large but finite we can obtain the lioms using perturbation theory in @xmath98 with @xmath97 having a similar role as the disorder @xmath50 had at weak coupling . we obtain a decay rate that scales as @xmath99 to leading order in the small parameter . this should be compared with the weak coupling rate @xmath100 $ ] . thus , as we vary @xmath97 from weak to strong coupling the maximal decay rate is expected to occur around @xmath101 @xcite . it is also interesting to consider the effect of single particle loss in the strong interaction limit . for simplicity we take @xmath96 and consider the dynamics at sufficiently short times when there are still few holes in the density wave . then , the initial state in the vicinity of the lost particle can be taken to be the perfect density wave . to lowest order in the hopping , the final state after the particle loss is @xmath102 with the renormalized single - fermion operator _ 2i = c_2i+w^i_-c^_2i-1c_2i-2c_2i + w^i_+c^_2i+1c_2i+2c_2i and @xmath103 . the first - order terms result in a decay of the imbalance at an initial rate that scales as @xmath104 to be contrasted with the decay rate @xmath105 obtained for weak interactions @xmath106 . hence , we find an enhancement of the decay of the imbalance by interactions that , for the case of single - particle loss , persists all the way to the limit of infinite @xmath97 . , @xmath40 , @xmath107 , and 250 disorder realizations . the inset shows the induced hopping processes that lead to diffusion as described by eq . . ] _ conclusions _ we have presented a general framework to study the dynamics in mbl systems coupled weakly to a dissipative markovian bath . making use of the local integrals of motion characterizing the mbl phase , we have mapped the quantum evolution to much simpler classical rate equations . the form of these equations depends on the type of coupling to the bath and the intrinsic interactions in the hamiltonian . compared to the quantum evolution the effective classical equations are feasible to tackle numerically and , with further approximations , even analytically . our effective model provides a simple explanation for the stretched exponential decay of observables in the localized phase . the dynamics of mbl systems weakly coupled to a bath is seen , from this perspective , to be similar to that of a classical glass @xcite . further study is needed to understand how different classes of disorder affect the pattern of relaxation of observables ( see si for comparison with the case of a quasi - periodic potential studied in recent experiments @xcite ) . our scheme of mapping the quantum dynamics to classical rate equations relies on the fact that the action of a local operator does not cause resonant transitions if the system is deep in the localized state . this follows from the large offset in energy associated with nearby integrals of motion . however , if the system is closer to the delocalization transition then resonances start to play an increasingly important role . it would therefore be very interesting to investigate how the relaxation behavior changes as the system is tuned toward the transition . this could give important information about the nature of the critical - thermal regions that destabilize the mbl state @xcite .
the appearance , in p - sige , of insulating behaviour between filling factors @xmath0 = 1 and 2 has been known for several years @xcite . although not always present it has been observed by several groups in samples from more than one source . similar behaviour has also been seen recently @xcite near filling factors 5/2 and 4 . there is , as yet , no generally accepted explanation for this phenomenon . two possibilities are that the insulating phase ( ip ) might be attributed to a premature and re - entrant hall insulating state @xcite or that it results from a combination of the unusual ordering of the landau levels and a long - range modulation of the potential @xcite . an alternative suggestion @xcite is that the ip appears when the large , exchange enhanced , zeeman splitting causes adjacent landau levels to cross and could be associated with an unusual spin texture or domain structure . results are presented here to support this view . it is shown ( a ) that the behaviour of the hall resistivity in the ip is not consistent with a hall insulating state and ( b ) that although no exotic spin texture states are expected @xcite at @xmath5 this is not necessarily the case for non - integer filling factors , particularly when disorder and screening are taken into account . figure 1 shows details of the temperature dependence of the shubnikov - de haas ( sdh ) oscillations in a p - sige sample @xcite with a density of @xmath6m@xmath7 and mobility of 1.65 m@xmath8(vs)@xmath9 . the ip which appears at @xmath0 = 3/2 is characterised by fixed points with @xmath10 ( see also figure 2a ) . they indicate a metal - insulator transition with scaling behaviour @xcite similar to that observed for transitions into the high field ( @xmath11 ) hall insulating state . separate , well defined , fixed points associated with integer quantum hall effect ( iqhe ) transitions can also be observed . these are shifted somewhat from the expected magnetic fields but the critical values ( in units of @xmath12 ) of @xmath13 = 0.65 and @xmath14 = 2.7 for the 3 @xmath15 2 transition and @xmath13 = 0.59 , @xmath14 = 1.6 for the 2 @xmath16 transition , are close to the expected values of ( 0.5,2.5 ) and ( 0.5,1.5 ) respectively . near b = 3 tesla the increase of @xmath17 with temperature is metallic - like , mirroring the behaviour around 4 tesla . this is interpreted as the precursor to another ip which is suppressed by the onset of the @xmath0 = 2 iqhe plateau before becoming fully developed . ( in a lower density sample this peak becomes fully insulating @xcite ) . a feature of the data is the behaviour of @xmath18 through the ip . this is not an easy measurement , because any sample inhomogeneity produces spurious @xmath17 terms which can be an order of magnitude larger than the genuine signal , but when these are removed by averaging data with normal and reversed magnetic fields , it can be seen that the hall resistance tends to vanish as the insulating behaviour strengthens . this is a general feature of the ip : figure 2b shows measurements in a slightly lower density sample and similar trends can also be identified in other data in the literature @xcite . a vanishing hall resistance is incompatible with the @xmath0= 3/2 ip being a re - entrant hall insulating state . the hall insulator @xcite appears when the fermi energy lies in the last landau level : in the iqhe regime it is characterised by a hall conductance that vanishes as @xmath19 so the hall resistivity remains approximately quantised at @xmath20 , in the fractional qhe regime it increases approximately as @xmath21 ( where n@xmath22 is the density ) with fqhe plateaus appearing over limited field ranges . the vanishing of @xmath18 therefore indicates the insulating phase is neither an iqhe nor a fqhe hall insulator . coulomb interaction and exchange effects are large in these p - sige samples . for example , for that shown in figure 1 , at @xmath0=2 , the exchange energy [ @xmath23 where @xmath24 is the magnetic length ] is 10mev , approximately five times larger than the cyclotron energy ( @xmath25 ) of 1.9 mev . the ( bare ) spin splitting in p- sige is relatively large , of order @xmath26 , and is suficiently enhanced by exchange that the @xmath27 and @xmath28 landau levels cross and the system becomes ferromagnetically polarised at @xmath0=2 ( see figure 3 ) . activation measurements in other p - sige samples @xcite confirm this is so . at integer filling factors , when the fermi energy lies in localised states between landau levels , screening is unimportant , but for non - integer filling factors , when the density of extended states at the fermi energy is large , it can significantly reduce the exchange enhancement of the spin splitting . a model calculation is presented that explores this situation . the calculation uses the formalism given by yarlagadda @xcite with the screening of the bare coulomb potential ( @xmath29 ) given by @xmath30 where @xmath31 is the density of states at the fermi level of the @xmath32 landau level and @xmath33 is an inter - landau level term . for any significant density of states at the fermi energy this last term is relatively small and will be ignored . the separation of the @xmath28 and @xmath27 landau levels is then given by @xmath34 where @xmath35 with @xmath36 the coupling between the n and m landau levels and @xmath37 etc partial filling factors . one can also calculate the total ( hartree fock ) energy of the system ( e@xmath38 ) as the integral over occupied states of @xmath39 where @xmath40 is the kinetic energy . realistic parameters for the sample shown in figure 1 were used : an effective mass of 0.2@xmath41 , a bare zeeman splitting of 0.65 @xmath42 and a ( gaussian ) landau level broadening given , according to the self consistent born approximation , by @xmath43 where the ( measured ) quantum mobility @xmath44 is 1.7m@xmath8(vs)@xmath9 . figure 4 shows the separation between the @xmath45 and @xmath46 landau levels as a function of @xmath37 , determined ( a ) from eqns . 2 and 3 and ( b ) from the constraint that @xmath47 . it is assumed the lowest , @xmath48 , level always remains full . valid solutions occur at the intersection of these two curves . at @xmath0=2 , the calculation without screening ( see figure 4a ) shows an unstable , degenerate solution and two stable solutions corresponding to either a ferromagnetic or paramagnetic ordering of the landau levels ( the former with the lowest energy ) . this essentially reproduces the result of giuliani and quinn @xcite who also showed that for non - degenerate landau levels one of these stable solutions will pre - empt any spin - density - like state . when screening is included the ferromagnetic configuration becomes even more strongly favoured but the other solutions are significantly altered , in particular the paramagnetic solution now involves degenerate landau levels and there is an increase in the total energy . when the filling factor moves away from two more significant differences appear ( see figure 4b ) . the density of states at the fermi energy is always non - zero , screening becomes important and the exchange enhancement of the spin splitting is reduced . under these conditions only one , degenerate , solution remains and the ferromagnetically polarised state can no longer exist . figure 5 shows the landau level separation as a function of filling factor . the ferromagnetically polarised state appears only at integer filling factors , when the screening is small . elsewhere , the levels stay close together and are degenerate with a predominantly paramagnetic alignment . the transition into the ferromagnetically polarised state is sharp near @xmath0=2 but much smoother near @xmath0= 1 and 3 . the model calculation can explain the existence of both a high field metal / insulator transition and separate , well defined , iqhe transitions . near integer filling factors , the exchange interaction ensures the landau levels are well separated and the iqhe transitions can then occur in the tail of an isolated landau level . away from integer filling factors , screening means the landau levels overlap which correlates with the appearance of the insulating ( or metallic ) behaviour . separate metal / insulator and iqhe transitions occur because the landau levels realign as the filling factor changes . in some circumstances , for example in lower mobility samples or at higher temperatures , this realignment may not occur and the high field ip and iqhe transitions will then merge and no longer be distinguishable . away from integer filling factors the only allowed solution corresponds to degenerate landau levels ( at least for a homogeneous system ) . although the hartree - fock energy in figure 4b appears relatively flat as a function of the relative populations of the two levels , a detailed examination shows there could be a small lowering of the energy if the system were to break up into domains . while the exact nature of such a state ( or some other kind of spin texture ) is uncertain such a mechanism would appear to be a candidate for the cause of the insulating behaviour . in high mobility , gaas based , 2deg samples , coulomb interaction effects manifest themselves as fractional quantum hall effect features . because of the difference in effective masses ( 0.2@xmath41 compared with 0.067@xmath41 ) and the ratio of the transport to quantum lifetimes ( approximately one in p - sige compared with 20 or so in n - gaas 2degs ) the landau level broadening in the p - sige sample investigated here is very similiar to that of a gaas based 2deg with a mobility of over 200 m@xmath8(vs)@xmath9 . the exchange energy depends only on magnetic field so the absence of fractional quantum hall features is a little puzzling . the essential difference is that in n - gaas heterostructures higher landau levels are well removed in energy , typically by several times the interaction energy , whereas in p - sige there are two degenerate landau levels at the fermi energy . this is consistent with the insulating phase ( and associated high field metal - insulator transition ) being a direct consequence of strong exchange interactions in degenerate landau levels suppressing more usual fqhe behaviour . the model calculation does not include a number of factors such as : the finite thickness of the hole gas , correlation effects , the existence of a mobility edge and the contribution of inter - landau level screening but the result , that screening causes the landau levels to stick together , is robust . a negative feed - back mechanism ( relating the exchange splitting of the landau levels to the screening ) ensures that the separation of the levels is small and relatively independent of exact values of the parameters , provided only that the _ unscreened _ exchange energy is large compared with the bare separation and that screening significantly reduces this exchange energy . in electron systems landau level crossings occur and can be tuned with parallel magnetic fields @xcite . they are identified by ` spikes ' in the shubnikov - de haas data with , in some cases , hysteresis . this is interpreted in terms of domain formation in these quantum hall ising ferromagnets . the question of spin textures in such systems has recently been studied theoretically by brey and tejedor @xcite . they find that a spin - orbit interaction , which directly couples the @xmath27 and @xmath28 states , provides a means of producing a gap in the energy spectrum of the charged excitations that cross domain walls . it is therefore interesting to note that in p - sige , where strain separates the two degenerate heavy - hole states at the brillouin zone centre , there is a large spin - orbit coupling with the holes having almost pure @xmath49 = 3/2 character . however , it should be also noted these systems are not necessarily directly equivalent to that discussed here which often involve relatively weakly coupled 2degs , either in a rather wide single quantum well or in two wells separated by a barrier . fixed points defining both iqhe transitions and a high field insulating phase can be separately identified in shubnikov - de haas measurements in a p - sige sample . further , it is found that the hall resistivity in the ip tends towards zero as the temperature is lowered , which implies the ip is not a re - entrant hall insulator . a model calculation is presented which explains these two types of behaviour in terms of the large exchange interaction which is screened when the fermi energy lies within a landau level but not when it lies in localised states between landau levels . insulating behaviour ( or more generally a metal- insulator transition ) co - incides with the overlap of degenerate landau levels overlap at the fermi energy . it is suggested therefore that the insulating behaviour could well be a direct consequence some form of domain structure or other kind of spin texture induced by a strong exchange interaction in degenerate landau levels . et al _ , phys . rev . lett . * 79*,4449 ( 1997 ) ; t. jungwirth _ et al _ , phys . . lett . * 81*,2328 ( 1998 ) ; v. piazza _ et al _ , nature ( london ) * 402*,638 ( 1999 ) ; e.p . de poortere _ et al _ , science * 290 * , 1546 ( 2000 )
shubnikov - de haas data is presented for a p - sige sample exhibiting strongly insulating behaviour at @xmath0 = 3/2 . in addition to the fixed points defining a high field metal- insulator transition into this phase separate fixed points can also be identified for the @xmath1 and @xmath2 integer quantum hall transitions . another feature of the data , that the hall resistivity approaches zero in the insulating phase , indicates it is not a re - entrant hall insulator . the behaviour is explained in terms of the strong exchange interactions . at integer filling factors these cause the @xmath3 and @xmath4 landau levels to cross and be well separated but at non - integer values of @xmath0 screening reduces exchange effects and causes the levels to stick together . it is suggested the insulating behaviour , and high field metal / insulator transition , is a consequence of the strong exchange interactions .
monolayer and bilayer graphene are distinct electronic materials . monolayer graphene is a sheet of carbon in a honeycomb lattice that is one atom thick , while bilayer graphene comprises two such sheets , with the first lattice @xmath2 above the second . since the first transport measurements @xcite in 2005 , we have come a long way in understanding the basic transport mechanisms of carriers in these new carbon allotropes . ( for recent reviews , see refs . ) . a unique feature of both monolayer and bilayer graphene is that the density of carriers can be tuned continuously by an external gate from electron - like carriers at positive doping to holes at negative doping . the behavior at the crossover depends strongly on the amount of disorder . in the absence of any disorder and at zero temperature , there are no free carriers at precisely zero doping . however , ballistic transport through evanescent modes should give rise to a universal minimum quantum limited conductivity @xmath3 in both monolayer @xcite and bilayer graphene.@xcite the `` ballistic regime '' should hold so long as the disorder - limited mean - free path is larger than the distance between the contacts . @xcite at finite temperature , the thermal smearing of the fermi surface gives a density @xmath4 for monolayer graphene . for ballistic transport in these monolayers , the conductivity @xmath5 for large @xmath6 , so @xmath7.@xcite in the absence of disorder , @xmath8 interpolates from the universal @xmath9 to the linear in @xmath10 regime following a function that depends only on @xmath11 ; ( @xmath12 is the fermi temperature ) . @xcite most experiments , however , are in the dirty or diffusive limit , which is characterized by a conductivity that is linear in density ( i.e. @xmath13 , with a mobility @xmath14 that is independent of both temperature and carrier density @xcite ) , and the existence of a minimum conductivity plateau @xcite in @xmath15 , with @xmath16 . @xmath17 is the root - mean - square fluctuation in carrier density induced by the disorder . in bilayer graphene , to our knowledge , all experiments are in the diffusive limit . the purpose of the current work is to calculate the temperature dependence of the minimum conductivity plateau in bilayer graphene . the temperature dependent conductivity of diffusive graphene monolayers is understood to depend largely on phonons , @xcite but monolayer and bilayer graphene are distinct electronic materials and phonons are not expected to be important for bilayer graphene transport at the experimentally relevant temperatures . an important difference between monolayer and bilayer graphene is the band structure near the dirac point . monolayer graphene has the conical band structure and a density of states that vanishes linearly at the dirac point . bilayer graphene has a constant density of states close to the dirac point from a hyperbolic dispersion . the tight - binding description for bilayer graphene @xcite results in a hyperbolic band dispersion @xmath18 , \label{eq : band}\ ] ] that is completely specified by two parameters , @xmath19 and @xmath20 ( where @xmath21 is planck s constant ) . for very small carrier density @xmath22 , one can approximate bilayer graphene as having a parabolic dispersion , although most experiments typically approach carrier densities as large as @xmath23 . the density of states for bilayer graphene is @xmath24,\ ] ] where the parabolic approximation keeps only the first term . understanding the temperature dependence of the conductivity minimum is complicated for two reasons . first , there is activation of both electron and hole carriers at finite temperature . second , the disorder induces regions of inhomogeneous carrier density ( i.e. puddles of electrons and holes ) . moreover , tuning the carrier density with a gate changes the ratio between electron - puddles and hole - puddles , until at very high density there is only a single type of carrier . the temperature dependence of the conductivity for bilayer graphene was studied in ref . using a coherent potential approximation . while this approach better captures the impurity scattering and electronic screening properties of graphene , it does not account for the puddle physics which is our main focus . reference modeled the temperature dependence of the dirac point conductivity by assuming that the graphene samples comprised just two big `` puddles '' each with the same number of carriers . in the appropriate limits , our results agree with these previous works . below we will provide a semi - analytic expression for the graphene conductivity by averaging over the random distribution of puddles with different carrier densities . this result is valid throughout the crossover from the dirac point ( where fluctuations in carrier density dominate ) to high density ( where these fluctuations are irrelevant ) , both with and without the thermal activation of carriers . = 1.0 given a microscopic model for the disorder , one can compute both @xmath14 and @xmath17 . shown in fig . [ fig : mobility ] are results for bilayer graphene mobility assuming both short - range and coulomb disorder with different approximations for the screening , and for both parabolic and hyperbolic dispersion relations . as seen from the figure , generically , coulomb impurities show a super - linear dependence on carrier density while short - range scattereres are sub - linear . similar to monolayer graphene,@xcite increasing the dielectric constant tends to decrease ( increase ) the scattering of electrons off long ( short ) range impurities , except in the over - screened and unscreened limits . all experiments to date find the mobility to be linear in gate voltage , so it is unclear what the dominant scattering mechanism in bilayer graphene is ( see also discussion in ref . ) . further experiments along the lines of refs . are needed . in what follows we take @xmath14 and @xmath17 to be parameters of the theory that can be determined directly from experiments : @xmath14 can be obtained from low temperature transport measurements and @xmath17 from local probe measurements.@xcite lacking such microscopic measurements for the samples we compare with , we treat @xmath17 as a fitting parameter , while taking @xmath14 from experiment . as a consequence of this parameterization , the results reported here do not depend on the microscopic details of the impurity potential , provided this parameterization reasonably characterizes the properties of the impurity potential . until more information about the important scattering centers is determined from experiment , all microscopic models will require a similar number of parameters such as the concentration of impurities @xmath25 and their typical distance @xmath26 from the graphene sheet . further , the results will disagree with experiment unless the choices give a constant mobility . a key assumption in this work is the applicability of effective medium theory ( emt ) , which describes the bulk conductivity @xmath27 of an inhomogeneous medium by the integral equation @xcite @xmath28 \frac{\sigma(n ) - \sigma_{\rm emt}}{\sigma(n ) + \sigma_{\rm emt } } = 0 . \label{eq : eqemt1}\ ] ] @xmath29 $ ] is the probability distribution of the carrier density in the inhomogeneous medium positive ( negative ) @xmath6 corresponds to ( electrons ) holes , and @xmath15 is the local conductivity of a small patch with a homogeneous carrier density @xmath6 . ignoring the denominator , eq . [ eq : eqemt1 ] gives @xmath27 equal to the average conductivity . the denominator weights the integral to cancel the build - up of any internal electric fields . the emt description has been shown to work well whenever the transport is semiclassical and quantum corrections and any additional resistance caused by the @xmath30 interfaces between the electron and hole puddles can be ignored.@xcite it is assumed that the band structure is not altered by the disorder , which is to be expected for the experimentally relevant disorder concentrations.@xcite since we are concerned with diffusive transport in the dirty limit , we expect that the emt results hold for bilayer graphene . = 1.0 = 1.0 to solve eq . [ eq : eqemt1 ] we make the additional assumption that the distribution function @xmath31 $ ] is gaussian centered at @xmath32 , ( i.e. the field effect carrier density induced by the back gate that is proportional to @xmath33 ) , with width @xmath17 . ( this assumption is justified both theoretically@xcite and empirically@xcite ) . our results are shown in fig . [ fig : main ] , where as discussed earlier , the temperature dependence comes from the smearing of the fermi surface . at first glance , it is not obvious that the results for clean bilayer graphene ( left panel of fig . [ fig : main ] ) and dirty bilayer graphene ( right panel ) are closely related . however , if we consider scaling the conductivity as @xmath34 , scaling temperature as @xmath35 , where we define @xmath36 , and scaling carrier density as @xmath37 , we find that for both the linear band dispersion @xmath38 and the parabolic band dispersion @xmath39 , the scaled functions @xmath40 each follow a universal curve . this is illustrated in fig . [ fig : scaling ] where we show the temperature dependence of the minimum conductivity . the results for the hyperbolic dispersion ( which is the correct approximation at experimentally relevant carrier densities ) , depends on an additional parameter @xmath41.@xcite the scaling function for the hyperbolic dispersion extrapolates from the parabolic theory at large @xmath42 becoming similar to the linear result for small @xmath42 . for the experimentally relevant regime @xmath43 the hyperbolic result depends only weakly on @xmath42 and is indistinguishable from the parabolic result for @xmath44 . = 1.0 this analysis suggests that @xmath45 , which can be taken directly from experiment , is not a function of @xmath14 , but only @xmath17 . we take results from a set of experiments in very different regimes ( see the inset of fig . [ fig : expts ] ) and choose @xmath17 to fix the value of @xmath46 at @xmath47 . then using @xmath48 to scale the temperature , all of the results lie on top of the theoretical curve computed using the hyperbolic dispersion , see fig . [ fig : expts ] . the theoretical curve with which they agree is distinct from similar curves calculated for a linear dispersion and for the purely parabolic dispersion at high @xmath0 . we note that the scaling function is more complicated than a line . the calculation reproduces not only the initial slope as a function of temperature , but the crossover to higher temperature behavior . for the parabolic dispersion , which agrees at low temperatures , the conductivity extrapolates from @xmath49 at low temperature to @xmath50 at high temperature , with a crossover temperature scale of @xmath51 . in the future , it should be possible to further test this agreement by measuring @xmath17 experimentally . @xcite = 1.0 one feature of fig . [ fig : main ] and fig . [ fig : expts ] is that for most of the experimentally relevant regime , the temperature dependence of the conductivity calculated using the parabolic approximation provides an adequate solution . this limit has been treated in contemporaneous work @xcite treating this problem with different approximations and reaching similar conclusions . to better understand the emergence of a universal scaling form , we consider the conductivity for a parabolic band dispersion . using the scaled variables defined above , we can manipulate eq . ( [ eq : eqemt1 ] ) into the dimensionless form @xmath52 \cosh \left [ z_g z\right ] \frac{h[z , t ] - { \bar \sigma}[z_g , t]}{h[z , t ] + { \bar \sigma}[z_g , t ] } = 0,\end{aligned}\ ] ] where @xmath53 and we have written the local conductivity as @xmath54 . below we calculate the dimensionless function @xmath55 assuming thermally activated carrier transport with constant @xmath17 and @xmath14 and explicitly show that it depends only on scaled variables @xmath37 and @xmath35 . with the analytical results for @xmath55 discussed below , this implicit equation can be solved either perturbatively or by numerical integration to give @xmath27 . the results of this calculation are shown in fig . [ fig : parabolic ] . to proceed , we calculate the function @xmath55 . for thermal activation of carriers , the chemical potential @xmath56 is determined by solving for @xmath57,@xcite where @xmath58,\end{aligned}\ ] ] where @xmath59 is the fermi - dirac function and @xmath60 is the boltzmann constant . for @xmath47 , only majority carriers are present , while for @xmath61 , activated carriers of both types are present in equal number . within the parabolic approximation , we find @xmath62 $ ] and @xmath63 . using @xmath64 , we obtain @xmath65.\ ] ] this demonstrates that eq . [ eq : emt ] depends only on the scaled variables , guaranteeing that @xmath66 is a function only of @xmath67 and @xmath68 as shown in fig . [ fig : parabolic ] . a similar analysis can be done for the hyperbolic dispersion . we find @xmath69 + 2y \frac{}{}\right . \nonumber \\ & & \mbox { } \left . + \frac{(t g \pi)^2\xi}{3 } + \xi y^2 \right ] , \label{eq : mu}\end{aligned}\ ] ] where @xmath70 , @xmath71 , and the scaled chemical potential @xmath72 is given by @xmath73 , \label{eq : ef } \end{aligned}\ ] ] where @xmath74 is the dilogarithm function . only for @xmath75 and @xmath76 does @xmath77 become independent of @xmath78 giving the universal scaling forms for linear and parabolic dispersions , respectively . in summary , we have developed an effective medium theory that captures the gate voltage and temperature dependence of the conductivity for bilayer graphene . the theory depends on two parameters : @xmath17 that sets the scale of the disorder , and @xmath14 the carrier mobility . these could be computed _ a priori _ by assuming a microscopic model for the disorder potential and its coupling to the carriers in graphene . alternatively , one could use an empirical approach where one uses experimental data at @xmath47 to determine the parameters and use the theory to predict the temperature dependence . our main finding is that experimental data taken from various groups collapse onto our calculated scaling function where the disorder sets the scale of the temperature dependence of the conductivity . this further suggests that even some suspended bilayer samples are still the the diffusive ( rather than ballistic ) transport regime .
assuming diffusive carrier transport , and employing an effective medium theory , we calculate the temperature dependence of bilayer graphene conductivity due to fermi surface broadening as a function of carrier density . we find that the temperature dependence of the conductivity depends strongly on the amount of disorder . in the regime relevant to most experiments , the conductivity is a function of @xmath0 , where @xmath1 is the characteristic temperature set by disorder . we demonstrate that experimental data taken from various groups collapse onto a theoretically predicted scaling function .
within the main paper , we have assumed that the chemoattractant concentration @xmath0 regulates the susceptibility of a cell to contact inhibition of locomotion @xmath58 , with @xmath113 . this models the stabilization of protrusions induced by contact interactions . this is consistent with the results of theveneau et al . @xcite , who find that protrusion stabilization is stronger in clusters than in single cells . however , very similar results can be found if we assume that @xmath114 is constant and the signal regulates the time required for the cell s polarity to relax , i.e. @xmath115 . in this case , the mean polarity of a cell is @xmath116 and we find @xmath117 where the mobility matrix @xmath30 is the same as in the main paper , @xmath118 . however , because @xmath14 varies over space , the fluctuations will also vary : @xmath119 , where @xmath120 is the mean signal across the cluster . for this reason , the chemotactic index in the @xmath14-regulation model will depend on @xmath121 , and will not be constant over a linear gradient . in addition , a single cell with a persistence time @xmath14 that depends on the chemoattractant level will undergo biased motion . this is shown in fig . [ fig : singletau ] below . this drift can be made smaller than the cil - driven cluster drift , as it is independent of @xmath114 , while the cluster drift is proportional to @xmath114 . develop a mean drift*. the mean @xmath3 and @xmath49 velocities for a cell with spatially varying @xmath14 are shown : @xmath122 , with @xmath123 , @xmath124 , @xmath2 . result is average over @xmath125 iterations , each started at the origin ; error bars indicate @xmath126 ^ 2 \rangle^{1/2}/\sqrt{n}$ ] . , width=302 ] we can compute the mobility matrix of the @xmath65-layer oligomers for arbitrary @xmath65 . our mobility matrix is given by eq . [ eq : matrix ] with @xmath16 . to simplify the calculation , we can make a few assumptions . first , we note that @xmath127 , but @xmath128 for the @xmath65-layer oligomer . we only need to calculate @xmath129 . the only cells @xmath4 in the sum of eq . [ eq : matrix ] that are nonzero are those around the boundary . @xmath130 does not depend on orientation , so we can compute the sum @xmath131 for one face of the oligomer ( fig . [ fig : qmer ] ) , then multiply by six . however , this double - counts the corner cells , so we must weight them by @xmath132 . we then find @xmath133/n(q)$ ] , where @xmath134 is the number of cells in the cluster . . for each of the configurations shown , nearest - neighbor cells have unit separation . a @xmath65-layer oligomer has @xmath136 cells . @xmath30 is given for the orientation shown in the left column ; other orientations may be found by transforming the mobility tensor ; @xmath70 ( see section [ sec : mobrotation ] below ) . [ cols="^ " , ]
many eukaryotic cells chemotax , sensing and following chemical gradients . however , experiments have shown that even under conditions when single cells can not chemotax , small clusters may still follow a gradient . this behavior has been observed in neural crest cells , in lymphocytes , and during border cell migration in drosophila , but its origin remains puzzling . here , we propose a new mechanism underlying this collective guidance " , and study a model based on this mechanism both analytically and computationally . our approach posits that the contact inhibition of locomotion ( cil ) , where cells polarize away from cell - cell contact , is regulated by the chemoattractant . individual cells must measure the mean attractant value , but need not measure its gradient , to give rise to directional motility for a cell cluster . we present analytic formulas for how cluster velocity and chemotactic index depend on the number and organization of cells in the cluster . the presence of strong orientation effects provides a simple test for our theory of collective guidance . cells often perform chemotaxis , detecting and moving toward increasing concentrations of a chemoattractant , to find nutrients or reach a targeted location . this is a fundamental aspect of biological processes from immune response to development . many single eukaryotic cells sense gradients by measuring how a chemoattractant varies over their length @xcite ; this is distinct from bacteria that measure chemoattractant over time @xcite . in both , single cells are capable of net motion toward higher chemoattractant . recent measurements of how neural crest cells respond to the chemoattractant sdf1 suggest that single neural crest cells can not chemotax effectively , but small clusters can @xcite . a more recent report shows that at low gradients , clusters of lymphocytes also chemotax without corresponding single cell directional behavior ; at higher gradients clusters actually move in the opposite direction to single cells @xcite . in addition , late border cell migration in the _ drosophila _ egg chamber may occur by a similar mechanism @xcite . these experiments strongly suggest that gradient sensing in a cluster of cells may be an _ emergent _ property of cell - cell interactions , rather than arising from amplifying a single cell s biased motion ; interestingly , some fish schools also display emergent gradient sensing @xcite . in fact , these experiments led to a collective guidance " hypothesis @xcite , in which a cluster of cells where each individual cell has no information about the gradient may nevertheless move directionally . in a sense that will become clear , cell - cell interactions allow for a measurement of the gradient across the entire cluster , as opposed to across a single cell . in this paper , we develop a quantitative model that embodies the collective guidance hypothesis . our model is based on modulation of the well - known contact inhibition of locomotion ( cil ) interaction @xcite , in which cells move away from neighboring cells . we propose that individual cells measure the local signal concentration and adjust their cil strength accordingly ; the cluster moves directionally due to the spatial bias in the cell - cell interaction . we discuss the suitability of this approach for explaining current experiments , and provide experimental criteria to distinguish between chemotaxis via collective guidance and other mechanisms where clusters could gain improvement over single - cell migration @xcite . these results may have relevance to collective cancer motility @xcite , as recent data suggest that tumor cell clusters are particularly effective metastatic agents @xcite . by contact inhibition of locomotion ( cil ) ; the strength of this bias is proportional to the local chemoattractant value @xmath0 , leading to cells being more polarized at higher @xmath1 . see text for details . * b , * one hundred trajectories of a single cell and * c , * cluster of seven cells . trajectories are six persistence times in length ( 120 min ) . scalebar is one cell diameter . the gradient strength is @xmath2 in these simulations , with the gradient in the @xmath3 direction.,width=321 ] we consider a cluster of cells exposed to a chemical gradient @xmath0 . we use a two - dimensional stochastic particle model to describe cells , giving each cell @xmath4 a position @xmath5 and a polarity @xmath6 . the cell polarity indicates its direction and propulsion strength : an isolated cell with polarity @xmath6 has velocity @xmath6 . the cell s motion is overdamped , so the velocity of the cell is @xmath6 plus the total physical force other cells exert on it , @xmath7 . biochemical interaction between cells alter a cell s polarity @xmath6 . our model is then : @xmath8 where @xmath9 are the intercellular forces of cell - cell adhesion and volume exclusion , and @xmath10 are gaussian langevin noises with @xmath11 , where the greek indices @xmath12 run over the dimensions @xmath13 . the first two terms on the right of eq . [ eq : polarity ] are a standard ornstein - uhlenbeck model @xcite : @xmath6 relaxes to zero with a timescale @xmath14 , but is driven away from zero by the noise @xmath15 . this corresponds with a cell that is orientationally persistent over a time of @xmath14 . we have introduced the last term on the right of eq . [ eq : polarity ] to describe contact inhibition of locomotion ( cil ) . cil is a well - known property of many cell types in which cells polarize away from cell - cell contact @xcite . we model cil by biasing @xmath6 away from nearby cells , toward @xmath16 , where @xmath17 is the unit vector pointing from cell @xmath18 to cell @xmath4 and the sum over @xmath19 indicates the sum over the neighbors of @xmath4 ( those cells within a distance @xmath20 cell diameters ) . while this is motivated by cil in neural crest , it is also a natural minimal model under the assumption that cells know nothing about their neighbors other than their direction @xmath21 . for cells along the cluster edge , the cil bias @xmath22 points outward from the cluster , but for interior cells @xmath22 is smaller or zero ( fig . [ fig : schematic]a ) . this is consistent with experimental observations that edge cells have a strong outward polarity , while interior cells have weaker protrusions @xcite . chemotaxis arises in our model if the chemoattractant @xmath0 changes a cell s susceptibility to cil , @xmath23 , @xmath24 . this models the result of @xcite that the chemoattractant sdf1 stabilizes protrusions induced by cil @xcite . we also assume that the cell s chemotactic receptors are not close to saturation - i.e. the response is perfectly linear . if cil is present even in the absence of chemoattractant ( @xmath25 ) , as in neural crest @xcite , i.e. @xmath26 , this will not significantly change our analysis . similar results can also be obtained if all protrusions are stabilized by sdf1 ( @xmath14 regulated by @xmath1 ) , though with some complications ( _ appendix _ , fig . a1 ) . _ analytic predictions for cluster velocity._our model predicts that while single cells do not chemotax , clusters as small as two cells will , consistent with @xcite . we can analytically predict the mean drift of a cluster of cells obeying eqs . [ eq : position]-[eq : polarity ] : @xmath27 where the approximation is true for shallow gradients , @xmath28 . @xmath29 indicates an average over the fluctuating @xmath6 but with a fixed configuration of cells @xmath5 . the matrix @xmath30 only depends on the cells configuration , @xmath31 where , as above , @xmath16 . eq . [ eq : shallow ] resembles the equation of motion for an arbitrarily shaped object in a low reynolds number fluid under a constant force @xmath32 @xcite : by analogy , we call @xmath30 the mobility matrix . " there is , however , no fluctuation - dissipation relationship as there would be in equilibrium @xcite . to derive eq . [ eq : shallow ] , we note that eq . [ eq : position ] states that , in our units , the velocity of a single cell is equal to the force on it ( i.e. the mobility is one ) . for a cluster of @xmath33 cells , the mean velocity of the cluster is @xmath34 times the total force on the cluster . as @xmath35 , the cluster velocity is @xmath36 . when the cluster configuration changes slowly over the timescale @xmath14 , eq . [ eq : polarity ] can be treated as an ornstein - uhlenbeck equation with an effectively time - independent bias from cil . the mean polarity is then @xmath37 , with gaussian fluctuations away from the mean , @xmath38 . the mean cell cluster velocity is @xmath39 in a constant chemoattractant field , @xmath40 , no net motion is observed , as @xmath41 . for linear or slowly - varying gradients @xmath28 , and we get eq . [ eq : shallow ] . _ cluster motion and chemotactic efficiency depend on cluster size , shape , and orientation. _ within our model , a cluster s motion can be highly anisotropic . consider a pair of cells separated by unit distance along @xmath42 . then by eq . [ eq : matrix ] , @xmath43 , @xmath44 , @xmath45 . if the gradient is in the @xmath3 direction , then @xmath46 and @xmath47 , where @xmath48 . cell pairs move toward higher chemoattractant , but their motion is along the pair axis , leading to a transient bias in the @xmath49 direction before the cell pair reorients due to fluctuations in @xmath6 ( fig . [ fig : pairs ] ) . we compare our theory for the motility of rigid cell clusters ( eq . [ eq : shallow ] ) with a simulation of eq . [ eq : position]-[eq : polarity ] with strongly adherent cell pairs with excellent agreement ( fig . [ fig : pairs ] ) . for the simulations in fig . [ fig : pairs ] and throughout the paper , we solve the model equations eqs . [ eq : position]-[eq : polarity ] numerically using a standard euler - maruyama scheme . we choose units such that the equilibrium cell - cell separation ( roughly 20 @xmath50 m for neural crest @xcite ) is unity , and the relaxation time @xmath51 ( we estimate @xmath52 minutes in neural crest @xcite ) . within these units , neural crest cell velocities are on the order of @xmath53 , so we choose @xmath54 this corresponds to a root mean square speed of an isolated cell being @xmath55 microns / minute . the typical cluster velocity scale is @xmath56 , which is 0.5 ( 0.5 microns / minute in physical units ) if @xmath2 and @xmath57 , corresponding to @xmath58 changing by 2.5% across a single cell at the origin . cell - cell forces @xmath9 are chosen to be stiff springs so that clusters are effectively rigid ( see _ appendix _ for details ) . depends strongly on the angle @xmath59 between the cell - cell axis and the chemotactic gradient . cell pairs also drift perpendicular to the gradient , @xmath60 . @xmath61 is the velocity scale ; @xmath2 . simulations are of eqs . [ eq : position]-[eq : polarity ] . we compute @xmath62 by tracking the instantaneous angle , then averaging over all velocities within the appropriate angle bin . error bars here and throughout are one standard deviation of the mean , calculated from a bootstrap . over @xmath63 trajectories of 6@xmath14 ( 120 minutes ) are simulated.,width=340 ] we can also compute @xmath30 and hence @xmath64 for larger clusters ( table s1 , _ appendix _ , fig . a2 ) . for a cluster of @xmath65 layers of cells surrounding a center cell , @xmath66 , with @xmath67 . a cluster with @xmath65 layers has @xmath68 cells ; thus the mean velocity of a @xmath65-layer cluster is given by @xmath69 , where @xmath70 is the angular average of @xmath30 . we predict that @xmath71 first increases with @xmath33 , then slowly saturates to @xmath72 . this is confirmed by simulations of the full model ( fig . [ fig : clustersize]a ) . we note that @xmath64 is an average over time , and hence orientation ( see below , _ appendix _ ) . we can see why @xmath64 saturates as @xmath73 by considering a large circular cluster of radius @xmath74 . here , we expect @xmath75 on the outside edge , where @xmath76 is a geometric prefactor and @xmath77 is the outward normal , with @xmath78 elsewhere . then , @xmath79 , independent of cluster radius @xmath74 . a related result has been found for circular clusters by malet - engra et al . @xcite ; we note that they do not consider the behavior of single cells or cluster geometry . the efficiency of cluster chemotaxis may be measured by chemotactic index ( @xmath80 ) , commonly defined as the ratio of distance traveled along the gradient ( the @xmath3 displacement ) to total distance traveled @xcite ; @xmath80ranges from -1 to 1 . we define @xmath81 , where the average is over both time and trajectories ( and hence over orientation ) . the chemotactic index @xmath80may also be computed analytically , and it depends on the variance of @xmath82 , which is @xmath83 . in our model , @xmath84 only depends on the ratio @xmath85 of mean chemotactic velocity to its standard deviation , @xmath86 where @xmath87 is a generalized laguerre polynomial . when mean cluster velocity is much larger than its fluctuations , @xmath88 and @xmath89 , but when fluctuations are large , @xmath90 and @xmath91 ( _ appendix _ , fig . a3 ) . together , eq . [ eq : shallow ] , eq . [ eq : ci ] and table s1 provide an analytic prediction for cluster velocity and @xmath80 , with excellent agreement with simulations ( fig . [ fig : clustersize ] ) . we note that @xmath71 only depends on cluster configuration , where @xmath48 , so @xmath92 collapses onto a single curve as the gradient strength is changed ( fig . [ fig : clustersize]a ) . by contrast , how @xmath80increases with @xmath33 depends on @xmath93 and @xmath94 ( eq . [ eq : ci ] , fig . [ fig : clustersize]b ) . in a cluster increases , the mean velocity @xmath64 increases with @xmath33 but then saturates ; the mean velocity can be collapsed onto a single curve by rescaling by @xmath61 . * b , * the chemotactic index @xmath80also saturates to its maximum value . black squares and lines are the orientationally - averaged drift velocity computed for rigid clusters by eq . [ eq : shallow ] and eq . [ eq : ci ] . colored symbols are full model simulations with strong adhesion . cell cluster shape may influence @xmath64 ( _ appendix _ fig . a4 ) ; our calculations are for the shapes in table s1 . error bars here are symbol size or smaller ; @xmath95 trajectories of @xmath96 are used for each point.,width=321 ] in our model , clusters can in principle develop a spontaneous rotation , but in practice this effect is small , and absent for symmetric clusters ( see _ appendix _ ) . _ motion in non - rigid clusters. _ while we studied near - rigid clusters above , our results hold qualitatively for clusters that are loosely adherent and may rearrange . cell rearrangements are common in many collective cell motions @xcite , but we note that in @xcite clusters are more rigid . we choose cell - cell forces @xmath9 to allow clusters to rearrange ( see _ appendix _ , @xcite ) , and simulate eqs . [ eq : position]-[eq : polarity ] . as in rigid clusters , @xmath64 increases and saturates , while @xmath80increases toward unity , though more slowly than a rigid cluster ( fig . [ fig : fluid]ab ) . clusters may fragment ; with increasing @xmath3 , @xmath58 increases and the cluster breaks up ( fig . [ fig : fluid]c ) . cluster breakup can limit guidance if @xmath97 is too large , clusters are not stable , and will not chemotax . in fig . [ fig : fluid]ab , we compute @xmath80and velocity by averaging over all cells , not merely those that are connected . if we track cells ejected from the cluster , they have an apparent @xmath98 , as they are preferentially ejected from the high-@xmath58 cluster edge ( _ appendix _ ) . experimental analysis of dissociating clusters may therefore not be straightforward . anisotropic chemotaxis is present in non - rigid pairs , though lessened because our non - rigid pairs rotate quickly with respect to @xmath14 ( _ appendix _ ) . in a cluster increases , the mean velocity @xmath64 increases with @xmath33 but then saturates . * b , * chemotactic index also approaches unity , but slower than in a rigid cluster . rigid cluster theory assumes the same cluster geometries as in fig . [ fig : clustersize ] . averages in * a - b * are over @xmath99 trajectories ( ranging from @xmath100 for @xmath101 to @xmath102 for @xmath103 ) , over the time @xmath104 to @xmath105 . * c , * breakdown of a cluster as it moves up the chemoattractant gradient . x marks the initial cluster center of mass , o the current center . @xmath106 , @xmath107 in this simulation.,width=340 ] _ distinguishing between potential collective chemotaxis models._our model explains how chemotaxis can emerge from interactions of non - chemotaxing cells . however , other possibilities exist for enhancement of chemotaxis in clusters . coburn et al . showed that in contact - based models , a few chemotactic cells can direct many non - chemotactic ones @xcite . if single cells are weakly chemotactic , cell - cell interactions could amplify this response or average out fluctuations @xcite . how can we distinguish these options ? in lymphocytes @xcite , the motion of single cells oppositely to the cluster immediately rules out simple averaging or amplification of single cell bias . more generally , the scaling of collective chemotaxis with cluster size does not allow easy discrimination . in fig . [ fig : clustersize ] , at large @xmath33 , @xmath64 and @xmath80saturate . as an alternate theory , suppose each cell chemotaxes noisily , e.g. @xmath108 , where @xmath109 are independent zero - mean noises . in this case , @xmath110 independent of @xmath33 , and @xmath111 , as in our large-@xmath33 asymptotic results and the related circular - cluster theory of @xcite . instead , we propose that orientation effects in small clusters are a good test of emergent chemotaxis . in particular , studying cell pairs as in fig . [ fig : pairs ] is critical : anisotropic chemotaxis is a generic sign of cluster - level gradient sensing . even beyond our model , chemotactic drift is anisotropic for almost all mechanisms where single cells do not chemotax , because two cells separated perpendicular to the gradient sense the same concentration . this leads to anisotropic chemotaxis unless cells integrate information over times much larger than the pair s reorientation time . by contrast , the simple model with single cell chemotaxis above leads to isotropic chemotaxis of pairs . how well does our model fit current experiments ? we find increasing cluster size increases cluster velocity and chemotactic index . this is consistent with @xcite , who see a large increase in taxis from small clusters ( @xmath112 cells ) to large , but not @xcite , who find that @xmath80is similar between small and large clusters , and note no large variations in velocity . this suggests that the minimal version of collective guidance as developed here can create chemotaxis , but does not fully explain the experiments of @xcite . there are a number of directions for improvement . more quantitative comparisons could be made by detailed measurement of single - cell statistics @xcite , leading to nonlinear or anisotropic terms in eq . [ eq : polarity ] . our description of cil has also assumed , for simplicity , that both cell front and back are inhibitory ; other possibilities may alter collective cell motion @xcite . we could also add adaptation as in the legi model @xcite to enable clusters to adapt their response to a value independent of the mean chemoattractant concentration . we will treat extensions of this model elsewhere ; our focus here is on the simplest possible results . in summary , we provide a simple , quantitative model that embodies a minimal version of the collective guidance hypothesis @xcite and provides a plausible initial model for collective chemotaxis when single cells do not chemotax . our work allows us to make an unambiguous and testable prediction for emergent collective guidance : pairs of cells will develop anisotropic chemotaxis . although there has been considerable effort devoted to models of collective motility @xcite , ours is the first model of how collective chemotaxis can emerge from single non - gradient - sensing cells via collective guidance and regulation of cil . bac appreciates helpful discussions with albert bae and monica skoge . this work was supported by nih grant no . p01 gm078586 , nsf grant no . dms 1309542 , and by the center for theoretical biological physics . bac was supported by nih grant no . f32gm110983 .
loop quantum cosmology ( lqc ) arises from the application of the more general theory of loop quantum gravity ( lqg ) to cosmology . one of the most important predictions of lqc is that in the homogeneous and isotropic friedmann - robertson - walker models the classical big bang singularity is avoided being replaced by a bouncing universe . more recently , it has been shown that the big bang singularity is also resolved for anisotropic bianchi type i ( bi ) @xcite , ii ( bii ) @xcite and ix @xcite . usually the matter source that was considered is a massless scalar field that play the role of internal time . the investigations have been extended to more complicated models including a perfect fluid , magnetic fields @xcite , cosmological strings @xcite . it is remarkable the fact that in all these studies it was observed that the bounce prediction is robust . the purpose of this paper is to investigate the dynamics of a bii string cosmological model in the presence of a magnetic field in the framework of lqc . we show that a bounce occurs in a collapsing magnetized bii string cosmological model , thus extending the known cases of singularity resolution . the plan of the paper is as follows : in sect . 2 we outline the classical equations of a bii string cosmological model in the presence of a magnetic field . in sect . 3 we discuss the quantum theory introducing the effective equations for the model . 4 is devoted to numerical calculations and it is shown that the classical singularity is resolved in the bii effective quantum dynamics . finally , in sect . 5 we summarize our results . the spacetime metric of bii model is given by @xmath0 where @xmath1 and @xmath2 are the directional scale factors . the parameter @xmath3 makes the difference between bi ( @xmath4 ) and bii ( @xmath5 ) spacetimes . having in view a comparison between the classical theory and the effective theory from lqc it is useful to rewrite the theory in terms of triads and connections . taking into account that the spatial manifold is non - compact , one needs to introduce a fiducial cell @xmath6 with the coordinate lengths @xmath7 and the fiducial volume @xmath8 @xcite . the fiducial metric is @xmath9 with the co - triads @xmath10 and triads @xmath11 in terms of the fiducial triads @xmath12 and co - triads @xmath13 a convenient parametrization of the phase space variables @xmath14 is @xmath15 without sum over index @xmath16 . the connection and triad components @xmath17 and @xmath18 satisfy the poisson bracket @xmath19 where @xmath20 0.2375 is barbero - immirzi parameter . choosing the lapse function @xmath21 , in the hamiltonian formulation of the model we have the hamiltonian constraint @xcite @xmath22 + { \mathcal h}_m v = 0 \ , , \end{split}\ ] ] where @xmath23 denotes the physical volume of the cell @xmath6 . @xmath24 depending on whether the frame @xmath25 is right or left handed . without any loss of generality , we will choose the orientation to be positive in the following . the hamiltonian for the matter contribution @xmath26 is proportional to the matter energy density @xmath27 the triads @xmath18 are related to the directional scale factors as @xmath28 assuming @xmath29 , i.e. @xmath30 with the positive orientation of the triads . the relation between the phase space variables @xmath31 and the metric variables are @xcite @xmath32 in terms of the hubble parameters @xmath33 where the dot represents the derivative with respect to the harmonic time . einstein s equations are derived from hamilton s equations : @xmath34 where @xmath35 . using the explicit form of the hamiltonian @xmath36 we have the following equations @xcite : [ hamp ] @xmath37 [ hamc ] @xmath38 from the above equations of motion it can be observed that the classical solutions posses the following constants of motion : [ const ] @xmath39 with @xmath40 constants . these equations allow us to find exact analytically solutions for @xmath41 and @xmath42 : [ p2p3 ] @xmath43 with @xmath44 the initial values at @xmath45 of these variables . in our model the matter density @xmath46 comprises the contribution of cosmological string density @xmath47 and the energy density of the magnetic field @xmath48 the energy momentum tensor for a system of cosmic strings and magnetic field in a comoving coordinate system is given by @xmath49 where @xmath47 is the rest energy density of strings with massive particles attached to them @xcite . it can be expressed as @xmath50 where @xmath51 is the rest energy of the particles attached to the strings and @xmath52 is the tension density of the system of strings . the four velocities @xmath53 and the direction of the string @xmath54 obey the relations @xmath55 the electromagnetic field @xmath56 is taken in the form given by lichnerowich @xcite . we assume that the magnetic field is homogeneous aligned along the @xmath57-direction , @xmath58 and consequently the energy density of the magnetic field is @xcite @xmath59 where @xmath60 is a constant and @xmath61 is the magnetic permeability of the medium . typically @xmath61 differs from unity only by a few parts in @xmath62 and in our numerical simulations we shall take @xmath63 . taking into account the conservation of the energy - momentum tensor , i.e. , @xmath64 , after a little manipulation one obtains @xcite : @xmath65 here we take into account that the conservation law for magnetic field fulfills identically . usually it is assumed that @xmath47 and @xmath52 are proportional @xcite : @xmath66 where the constant @xmath67 is @xmath68 for the so called geometric string , greater than @xmath68 for takabayasi string and @xmath69 for reddy string . the solution of with the proportionality relation is @xmath70 with @xmath71 a constant of integration . in lqc the connection variables @xmath31 do not have direct quantum analogues and are replaced by holonomies . the quantum effects are incorporated in the _ effective _ hamiltonian , @xmath72 , constructed from the classical one , @xmath73 , by replacing the connection components @xmath31 with sine functions : @xmath74 where @xmath75 are real valued functions of the triad coefficients @xmath18 . in what follows we shall use the so called @xmath76-scheme in which the parameters @xmath77 are chosen as follows @xcite : @xmath78 where @xmath79 is the area gap in the lqc with the planck length @xmath80 . the equations of motion that incorporate loop quantum modifications are deduced accordingly @xcite . for the equations for @xmath31 in r.h.s . we shall take into account the contribution of the matter density . the equations for the effective theory are given by poisson brackets with the hamiltonian constraint ( @xmath81 ) : @xmath82 @xmath83\nonumber\\ & \quad + r\frac{\alpha-1}{2\alpha } p_1^{-\frac{1+\alpha}{2\alpha } } p_2^{\frac{1+\alpha}{2\alpha}}p_3^{\frac{1+\alpha}{2\alpha } } + \frac{{\cal j}^2}{2\mu } p_1 p_2 p_3^{-1}\,,\end{aligned}\ ] ] @xmath84 -\eta^2(1+\gamma^2 ) \nonumber \\ & \left.\qquad - \eta(\bar\mu_1c_1\cos\bar\mu_1c_1 -3\sin\bar\mu_1c_1 ) \frac{}{}\right]\nonumber\\ & \quad + r\frac{\alpha+1}{2\alpha } p_1^{\frac{\alpha-1}{2\alpha } } p_2^{\frac{1-\alpha}{2\alpha}}p_3^{\frac{1+\alpha}{2\alpha } } + \frac{{\cal j}^2}{2\mu } p_1 p_3^{-1}\,,\end{aligned}\ ] ] @xmath85 -\eta^2(1+\gamma^2 ) \nonumber \\ & \left.\qquad -\eta(\bar\mu_1c_1\cos\bar\mu_1c_1 -3\sin\bar\mu_1c_1 ) \frac{}{}\right ] \nonumber\\ & \quad + r\frac{\alpha+1}{2\alpha } p_1^{\frac{\alpha-1}{2\alpha } } p_2^{\frac{1+\alpha}{2\alpha}}p_3^{\frac{1-\alpha}{2\alpha } } -\frac{{\cal j}^2}{2\mu } p_1 p_2 p_3^{-2}\,.\end{aligned}\ ] ] where @xmath86 . the complexity of these equations does not allow for analytic solutions and imposes numerical simulations . however , taking into account that @xmath87 is bounded by @xmath68 for all @xmath88 it is possible to infer an upper bound for the density of matter @xcite : @xmath89 let us note that in the case of bi cosmology , the matter density is bounded by 0.41@xmath90 . in our numerical analysis we choose the units : @xmath91 . the graphics are plotted as functions of the harmonic time ( lapse function @xmath21 ) . in order to compare the classical solutions with those of the effective equations we investigate the evolution of the directional hubble parameters , the matter density , the volume and the shear @xmath92\,.\ ] ] in the figures red solid line shows the volume scale @xmath93 blue dashed line presents the density of matter @xmath46 , and black dotted lines is for the shear @xmath94 . the initial conditions for effective and classical solutions should be chosen in accordance with the hamiltonian constraints . in order to simplify the analysis we choose as initial conditions @xmath95 and @xmath96 at the initial moment @xmath97 . in fig . [ pc ] we plot the admissible values of @xmath98 which provide an acceptable value for the initial data @xmath99 . first of all we choose the initial conditions corresponding to a classically collapsing universe approaching the classical big bang singularity . for this purpose we take all initial @xmath31 as being negative . in fig . [ 4103 ] we have plotted the classical evolution of volume scale @xmath100 , matter density @xmath101 and shear @xmath94 for a positive proportionality constant @xmath102 with nonvanishing magnetic field ( @xmath103 ) and string density ( @xmath104 ) . the fig . [ 1103 ] shows quantum analog of fig . [ 4003 ] is the classical picture of evolution with a negative proportionality constant @xmath105 . the quantum counterpart of fig . [ 4003 ] is given in fig . [ 1003 ] . as it is seen from the figs . [ 4103 ] - [ 1003 ] , while the classical evolution ends in big crunch in quantum case the spacetime avoids singularity and after attaining some minimum value the universe again begins to expand . it should be noted that the qualitative pictures of evolution are almost the same for positive or negative values of @xmath67 . accordingly , in what follows we shall present only the plots for a single value of the parameter @xmath67 , namely @xmath106 . in figs . [ 2103 ] and [ 1113 ] we present the effective quantum evolution of the bii universe with the initial conditions @xmath107 and @xmath99 of different signs . we remark the appearance of a few oscillations in the vicinity of the classical singularity and from which the solutions evolve smooth in time . in fig . [ 5103 ] we present the evolution of the classical evolution of volume scale @xmath100 , matter density @xmath101 and shear @xmath94 choosing all initial conditions @xmath108 . their quantum counterpart is plotted in fig . [ 2113 ] . again we get that the classical bii universe evolves from a singularity , while in the lqc approach all observable are finite . in the vicinity of the classical singularity the effective solution presents a bounce and subsequently the evolution is similar with the classical one . in this paper we analyzed the numerical solutions of the effective equations of the lqc dynamics for bii model . we have extended the effective lqc treatment of bii cosmologies by including cosmic strings and a homogeneous magnetic field . we considered the analytical and numerical solutions of bii model at the classical and effective level . the main objective of the paper was the investigation how the classical big bang singularity is resolved and how the effective equations evolve . for this purpose we chose as a set of observable quantities like volume , string density , shear expansion and studied their evolution numerically . we showed that a big bounce occurs in a collapsing magnetized bii string universe , thus extending the known cosmological models of singularity avoidance . after the bounce the universe enters a classical regime . the numerical simulations are quite sensitive to the initial conditions regarding the directional hubble parameters . choosing a negative @xmath107 , the numerical simulations are appropriate to describe the evolution towards the big bang singularity . in the classical case the evolution stops at the singularity , while in the quantum case we have bounces and the singularity is eluded . on the other hand the choice of the initial conditions with a positive @xmath107 is adequate to have in view the evolution after the big bang . in the classical case , starting from the vicinity of the singularity , the volume of the universe tends to infinity , while the string density is smaller and smaller . in the quantum case , after a few small bounces we have for large time the same behavior as in the classical case . the study of anisotropic models with different kinds of matter sources in the framework of lqc deserves further investigations . these studies will contribute to answer the challenge question if the bouncing non - singular behavior of the effective solutions is generic . 99 a. ashtekar and e. wilson - ewing , _ phys . d _ * 79 * , 083535 ( 2009 ) . a. ashtekar and e. wilson - ewing , _ phys . d _ * 80 * , 123532 ( 2009 ) . e. wilson - ewing , _ phys . d _ * 82 * , 043508 ( 2010 ) . r. maartens and k. vandersloot , _ arxiv : 0812.1889 [ gr - qc]_. v. rikhvitsky , b. saha and m. visinescu , _ astrophys . space sci . _ * 339 * , 371 ( 2012 ) . a. corichi and e. montoya , _ phys . d _ * 85 * , 104052 ( 2012 ) . b. gupt and p. singh , _ phys . d _ * 85 * , 044011 ( 2012 ) . letelier , _ phys . * 28 * , 2414 ( 1983 ) . a. lichnerowicz , _ relativistic hydrodynamics and magnetohydrodynamics _ , ( benjamin , new york , 1967 ) . b. saha , v. rikhvitsky and m. visinescu , _ cent . j. phys . _ * 8 * , 113 ( 2010 ) . b. saha and m. visinescu , _ astrophys . space sci . _ * 315 * , 99 ( 2008 ) . d. w. chiou and k. vandersloot , _ phys . d _ * 76 * , 084015 ( 2007 ) . d. w. chiou , _ phys . d _ * 76 * , 124037 ( 2007 ) .
the loop quantum cosmology of the bianchi type ii string cosmological model in the presence of a homogeneous magnetic field is studied . we present the effective equations which provide modifications to the classical equations of motion due to quantum effects . the numerical simulations confirm that the big bang singularity is resolved by quantum gravity effects . -24pt 1 cm
the study of strongly interacting quantum models by direct realization of an experimental system reproducing the model properties , an idea proposed by feynman @xcite , was realized in the past ten years with the production of bose - einstein condensates ( bec ) and their use as quantum simulators " @xcite . in particular , bec in conjunction with optical lattices are used to produce systems reproducing the physics of well known quantum statistical discrete models such as fermionic or bosonic hubbard models . used to study simple models of bosons @xcite or fermions @xcite at low temperature , the flexibility offered by these systems extends the range of interesting models to more exotic ones which can be treated both experimentally and theoretically . examples include systems with long range interactions @xcite , fermions with imbalanced populations @xcite , mixtures of different kinds of particles @xcite and spin-1 bosons with spin - independent and spin - dependent interactions which allow interplay between superfluidity and magnetism @xcite . furthermore , it is possible to study systems of bosons with two effective internal degrees of freedom on an optical lattice , the so - called spin-@xmath0 bosons " . such a system , with spin - dependent interactions , could be produced by applying a periodic optical lattice on a bosonic system with two triply degenerate internal energy levels . the optical potential applied would localise the atoms at the nodes of a regular network , but would also couple the internal states by @xmath1 or v virtual processes , thus leaving only two internal low energy degenerated states denoted @xmath2 and @xmath1 and realizing an effective spin-@xmath0 model @xcite . the presence of the spin - dependent interaction introduces a term in the hamiltonian which permits the conversion of two particles of one type into the other type and renders numerical simulations more difficult . this model is related , but not identical , to other models including p - band superfluid models @xcite and the bosonic kondo model @xcite . understanding the phase diagram and properties of the simpler spin-@xmath0 bosonic system takes us a step toward understanding the more elaborate , and more difficult to simulate , models . the spin-@xmath0 model has been extensively studied with mean - field theory ( mft ) at zero or finite temperatures and , in previous work , we explored its zero temperature behavior with quantum monte carlo ( qmc ) simulations in one @xcite and two @xcite dimensions for on - site repulsive interactions . at zero temperature , the phase diagrams obtained in one and two dimensions , with mft or qmc , are similar . generally speaking , at zero temperature , the system can adopt two different kinds of phases : insulating mott phases that appear for integer density , @xmath3 , and for large enough repulsion between particles and superfluid phases ( sf ) otherwise . the detailed nature of these phases depends on the interactions between the different kinds of particles . in the case where the repulsion between identical particles is smaller than between different particles ( @xmath4 in our previous work @xcite and in the following ) the superfluid is found to be polarized , that is , an imbalance develops in the populations of the two kinds of particles and one of the species becomes dominant . the @xmath5 mott phase is also polarized whereas the @xmath6 phase is not ( we did not study higher densities with qmc ) . in the opposite case , @xmath7 , all the phases are unpolarized . noteworthy is the presence of coherent exchange movements @xcite , where two particles of different types exchange their position in the @xmath5 mott phase in both cases , as well as in the @xmath6 mott phase for @xmath8 . finally , in one dimension , all zero temperature phase transitions were found to be continuous whereas in two dimensions and when @xmath9 is small enough , the @xmath6 mott - superfluid transition was predicted by mft to be first order near the tip of the mott lobe and continuous otherwise , whereas for larger @xmath10 the transition was predicted to be always continuous . this was confirmed by qmc simulations @xcite . related spin-1 models were studied using mft @xcite or qmc in one dimension @xcite and a similar spin 1/2 bosons model was also recently studied @xcite . in this paper , we will study the spin-@xmath0 model at finite temperature in two dimensions and compare with mft predictions . the results for finite system sizes and temperatures are relevant to experimental efforts to study this system . in section ii , we will introduce the model and the mft and qmc techniques used to study it . section iii and iv will be devoted to the presentation of the results obtained for the @xmath9 and @xmath11 cases , respectively . we will summarize these results and give some final remarks in section iv . the model we will study is the same we previously studied in the low temperature limit in one @xcite and two dimensions @xcite and previously introduced in @xcite . it is an extended hubbard model governed by the hamiltonian @xmath12 where operator @xmath13 ( @xmath14 ) destroys ( creates ) a boson of type @xmath15 on site @xmath16 of a two - dimensional square lattice of size @xmath17 . the operators @xmath18 measures the number of particles of type @xmath19 on site @xmath20 . the densities of particles of type 0 and @xmath1 are called @xmath21 and @xmath22 while the total density is called @xmath23 . the first term ( [ h1 ] ) of the hamiltonian is the kinetic term that lets particles hop from site @xmath20 to its nearest neighbours @xmath24 . the associated hopping energy @xmath25 sets the energy scale . a chemical potential @xmath26 is added if one works in the grand canonical ensemble . the second term ( [ h2 ] ) describes on - site repulsion between identical particles with a strength @xmath27 or between different particles with a strength @xmath28 . we will study both the positive and negative @xmath10 cases but will keep only repulsive interactions , that is @xmath29 , and a fixed moderate value of @xmath30 . the last term ( [ h3 ] ) provides a possibility to change the spins " of the particles : when two identical particles are on the same site , they can be transformed into two particles of the other type . it was demonstrated in @xcite that the matrix element associated with this conversion is @xmath31 . we are using a different sign for the term ( [ h3 ] ) compared to the articles where the model was originally introduced @xcite but we have shown in a previous work @xcite that this sign can indeed be chosen freely due to a symmetry of the model . the only term that couples different sites in the hamiltonian is the hopping term ( [ h1 ] ) . introducing the field @xmath32 , we replace the creation / destruction operators on site @xmath33 by their mean values @xmath34 , following the approach used in @xcite . the hamiltonian on site @xmath20 is then decoupled from neighboring sites and can be easily diagonalized numerically in a finite basis . the optimal value of the fields @xmath34 are then chosen by minimizing the grand canonical potential @xmath35 with respect to @xmath34 where @xmath36 is the grand canonical partition function . the system is in a superfluid phase when @xmath34 is nonzero , with superfluid density @xmath37 where @xmath38 is the total number of particles on site @xmath20 . @xmath39 is close to zero in the mott phase and much larger in the liquid phase . while this mft was shown to reproduce qualitatively the phase diagram at zero temperature @xcite , it is rather limited at finite temperature indeed , whenever the @xmath34 are zero the hopping parameter @xmath40 no longer plays a rle in the mft . then , while the mft can distinguish between sf and unordered phases , it does not correctly distinguish mott insulator ( mi ) regions from normal bose liquids ones , as the crossover boundary between those regions will not depend on @xmath40 and will be the same as in the @xmath41 case . to simulate this system , we used the stochastic green function algorithm ( sgf ) @xcite , an exact quantum monte carlo ( qmc ) technique that allows canonical or grand canonical simulations of the system at finite temperatures as well as measurements of many - particle green functions . in particular , this algorithm can simulate efficiently the spin - flip term in the hamiltonian . we studied sizes up to @xmath42 . the density @xmath3 is conserved in canonical simulations , but individual densities @xmath21 and @xmath22 fluctuate due to the conversion term eq . ( [ h3 ] ) . the superfluid density is given by fluctuations of the total winding number , @xmath43 , of the world lines of the particles @xcite @xmath44 the superfluid density can not be measured separately for @xmath2 and @xmath1 particles due to the conversion term @xcite . it , therefore depends on the total density not on the separate densities of the two species . we also calculate the one particle green functions @xmath45 with @xmath46 . @xmath47 measures the phase coherence of individual particles . in a strongly correlated system it is useful to study correlated movements of particles which can be done , for example , by studying two - particle green functions . we found @xcite that anticorrelated movements of particles govern the dynamics of particles inside mott lobes , as the particles of different types exchange their positions . the two - particles anti - correlated green function @xmath48 measures the phase coherence of such exchange movements as a function of distance . due to its definition , @xmath49 can not be larger that @xmath50 and is equal to @xmath51 if there is no correlation between the movements of 0 and @xmath1 particles . in two dimensions , at low but finite temperatures , we expect , in some cases , to observe a berezinsky - kosterlitz - thouless ( bkt ) phase transition and the different green functions to adopt a power law behavior at large distance @xmath52 , @xmath53 , characteristic of the appearance of a quasi long range order ( qlro ) in the phase , long range order ( lro ) being achieved only at @xmath54 ( @xmath55 varying between 1/4 and 0 as @xmath56 is lowered ) @xcite . in other words , at finite @xmath56 we can expect a superfluid where the phase is stiff but not ordered and not a bec whith an ordered phase . at high temperature , the green functions are of course expected to decay exponentially . on the finite size systems which we study ( @xmath57 ) , it is difficult to distinguish the qlro from a true lro , whereas one can easily distinguish between the qlro and exponentially decreasing regimes . this difficulty of distinguishing qlro from lro is also encountered in experiments where the sizes of systems that can be studied are typically of the same order as in our qmc simulations ( hundreds of particles ) . the finite size results are , therefore , directly relevant to experiments . to elucidate the properties the model , we formulate it in terms of spins using a schwinger bosons approach @xcite . defining the spin operators @xmath58 , @xmath59 , and @xmath60 the hamiltonian takes the form @xmath61\right ) \label{hspart2}\end{aligned}\ ] ] where @xmath62 is the hopping term eq . ( [ h1 ] ) . the terms in ( [ hspart1 ] ) are invariant under spin rotations . on the other hand , term ( [ hspart2 ] ) favors pseudo - spin correlations to develop along the @xmath63 axis if @xmath7 or in the @xmath64 plane for @xmath65 . we note that the total spin @xmath66 on a given site is not fixed but depends on the total number of particles on the site ( @xmath67 ) and will then fluctuate with this number . an order along @xmath68 is measured through the densities or through density - density correlations of 0 or @xmath1 particles , _ i.e. _ it corresponds to the polarization of the system . an order along the @xmath63 or @xmath69 axes is exposed through the behavior of @xmath49 which , in terms of spins , is equal to @xmath70 . our qmc algorithm allows the calculation of @xmath71 but does not give access to correlations along the @xmath63 and @xmath69 axes independently . we remark that the @xmath69 and @xmath68 axis have the same behavior , which means that we expect a spin qlro in the @xmath64 plane to appear at low enough temperature for @xmath65 in addition to the expected qlro of the global phase of the particles discussed earlier . on our finite size systems , this means that we should simultaneously observe a polarization and a qlro for @xmath49 . we will call such a phenomenon a quasi - polarization " ( qp ) . for @xmath65 and finite @xmath56 , we expect a qlro to develop along the @xmath63 axis and , consequently , no polarization but still a plateau at long distances in the function @xmath49 . at zero temperature , the results obtained with qmc and mft were in good qualitative agreement @xcite . the phase diagram , studied for densities up to two , exhibits three phases . the first two phases are incompressible mott phases obtained for integer densities @xmath5 and @xmath6 for large enough interactions @xmath27 . at zero temperature , the entire @xmath5 mott phase is polarized , that is the system sustains a spontaneous symmetry breaking and the density of one type of particles becomes dominant . this polarization can be understood in the framework of an effective spin-1/2 model @xcite as the coupling in the @xmath64 plane is stronger than along the @xmath63 axis . as expected , exchange movements of particles coexist with polarization in this phase and @xmath49 develop a long range phase coherence . the @xmath6 mott phase is unpolarized and show no sign of exchange moves . in terms of spin , this corresponds to the fact that , neglecting the kinetic term , the ground state for a given site is uniquely determined as the state with @xmath73 . the third possible phase is a polarized superfluid ( sf ) and occurs at any @xmath27 when the density is incommensurate and also at small @xmath27 for commensurate values . it is not possible to discuss this phase in terms of a simple effective spin degree of freedom as the number of particles on a site is not fixed . at zero temperature , the transition from the @xmath5 mott phase to the sf is continuous , whereas at the tip of the @xmath6 mott lobe , the transition to the sf is first order for small values of @xmath74 becoming second order for larger values . this was predicted by the mft @xcite and confirmed by qmc simulations @xcite . and @xmath6 mott regions ( mi ) for different values of the temperature @xmath76 and for different sizes @xmath77 . as the temperature increases , the mott lobes shrink and totally disappear for @xmath78 . outside the mott , the system crosses over to a normal bose liquid ( nbl ) and , eventually , transitions to a quasi - polarized superfluid ( qp - sf ) phase ( see below ) . in the limit of zero temperature ( here represented by the low temperature @xmath79 ) , there is a direct phase transition between a mott phase and a superfluid . [ mottlimits],width=321 ] for @xmath80 . the qmc simulations ( top ) show that a liquid ( nbl ) region appears between the quasi polarized superfluid ( qp - sf ) phase and the mott region . the mft ( bottom ) is not able to reproduce correctly this result as it does not take into account the kinetic term in the unordered ( mott or nbl ) phase . it predicts first order phase transition between the sf and the other phase at the tip of the @xmath5 and @xmath6 lobes : the region of coexistence of the two phases is limited by the dashed line and the redline with squares . [ phasediagram],title="fig:",width=321 ] for @xmath80 . the qmc simulations ( top ) show that a liquid ( nbl ) region appears between the quasi polarized superfluid ( qp - sf ) phase and the mott region . the mft ( bottom ) is not able to reproduce correctly this result as it does not take into account the kinetic term in the unordered ( mott or nbl ) phase . it predicts first order phase transition between the sf and the other phase at the tip of the @xmath5 and @xmath6 lobes : the region of coexistence of the two phases is limited by the dashed line and the redline with squares . [ phasediagram],title="fig:",width=321 ] to map the phase diagram at finite temperature with qmc , we determine the limit of the mott insulator regions by measuring the density as a function of @xmath26 and determining the boundaries of the plateaux indicating the almost incompressible regions . as explained in sec . iia , although the compressibility of these regions is very small , they are not strictly incompressible due to thermal fluctuations . the evolution from mi to nbl does not show any singularity and is then simply a crossover between two different limiting behaviors , incompressible and compressible , of the same unordered phase . strictly speaking , a truly incompressible mott phase exists only at @xmath54 but , following convention , we will continue to refer to this finite @xmath56 region as a mi . to define the crossover boundary between the mi and the nbl shown in phase diagrams , we use the following criterion : when the density deviates by 1% from the total integer density we consider that the system is no longer in the mi region but in the nbl region . figure [ mottlimits ] shows the limits of the mi regions for different temperatures . as expected , the mi regions progressively disappear as the temperature is increased . the limits of the superfluid phase are located by direct measurement of @xmath81 . when the system is compressible and has zero superfluid density , there is a normal bose liquid . [ phasediagram ] ( top ) shows the qmc phase diagram of the system at a constant finite temperature . as the sf and mi regions are destroyed due to thermal fluctuations , an intermediate nbl region appears . the mft used with success at @xmath54 , where it reproduces qualitatively the phase diagram , is unable to do so at finite temperature as explained in sec . figure [ phasediagram ] ( bottom ) shows the mft phase diagram where the afore mentioned problem clearly appears : the boundaries between the mi and nbl do not depend on the value of @xmath82 and the mft is unable to give correct predictions regarding this crossover . however , the boundaries of the sf are reasonably well reproduced . a surprising result is that the transition between the unordered phase and the sf appears to be first order at the tip of the @xmath5 and @xmath6 lobes in this mft approach . at zero temperature , only @xmath6 showed a first order transition . as will be shown below , for finite temperatures , this is in total contradiction with the results obtained by qmc simulations that show continuous phase transitions for @xmath5 and @xmath6 . so this mft provides incorrect description of the phase transition and of the position of the different regions at finite temperature . for fixed integer density at zero temperature , we have a quantum phase transition ( qpt ) between the mi and sf phases . as expected @xcite , when the temperature is increased from zero , an intermediate compressible unordered region appears between the superfluid phase and the mott region , namely the normal bose liquid region ( nbl ) . this is observed for @xmath5 ( fig . [ afrho1 ] ) and @xmath6 ( fig . [ afrho2 ] ) . and for different sizes showing the quantum phase transition point between the mott insulating phase ( mi ) and the superfluid at @xmath83 . the finite @xmath56 superfluid phase does not have bec in this two - dimensional system . the system is quasi - polarized throughout the superfluid phase , but it is not in the nbl and mi . the mott phase is polarized at @xmath54 and should be quasi - polarized at extremely small @xmath56 . the transition between the sf and the nbl is continuous and there is only a crossover separating nbl from mi . [ afrho1],width=321 ] and for different sizes showing the quantum phase transition point between the mi and the sf occurring only at @xmath83 . the superfluid phase is always quasi - polarized whereas the nbl and mi are not . the sf - nbl transition is continuous , whereas at @xmath54 the mi - sf quantum phase transition was found to be discontinuous for @xmath74 small enough @xcite . the dash line corresponds to the case studied in fig . [ histo].,width=321 ] as for the possible polarization of the different phases , we observe that the sf phase appears to be polarized at finite @xmath56 as it is at @xmath54 : the histogram of the density of one of the species shows two peaks at low and high densities ( see fig . [ histo ] ) . however , due to the continuous symmetry in the @xmath64 plane of the pseudo - spin part of the hamiltonian eq . ( [ hspart2 ] ) , no lro exists at finite @xmath56 . therefore , this apparent polarization is due to the finite size of the system , and is , in fact , a quasi - polarization . an intuitive way to understand this quasi - polarization is as follows : in the superfluid phase , the pseudo - spins in the @xmath64 plane are stiff and appear to be mostly aligned in the same direction on a finite lattice , such as the case here . but since the symmetry is not broken , this `` magnetization '' direction in the @xmath64 plane will drift and point in all directions . as the direction changes , so does the projection of this pseudo - spin on the @xmath68-axis . in other words , the polarization drifts too , and changes with time , giving the double peak structure to the polarization histogram , fig . [ histo ] . we note that experimental systems are typically of the same sizes as the ones we study here and , therefore , this polarization drift will be present in these experiments too : the particle content of the system will change as a function of time . this quasi - polarization disappears as @xmath56 increases when the system undergoes a thermal bkt phase transition into the nbl . in other words , the entire sf phase is quasi - polarized but the nbl is not ( see fig . [ histo ] ) as a function of temperature for @xmath84 , @xmath85 , @xmath6 and @xmath86 . when @xmath87 , the system is in the quasi - polarized sf phase . for @xmath88 , the system is in the unordered unpolarized phase . the temperature at which the quasi - polarization disappears is approximately the same as the temperature where @xmath81 becomes zero ( see the vertical dashed line in fig . [ afrho2 ] ) . only the sf phase is quasi - polarized.[histo],width=321 ] a histogram of the polarization in the @xmath5 mi region shows that as soon as the temperature is increased from zero , the polarization disappears and the populations become balanced . according to the effective pseudo - spin model @xcite , it is possible to observe quasi - polarization as in the sf phase . in this case , the coupling generating these spin correlations and the polarization of the mi is of order @xmath89 . however , even for temperatures as low as @xmath90 , the system is already in the regime where correlations decay exponentially and we do not observe any qlro at finite @xmath56 in the mott region . this is confirmed by the behavior of the green functions that is shown in fig . [ green ] . in fig . [ green](a ) we show the anticorrelated green function @xmath91 , eq . ( [ greenf ] ) , in the @xmath5 mi region as @xmath56 is increased . we see that as soon as @xmath56 becomes finite , @xmath49 decays exponentially with distance , contrary to its constant value at long distance observed at @xmath54 @xcite . this exponential decay of course persists in the nbl region . in fig . [ green](b ) we show @xmath92 , eq . ( [ green1 ] ) , for various @xmath56 values at @xmath6 and @xmath93 which , at @xmath54 , puts the system in the sf phase . we see that for low @xmath56 , @xmath94 decays slowly . this decay is expected to be a power law but the system size is too small to show that unambiguously . as @xmath56 is increased , the system transitions into the nbl where @xmath94 exhibits exponential decay clearly distinguishing the nbl and sf phases . a similar behavior is found for @xmath95 and @xmath49 , the latter being expected since it accompanies the presence of quasi - polarization . , eq . ( [ greenf ] ) , in the @xmath5 mi and @xmath94 , eq . ( [ green1 ] ) , in the sf as functions of distance @xmath96 for different temperatures , with @xmath97 and @xmath80 . ( a ) in the @xmath5 mi , @xmath49 decays exponentially at finite @xmath56 , unlike its @xmath54 behavior where it reaches a constant value @xcite . a regime of qlro is not observed and would occur only at very small @xmath56 . the exponential decay persists in the nbl at higher @xmath56 . ( b ) @xmath94 at @xmath6 for @xmath56 values taking the system from the sf to the nbl . same case as in fig . [ histo ] . in the sf , @xmath94 decays slowly as a power law , but decays exponentially in the nbl . similar behavior is found for @xmath95 and @xmath49 . [ green ] , width=321 ] at finite temperature , @xmath98 , the transitions between the sf and the nbl are continuous for @xmath5 as well as for @xmath6 ( and of course for all other densities ) . since this transition takes place at constant density and is therefore a phase - only transition , it is expected to be in the berezinsky - kosterlitz - thouless universality class ( bkt ) for our two - dimensional system . actually , below the critical temperature , we have two different quasi long range orders that occur : the global @xmath99 phase qlro associated with the superfluid behavior and the pseudo - spin one , associated with the quasi - ordering of the spins in the @xmath64 plane , _ i.e. _ the so - called quasi polarization . as explained previously , these two qlro appear simultaneously . since the transition is bkt , we first determined the transition temperature using the universal jump of the superfluid density @xcite , where , at the transition temperature , @xmath100 , we have @xmath101 . to observe this , we calculated @xmath81 as a function of temperature and determined @xmath102 graphically as the intersection of @xmath103 with @xmath104 ( see fig . [ univjump ] ) . we also calculated the specific heat @xmath105 and determined the transition temperatures @xmath106 as the temperature where @xmath105 reaches its maximum ( fig . [ univjump ] ) . our qmc simulations were done for @xmath84 because @xmath105 is extremely difficult to calculate at low temperature for @xmath107 for larger sizes . and specific heat @xmath105 as functions of temperature for different values of @xmath82 . the universal jump condition states that @xmath108 . the transition temperatures are calculated independently by determining the maximum of @xmath109 . [ univjump],width=321 ] .values of the transition temperatures for different values of @xmath82 for an @xmath110 system . @xmath102 is obtained from the universal jump while @xmath106 from the maximum of @xmath109 . the values are in agreement and show that the universal prediction is verified . the data are taken from fig . [ univjump ] . [ tab1 ] [ cols="^,^,^",options="header " , ] and @xmath6 mott regions ( mi ) for different values of the temperature @xmath76 and for different sizes @xmath77 in the @xmath7 case . as in the positive @xmath10 case , the mott phases disappear for @xmath111 . in the @xmath54 limit , there is a direct transition between the mi and sf.[mottlimit2],width=321 ] in the @xmath112 case . all phases are unpolarized and the sf - nbl transition at finite temperature as well as the mi - sf transition at zero temperature are continuous . a similar phase diagram is observed for @xmath6.[rho12_u2neg],width=321 ] ( a ) and @xmath6 ( b ) mott regions for negative @xmath10 . in both cases , the long range order of the anticorrelated movements that was present at @xmath54 is lost even for very small temperatures.[green_u2neg],width=321 ] we also examined the anticorrelated green function @xmath49 in both @xmath5 and @xmath6 mi regions at various temperatures . in the zero temperature limit , both these phases exhibit nonzero values of @xmath113 at long distances . this is expected in both cases , considering the pseudo - spin hamiltonian . neglecting the kinetic energy , for @xmath7 , the energy is minimized by maximizing @xmath114 on each site . for @xmath5 and @xmath6 , there are two degenerates states that achieve this : the @xmath115 for @xmath5 and the @xmath116 for @xmath6 . these degenerate ground states are coupled by second order hopping processes which lift this degeneracy through coherent anticorrelated movements @xcite . in fig . [ green_u2neg ] , we show @xmath49 in the @xmath5 and @xmath6 mi regions for different temperatures . as expected , the phase coherence once again completely disappears rapidly as the temperature is increased from zero and we do not observe any sign of quasi long range phase coherence in the mi region , as well as in the nbl phase . studying a bosonic spin-@xmath0 hubbard model at finite temperature and comparing to the @xmath54 case , we find that the effect of temperature is dramatically different depending on the phase we consider . the superfluid phases are essentially unchanged by raising the temperature : the long range order present at @xmath54 is transformed into qlro . on the other hand , the mi regions are drastically modified as the polarization that occurs in certain cases is almost immediately wiped out by thermal fluctuations : low temperature quasi polarized states are not found in the regime of temperatures we studied . this is due to the fact that the energy scales associated with the polarization of the system take different values in these different phases . in the mi regions , the polarization is due to the coupling between different low energy degenerate mott states , as emphasized by the pseudo - spin theory @xcite . these couplings are of order @xmath89 and the pseudo - spin quasi - ordering vanishes very rapidly with temperature . on the other hand , in the superfluid , the energy scale associated with the coupling of pseudo - spins is obviously much larger . while it is not possible to specify this scale as precisely as in the mott phases , due to the itinerant nature of the particles in the superfluid regime , a simple argument shows that it is of order @xmath10 : whenever the particles enter the superfluid phase and adopt delocalized states , they overlap ; there is an interaction cost which is then of order @xmath27 for identical particles and @xmath28 for different ones . this favors having the particles mostly of the same type for @xmath65 and leads to a quasi polarization . for @xmath7 , the interaction favors having a mixture of particles and gives a system without any sign of polarization . in both cases , the energy scale is typically @xmath117 the finite temperature phase diagram presented here is important for the proper interpretation of experimental realization of this and related systems using ultra - cold atoms loaded on optical lattices . such experiments are , of course , always at finite temperature . similarly to what happens for fermions , we observe that the spin correlations in the mott phases will be very difficult to access experimentally as the associated energy scales are very small and as the correlations are almost immediately wiped out by thermal fluctuations . on the other hand , due to the relatively small sizes of experimental systems and to its larger associated energies , the quasi - ordering of spins in the superfluid phase should be immediately visible experimentally and indistinguishable from true polarization of the system . in a finite size system , one would expect to observe a slow drift of the polarization as the @xmath64 symmetry of the system is restored over time . from a more technical point of view , we have elucidated the limitations of the mft commonly used in the literature . we found that it is unable to distinguish correctly the mi and nbl regions . the mft does predict reasonably well the nbl - sf boundaries but not the nature of the transition which is sometimes predicted to be of first order whereas direct qmc simulations , and symmetry considerations , show that it is in the bkt universality class . furthermore , we found that mft predicts direct first and second order transitions at finite @xmath56 between the mi and sf phases which qmc shows do not exist , ( see figs . [ afrho1 ] , [ afrho2 ] , [ rho12_u2neg ] ) . similar incorrect mft behavior was found for the spin-@xmath118 model and the same caveats should be applied for example in ref.@xcite . this work was supported by : the cnrs - uc davis epocal lia joint research grant ; by nsf grant oise-0952300 ; an aro award w911nf0710576 with funds from the darpa ole program . we would like to thank michael foss - feig and ana - maria rey for useful input and discussion . m. vengalattore , s.r . leslie , j. guzman , d.m . stamper - kurn , phys . lett . * 100 * , 170403 ( 2008 ) ; + m. vengalattore , j. guzman , s. r. leslie , f. serwane , and d.m . stamper - kurn , phys . a * 81 * , 053612 ( 2010 ) .
we study a two - species bosonic hubbard model on a two - dimensional square lattice by means of quantum monte carlo simulations and focus on finite temperature effects . we show in two different cases , ferro- and antiferromagnetic spin - spin interactions , that the phase diagram is composed of a superfluid phase and an unordered phase that can be separated into weakly compressible mott insulators regions and compressible bose liquid regions . the superfluid - liquid transitions are of the berezinsky - kosterlitz - thouless type whereas the insulator - liquid passages are crossovers . we analyse the pseudo - spin correlations that are present in the different phases , focusing particularly on the existence of a polarization in this system .
the vacuum of quantum chromodynamics ( qcd ) is a prominent nonperturbative system , whose strongly interacting nature persists even above the transition to the quark - gluon plasma . to get insight into its mechanism , spectral properties of the qcd dirac operator are very useful . a nonzero density of eigenvalues at zero gives rise to chiral symmetry breaking via the banks - casher formula @xcite . moreover , exact zero modes are related to the topological charge via index theorems . in recent years , localization properties of the dirac eigenmodes have attracted attention as they can be used to draw analogies to condensed matter phenomena : concepts like the mobility edge and anderson localization can be studied in qcd lattice simulations . in this spirit the chiral transition at finite temperature has been conjectured to be an anderson ( metal - insulator ) transition . this goes hand in hand with different random matrix theory ( rmt ) descriptions of the dirac spectra . in the low temperature phase , the existence of the chiral condensate connects qcd to chiral perturbation theory and random matrix theory ( becoming exact in the epsilon - regime ) , which explains the statistics of the low lying part of the dirac spectrum @xcite . the spectral gap in the high temperature phase , on the other hand , seems to call for the `` soft edge '' description of rmt , which , however , could not be supported by lattice data @xcite . instead , a transition to independent eigenmodes obeying poisson statistics has been suggested @xcite . a refined analysis by one of us has shown , that the bulk of the spectrum is still delocalized and subject to rmt , while the lowest lying eigenmodes display a transition to localization and poissonian behavior of the eigenvalues @xcite . this effect has been confirmed to be universal in the sense that it does not depend on the resolution of the lattice . observables , like e.g. the number of localized modes , rather scale with the physical volume . on the other hand , ref . @xcite found that localization is essentially a finite volume artifact . however , they used a different , less restrictive definition of localization and thus their results are not in conflict with the rest of the above cited literature . in this work we give another crucial ingredient of anderson localization in qcd , namely we identify the `` defects '' causing it . we show that at high temperature local polyakov loop fluctuations trap low - lying modes . the average polyakov loop as an order parameter of the deconfinement ( or center ) phase transition approaches 1 with increasing temperature . locally , however , the polyakov loop takes on other values , in particular close to other center elements , @xmath1 in the case of gauge group @xmath0 . the phase transition can actually be viewed as a percolation of the physical sector embracing such islands where the polyakov loop is close to other center elements @xcite . the locations of these polyakov loop fluctuations ( which are similar to weiss domains in ferromagnetism ) are correlated with maxima in the profile of low - lying dirac modes . we will demonstrate this for eigenmodes of both the overlap and staggered operator . similarities between the spectrum of the overlap and staggered dirac operator have already been found in the schwinger model @xcite and also in qcd @xcite . our present study is , however , the first one when similarity of staggered and overlap dirac eigen_modes _ is seen in lattice simulations and we consider this an important side result of our study . the observed localization can be understood via polyakov loops compensating the twist caused by the antiperiodic boundary conditions in the temporal direction . the corresponding matsubara frequency is effectively lowered ( locally ) which results in lower eigenvalues . actually , our finding has been inspired by a similar localization effect in the spectrum of the gauge - covariant laplace operator @xcite . the laplacian is the square of the dirac operator in the free case , so the twist picture applies . otherwise this operator does not share important chiral features like topological zero modes and condensates . likewise , our finding is consistent with the existence of a chiral condensate in the polyakov loop sector close to other center elements @xcite ( also needed for center symmetry breaking @xcite ) , which implies low dirac eigenvalues in islands of such polyakov loops . the connection of these islands to topological excitations like magnetic monopoles is attractive , but in its naive form contradicts the observed topological susceptibility quantitatively , see below . for the construction of random matrix models valid at high temperature we investigate the distribution of local polyakov loops and find them to be uncorrelated to a good approximation . hence , polyakov loops in fact provide the poissonian ingredient for the dirac spectrum . this is built into a novel anderson - like random matrix model through supplementing it by random matrix entries that represent nearest neighbor hoppings in three - dimensional space . we motivate this model and show , that with a few parameters it reproduces the main features of the dirac modes : chirality , spectral gap , rmt - poisson transition and localization ( to the analogue of local polyakov loops ) . our findings are based on quenched lattice simulations with the su(2 ) gauge group , we strongly believe that the described phenomena are present in more realistic gauge theories , too . + the paper is organized as follows . in the next section we describe the dirac spectra at high temperature including the rmt - poisson ( chaotic to integrable ) transition and the similarity of the staggered and overlap modes . [ sect polyakov ] is devoted to the connection between local polyakov loops and low - lying modes . in sect . [ sect interpretation ] thereafter we investigate two possible interpretations of this finding , effective matsubara frequencies and topological objects . in section [ sect rmt ] we introduce and explore our random matrix model and finally section [ sect conc ] contains our conclusions . ) and overlap ( blue ) dirac operator from the lowest @xmath2 eigenvalues . ] we analyze quenched su(2 ) lattice configurations generated with wilson action on a @xmath3 lattice at @xmath4 which amounts to a temperature of @xmath5 . the average polyakov loop of @xmath6 signals deconfinement ( by polyakov loop we refer to the trace of the products of all temporal links @xmath7 and we selected the physical sector of positive polyakov loops by hand ) . we measured the @xmath2 lowest eigenvalues with positive imaginary parts of the overlap @xcite ( with parameter @xmath8 cf . @xcite ) and staggered dirac operator on 1136 ( overlap)/ 3149 ( staggered ) configurations . for a set of 1102 configurations , we also measured the 12 lowest eigenmodes of both staggered and overlap operator . in all cases the quark mass was set to zero . the eigenvalues are ordered according to their imaginary parts and the corresponding eigenvalue densities . ] are plotted in [ fig spec density ] . they display a gap - like behavior with the eigenvalue density starting to differ from zero considerably at @xmath9 and @xmath10 , respectively . in addition , the overlap operator possesses exact zero modes , which we use to determine the topological charge of the configuration . to describe the rmt vs. poissonian behavior of the dirac spectra we will scan windows in the range of available eigenvalues and measure the level spacing distributions @xmath11 on unfolded eigenvalues @xcite , a typical quantity describing the eigenvalue statistics . the gaussian rmt ensembles provide predictions for it , which also apply to systems with chiral symmetry , depending only on the universality class . for gauge group @xmath0 the latter are the gaussian orthogonal ensemble ( goe ) for the overlap operator ( like for the continuum dirac operator ) and the gaussian symplectic ensemble ( gse ) for the staggered operator . the main difference of those @xmath11 formulae lies in the different repulsion strength of nearby eigenvalues , which results in a linear and quartic behavior of @xmath11 near @xmath12 , respectively . independent eigenvalues , on the other hand , lack such a repulsion and the unfolded level spacing is a poissonian distribution , i.e. @xmath13 . [ fig transition overlap ] shows the level spacing distribution of overlap eigenvalues ( see @xcite for staggered spectra ) . it clearly reveals that the level spacing agrees with the associated rmt predictions in the bulk and moves towards poissonian when the spectral window is shifted towards lower eigenvalues . to observe such a level _ spacing _ at least a few independently ( poissonian ) distributed eigenvalues are needed on each configuration . since independent modes occur only at the very low end of the spectrum where the spectral density is low , large enough volumes are required for that . in @xcite also the independence of these data of the lattice spacing has been demonstrated for the staggered case . the properties of the independent and localized modes at the lower ends of the spectra are the main subject of the rest of the paper . we therefore check first , whether the modes of the overlap and the staggered operator see similar physical effects , meaning that their profiles are correlated and localized to similar locations . vs. @xmath14 , on a @xmath15 configuration over the whole lattice . data points with small amplitudes as indicated by the circle have been excluded to avoid overcrowding the plot . ] first of all we remark that for every overlap zero mode we find a staggered eigenvalue with unusually small value . in the topological sector @xmath15 , the average smallest eigenvalue is 0.175(1 ) , whereas it is 0.109(3 ) for @xmath16 ( and 0.098(8)/0.141(6 ) for @xmath17 , where we have two small eigenvalues ) . this can also be seen in [ fig polloop modes one ] bottom . next , the scatter plot of [ fig scatter overlap staggered ] gives a strong indication for the correlation of the local mode amplitudes in a typical example configuration , which is further visualized by the two - dimensional profiles in [ fig profile modes ] . , in a lattice plane where the overlap mode takes on its maximum . ( the absolute maximum of the staggered mode is separated from the overlap one by @xmath18 lattice spacings in the remaining directions.),title="fig : " ] + , in a lattice plane where the overlap mode takes on its maximum . ( the absolute maximum of the staggered mode is separated from the overlap one by @xmath18 lattice spacings in the remaining directions.),title="fig : " ] in order to quantify the similarity and localization of two modes we propose the following `` interlocalization '' @xmath19 where @xmath20 is the number of lattice sites , and @xmath21 is the absolute value ( @xmath22-norm ) of the @xmath23th overlap / staggered eigenmode summed up over gauge and in the case of overlap also spinor indices at lattice site @xmath24 . this is a positive quantity that receives large contributions when both modes are considerably large at some locations . moreover , for two identical modes it becomes their inverse participation ratio ( ipr ) . the latter is a well - known measure for the localization , taking on a value of @xmath20 for modes localized on a single point ( on the lattice ) and @xmath25 for constant modes . in fact , we find @xmath26 also for two normalized modes with independent gaussian distributed amplitudes at each site . only modes that are similar _ and _ localized generate large values of @xmath27 . we utilize @xmath27 for matching the overlap and staggered modes . we start by taking the lowest overlap mode and pair it with the staggered mode that has the largest interlocalization with it . going up in the overlap spectrum we continue this matching procedure in the same way , but use only those staggered modes that have not yet been paired up with a lower overlap mode . in [ fig matched modes interlocalisation ] we plot the interlocalization values for the lowest modes matched in this way as a function of the corresponding overlap eigenvalue @xmath28 . from this plot it is clear that the zero modes ( of the overlap operator , near - zero modes of the staggered operator ) are matching close to perfectly . the value of 300 is actually in the same order of magnitude as the ipr of the individual overlap zero modes . then @xmath27 drops quickly and after a few modes it reaches the reference value @xmath25 discussed above . as an appetizer of our main finding we show in [ fig profile polloops ] the polyakov loops ( of one example configuration ) in the lattice plane , where the lowest overlap and staggered eigenmodes take on their maximum , as shown in [ fig profile modes ] . the polyakov loop is dominated by uv fluctuations at the scale of the lattice spacing as almost every lattice observable . therefore it is virtually impossible to see any structures in it . we applied 6 sweeps of ape smearing @xcite with @xmath29 to the configuration , which leads to a smoother polyakov loop landscape . indeed , an island of polyakov loop with opposite sign emerges at the location of the maximum of the lowest dirac mode on the original unsmeared configuration . a similar profile becomes visible after simply averaging the ( traced ) polyakov loops with their neighbors . . by the naked eye nothing seems particular in this plane for the unsmeared case , whereas the smeared polyakov loop actually takes its minimum ( -0.68 , compared to an average smeared polyakov loop of 0.80 ) at the hotspot visible in the fermion modes in fig . [ fig profile modes ] . ( a correlation of the unsmeared polyakov loop to fermion modes is exposed by virtue of statistical measurements , see text.),title="fig : " ] + . by the naked eye nothing seems particular in this plane for the unsmeared case , whereas the smeared polyakov loop actually takes its minimum ( -0.68 , compared to an average smeared polyakov loop of 0.80 ) at the hotspot visible in the fermion modes in fig . [ fig profile modes ] . ( a correlation of the unsmeared polyakov loop to fermion modes is exposed by virtue of statistical measurements , see text.),title="fig : " ] let us stress , that the lower panel of [ fig profile polloops ] is the only occasion that we present a smeared result . we now return to correlation functions of unsmeared polyakov loops for the rest of this paper . to check the correlation of polyakov loop islands and low dirac modes in a quantitative way , we define `` polyakov loops as averaged by a particular mode '' , i.e. polyakov loops weighted with the density of a normalized dirac mode , cf . @xcite , @xmath30 this quantity is restricted to the interval @xmath31 $ ] , just like the polyakov loop @xmath32 . it is clear that a wave - like mode @xmath33 with approximately constant amplitude yields an @xmath34 close to the average polyakov loop @xmath35 , whereas a strongly localized mode picks the polyakov loop in that region . if the latter happens to be an island of `` wrong '' polyakov loops , @xmath34 tends to zero or even becomes negative . [ fig polloop modes one ] shows the ratio of @xmath34 averaged over different configurations and the average polyakov loop , both for the low - lying overlap and low - lying staggered modes . the polyakov loops averaged by the low - lying modes are indeed smaller than on average . higher modes , on the contrary , tend to see the average polyakov loop . from ( [ eqn new probability ] ) , in the @xmath15 sector compared to the polyakov loop distribution and the haar measure ( valid in the low temperature phase ) . ] the connection between local polyakov loops and low - lying dirac modes is confirmed from a slightly different perspective by the following `` polyakov loop distribution as seen by a mode '' . for that we weight the probability of polyakov loops @xmath32 by the amplitudes of the low - lying modes at those positions @xmath24 where @xmath36 , @xmath37 in continuous notion with some smeared delta - function @xmath38 ( in practice we rescale histograms ) . the quantity @xmath34 is recovered by the @xmath32-expectation value with this probability , @xmath39 thus the probability @xmath40 visualizes how the global quantity @xmath34 is generated by modifying the distributions of the local polyakov loops . in [ fig polloop modes two ] we show this probability for the lowest modes of the overlap operator , @xmath41 , on @xmath15 configurations . one can clearly see that low fermion modes enhance low polyakov loops down to @xmath42 . in a background of constant temporal and vanishing spatial gauge fields the influence of the polyakov loop on the dirac spectra is very clear . we first diagonalize the polyakov loop introducing its phase @xmath43 , @xmath44\ , . \label{eqn polloop phase second}\end{aligned}\ ] ] we gauge it into the last time slice , where it effectively changes the temporal boundary condition . the lowest modes of the free dirac operator in the presence of this polyakov loop are constant in space and plane waves in time , @xmath45 . the quantum numbers @xmath46 are governed by a combination of the antiperiodic temporal boundary condition for fermions and the polyakov loop phase , namely @xmath47 , where the different signs emerge from the different color components , see ( [ eqn polloop phase first ] ) . the eigenvalues of the free dirac operator are these numbers multiplied by the temperature . the lowest ones , @xmath48 we name _ effective matsubara frequencies_. these hold in the limit @xmath49 , whereas on the lattice one has @xmath50 at high temperatures the polyakov loop ( at fixed lattice spacing ) becomes trivial , @xmath51 , hence @xmath52 and the matsubara frequency is @xmath53 . a more realistic estimate is obtained by using the average polyakov loop at our temperature @xmath54 , from which we obtain the effective matsubara frequency in lattice units ( @xmath55 ) as @xmath56 which is the same order of magnitude as the lower end of the bulk of eigenvalues we measured ( consistent with the findings of @xcite ) . the main point of these considerations is that `` wrong '' polyakov loops , @xmath57 with @xmath58 , would lead to a vanishing effective matsubara frequency , @xmath59 , and thus to the lowest dirac eigenvalues . of course , in realistic configurations one has to take into account that the polyakov loop varies in space and that nontrivial spatial links are present . both will change the dirac eigenvalues away from the free ones . nonetheless , the tendency that `` wrong '' polyakov loops give rise to smaller eigenvalues persists and explains our finding about their pinning nature . topological excitations of the gauge field and their zero modes are an attractive hypothesis to explain low - lying dirac modes in yang - mills theory ( and qcd ) . the chiral condensate at zero temperature is thought of as due to instantons of realistic size and density and the first conjecture about the metal - insulator transition at finite temperature was based on instanton ensembles @xcite . the natural topological excitations at finite temperature are magnetic monopoles . they appear as selfdual or antiselfdual solutions of the yang - mills equations of motion at finite temperature . as they are also electrically charged , they are called dyons . dyons can be constituents of calorons ( finite temperature instantons ) @xcite , but may exist in isolation as well @xcite . in gauge group @xmath0 , there are two dyons ( and two antidyons ) with opposite magnetic charge . one sort of these dyons is characterized by properties that exactly match our findings : the polyakov loop at their core is @xmath1 and they possess zero modes with the physical antiperiodic boundary conditions @xcite ( just like the constant configurations discussed above ) . the other sort has polyakov loop @xmath60 or @xmath61 by definition . ] and zero modes with periodic boundary conditions . in dilute ensembles of dyons most of the zero modes should remain low - lying modes . then the first sort of dyons could explain the localized modes we analyzed , while the second sort could be responsible for low - lying periodic modes . at a positive average polyakov loop one expects fluctuations to @xmath1 much less frequent than those to @xmath60 , cf . [ fig polloop properties ] . hence the first sort of dyons could give independent low - lying modes , while the second sort could yield a condensate at periodic boundary conditions . this different appearance can be made quantitative by the different fractions of unit topological charge the two dyon sorts have , which are such that the @xmath1 dyons are indeed heavier as they have a larger ( classical ) action , see @xcite for lattice evidence of this picture and @xcite for a simple model . from the relation of magnetic monopoles to both low - lying modes and topological charge a crucial test of this picture is to compare these two quantities at given temperature and volume . above the finite temperature transition fluctuations of the topological charge decrease sharply . this can be clearly seen by looking at how the topological susceptibility decreases at higher temperatures . as a result , at high temperature , topological objects presumably form a dilute gas of non - interacting objects . from the index of the overlap dirac operator we have full information about the fluctuations of the _ total topological charge_. assuming that in the dilute gas topological objects are uncorrelated , this can be used to compute the density of topological objects that in turn can be compared to the density of localized poissonian dirac eigenmodes . in table [ tab : charge_distribution ] we show the probability of different topological sectors extracted from the index of the overlap dirac operator . from the low occurrence of the topological charge sector @xmath62 and in particular of the sector @xmath63 it is indeed clear that topological objects form a very dilute gas . in these volumes it very rarely happens that there is more than one topological object on any given configuration . since we need only an order of magnitude estimate of the density of topological objects we will ignore the probability of two or more objects occurring on any single configuration . in this approximation we can shortcut the calculation of the density of topological objects , and the average number of topological objects per configuration is just given by the probability of the @xmath16 sector . on our volumes it is between @xmath64 and @xmath65 which is clearly far too small to account for the few localized poissonian modes we found on average per configuration . this comparison rules out models based on uncorrelated gluonic objects that carry both ( o(1 ) ) topological charge and ( antiperiodic ) zero modes . however , it does not exclude combinations of topological objects in which the topological charge cancels , like instanton - antiinstanton molecules originally suggested to be present with light dynamical quarks @xcite or molecules of dyons called bions carrying one near zero mode @xcite . .the probability of different charge sectors in the @xmath66 ( first two columns ) and @xmath67 ensemble ( last three columns ) . the probabilities of charge sectors with the same magnitude but opposite sign have been added . [ tab : charge_distribution ] [ cols="^,^,^,^,^,^",options="header " , ] @xmath68 as a function of the lattice distance @xmath69 ( containing the same information as the free energy ) . bottom : the level spacing distribution ( of the polyakov loop trace @xmath32 ) compared to the poisson distribution.,title="fig : " ] + @xmath68 as a function of the lattice distance @xmath69 ( containing the same information as the free energy ) . bottom : the level spacing distribution ( of the polyakov loop trace @xmath32 ) compared to the poisson distribution.,title="fig : " ] transitions between correlated and uncorrelated eigenvalues like the one observed in the lattice data can be described by rmt models in various ways . one of the simplest possible anstze is to start with a diagonal matrix , whose entries are uncorrelated random numbers , and add a matrix taken from one of the gaussian ensembles . this model shows the desired transition in between parts of its spectrum with different eigenvalue density , however , the interpolating spacing distributions are different from the ones in the spectrum of the staggered operator @xcite . a possible explanation is the sparseness of this operator and the spatial information that is contained in the next - neighbor interactions . in contrast , the full matrices from the gaussian ensembles blindly connect all diagonal elements with equal strength . we construct a better suited random matrix model , based on sparse matrices , in the following . this model can be nicely motivated by our previous findings . concerning the polyakov loop , which will be connected to a random potential , three properties are relevant : ( i ) the distribution of local polyakov loops extends to negative values and thus small effective matsubara frequencies , cf . [ fig polloop modes two ] , ( ii ) polyakov loops become independent quickly with their distance and therefore ( iii ) the level spacings of the polyakov loop trace @xmath32 neglecting the spatial information are distributed according to a poissonian distribution . the last two properties are depicted in [ fig polloop properties ] . hence the polyakov loops could provide the poissonian ingredient for the statistics of the dirac eigenvalues . to become more concrete , we remind the reader of the definition of the staggered dirac operator , @xmath70\,,\ ] ] with @xmath71 and @xmath72 . we split this operator in the temporal and spatial part @xmath73 @xmath74 contains the hopping terms in the temporal direction , whereas spatial hoppings are included in @xmath75 . the temporal part has a block - diagonal structure , with one block for each spatial site . we want to approximate each block and thus the whole dirac operator by restricting it to the subspace of the smallest eigenvalue quadruplet of the temporal operator at each spatial lattice site . the quadruplet consists of two eigenvalue pairs with opposite sign . the plus - minus degeneracy is necessary to conserve chiral symmetry , whereas the exact two - fold degeneracy has to be kept in order to have the right rmt universality class we find in the su(2 ) staggered spectra . in this basis , the temporal part of the dirac operator can be brought in the form @xmath76 with @xmath77 which resemble the effective matsubara frequencies on the lattice , as given in ( [ eqn eff mf latt ] ) , however as a function of the local rather than averaged polyakov loop phase @xmath78 . therefore , @xmath79 is a diagonal matrix with very weakly coupled entries , as argued above . physically speaking , the temporal part represents a ( chiral ) random potential . the spatial part becomes in the restricted basis @xmath80 where the 2-by-2 matrices @xmath81 and @xmath82 can be shown to represent real quaternions . these have a concrete meaning : @xmath81 connects eigenvalue pairs of equal sign , and is therefore responsible for the gse - like level repulsion between nearest neighbors . @xmath82 on the other hand generates the gap around zero in the spectrum as it connects eigenvalues of different signs . we propose a random matrix model based on matrices @xmath83 that consist of two parts . @xmath84 has the same diagonal structure like @xmath79 , i.e. @xmath85 where the diagonal entries @xmath86 are random numbers constructed with an overall scale @xmath87 and a random angle @xmath88 $ ] . these quantities are equivalent to the effective matsubara frequencies @xmath89 in the continuum , ( [ eqn eff mf ] ) , the temperature @xmath90 and the angle @xmath78 of the local polyakov loop , respectively . for @xmath91 we have taken the empirical distribution of the angle @xmath92 of local polyakov loops , by converting the fine histogram in [ fig polloop modes two ] accordingly . this yields an asymmetric distribution of @xmath91 between @xmath93 and @xmath94 with a maximum below @xmath95 ( i.e. at positive polyakov loop ) . its most important feature , however , seems to be the tail towards the `` trapping '' @xmath96 ( negative polyakov loop locally ) , since we have observed that most of the features of the model persist when using the haar measure @xmath97 as distribution [ not shown ] . for the spatial part , one first of all needs to fix the periodicity of the underlying space , i.e. an integer @xmath98 such that @xmath99 is identified with @xmath100 , with unit vectors @xmath101 in each of the spatial directions . @xmath102 , like @xmath103 , has nonvanishing entries only at positions that connect next neighbors in that space . its blocks @xmath104 consist of random real quaternions @xmath105 and @xmath106 , that are rmt counterparts of @xmath81 and @xmath82 from ( [ sp0dirac ] ) . we have taken them to be gaussian distributed around zero with their mean deviations @xmath107 as parameters of the model . by rescaling all the random matrices @xmath108 it is clear , that only the ratios of the scales @xmath87 , @xmath109 and @xmath110 are relevant parameters . to fix them we have measured the ratio of the average determinants of the empirical quaternionic hopping terms finding @xmath111 and took this ratio over for the ratio of @xmath112 . the remaining ratio @xmath113 ( @xmath114 ) was put in by hand to obtain desired properties of the rmt model , namely a gap at zero and a transition between a poissonian and a gse spacing distribution . the eigenvalue density and spacing distribution of this model are plotted in [ sparsermtspacfigure ] for a spatial extent @xmath115 and @xmath116 . we observe a similar transition like in the spectrum of the staggered dirac operator . another feature that this model shares with the staggered operator is the increasing localization of eigenmodes as the corresponding eigenvalues decrease , quantified by the ipr s in [ sparsermtlocfigure ] top . in order to measure the correlation of the lowest eigenmodes of this rmt model to the diagonal entries , we again define the latter `` as averaged by a particular mode '' @xmath117 one can see in [ sparsermtlocfigure ] bottom , that the low modes are indeed localized to `` islands '' of low @xmath118 , which are equivalent to low polyakov loops . hence this important effect is shared by our random matrix model , too . in the present paper we found a possible explanation for the emergence of localized poissonian modes at the low end of the high temperature qcd dirac spectrum . we showed that the localized modes are strongly correlated with large fluctuations of the polyakov loop . this lowers the effective matsubara frequencies for modes concentrated there . we argued that the lowest part of the dirac spectrum consists of this type of eigenmodes . we verified this picture for eigenmodes of both the staggered and the overlap dirac operator . as a side result we also demonstrated that the spatial structure of the lowest overlap and staggered modes is highly correlated . this shows that different discretizations of the dirac operator are sensitive to the same type of gauge field fluctuations . we also looked at the topological charge fluctuations as given by the index of the overlap . assuming that at high temperature topological objects form a dilute gas and are uncorrelated , we could safely rule them out in creating the localized low modes . finally , we proposed a dimensionally reduced random matrix model . it is based on sparse matrices encoding the three dimensional nature of the problem through nearest neighbor couplings from a lattice dirac operator . to the best of our knowledge this is an example of a new kind of random matrix models for qcd , where so far only full matrices have been used . beside chirality and the spectral gap our model reproduces the transition from localized poissonian to delocalized random matrix type modes observed in the lattice dirac spectrum as well as the correlation of the localized modes to `` islands '' of low on site `` potential '' . it is instructive to compare the dirac operator to the hamiltonian of anderson type models , see @xcite for reviews . in the latter case usually diagonal ( on site ) disorder is responsible for creating the transition to localized eigenmodes . in the case of the dirac operator , the on site terms do not seem to be relevant . in fact , in the staggered operator they are exactly zero . however , in our dimensionally reduced three dimensional effective model the non - zero fluctuating on site terms are dynamically generated by fluctuations of the local matsubara frequency resulting from fluctuations of the polyakov loop . in this way our dimensionally reduced effective random matrix model is analogous to the anderson model . it would be interesting to study further how the presence and details of the transition depend on the parameters and matrix size defining the random matrix ensemble . we thank helpful discussions to jacques bloch , antonio m. garcia - garcia , ferenc pittler , mithat nsal and tilo wettig . fb and ss are supported by dfg ( br 2872/4 - 2 ) and tgk by eu grant ( fp7/2007 - 2013)/erc n@xmath119208740 . s. chandrasekharan and n. h. christ , nucl . suppl . * 47 * , 527 ( 1996 ) , hep - lat/9509095 . m. a. stephanov , phys . b * 375 * , 249 ( 1996 ) , hep - lat/9601001 . p. n. meisinger and m. c. ogilvie , phys . b * 379 * , 163 ( 1996 ) , hep - lat/9512011 . s. chandrasekharan and s. z. huang , phys . d * 53 * , 5100 ( 1996 ) , hep - ph/9512323 . c. gattringer , p. e. l. rakow , a. schfer and w. sldner , phys . d * 66 * , 054502 ( 2002 ) , hep - lat/0202009 . c. gattringer , s. schaefer , nucl . phys . * b654 * , 30 ( 2003 ) , hep - lat/0212029 .
we investigate low - lying fermion modes in @xmath0 gauge theory at temperatures above the phase transition . both staggered and overlap spectra reveal transitions from chaotic ( random matrix ) to integrable ( poissonian ) behavior accompanied by an increasing localization of the eigenmodes . we show that the latter are trapped by local polyakov loop fluctuations . islands of such `` wrong '' polyakov loops can therefore be viewed as defects leading to anderson localization in gauge theories . we find strong similarities in the spatial profile of these localized staggered and overlap eigenmodes . we discuss possible interpretations of this finding and present a sparse random matrix model that reproduces these features .
recent studies have shown that machine - learned tree - based models , combined with ensemble techniques , are highly effective for building web ranking algorithms @xcite within the `` learning to rank '' framework @xcite . beyond document retrieval , tree - based models have also been proven effective for tackling problems in diverse domains such as online advertising @xcite , medical diagnosis @xcite , genomic analysis @xcite , and computer vision @xcite . this paper focuses on runtime optimizations of tree - based models that take advantage of modern processor architectures : we assume that a model has already been trained , and now we wish to make predictions on new data as fast as possible . although exceedingly simple , tree - based models do not efficiently utilize modern processor architectures due to the prodigious amount of branches and non - local memory references in standard implementations . by laying out data structures in memory in a more cache - conscious fashion , removing branches from the execution flow using a technique called predication , and micro - batching predictions using a technique called vectorization , we are able to better exploit modern processor architectures and significantly improve the speed of tree - based models over hard - coded if - else blocks . our experimental results are measured in nanoseconds for individual trees and microseconds for complete ensembles . a natural starting question is : do such low - level optimizations actually matter ? does shaving microseconds off an algorithm have substantive impact on a real - world task ? we argue that the answer is _ yes _ , with two different motivating examples : first , in our primary application of learning to rank for web search , prediction by tree - based models forms the inner loop of a search engine . since commercial search engines receive billions of queries per day , improving this tight inner loop ( executed , perhaps , many billions of times ) can have a noticeable effect on the bottom line . faster prediction translates into fewer servers for the same query load , reducing datacenter footprint , electricity and cooling costs , etc . second , in the domain of financial engineering , every nanosecond counts in high frequency trading . orders on nasdaq are fulfilled in less than 40 microseconds . firms fight over the length of cables due to speed - of - light propagation delays , both within an individual datacenter and across oceans @xcite . thus , for machine learning in financial engineering , models that shave even a few microseconds off prediction times present an edge . we view our work as having the following contributions : first , we introduce the problem of _ architecture - conscious _ implementations of machine learning algorithms to the information retrieval and data mining communities . although similar work has long existed in the database community @xcite , there is little research on the application of architecture - conscious optimizations for information retrieval and machine learning problems . second , we propose novel implementations of tree - based models that are highly - tuned to modern processor architectures , taking advantage of cache hierarchies and superscalar processors . finally , we illustrate our techniques in a standard , widely - accepted , learning - to - rank task and show significant performance improvements over standard implementations and hard - coded if - else blocks . we begin with an overview of modern processor architectures and recap advances over the past few decades . the broadest trend is perhaps the multi - core revolution @xcite : the relentless march of moore s law continues to increase the number of transistors on a chip exponentially , but experts widely agree that we are long past the point of diminishing returns in extracting instruction - level parallelism in hardware . instead , adding more cores appears to be a better use of increased transistor density . since prediction is an embarrassingly parallel problem , our techniques can ride the wave of increasing core counts . a less - discussed , but just as important trend over the past two decades is the so - called `` memory wall '' @xcite , where increases in processor speed have far outpaced improvements in memory latency . this means that ram is becoming slower relative to the cpu . in the 1980s , memory latencies were on the order of a few clock cycles ; today , it could be several hundred clock cycles . to hide this latency , computer architects have introduced hierarchical cache memories : a typical server today will have l1 , l2 , and l3 caches between the processor and main memory . cache architectures are built on the assumption of reference locality that at any given time , the processor repeatedly accesses only a ( relatively ) small amount of data , and these fit into cache . the fraction of memory accesses that can be fulfilled directly from the cache is called the _ cache hit rate _ , and data not found in cache is said to cause a _ cache miss_. cache misses cascade down the hierarchy if a datum is not found in l1 , the processor tries to look for it in l2 , then in l3 , and finally in main memory ( paying an increasing latency cost each level down ) . managing cache content is a complex challenge , but there are two main principles that are relevant to a software developer . first , caches are organized into cache lines ( typically 64 bytes ) , which is the smallest unit of transfer between cache levels . that is , when a program accesses a particular memory location , the entire cache line is brought into ( l1 ) cache . this means that subsequent references to nearby memory locations are very fast , i.e. , a cache hit . therefore , in software it is worthwhile to organize data structures to take advantage of this fact . second , if a program accesses memory in a predictable sequential pattern ( called striding ) , the processor will prefetch memory blocks and move them into cache , before the program has explicitly requested the memory locations ( and in certain architectures , it is possible to explicitly control prefetch in software ) . there is , of course , much more complexity beyond this short description ; see @xcite for an overview . the database community has explored in depth the consequences of modern processor architectures for relational query processing @xcite . in contrast , these issues are underexplored for information retrieval and data mining applications . this is one of the first attempts at developing architectural - conscious runtime implementations of machine learning algorithms . researchers have explored scaling the _ training _ of tree - based models to massive datasets @xcite , which is of course an important problem , but orthogonal to the issue we tackle here : given a trained model , how do we make predictions quickly ? another salient property of modern cpus is pipelining , where instruction execution is split between several stages ( modern processors have between one to two dozen stages ) . at each clock cycle , all instructions `` in flight '' advance one stage in the pipeline ; new instructions enter the pipeline and instructions that leave the pipeline are `` retired '' . pipeline stages allow faster clock rates since there is less to do per stage . superscalar _ cpus add the ability to dispatch multiple instructions per clock cycle ( and out of order ) provided that they are independent . pipelining suffers from two dangers , known as `` hazards '' in vlsi design terminology . _ data hazards _ occur when one instruction requires the result of another ( that is , a data dependency ) . this happens frequently when dereferencing pointers , where we must first compute the memory location to access . subsequent instructions can not proceed until we actually know what memory location we are accessing the processor simply stalls waiting for the result ( unless there are other independent instructions that can be executed ) . _ control hazards _ are instruction dependencies introduced by if - then clauses ( which compile to conditional jumps in assembly ) . to cope with this , modern processors use _ branch prediction techniques_in short , trying to predict which code path will be taken . however , if the guess is not correct , the processor must `` undo '' the instructions that occurred after the branch point ( `` flushing '' the pipeline ) . the impact of data and control hazards can be substantial : an influential paper in 1999 concluded that in commercial rdbmses at the time , almost half of the execution time is spent on stalls @xcite . which is `` worse '' , data or control hazards ? not surprisingly , the answer is , it depends . however , with a technique called predication @xcite , which we explore in our work , it is possible to convert control dependencies into data dependencies ( see section [ section : approach ] ) . whether predication is worthwhile , and under what circumstances , remains an empirical question . another optimization that we adopt , called vectorization , was pioneered by database researchers @xcite : the basic idea is that instead of processing a tuple at a time , a relational query engine should process a `` vector '' ( i.e. , batch ) of tuples at a time to take advantage of pipelining . our work represents the first application of vectorization to optimizing machine learning algorithms that we are aware of . beyond processor architectures , the other area of relevant work is the vast literature on learning to rank @xcite , application of machine learning techniques to document ranking in search . our work uses gradient - boosted regression trees ( gbrts ) @xcite , a state - of - the - art ensemble method . the focus of most learning - to - rank research is on learning effective models , without considering efficiency , although there is an emerging thread of work that attempts to better balance both factors @xcite . in contrast , we focus exclusively on runtime ranking performance , assuming a model that has already been trained ( by other means ) . in this section we describe various implementations of tree - based models , starting from two baselines and progressively introducing architecture - conscious optimizations . we focus on an individual tree , the runtime execution of which involves checking a predicate in an interior node , following the left or right branch depending on the result of the predicate , and repeating until a leaf node is reached . we assume that the predicate at each node involves a feature and a threshold : if the feature value is less than the threshold , the left branch is taken ; otherwise , the right branch is taken . of course , trees with greater branching factors and more complex predicate checks can be converted into an equivalent binary tree , so our formulation is general . note that our discussion is agnostic with respect to the predictor at the leaf node , be it a boolean ( in the classification case ) , a real ( in the regression case ) , or even an embedded sub - model . we assume that the input feature vector is densely - packed in a floating - point array ( as opposed to a sparse , map - based representation ) . this means that checking the predicate at each tree node is simply an array access , based on a unique consecutively - numbered i d associated with each feature . object : as a high - flexibility baseline , we consider an implementation of trees with nodes and associated left and right pointers in c++ . each tree node is represented by an object , and contains the feature i d to be examined as well as the decision threshold . for convenience , we refer to this as the object implementation . in our mind , this represents the most obvious implementation of tree - based models that a software engineer would come up with and thus serves as a good point of comparison . this implementation has two advantages : simplicity and flexibility . however , we have no control over the physical layout of the tree nodes in memory , and hence no guarantee that the data structures exhibit good reference locality . prediction with this implementation essentially boils down to pointer chasing across the heap : when following either the left or the right pointer to the next tree node , the processor is likely to be stalled by a cache miss . codegen : as a high - performance baseline , we consider statically - generated if - else blocks . that is , a code generator takes a tree model and directly generates c code , which is then compiled and used to make predictions . for convenience , this is referred to as the codegen implementation . this represents the most obvious performance optimization that a software engineer would come up with , and thus serves as another good point for performance comparison . we expect this approach to be fast . the entire model is statically specified ; machines instructions are expected to be relatively compact and will fit into the instruction cache , thus exhibiting good reference locality . furthermore , we leverage decades of compiler optimizations that have been built into gcc . note that this eliminates data dependencies completely by converting them all into control dependencies . the downside , however , is that this approach is inflexible . the development cycle now requires more steps : after training the model , we need to run the code generation , compile the resulting code , and then link against the rest of the system . this may be a worthwhile tradeoff for a production system , but from the view of rapid experimentation and iteration , the approach is a bit awkward . struct : the object approach has two downsides : poor memory layout ( i.e. , no reference locality and hence cache misses ) and inefficient memory utilization ( due to object overhead ) . to address the second point , the solution is fairly obvious : get rid of c++ and drop down to c to avoid the object overhead . we can implement each node as a struct in c ( comprising feature i d , threshold , left and right pointers ) . we construct a tree by allocating memory for each node ( malloc ) and assigning the pointers appropriately . prediction with this implementation remains an exercise in pointer chasing , but now across more memory - efficient data structures . we refer to this as the struct implementation . struct@xmath0 : an improvement over the struct implementation is to physically manage the memory layout ourselves . instead of allocating memory for each node individually , we allocate memory for all the nodes at once ( i.e. , an array of structs ) and linearize the tree in the following way : the root lies at index 0 . assuming a perfectly - balanced tree , for a node at index @xmath1 , its left child is at @xmath2 and its right child is at @xmath3 . this is equivalent to laying out the tree using a breadth - first traversal of the nodes . the hope is that by manually controlling memory layout , we can achieve better reference locality , thereby speeding up the memory references . this is similar to the idea behind css - trees @xcite used in the database community . for convenience we call this the struct@xmath0 implementation . one nice property of retaining the left and right pointers in this implementation is that for unbalanced trees ( i.e. , trees with missing nodes ) , we can more tightly pack the nodes to remove `` empty space '' ( still following the layout approach based on breadth - first node traversal ) . thus , the struct@xmath0 implementation occupies the same amount of memory as struct , except that the memory is contiguous . pred : the struct@xmath0 implementation tackles the reference locality problem , but there remains one more issue : the presence of branches ( resulting from the conditionals ) , which can be quite expensive to execute . branch mispredicts may cause pipeline stalls and wasted cycles ( and of course , we would expect many mispredicts with trees ) . although it is true that speculative execution renders the situation far more complex , removing branches may yield performance increases . a well - known trick in the compiler community for overcoming these issues is known as predication @xcite . the underlying idea is to convert control dependencies ( hazards ) into data dependencies ( hazards ) , thus altogether avoiding jumps in the underlying assembly code . here is how predication is adapted for our case : we encode the tree as a struct array in c , nd , where nd[i].fid is the feature i d to examine , and nd[i].theta is the threshold . we assume a fully - branching binary tree , with nodes laid out via breadth - first traversal ( i.e. , for a node at index @xmath1 , its left child is at @xmath2 and its right child is at @xmath3 ) . to make the prediction , we probe the array in the following manner : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ .... i = ( i<<1 ) + 1 + ( v[nd[i].fid ] > = nd[i].theta ) ; i = ( i<<1 ) + 1 + ( v[nd[i].fid ] > = nd[i].theta ) ; ... .... _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we completely unroll the tree traversal loop , so the above statement is repeated @xmath4 times for a tree of depth @xmath4 . at the end , @xmath1 contains the index of the leaf node corresponding to the prediction ( which we look up in another array ) . one final implementation detail : we hard code a prediction function for each tree depth , and then dispatch dynamically using function pointers . note that this approach assumes a fully - balanced binary tree ; to cope with unbalanced trees , we expand by inserting dummy nodes . vpred : predication eliminates branches but at the cost of introducing data hazards . each statement in pred requires an indirect memory reference . subsequent instructions can not execute until the contents of the memory locations are fetched in other words , the processor will simply stall waiting for memory references to resolve . therefore , predication is entirely bottlenecked on memory access latencies . a common technique adopted in the database literature to mask these memory latencies is called _ vectorization _ @xcite . applied to our task , this translates into operating on multiple instances ( feature vectors ) at once , in an interleaved way . this takes advantage of multiple dispatch and pipelining in modern processors ( provided that there are no dependencies between dispatched instructions , which is true in our case ) . so , while the processor is waiting for the memory access from the predication step on the first instance , it can start working on the second instance . in fact , we can work on @xmath5 instances in parallel . for @xmath6 , this looks like the following , working on instances i0 , i1 , i2 , i3 in parallel : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ .... i0 = ( i0<<1 ) + 1 + ( v[nd[i0].fid ] > = nd[i0].theta ) ; i1 = ( i1<<1 ) + 1 + ( v[nd[i1].fid ] > = nd[i1].theta ) ; i2 = ( i2<<1 ) + 1 + ( v[nd[i2].fid ] > = nd[i2].theta ) ; i3 = ( i3<<1 ) + 1 + ( v[nd[i3].fid ] > = nd[i3].theta ) ; i0 = ( i0<<1 ) + 1 + ( v[nd[i0].fid ] > = nd[i0].theta ) ; i1 = ( i1<<1 ) + 1 + ( v[nd[i1].fid ] > = nd[i1].theta ) ; i2 = ( i2<<1 ) + 1 + ( v[nd[i2].fid ] > = nd[i2].theta ) ; i3 = ( i3<<1 ) + 1 + ( v[nd[i3].fid ] > = nd[i3].theta ) ; ... .... _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in other words , we traverse one layer in the tree for four instances at once . while we re waiting for v[nd[i0].fid ] to resolve , we dispatch instructions for accessing v[nd[i1].fid ] , and so on . hopefully , by the time the final memory access has been dispatched , the contents of the first memory access are available , and we can continue without processor stalls . again , we completely unroll the tree traversal loop , so each block of statements is repeated @xmath4 times for a tree of depth @xmath4 . at the end , @xmath1 contains the index of the leaf nodes corresponding to the prediction for @xmath5 instances . setting @xmath5 to 1 reduces this model to pure predication ( i.e. , no vectorization ) . note that the optimal value of @xmath5 is dependent on the relationship between the amount of computation performed and memory latencies we will determine this relationship empirically . for convenience , we refer to the vectorized version of the predication technique as vpred . given that the focus of our work is efficiency , our primary evaluation metric is prediction speed . we define this as the elapsed time between the moment a feature vector ( i.e. , a test instance ) is presented to the tree - based model to the moment that a prediction ( in our case , a regression value ) is made for the instance . to increase the reliability of our results , we conducted multiple trials and report the mean and variance . we conducted two sets of experiments : first , using synthetically - generated data to quantify the performance of individual trees in isolation , and second , on standard learning - to - rank datasets to verify the performance of full ensembles . all experiments were run on a red hat linux server , with intel xeon westmere quad - core processors ( e5620 2.4ghz ) . this architecture has a 64 kb l1 cache per core , split between data and instructions ; a 256 kb l2 cache per core ; and a 12 mb l3 cache shared by all cores . code was compiled with gcc ( version 4.1.2 ) using optimization flags -o3 -fomit - frame - pointer -pipe . all code ran single - threaded . the synthetic data consisted of randomly generated trees and randomly generated feature vectors . each intermediate node in a tree has two fields : a feature i d and a threshold on which the decision is made . each leaf is associated with a regression value . construction of a random tree of depth @xmath4 begins with the root node . we pick a feature i d at random and generate a random threshold to split the tree into left and right subtrees . this process is recursively performed to build each subtree until we reach the desired tree depth . when we reach a leaf node , we generate a regression value at random . note that our randomly - generated trees are fully - balanced , i.e. , a tree of depth @xmath4 has @xmath7 leaf nodes . once a tree has been constructed , the next step is to generate random feature vectors . each random feature vector is simply a floating - point array of length @xmath8 ( @xmath9 number of features ) , where each index position corresponds to a feature value . we assume that all paths in the decision tree are equally likely ; the feature vectors are generated in a way that guarantees an equal likelihood of visiting each leaf . to accomplish this , we take one leaf at a time and follow its parents back to the root . at each node , we take the node s feature i d and produce a feature value based on the position of the child node . that is , if the child node we have just visited is on the left subtree we generate a feature value that is smaller than the threshold stored at the current parent node ; otherwise we generate a feature value larger than the threshold . we randomize the order of instances once we have generated all the feature vectors . to avoid any cache effects , our experiments are conducted on a large number of instances ( 512k ) . given a random tree and a set of random feature vectors , we ran experiments to assess the various implementations of tree - based models described in section [ section : approach ] . to get a better sense of the variance , we performed 5 trials ; in each trial we constructed a new random binary tree and a different randomly - generated set of feature vectors . to explore the design space , we conducted experiments with varying tree depths @xmath10 and varying feature sizes @xmath11 . in addition to randomly - generated trees , we conducted experiments using standard learning - to - rank datasets , where training , validation , and test data are provided . using the training and validation sets we learned a complete tree - ensemble ranking model , and evaluation is then carried out on test instances to determine the speed of the various implementations . these experiments assess performance in a real - world application . we used gradient - boosted regression trees ( gbrts ) @xcite to train a learning - to - rank model . gbrts are ensembles of regression trees that yield state - of - the - art effectiveness on learning - to - rank tasks . the learning algorithm sequentially adds new trees to the ensemble that best account for the remaining regression error ( i.e. , the residuals ) . we used the open - source jforests implementation of lambdamart to optimize ndcg @xcite . although there is no way to precisely control the depth of each tree , we can adjust the size distribution of the trees by setting a cap on the number of leaves ( which is an input parameter to the learner ) . we used two standard learning - to - rank datasets : letor - mq2007 and mslr - web10k . both are pre - folded , providing training , validation , and test instances . table [ table : datasets ] shows the dataset sizes and the numbers of features . to measure variance , we repeated experiments on all five folds . note that mq2007 is much smaller and is considered by many in the community to be outdated . .average number of training , validation , and test instances in our learning - to - rank datasets , along with the number of features . [ table : datasets ] [ cols="<,>,>,>,>",options="header " , ] the values of @xmath8 ( number of features ) in our synthetic experiments are guided by these learning - to - rank datasets . we selected feature sizes that are multiples of 16 ( 4-byte floats ) so that the feature vectors are integer multiples of cache line sizes ( 64 bytes ) : @xmath12 roughly corresponds to letor features and is representative of a small feature space ; @xmath13 corresponds to mslr and is representative of a medium - sized feature space . we introduced a third condition @xmath14 to capture a large feature space condition . in this section we present experimental results , beginning with evaluation on synthetic data and then on learning - to - rank datasets . we begin by focusing on the first five implementations described in section [ section : approach ] ( leaving aside vpred for now ) , using the procedure described in section [ section : experimental_setup : synthetic ] . the prediction time per randomly - generated test instance is shown in figure [ figure : results : synthetic ] , measured in nanoseconds . the balanced randomly - generated trees vary in terms of tree depth _ d _ , and each bar chart shows a separate value of _ f _ ( number of features ) . time is averaged across five trials and error bars denote 95% confidence intervals . it is clear that as trees become deeper , prediction speeds decrease overall . this is obvious since deeper trees require more feature accesses and predicate checks , more pointer chasing , and more branching ( depending on the implementation ) . first , consider the high - flexibility and high - performance baselines . as expected , the object implementation is the slowest ( except for pred with @xmath14 ) . it is no surprise that the c++ implementation is slow due to the overhead from classes and objects ( recall the other implementations are in c ) . the gap between object and struct , which is the comparable c implementation , grows with larger trees . also as expected , the codegen implementation is very fast : with the exception of @xmath12 , hard - coded if - else statements are faster or just as fast as all other implementations , regardless of tree depth . comparing struct@xmath0 with struct , we observe no significant improvement for shallow trees , but a significant speedup for deep trees . recall that in struct@xmath0 , we allocate memory for the entire tree so that it resides in a contiguous memory block , whereas in struct we let allocate memory however it chooses . this shows that reference locality is important for deeper trees . finally , turning to the pred condition , we observe a very interesting behavior . for small feature vectors @xmath12 , the technique is actually faster than codegen . this shows that for small feature sizes , predication helps to overcome branch mispredicts , i.e. , converting control dependencies into data dependencies increases performance . for @xmath13 , results are mixed compared to codegen , struct , and struct@xmath0 : sometimes faster , sometimes slower . however , for large feature vectors ( @xmath14 ) , the performance of pred is terrible , even worse than the object implementation . we explain this result as follows : pred performance is entirely dependent on memory latency . when traversing the tree , it needs to wait for the contents of memory before proceeding . until the memory references are resolved , the processor simply stalls . with small feature vectors , we get excellent locality : 32 features take up two 64-byte cache lines , which means that evaluation incurs at most two cache misses . since memory is fetched by cache lines , once a feature is accessed , accesses to all other features on the same cache line are essentially `` free '' . locality decreases as the feature vector size increases : the probability that the predicate at a tree node accesses a feature close to one that has already been accessed goes down . thus , as the feature vector size grows , the pred prediction time becomes increasingly dominated by stalls waiting for memory fetches . the effect of this `` memory wall '' is evident in the other implementations as well . we observe that the performance differences between codegen , struct , and struct@xmath0 shrink as the feature size increases ( whereas they are more pronounced for smaller feature vectors ) . this is because as feature vector size increases , more and more of the prediction time is dominated by memory latencies . how can we overcome these memory latencies ? instead of simply stalling while we wait for memory references to resolve , we can try to do other useful computation this is exactly what vectorization is designed to accomplish . in section [ section : approach ] , we proposed _ vectorization _ of the predication technique in order to mask memory latencies . the idea is to work on @xmath5 instances ( feature vectors ) at the same time , so that while the processor is waiting for memory access for one instance , useful computation can happen on another . this takes advantage of pipelining and multiple dispatch in modern superscalar processors . the effectiveness of vectorization depends on the relationship between time spent in actual computation and memory latencies . for example , if memory fetches take only one clock cycle , then vectorization can not possibly help . the longer the memory latencies , the more we would expect vectorization ( larger batch sizes ) to help . however , beyond a certain point , once memory latencies are effectively masked by vectorization , we would expect larger values of @xmath5 to have little impact . in fact , values that are too large start to bottleneck on memory bandwidth and cache size . in figure [ figure : results : vectorization ] , we show the impact of various batch sizes , @xmath15 , for the different feature sizes . note that when @xmath5 is set to 1 , we evaluate one instance at a time , which reduces to the pred implementation . prediction speed is measured in nanoseconds and normalized by batch size ( i.e. , divided by @xmath5 ) , so we report _ per - instance _ prediction time . for @xmath12 , @xmath16 yields the best performance ; for @xmath13 , @xmath17 yields the best performance ; for @xmath14 , @xmath18 all provide approximately the same level of performance . these results are exactly what we would expect : since memory latencies increase with larger feature sizes , a larger batch size is needed to mask the latencies . with the combination of vectorization and predication , vpred becomes the fastest of all our implementations on the synthetic data . comparing figures [ figure : results : synthetic ] and [ figure : results : vectorization ] , we see that vpred ( with optimal vectorization parameter ) is actually faster than codegen . table [ table : relative_speed ] summarizes this comparison . vectorization is up to 70% faster than the non - vectorized implementation ; vpred can be twice as fast as codegen . in other words , we retain the best of both worlds : speed and flexibility , since the vpred implementation does not require code recompilation . having evaluated different implementations on synthetic data , we move on to learning - to - rank datasets using tree ensembles . as previously described , we used the implementation of lambdamart by ganjisaffar et al . once a model has been trained and validated , we evaluate on the test set to measure prediction speed . since the datasets come pre - folded five ways , we repeated our experiments five times and report mean and variance across the runs . to handle ensembles in our implementations , we simply add an outer loop to the algorithm that iterates over individual trees in the ensemble . note that ganjisaffar et al . actually construct multiple ensembles , each built using a random bootstrap of the training data ( i.e. , _ bagging _ multiple boosted ensembles ) . in this work , we do not adopt this procedure because bagging is embarrassingly parallel from the runtime execution perspective and hence not particularly interesting . for learning parameters , we used values recommended by ganjisaffar et al . , with the exception of max leaves ( see below ) . feature and data sub - sampling parameters were set to 0.3 , minimum percentage of observations per leaf was set to 0.5 , and the learning rate was set to 0.05 . in terms of performance , shallower trees are naturally preferred . but what is the relationship between tree depth and ranking effectiveness ? tree depth with our particular training algorithm can not be precisely controlled , but can be indirectly influenced by the maximum number of leaves on an individual tree ( an input to the learner ) . table [ table : effectiveness ] shows the average ndcg values ( at different ranks ) measured across five folds on the letor and mslr datasets with different values of this parameter , similar to the range of values explored in @xcite . statistical significance was tested using the wilcoxon test ( @xmath19-value 0.05 ) ; none of the differences on the letor dataset were significant . for each condition , we also report the average depth of the trees that were actually learned . the average tree depth is computed for every ensemble and then averaged across the five folds ; variance is presented in parentheses . results show that for letor , tree depth makes no significant difference on ndcg , whereas larger trees yield better results on mslr ; however , there appears to be little difference between 50 and 70 max leaves . the results make sense : to exploit larger feature spaces we need trees with more nodes . since many in the community consider the letor dataset to be out of date with an impoverished feature set , more credence should be given to the mslr results . turning to performance results , figure [ figure : results : actual ] illustrates per - instance prediction speed for various implementations on the learning - to - rank datasets . note that this is on the entire ensemble , with latencies now measured in microseconds instead of nanoseconds . as described above , the trees were trained with different settings of max leaves ; the _ x_-axis plots the tree depths from table [ table : effectiveness ] . in this set of experiments , we made use of the vpred approach with the vectorization parameter set to @xmath20 for letor and @xmath21 for mslr . results from the synthetic datasets mostly carry over to these learning - to - rank datasets . object is the slowest implementation and struct is slightly faster . on the letor dataset , struct is only slightly slower than struct@xmath0 , but on mslr , struct@xmath0 is faster than struct by a larger margin in most cases . vpred outperforms all other techniques , including codegen on mslr , but is slower than codegen on letor ( except for the shallowest trees ) . however , note that in terms of ndcg , table [ table : effectiveness](a ) shows no difference in effectiveness , so there is no advantage to building deeper trees for letor . the conclusion appears clear : for tree - based ensembles on real - world learning - to - rank datasets , we can achieve the best of both worlds . with a combination of predication and vectorization , we can make predictions faster than statically - generated if - else blocks , yet retain the flexibility in being able to specify the model dynamically , which enables rapid experimentation . our experiments show that predication and vectorization are effective techniques for substantially increasing the performance of tree - based models , but one potential objection might be : are we measuring the right thing ? in our experiments , prediction time is measured from when the feature vector is presented to the model to when the prediction is made . critically , we assume that features have already been computed . what about an alternative architecture where features are computed lazily , i.e. , only when the predicate at a tree node needs to access a particular feature ? this alternative architecture , where features are computed on demand , is difficult to study since results will be highly dependent on the implementation of feature extraction which in turn depends on the underlying data structures ( layout of the inverted indexes ) , compression techniques , and how computation - intensive the features are . however , there is a much easier way to study this issue we can trace the execution of the full tree ensemble and keep track of the fraction of features that are accessed . if during the course of making a prediction , most of the features are accessed , then there is little waste in computing all the features first and then presenting the complete feature vector to the model . table [ table : features ] shows the average fraction of features accessed in the final learned models for both learning - to - rank datasets , with different max leaves configurations . it is clear that , for both datasets , most of the features are accessed during the course of making a prediction , and in the case of the mslr dataset , nearly all the features are accessed all the time ( especially with deeper trees , which yield higher effectiveness ) . therefore , it makes sense to separate feature extraction from prediction . in fact , there are independent compelling reasons to do so : a dedicated feature extraction stage can benefit from better reference locality ( when it comes to document vectors , postings , or whatever underlying data structures are necessary for computing features ) . interleaving feature extraction with tree traversal may lead to `` cache churn '' , where a particular data structure is repeatedly loaded and then displaced by other data . returning to a point in the introduction : do these optimizations actually matter , in the broader context of real - world search engines ? this is of course a difficult question to answer and highly dependent on the actual search architecture , which is a complex distributed system spanning hundreds of machines or more . here , we venture some rough estimates . from figure [ figure : results : actual](b ) , the mslr dataset , we see that compared to codegen , vpred reduces per - instance prediction time from around 40@xmath22 to around 25@xmath22 ( for max leaves setting of 50 ) ; this translates into a 38% reduction in latency per instance . in a web search engine , the learning to rank algorithm is applied to a candidate list of documents that is usually generated by other means ( e.g. , scoring with bm25 and a static prior ) . the exact details are proprietary , but the published literature does provide some clues . for example , cambazoglu et al . @xcite ( authors from yahoo ! ) experimented with reranking 200 candidate documents to produce the final ranked list of 20 results ( the first two pages of search results ) . from these numbers , we can compute the per - query reranking time to be 8ms using the codegen approach and 5ms with vpred . this translates into an increase from 125 queries per second to 200 queries per second on a single thread for this phase of the search pipeline . alternatively , gains from faster prediction can be leveraged to rerank more results or take advantage of more features . this simple estimate suggests that our optimizations can make a noticeable difference in web search , and given that our techniques are relatively simple the predication and vectorization optimizations definitely seem worthwhile . during the course of our experiments , we noticed that two assumptions of our implementations did not appear to be fully valid . first , the pred and vpred implementations assume fully - balanced binary trees ( i.e. , every node has a left and a right child ) . in contrast , recall that struct@xmath0 makes no such assumption because with the left and right pointers we can tightly pack the tree nodes . the fully - balanced tree assumption does not turn out to be valid for gbrts the learner does not have a preference for any particular tree topology , and so the trees are unbalanced most of the time . to compensate for this , the pred and vpred implementations require insertion of dummy nodes to create a fully - balanced tree . second , we assume that all paths are equally likely in a tree , i.e. , that at each node , the left and right branches are taken with roughly - equal frequency . we noticed , however , that this is often not the case . to the extent that one branch is favored over another , branch prediction provides non - predicated implementations ( i.e. , if - else blocks ) an advantage , since branch prediction will guess correctly more often , thus avoiding pipeline flushes . one promising future direction to address the above two issues is to adapt the model learning process to prefer balanced trees and predicates that divide up the feature space evenly . we believe this can be incorporated into the learning algorithm as a penalty , much in the same way that regularization is performed on the objective in standard machine learning . thus , it is perhaps possible to jointly learn models that are both fast and good , as in the recently - proposed `` learning to _ efficiently _ rank '' framework @xcite . modern processor architectures are incredibly complex because technological improvements have been uneven . this paper focuses on one particular issue : not all memory references are equally fast , and in fact , latency can differ by an order of magnitude . there are a number of mechanisms to mask these latencies , although it largely depends on developers knowing how to exploit these mechanisms . the database community has been exploring these issues for quite some time now , and in this respect the information retrieval , machine learning , and data mining communities are behind . in this paper , we demonstrate that two relatively simple techniques , predication and vectorization , along with more efficient memory layouts , can significantly accelerate prediction performance for tree - based models , both on synthetic data and on real - world learning - to - rank datasets . our work explores architecture - conscious implementations of a particular machine learning model but we believe there are plenty of similar opportunities in other areas of machine learning as well . this work has been supported by nsf under awards iis-0916043 , iis-1144034 , and iis-1218043 . any opinions , findings , conclusions , or recommendations expressed are the authors and do not necessarily reflect those of the sponsor . the first author s deepest gratitude goes to katherine , for her invaluable encouragement and wholehearted support . the second author is grateful to esther and kiri for their loving support and dedicates this work to joshua and jacob . a. criminisi , j. shotton , and e. konukoglu . decision forests : a unified framework for classification , regression , density estimation , manifold learning and semi - supervised learning . , 7(23):81227 , 2011 .
tree - based models have proven to be an effective solution for web ranking as well as other problems in diverse domains . this paper focuses on optimizing the runtime performance of applying such models to make predictions , given an already - trained model . although exceedingly simple conceptually , most implementations of tree - based models do not efficiently utilize modern superscalar processor architectures . by laying out data structures in memory in a more cache - conscious fashion , removing branches from the execution flow using a technique called predication , and micro - batching predictions using a technique called vectorization , we are able to better exploit modern processor architectures and significantly improve the speed of tree - based models over hard - coded if - else blocks . our work contributes to the exploration of _ architecture - conscious _ runtime implementations of machine learning algorithms .
simple theoretical models of spherical gravitational lenses predict that spiral galaxies produce only 1020% of gravitational lenses ( turner , ostriker & gott 1984 ; fukugita & turner 1991 ; maoz & rix 1993 ; kochanek 1991 , 1993 , 1996a ) . the prediction is roughly consistent with observations ( for a summary , see keeton & kochanek 1996 ; keeton , kochanek & falco 1997b ) . specifically , of @xmath1 known lenses , only b 0218 + 357 ( odea et al . 1992 ; patnaik et al.1993 ) is produced by a galaxy unambiguously identified as a distant spiral galaxy based on its colors and mass - to - light ratio as well as the presence of hi and molecular gas and strong faraday rotation ( patnaik et al . 1993 ; carilli , rupen & yanny 1993 ; browne et al.1993 ; keeton et al . the lens q2237 + 0305 ( huchra et al.1985 ) , found as part of a redshift survey , is a special case of lensing by the bulge of a nearby spiral galaxy ( @xmath2 ) . the radio ring pks 1830@xmath3211 ( rao & subrahmanyan 1988 ; jauncey et al.1991 ) shows both hi and molecular absorption features ( lovell et al . 1996 ; wiklind & combes 1996 ) and thus may have a spiral lens galaxy , but because the absorption features are at different redshifts and there is no optical identification of the lens galaxy this lens is still not understood . for the remaining lenses , the lens galaxy is generally more consistent with an early - type galaxy , with the exception of mg 0414 + 0534 ( hewitt et al . 1992 ) whose red color matches no standard galaxy type ( lawrence et al . 1995 ; keeton et al . 1997b ) . models of individual lenses and the observed numbers of 4-image lenses seem to require mean axis ratios somewhat flatter than expected for early - type galaxies ( king et al . 1996 ; kochanek 1996b ; keeton , kochanek & seljak 1997a ) . the apparent discrepancy may be due to difficulties interpretating the axis ratios of lens models , which is complicated by the effects of external tidal shears from neighboring galaxies and clusters ( e.g. hogg & blandford 1994 ; schechter et al . 1997 ) , and by the possibility that dark halos may be flatter than the light ( e.g. dubinski & carlberg 1991 ) . an alternate possibility is that spherical models may grossly underestimate the number of spiral galaxy lenses by not adequately representing real galaxies , which have not only round halos but also flat disks . because errors in estimating the expected number of lenses can bias inferences about the cosmological model based on the statistics of gravitational lenses , there is growing interest in studying lensing by spirals using models that better represent real galaxies . there are as yet no treatments of lensing by spiral galaxies using a model that includes both a realistic disk and a halo . the spherical models treated spiral galaxies as singular isothermal spheres ( sis ) normalized by their rotation curves , so they represented diskless , pure halo models . models using ellipsoidal densities ( kassiola & kovner 1993 ; kormann , schneider & bartelmann 1994ab ; kochanek 1996b ; keeton et al.1997a ) can be interpreted as projections of disk galaxies without halos , although they are not generally viewed as such . pure disk models are poor representations of spiral galaxies because they neglect the dynamically important dark halos that may not be spherical but are certainly not as flat as disks ( see reviews by ashman 1992 and rix 1996 ) . in addition , pure disk models with a flat rotation curve predict that the cross section diverges as the disk becomes edge - on , and that the divergent cross section is dominated by an image geometry consisting of two bright images offset from the center of the galaxy and straddling the projected disk . among point image lenses we see only lenses consisting of two or four images surrounding the center of the galaxy ( see keeton & kochanek 1996 for a summary ) , and the absence of the `` disk '' image geometry is direct evidence for rounder halos . what makes lensing by realistic spiral galaxies interesting , then , is not the effects of the thin disk , because the properties of pure disk models were already understood from studies of ellipsoidal lenses , but the effects of the halo in suppressing the divergent cross section and unphysical image geometry of a pure disk model . the fact that the observational data are not consistent with pure disk models means that spiral gravitational lenses will provide a useful probe of the balance between the disk and the halo in spiral galaxies . in our own galaxy , the constraint on the local surface mass density of the disk of @xmath4 ( kuijken & gilmore 1991 ; bahcall , flynn & gould 1992 ; also see sackett 1996b ) is one of the weakest links in understanding the mass distribution of the galaxy and interpreting the results of the lmc and galactic bulge microlensing searches ( e.g. alcock et al . 1995 ) . in external galaxies , the decomposition of rotation curves between the disk and the halo is usually degenerate and standard models assume a `` maximal disk '' to derive lower bounds on the halo contributions ( e.g. van albada & sancisi 1986 ) . thus any new constraint on the relative contributions of the disk and the halo in spiral galaxies has significance well beyond its particular effects on gravitational lensing . recently maller , flores & primack ( 1997 ) and wang & turner ( 1997 ) began to explore the effects of combining a disk with a halo by embedding a constant surface density , finite radius disk in a spherical isothermal halo . maller et al . ( 1997 ) examined the ability of the model to fit b 1600 + 434 ( jackson et al . 1995 ) , a two - image lens that jaunsen & hjorth ( 1997 ) suggested is a spiral galaxy . wang & turner ( 1997 ) examined the inclination - averaged cross section to see if the spherical models systematically underestimate the number of lenses produced by spirals . the constant density disk model is analytically tractable , but the mass density and rotation curve bear little resemblance to a real galaxy , and the sharp disk edge introduces peculiar features in the lensing properties . here we introduce several simple , physically reasonable models for lensing by spiral galaxies by combining halo , disk , and bulge components to produce nearly flat rotation curves . in 2 we describe the halo , disk , and bulge components and discuss their lensing properties . in 3 and 4 we combine the components into realistic models and study the effects of inclination and the shapes and masses of the halo , disk , and bulge on the lensing cross section , optical depth , and image geometries . in 5 we summarize our results and discuss their implications for lensing statistics , galactic structure , and the statistics of damped ly@xmath5 absorbers . we build realistic models for spiral galaxies by embedding a thin disk and possibly a central bulge in a dark matter halo . we can describe both disky and spheroidal components by using an oblate spheroid with axis ratio @xmath6 , and then letting @xmath7 for an infinitely thin disk or @xmath8 for a moderately flattened halo or bulge . an oblate spheroid projects to an ellipsoidal density distribution with projected axis ratio @xmath9 , where @xmath10 is the standard inclination angle ( @xmath11 is face - on and @xmath12 is edge - on ) . in the limit of an infinitely thin disk , a surface mass distribution @xmath13 projects to an ellipsoid with surface density @xmath14 where @xmath15 . thus the ellipsoidal gravitational lens models used by kassiola & kovner ( 1993 ) , kormann et al . ( 1994ab ) , kochanek ( 1996b ) , and keeton et al . ( 1997a ) can be viewed in the traditional way as models of early - type galaxies ( projections of three - dimensional ellipsoids ) , or can be reinterpreted as pure disk models of spiral galaxies ( projections of a two - dimensional disk ) . however , the dynamical normalization differs for the two interpretations ; we focus on the disk interpretation and occasionally discuss its relation to the early - type galaxy interpretation . a simple building block for galaxies with flat rotation curves is the softened , oblate , isothermal density distribution . the density and rotation curve for this model are @xmath16 \right\ } , \end{aligned}\ ] ] where @xmath17 is a core radius , @xmath18 is the eccentricity of the mass distribution , and the model is normalized so that asymptotically @xmath19 . the sis model corresponds to the limit @xmath20 and @xmath21 . the projected surface mass density in units of the critical surface mass density for lensing is @xmath22^{-1/2 } \label{eq : mestsig}\ ] ] where @xmath23 , @xmath24 is the critical radius of a singular isothermal sphere with rotation velocity @xmath25 , and @xmath26 and @xmath27 are comoving distances from the observer to the source and from the lens to the source , respectively . the lensing potential , deflection , and magnification produced by the lens are @xmath28^{1/2 } + \mbox{constant } , \\ \alpha,_x & = & { b_i \over ( 1-q^2)^{1/2 } } \tan^{-1 } \left[{(1-q^2)^{1/2 } x \over \psi+s } \right ] , \\ \alpha,_y & = & { b_i \over ( 1-q^2)^{1/2 } } \tanh^{-1 } \left[{(1-q^2)^{1/2 } y \over \psi+q^2 s}\right ] , \\ m^{-1 } & = & 1 - { b_i \over \psi } - { b_i^2 s \over \psi \left[(\psi+s)^2+(1-q^2 ) x^2\right]}\ , \end{aligned}\ ] ] where @xmath29 . these equations are identical ( up to an overall normalization factor ) to those derived in previous treatments of the softened isothermal ellipsoid ( e.g. kassiola & kovner 1993 ; kormann et al . 1994a ) , but the analytic forms are simpler . the normalization is such that a softened isothermal ellipsoid usually written as @xmath30^{-1/2}$ ] can be written in the form above by identifying @xmath31 , @xmath32 , and @xmath33 . in the limit of an infinitely thin disk ( @xmath34 ) , the isothermal model becomes a disk with the surface density and rotation curve pair @xmath35.\ ] ] we call this model a softened mestel disk , because in the limit @xmath36 it becomes a mestel ( 1963 ) disk , the surface density distribution producing a flat rotation curve . the lensing potential of the softened mestel disk is @xmath37 , and the deflection scale @xmath38 is the limit of @xmath39 as @xmath40 . because a disk requires less mass than a spherical distribution to produce a given rotation velocity , one immediate difference between a mestel disk and a singular isothermal sphere is that a face - on mestel disk has image separations smaller by @xmath41 and a cross section smaller by @xmath42 . as the inclination increases the lensing cross section grows , and it diverges for an edge - on mestel disk . we will study this divergence in 3.2 ; the main result is that the cross section diverges not because the model neglects the finite thickness of the disk , but rather because the total mass of the disk diverges . one way to avoid the divergence is to smoothly truncate the mestel disk by using the difference of two mestel disk models , @xmath43 where the truncation radius @xmath44 is larger than the core radius @xmath17 . the surface density of the truncated mestel disk model is constant for @xmath45 ( rising rotation curve ) , declines as @xmath46 for @xmath47 ( flat rotation curve ) , and declines as @xmath48 for @xmath49 ( keplerian rotation curve ) . the truncated mestel disk has a finite mass of @xmath50 . note that if the truncated model is round rather than flat , its density @xmath51 is similar to a jaffe ( 1983 ) model , @xmath52 . a second useful building block is an unnamed density distribution with @xmath53 asymptotically . in terms of the total mass @xmath54 , the density and rotation curve are @xmath55 - { \qq { } s r^3 \over ( r^2+s^2)(r^2+e^2 s^2 ) } \right\ } .\end{aligned}\ ] ] in projection the surface mass density , lensing potential , deflection , and magnification are @xmath56^{-3/2 } , \label{eq : kuzsig } \\ \phi_k(s,\qq { } ) & = & b_k^2 \ln \left [ ( \psi+s)^2 + ( 1-q^2 ) x^2 \right]^{1/2 } + \mbox{constant } , \\ \alpha,_x & = & { b_k^2 x \over \psi } { \psi+q^2 s \over ( \psi+s)^2+(1-q^2)x^2 } \ , \\ \alpha,_y & = & { b_k^2 y \over \psi } { \psi+s \over ( \psi+s)^2+(1-q^2)x^2 } \ , \\ m^{-1 } & = & 1 - { b_k^2 q^2 s \over \psi^3 } \,-\ , { b_k^4 q^2 s \over \psi^3 \left[(\psi+s)^2+(1-q^2)x^2\right ] } \nonumber \\ & & \,-\ , { b_k^4 \left[\psi^2(\psi+s)^2-s(2\psi+s)(\psi+q^2 s)^2\right ] \over \psi^4 \left[(\psi+s)^2+(1-q^2)x^2\right]^2 } \ , \end{aligned}\ ] ] where the deflection scale @xmath57 is related to the mass by @xmath58 . in the limit of an infinitely thin disk ( @xmath34 ) , the model corresponds to a kuzmin ( 1956 ) or toomre ( 1962 ) model i disk and can be used to approximate an exponential disk . it has the same mass and central surface density as an exponential disk of the form @xmath59 , and the rotation curves differ by at most @xmath60 . a true exponential disk in projection requires numerical integrals , making it cumbersome to use . where a cosmological model is required we adopt @xmath61 and @xmath62 . we first consider models consisting of a truncated mestel disk embedded in an oblate isothermal halo . a mestel disk has a surface mass density that falls off as @xmath63 while observed spiral galaxies have luminosity densities that fall off as @xmath64 , so a mestel disk can not represent a galaxy with a constant mass - to - light ratio in the disk . nevertheless , the mestel disk is interesting to study because it is the simplest disk system with a flat rotation curve . for simplicity , we let the disk be singular ( @xmath21 ) , so its only scale length is the truncation radius @xmath65 . we place the disk in a softened isothermal halo and tune the ratio of the halo scale radius @xmath66 to the disk truncation radius @xmath65 to produce a flat rotation curve ; gives typical values of the ratio for a rotation curve that is flat to better than @xmath67 . the inner rotation curve ( @xmath68 ) is supported entirely by the disk , so this is a `` maximal disk '' model for a spiral galaxy ( e.g. van albada & sancisi 1986 ) . it is not known whether most spiral galaxies have maximal disks , although it is generally believed that our galaxy has a disk that is only @xmath69 of maximal ( bahcall 1984 ; kuijken & gilmore 1989 ; van der kruit 1989 ; kuijken & gilmore 1991 ; kuijken 1995 ; but see sackett 1996b for a recent rebuttal ) . we allow for a submaximal disk , i.e. for some of the inner rotation curve to be supported by a dark matter halo , by embedding the disk+softened halo system in a singular isothermal halo . the overall lensing model is then @xmath70 + ( 1-f_d ) \phi_i(0,\qq{h}),\ ] ] where the `` disk fraction '' @xmath71 is the fraction of the inner rotation curve supplied by the disk , and @xmath72 and @xmath73 are the three - dimensional axis ratios of the disk and halo , respectively . an infinitely thin disk has @xmath74 and a spherical halo has @xmath75 . the projected axis ratios of the disk and halo are @xmath76 and @xmath77 . the model contains the limits of a pure mestel disk ( @xmath78 and @xmath79 ) and a pure isothermal halo ( either @xmath80 or @xmath81 ) . it has no bulge component and a singular central surface density . given a rotation velocity @xmath25 , it is convenient to normalize the length scales by the critical radius @xmath82 of the sis model with the same circular velocity , which yields lensing cross sections in units of the sis cross section ( @xmath83 ) and thus indicates whether including the disk increases or decreases the cross section . we choose values for the disk and halo axis ratios @xmath72 and @xmath73 , the disk truncation radius @xmath65 , and the disk fraction @xmath71 , and finally determine the halo core radius @xmath66 from the value of @xmath84 that gives a flat rotation curve ( see ) . for the physical normalization we can compare the model with the galaxy ; we use iau value of the circular velocity @xmath85 ( kerr & lynden - bell 1986 ) and a consensus value for the solar radius @xmath86 that is slightly smaller than the iau value of @xmath87 ( see the review by reid 1993 ) . the surface mass density of the disk at @xmath88 is @xmath89 ^ 2 \left[{8{\mbox { kpc}}\over r_0}\right ] \left [ 1 - { r_0 \over ( r_0 ^ 2 + a_d^2)^{1/2 } } \right]\ , m_\odot{\mbox { pc}}^{-2},\ ] ] where @xmath90 is the eccentricity of the disk . local estimates of the surface mass density of the disk are @xmath91 with more of a consensus toward low values ( kuijken & gilmore 1991 ; bahcall et al . 1992 ; also see sackett 1996b ) , so we must choose @xmath92 , and we should reduce the disk fraction @xmath71 if @xmath93 . the physical scale @xmath94 is related to the dimensionless ratio @xmath95 appearing in the lens models by @xmath96 \left[{220{\mbox { km~s$^{-1}$}}\over \theta_0}\right]^2 { 2 r_h ( 1+z_l ) d_{os } \over d_{ol } d_{ls } } , \label{eq : abar}\ ] ] where @xmath97 is the hubble radius , @xmath98 , @xmath26 , and @xmath27 are comoving distances to the lens , to the source , and from the lens to the source , respectively ( with @xmath99 $ ] for @xmath61 ) , and @xmath100 . with @xmath78 , the disk dominates the inner rotation curve of the model provided @xmath101 , which is always true because the minimum value of the cosmological distance ratio is @xmath102 . we first consider maximal disk models ( @xmath78 ) , so the inner rotation curve is supported entirely by the disk . the lensing properties of the model depend strongly on both the inclination ( through the axis ratio @xmath103 ) and the size ( truncation radius @xmath65 ) of the disk . illustrate the the critical curves , caustics , image geometries , and cross sections as functions of @xmath65 and @xmath103 . for a face - on galaxy ( @xmath104 ) , the model is strictly circular and the only multiple image geometry has three images ( with one trapped and demagnified in the singular core of the disk ) . a non - axisymmetric galaxy would also have a 5-image cross section , but studies of face - on spiral galaxies indicate that they have axis ratios @xmath105 ( see the review by rix 1996 ) so the 5-image cross section would be small . we noted in 2 that a disk requires considerably less mass than a spherical halo to produce a given rotation velocity , so the critical radius of a pure mestel disk ( @xmath79 ) is smaller than the corresponding sis by @xmath106 . thus a face - on mestel disk has a cross section @xmath107 , making it less efficient than an sis at producing multiple images . as we truncate the mestel disk , however , the isothermal halo supporting the outer rotation curve begins to increase the cross section , so @xmath108 depends on the truncation radius @xmath65 and varies from @xmath109 for @xmath79 to unity for @xmath110 . for a modestly inclined galaxy ( @xmath111 ) , the tangential critical line becomes elongated and produces an `` astroid '' caustic corresponding to standard 4-image geometries ( see schneider , ehlers & falco 1992 ) , with a fifth image trapped in the singular core of the disk . as the inclination increases ( @xmath103 decreases ) , the tangential critical line becomes even more elongated and the astroid caustic pierces the radial caustic . the region inside the astroid caustic but outside the radial caustic corresponds to a configuration of three images on one side of the center of the galaxy . the middle image fades as the inclination increases ( @xmath103 decreases ) , resulting in a geometry with two bright images off to one side of the galactic center and straddling the projected disk . this image geometry , which we refer to as the `` disk '' image geometry , has not been observed . for a nearly edge - on galaxy ( @xmath112 ) the tangential critical line consists of a central round region with a narrow `` spike '' extending out the @xmath113-axis , and the cross section diverges . there are two elements of the divergence . the first is the divergence of the radial caustic as @xmath114 and the surface mass density becomes a line density . for a disk with a core radius @xmath17 that is small compared with the disk truncation radius @xmath65 and the halo scale radius @xmath66 , the radial caustic is determined entirely by the central part of the disk and is independent of @xmath65 and @xmath66 . the radial caustic moves up the @xmath115-axis as @xmath116 and the 3-image cross section diverges logarithmically . the divergence is unobservable because most of the large cross section corresponds to image geometries where the fluxes differ by orders of magnitude . this is analogous to the divergent cross section of a point mass lens ( see schneider et al . 1992 ) , which is formally infinite only because it allows images to pass arbitrarily close to a singular mass distribution and to be arbitrarily faint . in practice spiral galaxy disks are observed to have a finite thickness ; for example guthrie ( 1992 ) found a mean axis ratio of @xmath117 in a sample of edge - on spiral galaxies . a finite disk thickness prevents the mass density from becoming a singular line density and hence the cross section from diverging . the second element is the divergence of the astroid caustic as the mass of the disk diverges . in the limits of an edge - on galaxy ( @xmath118 ) or a pure mestel disk ( @xmath119 ) the asymptotic cross sections are @xmath120 where @xmath121 is the area of the astroid caustic . results for intermediate regimes are shown in . an edge - on disk ( @xmath118 ) has an astroid cross section @xmath122 , where @xmath123 is the disk mass , so that the astroid cross section is finite provided @xmath123 is finite . the astroid cross section of a pure mestel disk diverges because the mass diverges , not because the disk is infinitely thin . although the cross section is finite for a truncated disk , it can still be quite large compared to an sis model . because most of the astroid lies outside the radial caustic , a nearly edge - on disk is dominated by the `` disk '' image geometry . in order to produce a realistic disk model and to avoid the unphysical logarithmic divergence of the radial caustic , we henceforth give the disk a finite thickness by making it an oblate spheroid with a small but non - zero @xmath72 . spheroids are not ideal representations of the exponential vertical structure of disks , but because the details of the disk thickness matter only for inclinations with @xmath124 we use spheroids for analytic simplicity . we can characterize the expected contribution of spiral galaxies to lensing statistics by computing cross sections and optical depths averaged over inclination . in doing so we neglect the magnification bias that , if included , would tend to reduce the inclination dependence of the 5-image and disk geometry cross sections because the mean magnification is higher when the cross section is lower . total probabilities , however , stay roughly proportional to the optical depth ( see wallington & narayan 1993 ; kochanek 1996b ; keeton et al . . shows the inclination - averaged cross section as a function of the disk truncation radius @xmath95 and the disk fraction @xmath71 , for a disk with thickness @xmath125 in a spherical halo . shows the corresponding optical depth as a function of @xmath94 . note that @xmath94 is related to @xmath95 by the redshift - dependent factor given in , so integrating over redshift to obtain the optical depth is equivalent to integrating over @xmath95 . somewhat surprisingly , although the face - on cross section is small and the edge - on cross section is large , the inclination - averaged cross section and optical depth ( in units of the sis values ) are near unity . in other words , the disk+halo model does not significantly increase the number of lenses expected from spiral galaxies over the simple spherical sis model . the number of lenses can be increased by @xmath126 only if @xmath65 is large and @xmath71 is near unity , corresponding to a disk that is much more massive than in the galaxy ( figures [ fig : msig]a and [ fig : mtau]a ) . moreover , many of the additional lenses have the unobserved `` disk '' image geometry ( figures [ fig : msig]c and [ fig : mtau]c ) . thus the disk+halo models that predict significantly more lenses than the spherical models are physically implausible , while models with a reasonable disk mass increase the total number of expected lenses by @xmath0 . evidence from observations and from n - body simulations suggests that the dark halos of spirals are not spherical ( see the reviews by rix 1996 and sackett 1996a ) , so in we consider the effects of of flattening the halo . we also consider making the disk both thicker and thinner . changing the disk thickness has little effect on the 5-image lens fraction ( @xmath127 ) but significantly changes the `` disk '' lens fraction ( @xmath128 , not shown ) and the total optical depth . this is because a thicker disk ( larger @xmath72 ) rules out the thin edge - on models that increase the cross section with numerous `` disk '' lenses . by contrast , making the halo oblate has little effect on the total optical depth but significantly changes the 5-image and `` disk '' lens fractions . this makes sense because flattening the halo increases the net flattening of the system , thus causing more 5-image and `` disk '' lenses , while reducing the halo mass needed to produce the same rotation curve . apparently the two effects conspire to keep the total optical depth unchanged , suggesting that flattening the halo even as much as 3:1does little to increase the total number of lenses . with any reasonably shaped halo the only way to increase the number of lenses by @xmath69 compared to the simple sis model is to let the mass of the system be dominated by the disk . in addition to studying the expected number of lenses , we can also study their distribution with inclination . we use the optical depth distribution @xmath129 to estimate the number of 5-image lenses produced by a galaxy with inclination @xmath10 . although the distribution of spiral galaxies should be uniform in @xmath130 , we know from 3.2 that the optical depth is dominated by nearly edge - on systems . a shows the median value of @xmath130 for a model with @xmath125 and a 2:1 flattened halo . the results depend on @xmath72 and @xmath73 , but in most of the parameter space we have considered the median value is less than @xmath131 . in other words , because of the strong dependence of the cross section on the inclination , more than half of 5-image lenses produced by spiral galaxies should come from systems with @xmath132 , i.e. systems within @xmath133 of edge - on . the cross section for 5-image lenses is strongly correlated with the cross section for `` disk '' lenses because both image geometries are associated with the astroid caustic . the edge - on galaxies that produce most of the 5-image lenses also produce `` disk '' lenses , although the ratio of `` disk '' to 5-image lenses depends on the disk thickness . for example , a disk with thickness @xmath125 and a reasonable disk mass produces about half as many `` disk '' lenses as 5-image lenses ( see c d ) , while a thicker disk eliminates thin edge - on models and hence reduces the number of `` disk '' lenses . we noted in 2 that disk lens models are closely related to ellipsoid lens models for early - type galaxies . one way to think about the relation is to compare the fractions of 5-image lenses they produce . b shows the axis ratio of the singular isothermal ellipsoid ( sie ) that produces the same fraction of 5-image lenses ( i.e. the same @xmath127 ) as the inclination - averaged disk+halo model with @xmath125 and @xmath134 . the results depend on @xmath72 and @xmath73 , but most models with a plausible disk mass have @xmath135 between @xmath136 and @xmath137 , with the flatter halos giving the lower values . in other words , in terms of the 5-image lens fraction , inclination - averaged spiral galaxies correspond roughly to e4e6 elliptical galaxies . one key difference , though , is that the elliptical galaxies would not produce lenses with the `` disk '' image geometry . real spiral galaxies have exponential disks and central bulges , so the inner regions are not well described by the mestel disk models of 3 . we now use a kuzmin disk as an approximation to an exponential disk , and we embed the disk in an isothermal halo to obtain the lensing model @xmath138 where @xmath139 is the scale length of the exponential disk , and @xmath66 is the scale radius of the halo . in 3 we examined the effects of varying the disk thickness @xmath72 and the halo oblateness @xmath73 , so for simplicity we use a thin disk with @xmath125 and a 2:1 flattened halo ( @xmath134 ) . we have four remaining parameters ( the scale lengths @xmath139 and @xmath66 , the disk mass @xmath123 , and the asymptotic circular velocity @xmath25 ) , but by requiring that the disk+halo rotation curve be as flat as possible we can fix two ratios , @xmath140 with these constraints , the rotation curve starts at zero , rises to a peak @xmath141 above @xmath25 at @xmath142 , falls to a minimum @xmath143 below @xmath25 at @xmath144 , and then slowly asymptotes to @xmath25 . we can then normalize the model using the galaxy as in 3 , which we take to have scale length @xmath145 ( see sackett 1996b ) . the local surface density and the total disk mass are then @xmath146 ^ 2 \left[{3.5{\mbox { kpc}}\over r_d}\right ] \left[{1+(8/3.5)^2 \over 1+(r_0/r_d)^2}\right]^{3/2 } m_\odot{\mbox { pc}}^{-2 } , \\ m_d & = & 10^{11 } \left[{\theta_0\over 220{\mbox { km~s$^{-1}$}}}\right]^2 \left[{r_d\over 3.5{\mbox { kpc}}}\right ] m_\odot , \end{aligned}\ ] ] so the disk is significantly more massive than the estimate of @xmath147 for our galaxy ( e.g. bahcall 1986 ; binney & tremaine 1987 ) , but the local surface mass density is consistent with the estimates of @xmath148 ( kuijken & gilmore 1991 ; bahcall et al . 1992 ) . combining the surface mass densities for the halo and disk from , the central surface mass density in units of the critical density for lensing is @xmath149 . \label{eq : kap0}\ ] ] a circular system is `` supercritical , '' i.e. can produce multiple images , only if @xmath150 ( see schneider et al . if we normalize the disk and halo as above and consider source and lens redshifts @xmath151 and @xmath152 then we have @xmath153 so @xmath154 does not exceed unity until @xmath155 or @xmath156 . with a true exponential disk the same analysis yields @xmath157 and @xmath158 , so @xmath154 does not exceed unity until @xmath159 . thus nearly face - on systems ( normalized to our galaxy ) are subcritical and can not produce multiple images . modestly inclined systems are just barely supercritical , so although they can produce multiple images their cross section for lensing is small . thus the low central surface density of the disk means that the bulge plays a crucial role in gravitational lensing by spiral galaxies . the bulge of our galaxy is well described by a de vaucouleurs ( 1948 ) @xmath160 law ( e.g. bahcall 1986 ) , but lensing by a de vaucouleurs model is impractical because it requires five independent numerical integrals at every position . we could approximate the bulge with a modified hubble profile @xmath161^{-3/2}$ ] , but the bulge mass @xmath162 would diverge logarithmically and we would be unable to characterize the bulge by its mass . so as a simple way to examine the qualitative effects of a central bulge with a finite mass , we use a bulge with the @xmath53 profile discussed in 2 . the total lens model is then @xmath163 where @xmath164 is the scale radius and @xmath165 the axis ratio for the bulge . for simplicity we assume a fixed value for @xmath165 . a self - consistent disk+bulge model requires a flattened bulge ( e.g. monet , richstone & schechter 1981 ) , so without attempting to build a self - consistent model we fix @xmath166 for a 2:1 flattened bulge . shows the inclination - averaged cross section for lensing as a function of the bulge mass @xmath167 and scale length @xmath164 . the bulge strongly affects the cross section , primarily by controlling the central surface density . with a diffuse , low mass bulge ( large @xmath164 and small @xmath167 ) , the system is barely supercritical and the cross section is nonzero but small . as the bulge becomes massive and concentrated ( @xmath164 decreases and @xmath167 increases ) , the cross section increases dramatically . the divergent cross section is misleading , though , because it is a point mass divergence ( see schneider et al.1992 ) . including magnification bias and limits on detectable flux ratios would reduce the cross section to a reasonable value . in addition to increasing the central surface density , the bulge also circularizes the center of the galaxy . as a result , the 5-image and `` disk '' lenses that are associated with a flattened system become less significant as the bulge becomes more dominant ( see c thus the bulge can regulate the numbers of 5-image and `` disk '' lenses , and an analysis of the distribution of image geometries in a sample of spiral lenses must account for this effect . not all of the bulge parameter space in is physically reasonable . a concentrated bulge produces an unphysical mass distribution whose rotation curve has a strong central peak ; in a we show where the peak in the rotation curve due to the bulge is 20% higher than the asymptotic circular velocity @xmath25 . conversely , a low mass bulge can not support the inner rotation curve ; in a we also show where the circular velocity at @xmath168 is only 80% of the asymptotic value . if we require that the inner rotation curve not deviate by more than 20% from @xmath25 , then shows that the inclination - averaged cross section remains comparable to or smaller than the sis result . the traditional approach to gravitational lensing by spiral galaxies ( turner et al . 1984 ; fukugita & turner 1991 ; maoz & rix 1993 ; kochanek 1991 , 1993 , 1996a ) neglected the disk and used the dark halo alone to estimate that only 1020% of gravitational lenses should be produced by spiral galaxies . recent ellipsoidal lens models ( kassiola & kovner 1993 ; kormann et al . 1994ab ; kochanek 1996b ; keeton et al . 1997a ) can be reinterpreted as projections of disks to show that pure disk models viewed nearly edge - on can sharply increase the number of lenses compared with the pure halo models , but that most of the additional lenses have an unphysical `` disk '' image geometry with two bright images off to one side of the galactic center and straddling the projected disk . thus it is important to have a halo to regulate the unphysical effects of a disk . maller et al . ( 1997 ) and wang & turner ( 1997 ) recently considered a constant surface density , finite radius disk in a spherical isothermal halo , but their model had a mass distribution and rotation curve very different from real galaxies . we have constructed physically plausible models by combining disk , halo , and bulge components normalized to produce a nearly flat rotation curve . we considered two classes of models : a truncated mestel ( 1963 ) disk , which has dark matter in the disk , in an isothermal halo ; and a kuzmin ( 1956 ) disk as an approximation to an exponential disk , with a central bulge and an isothermal halo . these models reveal four distinctive features of lensing by spirals . 1 . proper dynamical normalization of the models is important . a disk requires less mass than a spherical halo to produce the same rotation curve , so a disk model can have a lensing cross section significantly smaller than the corresponding halo model . for example , the cross section of a face - on mestel disk is only 41% of the sis cross section , and the cross section of a face - on exponential disk in an isothermal halo can be small or even zero ( depending on the mass of the bulge ) . 2 . the disk makes the lensing effects sensitive to the inclination . the cross section increases dramatically with inclination and is dominated by nearly edge - on models . for example , more than half of all 5-image lenses produced by a mestel disk galaxy come from galaxies within @xmath133 of edge - on . the cross section for 5-image lenses is correlated with the cross section for lenses with the unobserved `` disk '' image geometry , so edge - on galaxies also produce significant numbers of `` disk '' lenses , although the disk thickness and a bulge offer ways to control the ratio of the two geometries . 3 . despite the inclination effects , disk+halo models averaged over inclination do not significantly increase the cross section compared with pure halo models . the constant density disk model of wang & turner ( 1997 ) predicted qualitatively that the disk can increase the cross section by at most @xmath69 , and our models normalized to produce a given rotation curve restrict the increase to @xmath169 . our models show that increasing the cross section or optical depth by even @xmath69 requires a very massive disk that dominates the dark halo , in conflict with observations that dark halos contribute significantly to spiral galaxy dynamics ( see the review by ashman 1992 ) . our conclusion is insensitive to the shape of the halo , even for a halo that is 3:1 flattened . flattening of the halo changes the fraction of lenses with 5-image or `` disk '' image geometries but has little effect on the total number of lenses . a central bulge plays a crucial role in lensing by spiral galaxies with exponential disks , because an exponential disk normalized to our galaxy has a face - on central surface density too small to produce multiple images . the bulge raises the face - on central surface density enough to allow multiple imaging , with a diffuse bulge producing a small lensing cross section and a concentrated bulge producing a large cross section . the bulge also circularizes the center of the galaxy , diluting the effects of an edge - on disk and reducing the number of 5-image and `` disk '' lenses . replacing the bulge with a triaxial bar would give a face - on galaxy a small 5-image cross section but would otherwise have little effect . our calculations neglected magnification bias so that we could perform large parameter surveys . because mean magnifications tend to be large when cross sections are small , magnification bias would tend to reduce the inclination dependence of the 5-image and `` disk '' image cross sections ( see schneider et al . in addition , magnification bias would increase the ratio of 5-image and `` disk '' lenses ( which have larger mean magnifications ) to 3-image lenses ( which have smaller mean magnifications ) . thus magnification bias is important for comparison to any observational sample . it should not , however , significantly affect the total cross section or the ratio of `` disk '' lenses to 5-image lenses . these results suggest that lensing by spiral galaxies can provide a new constraint on the structure of spiral galaxies . at present the balance between disk and halo masses and the shapes of halos are poorly known . for example , the contribution of the galactic disk to the rotation curve is not precisely known ( kuijken & gilmore 1991 ; bahcall et al.1992 ; also see sackett 1996b ) , and it has been suggested that explaining the microlensing optical depth toward the galactic bulge requires a disk that is heavier and closer to maximal than conventionally thought ( e.g. alcock et al . a sample of spiral galaxy lenses would constrain the relative masses and shapes of disks and halos , particularly if combined with hst images to determine the inclination of the disk . discovering a lens with the `` disk '' image geometry would strongly constrain the disk surface mass density , while the continued absence of `` disk '' lenses would rule out disks with a surface density significantly higher than our galaxy . unfortunately , a sample of spiral lenses may be difficult to obtain because they should contribute only 1020% of all lenses , and because the small image separations and extinction in the lens galaxies may bias optical surveys against finding them . our results also have implications for the predicted correlation between gravitational lensing and damped ly@xmath5 absorption . damped ly@xmath5 systems are thought to be associated with galactic disks ( wolfe 1988 , 1995 ) and may thus produce lensing effects in background quasars . bartelmann & loeb ( 1996 ) and smette , claeskens & surdej ( 1997 ) have pointed out that lensing can affect the statistics of damped ly@xmath5 absorbers through magnification bias and by modifying the impact parameter . these analyses used the sis lens model and thus neglected inclination effects in the lensing properties , although they did include inclination effects in the hi column density . the strong inclination dependence of the lensing properties must be taken into account in order to properly treat the effects of lensing on the statistics of damped ly@xmath5 absorbers . finally , our results suggest that spiral galaxies can not explain the weak discrepancy between observed lens galaxy axis ratios and the model axis ratios required to explain individual lenses and the statistics of 4-image lenses ( king et al . 1996 ; kochanek 1996b ; keeton et al . while it is true that edge - on spiral galaxies can produce many 4-image lenses , the absence of observed `` disk '' lenses indicates that spirals do not contribute significantly to the observational sample . moreover , the disk does not substantially increase the total cross section compared with the sis model . thus the only way to increase the fraction of lenses due to spirals is to change the ratio of spiral to early - type galaxy number densities . kauffmann , charlot & white ( 1996 ) have offered evidence that evolution may reduce the number of early - type galaxies by as much as a factor of 23 at @xmath170 . however most lens galaxies are closer than @xmath170 , and it seems unlikely that number evolution could change the ratio of spirals to early - types by the factor of @xmath171@xmath172 that would be required to make spiral galaxies dominate lens samples .
we study gravitational lensing by spiral galaxies , using realistic models consisting of halo , disk , and bulge components combined to produce a flat rotation curve . proper dynamical normalization of the models is critical because a disk requires less mass than a spherical halo to produce the same rotation curve a face - on mestel disk has a lensing cross section only 41% as large as a singular isothermal sphere with the same rotation curve . the cross section is sensitive to inclination and dominated by edge - on galaxies , which produce lenses with an unobserved 2-image geometry and a smaller number of standard 5-image lenses . unless the disk is unreasonably massive , disk+halo models averaged over inclination predict @xmath0 more lenses than pure halo models . finally , models with an exponential disk and a central bulge are sensitive to the properties of the bulge . in particular , an exponential disk model normalized to our galaxy can not produce multiple images without a bulge , and including a bulge reduces the net flattening of edge - on galaxies . the dependence of the lensing properties on the masses and shapes of the halo , disk , and bulge means that a sample of spiral galaxy lenses would provide useful constraints on galactic structure . # 1q_3#1 # 1eq . ( [ eq:#1 ] ) # 1#2eqs . ( [ eq:#1 ] ) and ( [ eq:#2 ] ) # 1figure [ fig:#1 ] # 1#2figures [ fig:#1 ] and [ fig:#2 ] # 1table [ tab:#1 ]
a major failure of a solid or a dynamic catastrophe can often be viewed as a phase transition from an unbroken or non - chaotic phase to a broken or chaotic phase . some obvious correlations existing in the system before the failure , are lost in the failure process . however , the equivalence can be made precise and in fact the critical behavior goes to the bone of any failure or catastrophic phenomena . several dynamical models of cooperative failure dynamics @xcite are now well - studied and their criticality at the global failure point are well established . here , we have reviewed some of the numerical studies of precursor behavior in two self - organising dynamical @xcite models as one approaches the self - organised critical ( soc ) point of the avalanches in the lattice sandpile models . the critical behavior of a random fiber bundle model @xcite of failure or fracture of a solid under global load sharing is now very precisely demonstrated . the dynamics of failure also become critically slow there and all these have been demonstrated analytically . the precursors of the global failure in the fiber bundle models can therefore be discussed analytically . the resulting universality of the surface roughness in any such fracture process has also been well documented and analysed . the two - fractal overlap model @xcite is a very recent modeling approach of earthquake dynamics . the time series of overlap magnitudes obtained , when one fractal slides over the other with uniform velocity , represents the model seismic activity variations . this time series analysis suggests some precursors of large events in this model . a ` pile ' of dry sand is a unique example soc system in nature . a growing sandpile gradually comes to a ` quasi - stable ' state through its self - organising dynamics . this ` quasi - stable ' state is called the critical state of the system as the system exhibits power law behavior there . because of avalanches of all sizes , this critical point in the pile is also a catastrophic one . the dynamics of growing sandpiles are successfully modeled by the lattice model @xcite of bak , tang and wisenfeld ( btw ) in 1987 . a stochastic version of sand pile model has been introduced by manna @xcite which also shows soc , although it belongs to a different universality class . both the models have been studied extensively at their criticality . here , we study the sub - critical behavior of both the models and look for the precursors of the critical state . let us consider a square lattice of size @xmath0 . at each lattice site @xmath1 , there is an integer variable @xmath2 which represents the height of the sand column at that site . a unit of height ( one sand grain ) is added at a randomly chosen site at each time step and the system evolves in discrete time . the dynamics starts as soon as any site @xmath1 has got a height equal to the threshold value ( @xmath3= @xmath4 ) : the site topples , i.e. , @xmath2 becomes zero there , and the heights of the four neighboring sites increase by one unit @xmath5 if , due to this toppling at site @xmath1 , any neighboring site become unstable ( its height reaches the threshold value ) , they in turn follow the same dynamics . the process continues till all sites become stable ( @xmath6 @xmath3 for all @xmath1 ) . when toppling occurs at the boundary of the lattice ( four nearest neighbors are not available ) , extra heights get off the lattice and are removed from the system . with a very slow but steady rate of addition of unit height ( sand grain ) at random sites of the lattice , the avalanches get correlated over longer and longer ranges and the average height ( @xmath7 ) of the system grows with time . gradually the correlation length ( @xmath8 ) becomes of the order the system size @xmath9 . here , on average , the additional height units start leaving the system as the system approaches toward a critical average height @xmath10 and the average height @xmath7 remains stable there ( see fig . 1(a ) ) . also the system becomes critical here ( for @xmath11 @xmath12 @xmath13 ) as the distributions of the avalanche sizes and the corresponding life times follow robust power laws @xcite . * _ figure 1 . _ * the growth of average height @xmath7 against the number of iterations of adding unit heights @xmath14 . eventually @xmath7 settles at @xmath10 . in the inset , we show the finite size behavior of the critical height @xmath10 , obtained from simulation results for different @xmath9 . ( a ) for the btw model ; ( b ) for the manna model . .2 in here , a finite size scaling fit @xmath15 ( obtained by setting @xmath8 @xmath16 @xmath17 ) , where c is a constant , with @xmath18 gives @xmath19 ( see inset of fig . similar finite size scaling fit with @xmath20 gave @xmath21 in earlier large scale simulations @xcite . btw model is a deterministic one . manna proposed the stochastic sand - pile model @xcite by introducing randomness in the dynamics of sandpile growth . here , the critical height is @xmath22 . therefore at each toppling the rejected two grains choose their host among the four available neighbors randomly with equal probability . after constant adding of sand grains , the system ultimately settles at a critical state . we consider now the manna model on a square lattice of size @xmath0 , where the sites can be either empty or occupied with unit height i.e. , the height variables can have binary states @xmath23 or @xmath24 . a site is chosen randomly and one height is added at that site . if the site is initially empty , it gets occupied : @xmath25 if the chosen site is previously occupied then a toppling or ` hard core interaction ' rejects both the heights from that site : @xmath26 and each of these two rejected heights stochastically chooses its host among the @xmath4 neighbors of the toppled site . the toppling can happen in chains if any chosen neighbor was previously occupied and thus cascades are created . after the system attains stable state ( dynamics stopped ) , a new site is chosen randomly and unit height is added to it . thus the system evolves in discrete time steps . here again the boundary is assumed to be completely absorbing so that heights can leave the system due to the toppling at the boundary . with a slow rate of addition of heights ( sand grains ) at random sites , initially the average height of the system grows with time and soon the system approaches toward a critical average height @xmath27 ( see fig . 1(b ) ) . here also the critical average height @xmath27 has a finite size dependence and a similar finite size scaling fit @xmath28 gives @xmath18 and @xmath29 ( see inset of fig . this is close to an earlier estimate @xmath30 @xcite , made in a somewhat different version of the model . the avalanche size distribution has got power laws similar to the btw model , at this self - organised critical state at @xmath31 . however the exponents seem to be different @xcite for this stochastic model , compared to those of btw model . at an average height @xmath7 ( @xmath32 ) , when all sites of the system have become stable ( dynamics have stopped ) , a fixed number of height units @xmath33 ( pulse of sand grains ) is added at any central point of the system . just after this addition , the local dynamics starts and it takes a finite time or iterations ( @xmath34 ) to return back to the stable state ( @xmath35 for all @xmath1 ) after several toppling events @xmath36 ; and the disturbance spreads over a length @xmath37 , the correlation length of the system @xcite . all these three parameters are seen @xcite to diverge as @xmath7 approaches the critical height @xmath27 from below following power laws @xmath38 , where @xmath39 ; @xmath40 @xmath41@xmath42 , where @xmath43 ; @xmath8 @xmath16 @xmath44 , where @xmath45 ( see fig . * _ figure 2 . _ * the variations of the precursors with @xmath7 ( @xmath32 ) in the btw model for different system sizes : @xmath46 ( plus ) @xmath47 ( cross ) and @xmath48 ( open circle ) . ( a ) for relaxation time @xmath49 ; in the inset @xmath50 is plotted against @xmath7 . ( b ) for the total number of topplings @xmath40 ; inset shows @xmath51 versus @xmath7 plot . ( c ) for the correlation length @xmath8 ; in the inset , @xmath52 is plotted against @xmath7 . the precursor parameters in manna model have been measured using similar method as in btw model . here the pulse of height is @xmath53 and is added to any arbitrary central site . we get @xcite exactly similar power law behaviors for all the parameters as the critical point is approached from below : @xmath38 , where @xmath39 ; @xmath40 @xmath41@xmath42 , where @xmath43 ; @xmath8 @xmath16 @xmath44 , where @xmath45 ( see fig . * _ figure 3 . _ * the variations of the precursors with @xmath7 ( @xmath32 ) in the manna model for different system sizes : @xmath46 ( plus ) @xmath47 ( cross ) and @xmath48 ( open circle ) . ( a ) for relaxation time @xmath49 ; in the inset @xmath50 is plotted against @xmath7 . ( b ) for the total number of topplings @xmath40 ; inset shows @xmath51 versus @xmath7 plot . ( c ) for the correlation length @xmath8 ; in the inset , @xmath52 is plotted against @xmath7 . the monte carlo studies showed that these response parameters like the relaxation time @xmath34 , size of the damage @xmath36 and its radial size @xmath37 , all tend to diverge , following the above mentioned robust power laws , in both the models . precise knowledge of these power laws can therefore help estimating the critical or catastrophic point ( @xmath54 ) by extrapolating the inverse power of these quantities for informations at @xmath55 ( @xmath56 ) ( see insets of figs . 2 and 3 ) . similar critical slowing down phenomena should be of interest in the context of magnetohydrodynamics of plasma @xcite . the fiber bundle model was introduced by peirce @xcite in the context of testing the strength of cotton yarns . fiber bundles are of two classes with respect to the time dependence of fiber strength : the ` static ' bundles contain fibers whose strengths are independent of time , where as the ` dynamic ' bundles are assumed to have time dependent elements to capture the creep rupture and fatigue behaviors . according to the load sharing rule , fiber bundles are being classified into two groups : global load - sharing ( democratic ) bundles and local load - sharing bundles . in democratic bundles intact fibers bear the applied load equally and in local load - sharing bundles the terminal load of the failed fiber is given equally to all the intact neighbors . with steadily increasing load , the fiber bundles approach their respective failure point obeying a dynamics determined by the load sharing rule . the phase transition @xcite and dynamic critical behavior of the fracture process in such democratic bundles has been established through recursive formulation @xcite of the failure dynamics . the exact solutions of the recursion relations in the global load - sharing case suggest universal values of the exponents involved . we discuss here the failure of a static fiber bundles under global load sharing ( democratic bundles ) with steadily increasing load on the bundles . we show analytically the variation of associated precursor parameters with the applied stress which help to estimate the failure point accurately . the model assumes equal load sharing , i.e. , the intact fibers share the applied load equally . the strength of each of the fibers in the bundle is determined by the stress value ( @xmath57 ) it can bear , and beyond which it fails . the strength of the fibers are taken from a randomly distributed normalised density @xmath58 within the interval @xmath59 and @xmath60 such that @xmath61 the global load sharing assumption neglects ` local ' fluctuations in stress ( and its redistribution ) and renders the model as a mean - field one . .1 in the breaking dynamics starts when an initial stress @xmath62 ( load per fiber ) is applied on the bundle . the fibers having strength less than @xmath62 fail instantly . due to this rupture , total number of intact fibers decreases and now these intact fibers have to bear the applied load on the bundle . hence effective stress on the fibers increases and this compels some more fibers to break . these two sequential operations , the stress redistribution and further breaking of fibers continue till an equilibrium is reached , where either the surviving fibers are strong enough to bear the applied load on the bundle or all fibers fail . this breaking dynamics can be represented by recursion relations in discrete time steps . let @xmath63 be the fraction of fibers in the initial bundle that survive after time step @xmath64 , where time step indicates the number of ( elemental ) stress redistributions . then the redistributed load per fiber after @xmath64 time step becomes @xmath65 and after @xmath66 time steps the surviving fraction of fiber is @xmath67 where @xmath68 is the cumulative probability of corresponding density distribution @xmath58 : @xmath69 using now eqns . ( 1.5 ) and ( 1.6 ) we can write the recursion relations which show how @xmath70 and @xmath63 evolve in discrete time : @xmath71 and @xmath72 the recursion relations ( 1.5 ) and ( 1.6 ) represent the basic dynamics of failure in global load sharing models . at the equilibrium or steady state @xmath73 and @xmath74 . this corresponds to a fixed point of the recursive dynamics . ( 1.6 ) can be solved at the fixed point for some particular distribution of @xmath58 and these solutions near @xmath75 ( or @xmath76 ) give the detail features of the failure dynamics of the bundle . .2 in we choose first the uniform density of fiber strength distribution to solve the recursive failure dynamics of democratic bundle . here , the cumulative probability becomes @xmath77 therefore @xmath63 follows a simple recursion relation ( following eqn . ( 1.6 ) ) @xmath78 .1 in * _ figure 4 . _ * three simple models considered here assume ( a ) uniform , ( b ) linearly increasing and ( c ) linearly decreasing density @xmath58 of the fiber strength distribution up to a cutoff strength . .1 in at the equilibrium state ( @xmath79 ) , the above relation takes a quadratic form of @xmath75 : @xmath80 the solution is @xmath81 here @xmath82 is the critical value of the initial applied stress beyond which the bundle fails completely . the solution with ( @xmath83 ) sign is the stable one , whereas the one with ( @xmath84 sign gives unstable solution . the quantity @xmath85 must be real valued as it has a physical meaning : it is the fraction of the original bundle that remains intact under a fixed applied stress @xmath62 when the applied stress lies in the range @xmath86 . clearly , @xmath87 ( putting @xmath88 in eqn . ( 1.13 ) ) . therefore the stable solution can be written as @xmath89 for @xmath90 we can not get a real - valued fixed point as the dynamics never stops until @xmath91 , when the bundle breaks completely . it may be noted that the quantity @xmath92 behaves like an order parameter that determines a transition from a state of partial failure ( @xmath93 ) to a state of total failure ( @xmath90 ) @xcite : @xmath94 to study the dynamics away from criticality ( @xmath95 from below ) , we replace the recursion relation ( 1.11 ) by a differential equation @xmath96 close to the fixed point we write @xmath97 + @xmath98 ( where @xmath99 ) . this , following eq . ( 1.13 ) , gives @xcite @xmath100 where @xmath101 $ ] . near the critical point we can write @xmath102 therefore the relaxation time diverges following a power - law as @xmath95 from below @xcite . one can also consider the breakdown susceptibility @xmath103 , defined as the change of @xmath85 due to an infinitesimal increment of the applied stress @xmath62 @xcite @xmath104 from eqn . hence the susceptibility diverges as the applied stress @xmath62 approaches the critical value @xmath105 . such a divergence in @xmath103 had already been observed in the numerical studies @xcite . here , the cumulative probability becomes @xmath106 therefore @xmath63 follows a recursion relation ( following eq . ( 1.6 ) ) @xmath107 at the fixed point ( @xmath79 ) , the above recursion relation can be represented by a cubic equation of @xmath75 @xmath108 solving the above equation we get @xcite the value of critical stress @xmath109 which is the strength of the bundle for the above fiber strength distribution . here , the order parameter , susceptibility , relaxation time all follow the same power laws ( eqns . ( 1.15 ) , ( 1.18 ) and ( 1.19 ) ) observed for uniform strength distribution . in this case , the cumulative probability becomes @xmath110 and @xmath63 follows a recursion relation ( following eqn . ( 1.6 ) ) @xmath111 at the fixed point ( @xmath79 ) , the above recursion relation can be represented by a cubic equation of @xmath75 @xmath112 solution of the above equation gives @xcite the value of critical stress @xmath113 which is the strength of the bundle . here also , the precursor parameters , the order parameter , susceptibility and relaxation time , all follow the same power laws ( eqns . ( 1.15 ) , ( 1.18 ) and ( 1.19 ) ) observed for uniform strength distribution . thus the democratic fiber bundles ( for different fiber strength distributions ) show phase transition from a state of partial failure to total failure with a well defined precursors ( order parameter , susceptibility and relaxation time ) which show similar power law variation on the way as the critical point is approached ; characterized by the universal values of the associated exponents ( @xmath114 , @xmath115and @xmath116 ) . all the above scaling behaviors represented by eqns . ( 1.15 ) , ( 1.18 ) and ( 1.19 ) for stress ( @xmath117 ) below the global failure stress ( @xmath118 ) of the bundle suggests that a prior knowledge of the responses like the fraction of failed fibers , relaxation time etc . can be extrapolated following the above power laws to estimate the global failure point @xmath118 . the two - fractal overlap model of earthquake @xcite is a very recent modeling attempt . such models are all based on the observed ` plate tectonics ' and the fractal nature of the interface between tectonic plates and earth s solid crust . the statistics of overlaps between two fractals is not studied much yet , though their knowledge is often required in various physical contexts . for example , it has been established recently that since the fractured surfaces have got well - characterized self - affine properties @xcite , the distribution of the elastic energies released during the slips between two fractal surfaces ( earthquake events ) may follow the overlap distribution of two self - similar fractal surfaces @xcite . chakrabarti and stinchcombe @xcite had shown analytically that for regular fractal overlap ( cantor sets and carpets ) the contact area distribution follows a simple power law decay . the two fractal overlap magnitude changes in time as one fractal moves over the other . the overlap ( magnitude ) time series can therefore be studied as a model time series of earthquake avalanche dynamics @xcite . here , we study the time ( @xmath64 ) variation of contact area ( overlap ) @xmath119 between two well - characterized fractals having the same fractal dimension as one fractal moves over the other with constant velocity . we have chosen only very simple fractals : regular and random cantor sets , gaskets and percolating cluster . we analyse the time series data of cantor set overlaps only to find the prediction possibility of large events ( occurrence of large overlaps ) . we show that the time series @xmath119 obtained by moving one fractal uniformly over the other ( with periodic boundary condition ) has some features which can be utilized to predict the ` large events ' . this is shown utilizing the discrete or a ` quantum ' nature of the integrated ( cumulative ) overlap over time . .2 in we now discuss the two - fractal overlap sizes . let us take two identical cantor sets at finite generation and slide one over the other , assuming periodic boundary condition . two such cases for a regular cantor sets as well as a gasket are shown in fig . similar cases for random fractals are shown in fig . .3 in .1 in .6 in .1 in * _ figure 5 . _ * ( a ) a regular cantor set of dimension @xmath120 ; only three finite generations are shown . ( b ) the overlap of two identical ( regular ) cantor sets , at @xmath121 , when one slips over other ; the overlap sets are indicated within the vertical lines , where periodic boundary condition has been used . ( c ) a regular gasket of dimension @xmath122 at the @xmath123th generation . ( d ) the overlap of two identical regular gaskets at same generations ( @xmath124 ) is shown as one is translated over the other ; periodic boundary condition has been used for the translated gasket . .3 in .1 in .6 in .1 in * _ figure 6 . _ * ( a ) a random cantor set of dimension @xmath120 ; only three finite generations are shown . ( b ) overlap of two random cantor sets ( at @xmath121 ; having the same fractal dimension ) in two different realisations . the overlap sets are indicated within the vertical bars . ( c ) a random realisation of a gasket of dimension @xmath122 at @xmath123th generation . ( d ) the overlap of two random gaskets of same dimension and of same generation but generated in different realisations . .1 in next , we study the overlap distribution of two well - characterised random fractals ; namely the percolating fractals @xcite . it seems , although many detailed features of the clusters will change with the changes in ( parent ) fractals , the subtle features of the overlap distribution function remains unchanged . .3 in .1 in * _ figure 7 . _ * the overlap between two percolating clusters ; ( a ) and ( b ) are two typical realisations of the same percolating fractal on square lattice ( @xmath125 ) and ( c ) their overlap set . note , the overlap set need not be a connected one . .1 in our earlier study @xcite on the overlap statistics for regular and random cantor sets , gaskets and percolating clusters indicated a universal scaling behavior of the overlap or contact area ( @xmath126 ) distributions @xmath127 for all types of fractal set overlaps mentioned : @xmath128 where @xmath9 denotes the finite size of the fractal and the exponent @xmath129 ; @xmath130 being the mass dimension of the fractal and @xmath131 is the embedding dimension . also the overlap distribution @xmath127 , and hence the scaled distribution @xmath132 , are seen to decay with @xmath126 or @xmath133 following a power law ( with exponent value equal to the embedding dimension of the fractals ) for both regular and random cantor sets and gaskets : @xmath134 however , for the percolating clusters @xcite , the overlap size distribution takes a gaussian form . we consider now the time series obtained by counting the overlaps @xmath119 as a function of time as one cantor set moves over the other ( periodic boundary condition is assumed ) with uniform velocity . the analysis presented here follows our previous study @xcite . the time series are shown in fig . 8 , for finite generations of cantor sets of dimensions @xmath120 and @xmath135 respectively . .2 in .1 in * _ figure 8 . _ * the time ( @xmath64 ) series data ofoverlap size ( @xmath126 ) for regular cantor sets : ( a ) of dimension @xmath120 , at @xmath136th generation : ( b ) of dimension @xmath135 , at @xmath137th generation . the obvious periodic repeat of the time series comes from the assumed periodic boundary condition of one of the sets ( over which the other one slides ) . now we calculate the cumulative overlap size @xmath138 and plot that against time in fig . note , that ` on average ' @xmath139 is seen to grow linearly with time @xmath64 for regular as well as random cantor sets . this gives a clue that instead of looking at the individual overlaps @xmath119 series one may look for the cumulative quantity . we observe @xmath140 , where @xmath141 is dependent on the fractal . this result is even more prominent in the case of cantor sets with @xmath142 @xmath135 . .2 in .2 in .1 in * _ figure 9 . _ * the cumulative overlap @xmath143 versus time ; for pure cantor sets : ( a ) of dimension @xmath120 ( at @xmath136th generation ) and ( b ) of dimension @xmath135 ( at @xmath137th generation ) . the dotted line corresponds to those for two identical but random cantor sets . in ( b ) the two lines fall on each other . we first identify the ` large events ' occurring at time @xmath144 in the @xmath119 series , where @xmath145 , a pre - assigned magnitude . we then look for the cumulative overlap size @xmath146 , @xmath147 @xmath148 , where the successive large events occur at times @xmath149 and @xmath144 . the behavior of @xmath150 with time is shown in fig . 10 for regular cantor sets with @xmath151 @xmath120 at generation @xmath152 . similar results are also given for cantor sets with @xmath151 @xmath135 at generation @xmath153 in fig . @xmath154 is seen to grow almost linearly in time up to @xmath155 ) after which it drops down to zero . it appears that there are discrete ( quantum ) values of @xmath156 ) . one can therefore anticipate large events when the cumulative seismic overlap @xmath154 , since the last major event at @xmath149 , grows linearly with time @xmath147 to the discrete levels @xmath157 , specific to the series of events . no major event occurs during the growth period of @xmath158 for the intermediate values . .3 in .1 in * _ figure 10 . _ * the cumulative overlap size variation with time ( for regular cantor sets of dimension @xmath120 , at @xmath136th generation ) , where the cumulative overlap has been reset to @xmath59 value after every big event ( of overlap size @xmath159 where @xmath160 ( a ) and @xmath161 ( b ) respectively ) . .2 in * _ figure 11 . _ * the cumulative overlap size variation with time ( for regular cantor sets of dimension @xmath135 , at @xmath137th generation ) , where the cumulative overlap has been reset to @xmath59 value after every big event ( of overlap size @xmath159 where @xmath162 ( a ) and @xmath163 ( b ) respectively ) . in all the dynamical models of sandpile avalanches ( discussed in section 2 ) , we find that the growing correlations in the dynamics of constituent elements manifest themselves as various precursors . the number of topplings @xmath40 , relaxation time @xmath49 and the correlation length @xmath8 , in both btw and manna model , grow and diverge following power laws as the systems approach their respective critical points @xmath27 from below ( see figs . 2 and 3 ) : @xmath40 @xmath41@xmath42 , @xmath164 and @xmath8 @xmath16 @xmath44 . for two dimensional systems , we find numerically here @xmath165 , @xmath166 and @xmath18 for both btw and manna model . basically , we studied the behavior for @xmath167 , the precursor behavior , where @xmath8 is necessarily finite . as we add here the tiny pulse at some central site of a relatively large system , the boundary effect can not be really felt because of the smallness of @xmath8 compared to @xmath9 for most values of @xmath7 . this explains the lack of finite size effect in our precursor studies . for the global failure or fracture in fiber bundle models ( discussed in section 3 ) , we find that the breakdown susceptibility @xmath103 ( giving the increment in the number of broken fibers for an infinitesimal increment of load on the bundle ) and the corresponding relaxation time @xmath49 ( required for the bundle to stabilise , after successive failures of the fibers ) , both diverge as the external load or stress approaches its global failure point @xmath82 from below : @xmath49 @xmath16 @xmath168 @xmath169@xmath170 and @xmath103 @xmath16 @xmath168 @xmath169@xmath171 ; @xmath172 . these results are of course analytically derived for different fiber strength distributions and the universality class of the model has been confirmed . if one cantor set moves uniformly over another , the overlap between the two fractals change quasi - randomly with time ( see eg . , fig . 8) . these time series are taken as model seismic activity variations in such two - fractal overlap model of earthquake . the overlap size distribution was argued @xcite and shown @xcite to follow power law decay . we showed numerically ( see section 4 ) that if one fixes a magnitude @xmath173 of the overlap sizes @xmath126 , so that overlaps with @xmath174 are called ` events ' ( or earthquake ) , then the cumulative overlap @xmath175 , @xmath176 @xmath177 , ( where two successive events of @xmath174 occur at times @xmath149 and @xmath144 ) grows linearly with time up to some discrete quanta @xmath178 , where @xmath179 is the minimal overlap quantum , dependent on @xmath173 and @xmath180 ( the generation number ) . here @xmath181 is an integer ( see figs . 10 , 11 ) @xcite . although our results here are for regular fractals of finite generation @xmath180 , the observed discretisation of the overlap cumulant @xmath150 with the time limit set by @xmath180 , is a robust feature and can be seen for larger time series for larger generation number @xmath180 . this part of the model study therefore indicates that one can note the growth of the cumulant seismic response @xmath182 , rather than the seismic event strength @xmath119 , and anticipate some big events as the response reaches the discrete levels @xmath157 , specific to the series of events . some prior knowledge of the precursors and their precise behavior ( power laws etc . ) , as discussed here for three different kinds of failure models of cooperatively interacting dynamical systems which are prone to major failures , are therefore plausible . these precursor responses should help estimating the location of the global failure or critical point using proper extrapolation of the above quantities , which are available long before the failure occurs . the usefulness of such precursors can hardly be overemphasized . 1 b. k. chakrabarti and l. g. benguigui , _ statistical physics of fracture and breakdown in disorder systems _ , oxford univ . press , oxford ( 1997 ) ; h. j. herrmann and s. roux ( eds ) , _ statistical models for the fracture of disordered media _ , north holland , amsterdam ( 1990 ) ; p. bak , _ how nature works , _ oxford univ . press , oxford ( 1997 ) . f. t. peires , j. textile inst . * 17 * , t355 ( 1926 ) ; h. e. daniels , proc . london a * 183 * 405 ( 1945 ) ; s. l. phoenix and r. l. smith , int . j. solid struct . * 19 * , 479 ( 1983 ) ; r. da silveira , am . * 67 * 1177 ( 1999 ) ; e. altus , mech . * 34 * 257 ( 2002 ) ; s. pradhan and b. k. chakrabarti , proc . dmtas ( chennai , march 2003 ) , int . b ( in press ) , cond - mat/0307734 . b. k. chakrabarti , r. b. stinchcombe , physica a , * 270 * 27 ( 1999 ) . p. bak , c. tang and k. wiesenfeld , phys . lett . * 59 * 381 ( 1987 ) ; d , dhar , phys . rev . lett . * 64 * 1613 ( 1990 ) . s. s. manna , j. phys . a : math . gen . * 24 * l363 ( 1991 ) ; d. dhar , physica a * 270 * 69 ( 1999 ) . m. acharyya and b. k. chakrabarti , physica a * 224 * 254 ( 1996 ) ; phys . e * 53 * 140 ( 1996 ) . s. pradhan and b. k. chakrabarti , phys . e * 65 * , 016113 ( 2001 ) . s. ortolani and d. d. schnack , _ magnetohydrodynamics of plasma relaxation , world scientific publ . _ , singapore ( 1993 ) . s. pradhan , p. bhattacharyya and b. k. chakrabarti , phys . e * 66 * , 016116 ( 2002 ) ; p. bhattacharyya , s. pradhan and b. k. chakrabarti , phys . rev . e * * 6**7 , 046122 ( 2003 ) . b. b. mandelbrot , _ the fractal geometry of nature _ freeman , san francisco ( 1982 ) ; b. b. mandelbrot , d. e. passoja , a. j. pullay , nature * 308 * 721 ( 1984 ) ; a. hansen and j. schmittbuhl , phys . lett . * 90 * , 045504 ( 2003 ) ; j. o. h. bakke , j. bjelland , t. ramstad , t. stranden , a. hansen and j. schmittbuhl , proc . uasp03 ( kolkata , march 2003 ) , physica scripta * t106 * 65 ( 2003 ) . s. pradhan , b. k. chakrabarti , p. ray and m. k. dey , proc . uasp03 ( kolkata , march 2003 ) , physica scripta * t106 * 77 ( 2003 ) . j. m. carlson , j. s. langer and shaw , rev . * 62 * 2632 ( 1989 ) . see e.g. , d. stauffer and a. aharony , _ introduction to percolation theory _ , taylor and francis , london ( 1994 ) . s. pradhan , p. choudhuri and b. k. chakrabarti , in _ proc . nato conf . _ ( cmds-10 , soresh , israel , july 2003 ) , _ continuum models of discrete systems _ , eds . d. j. bergman and e. inan , kluwer acad . pub . , new york , cond - mat/0307735 .
we review here briefly the nature of precursors of global failures in three different kinds of many - body dynamical systems . first , we consider the lattice models of self - organised criticality in sandpiles and investigate numerically the effect of pulsed perturbations to the systems prior to reaching their respective critical points . we consider next , the random strength fiber bundle models , under global load sharing approximation , and derive analytically the partial failure response behavior at loading level less than its global failure or critical point . finally , we consider the two - fractal overlap model of earthquake and analyse numerically the overlap time series data as one fractal moves over the other with uniform velocity . the precursors of global or major failure in all three cases are shown to be very well characterized and prominent . .3 in saha institute of nuclear physics , 1/af bidhan nagar , kolkata 700064 , india fracture , earthquake , avalanches in sandpile , self - organised criticallity , fiber bundle , fractals , cantor sets . .2 in
observations as well as models of galactic chemical evolution suggest that the metallicity at the center of our galaxy , where the density of matter is high , can reach up to 3 to 5 times the solar metallicity ( mcwilliam & rich 1994 , simpson et al . other objects such as elliptical galaxies , or maybe even quasars ( korista et al . 1996 ) , also show evidences of very metal - rich stellar populations . the metallicity affects the evolution of stars mainly through its impact on the radiative opacities , the equation of state , the nuclear reaction rates and the stellar mass loss rates . the impact of the metallicity on the stellar properties has been well studied for @xmath13 on base of extensive stellar model calculations ( see e.g. schaller et al . 1992 , hereafter paper i , and maeder & conti 1994 ) . in contrast , they are still poorly known at metallicities much higher than solar . recent calculations have been performed at metallicities five times solar ( fagotto et al . 1994 , mowlavi et al . mowlavi et al . ( 1997 ) notice that those very metal - rich models present specific properties which are distinct from the ones at @xmath14 . in particular , they are _ hotter _ and _ more luminous _ that those at solar metallicity , contrary to expectations . it is the purpose of this paper to analyze in detail some of these properties of very metal - rich stars . these could have important consequences on the observable properties of high metal - rich galaxies . the study is based on the extensive grids of stellar models computed by the geneva group with @xmath15 . it covers the main phases of stellar evolution , excluding the latest stages of evolution . the core helium burning phase of very metal - rich low - mass ( @xmath16 ) stars , important for the study of elliptical galaxies , is not considered in this paper as it has been the subject of existing papers in the literature ( see , for example , dorman , rood & oconnell , 1993 , fagotto et al . 1994 ) . a general discussion on the dependence of stellar properties on chemical composition is presented in sect . [ sect : chemical composition ] . section [ sect : models ] summarizes the model ingredients used in the computation of the geneva grids . the stellar properties during the main sequence ( ms ) and the core helium burning ( cheb ) phase are analyzed in sects . [ sect : ms ] and [ sect : cheb ] , respectively . wolf - rayet stars in very metal - rich environments are briefly discussed in sect . [ sect : wr ] . conclusions are drawn in sect . [ sect : conclusions ] . chemical composition is traditionally described by @xmath17 , @xmath1 and @xmath0 , which are the mass fractions of hydrogen , helium and heavier elements , respectively . in the course of galactic evolution , the interstellar matter is progressively enriched in heavy elements , and @xmath0 increases through time . it is evident that helium is also enriched , the increase of its interstellar abundance with respect to @xmath0 being conveniently described by a @xmath18 law such that @xmath19 ( where @xmath20 is the primordial he abundance ) . at low metallicities ( @xmath21 ) , @xmath1 remains practically constant [ an absolute variation @xmath22 of the metallicity induces an absolute variation of he of @xmath23 , so that @xmath24 for @xmath25 , and the stellar properties _ as a function of metallicity _ are thus expected to be determined mainly by @xmath0 . at high metallicities ( @xmath26 ) , on the other hand , @xmath1 increases ( or , alternatively , @xmath17 decreases ) significantly with @xmath0 ( @xmath1 passes from 0.30 to 0.48 when @xmath0 increases from 0.02 to 0.1 , see table [ tab : grids ] ) . in those conditions , both @xmath0 _ and _ @xmath1 ( and @xmath27 ) determine the stellar properties as a function of metallicity . chemical composition influences the stellar structure basically through three contributions : the mean molecular weight @xmath28 , the radiative opacity @xmath29 , and the nuclear energy production @xmath30 . the metallicity - dependent mass loss rate @xmath31 further affects stellar evolution as a function of @xmath0 . the mean molecular weight acts in the hydrostatic equilibrium of the star . let us imagine a _ thought experiment _ in which @xmath28 is arbitrarily increased over the whole star , all other quantities such as @xmath29 or @xmath32 being unaltered ( though keeping , of course , their density and temperature dependences ) . the reduction in the gas pressure leads to an overall contraction ( and heating ) until hydrostatic equilibrium is re - established . as a result , the stellar radius decreases and the surface temperature @xmath33 increases . the surface luminosity @xmath34 , on the other hand , increases with @xmath28 since @xmath35 [ in the case of radiative transfer , with @xmath36 and @xmath37 being the luminosity and temperature at radius @xmath38 , respectively ] . in summary , _ an increase in @xmath28 tends to increase both @xmath34 and @xmath33_. this @xmath28-effect is undetectable in low - metallicity stars , where @xmath28 is essentially constant as a function of metallicity . at metallicities higher than about solar , however , @xmath28 increases as a result of increasing @xmath1 [ @xmath0 remains unimportant in this respect for all realistic values of @xmath0 , since @xmath39 . and indeed , it is shown in sections [ sect : ms ] and [ sect : cheb ] that very metal - rich models are more luminous and hotter than solar - metallicity ones . it has to be noted that the @xmath18 law adopted in metal - rich models has an important influence in this respect , as it fixes @xmath1 and thus @xmath28 . lccccll grid@xmath40 & @xmath0 & @xmath17 & @xmath1 & @xmath31 & interior opacities & low t opacities + i & 0.001 & 0.756 & 0.243 & standard & rogers & iglesias ( 1992 ) & kurucz ( 1991 ) + iii & 0.004 & 0.744 & 0.252 & standard & iglesias et al . ( 1992 ) & kurucz ( 1991 ) + ii & 0.008 & 0.728 & 0.264 & standard & rogers & iglesias ( 1992 ) & kurucz ( 1991 ) + i & 0.020 & 0.680 & 0.300 & standard & rogers & iglesias ( 1992 ) & kurucz ( 1991 ) + iv & 0.040 & 0.620 & 0.340 & standard & iglesias et al . ( 1992 ) & kurucz ( 1991 ) + vii & 0.100 & 0.420 & 0.480 & standard & iglesias & rogers ( 1996 ) & alexander & fergusson ( 1994 ) + v & @xmath41 & & 2@xmath42standard & iglesias et al . ( 1992 ) & kurucz ( 1991 ) + @xmath40 = iii= : charbonnel et al . ( 1993 , paper iii ) ; = vii= @xmath40 i : schaller et al . ( 1992 , paper i ) ; iv : schaerer et al . ( 1993 , paper iv ) ; + ii : schaerer et al . ( 1993 , paper ii ) ; v : meynet et al . ( 1994 , paper v ) ; + iii : charbonnel et al . ( 1993 , paper iii ) ; vii : mowlavi et al . ( 1997 , paper vii ) . + the radiative opacity acts in the ( radiative ) transport of energy . let us consider again a _ thought experiment _ in which the opacity is arbitrarily increased throughout the star , all other quantities ( such as @xmath28 or @xmath30 ) being unaltered . the hydrostatic structure remains unaffected to first approximation . in particular the radius and temperature profiles are unchanged . the energy flux , on the other hand , decreases with increasing @xmath29 [ since @xmath43 for radiative transfer ] , leading to a decrease in the surface luminosity . as the radius is unaffected to first order , the surface temperature decreases too . in summary , _ an increase in @xmath29 tends to decrease both @xmath34 and @xmath33_. in general , the opacities increase with increasing metallicity . this is the case for the bound - free ( see eq . 16.107 in cox & giuli 1969 ) and free - free ( see eq . 16.95 in cox & giuli 1969 ) transitions . in contrast , electron scattering depends only on @xmath17 [ @xmath44 , which is about constant at @xmath45 and decreases with @xmath0 at @xmath46 . as seen in sect . [ sect : ms ] , these considerations explain the stellar properties as a function of @xmath0 at sub - solar metallicities . the nuclear energy production @xmath30 sustains the stellar luminosity . if @xmath30 is arbitrarily increased , the central regions of the star expand , leading to a decrease in the temperatures and densities , and to an increase in the stellar radius . thus , _ an increase in @xmath30 tends to decrease both @xmath34 and @xmath33 _ , as well as the central temperature @xmath47 and density @xmath48 . the way @xmath30 depends on metallicity is closely related to the mode of nuclear burning . when the cno cycles are the main mode of h burning , @xmath30 increases with @xmath0 . in contrast , when the main mode of burning is the pp chain , @xmath30 is related to @xmath17 . it is thus about independent of @xmath0 at @xmath21 , and decreases with increasing @xmath0 at @xmath26 . as shown in sect . [ sect : ms ] , these properties determine the behavior of @xmath47 and @xmath48 in ms stars as a function of @xmath0 . when mass loss is driven by radiation , kudritzki et al . ( 1989 ) showed that @xmath31 in o stars is proportional to @xmath49 . as a result , the effects of mass loss dominate in more metal - rich stars . mass loss reduces the stellar mass , modifying thereby the post - ms stellar properties ( and even during the ms for the most massive ones ) . it also turns the ( massive ) star into a wolf - rayet star at an earlier stage than it would at lower metallicity . this is analysed in more detail in sect . [ sect : wr ] . at @xmath21 , @xmath0 is the main factor acting through the @xmath29-effect . the adopted @xmath18 law is thus not influential . at @xmath26 , on the other hand , @xmath1 plays a dominant role mainly through the @xmath28-effect , while @xmath0 acts on the mass loss rate . the adopted @xmath18 law is then important . these combined effects of @xmath28 , @xmath29 and @xmath30 on the stellar surface properties can be estimated more quantitatively with a semi - analytical approach using homology relations . this is developed in the appendix for a star . a more detailed analysis of stellar properties with metallicity , however , can only be done with the help of stellar model calculations . such analysis is provided in sects . [ sect : ms ] and [ sect : cheb ] using the geneva grids of models , whose ingredients are first summarized in the next section . the stellar grids published by the geneva group from @xmath50 to 0.1 ( summarized in table [ tab : grids ] ) constitute a homogeneous set of models . they are computed with a value of @xmath51 for the ratio of the mixing length @xmath52 in convective regions to the pressure scale height @xmath53 ( determined in order to match the sun and the location of the red giant branch in the hr diagram ) , and a moderate core overshooting distance of @xmath54 beyond the classical convective core boundary as defined by the schwarzschild criterion ( in order to reproduce the ms width in the hr diagram ) . we refer to paper i ( see table [ tab : grids ] for the references of the papers ) for a general description of the physical ingredients . we mention below only the specific ingredients which differ from one set of calculations to another . the primordial helium abundance @xmath20 is taken equal to 0.24 ( cf . audouze 1987 ) for all models . the relative ratio of helium to metal enrichment , on the other hand , is @xmath55 at @xmath56 , 2.5 at @xmath57 , and 2.4 at @xmath4 . at low metallicity , the model output is not very sensitive to the adopted value for @xmath18 . at @xmath26 , however , the model output is sensitive to @xmath18 ( see sect . [ sect : chemical composition ] ) . the values adopted at @xmath57 and 0.1 are consistent with the currently admitted value of @xmath58 ( see peimbert 1995 and references therein ) , but the uncertainties are still very high ( maeder 1998 ) . the grids presented in papers i to vii are all calculated with the opal opacities . they use , however , different opal tables according to their availability at the time of each calculation , which are summarized in table [ tab : grids ] . the low temperature opacities are taken from kurucz ( 1991 ) in all grids , except in paper vii which uses the alexander & ferguson ( 1994 ) tables . the mass loss prescription is of crucial importance for the evolution of the models . a general dependence of @xmath59 , as suggested by the models of kudritzki et al . ( 1989 ) , is adopted for all the stars except the wr stars . a reimers law ( 1975 ) is adopted in all grids for low- and intermediate - mass stars . for massive stars , the situation is more delicate . in a first set of calculations ( papers i to iv ) , mass loss rates from de jager et al . ( 1988 , ` standard ' mass loss rates ) were used . a second set of calculations ( paper v ) were then performed with enhanced mass loss rates during the pre - wr and wnl phases ( by a factor of two over the ` standard ' ones ) . the latter models were found to better reproduce many observational constraints concerning massive stars ( maeder & meynet 1994 ) , and we use them in this paper too . at @xmath4 , however , the standard mass loss rates for massive stars are used ( paper vii ) . indeed , we expect the mass loss rates driven by radiation pressure to be probably saturated at such high metallicities . the conclusions are , however , qualitatively not affected by this choice . let us recall that at @xmath4 , the criterion defining a star to be of the wolf - rayet type could be different from that at lower metallicities . however , in the absence of any observational counterpart of those objects at such high metallicities , we keep the same criterion as in paper i , i.e. we consider that a star enters the wr stage when its surface hydrogen mass fraction becomes lower than 0.4 and its effective temperature is higher than 10000 k. these values are actually not crucial for the analysis of wr populations since the characteristic time - scale for entering the wr stage is very short compared to the wr lifetimes . the mass loss rates of these objects are then accordingly taken as in paper i. the arguments presented in sect . [ sect : chemical composition ] enable to understand the location of the zero - age main sequence ( zams ) in the hertzsprung - russell ( hr ) diagram as a function of metallicity . that location is given in fig . [ fig : hr ] for stars of 1 , 3 and at metallicities ranging from @xmath50 to 0.1 . [ [ metallicities - below - simeq-0.04 ] ] metallicities below @xmath60 : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + at @xmath61 , the surface properties ( as a function of @xmath0 ) are determined by the @xmath29-effect ( see sect . [ sect : kappa - effect ] ) . the surface temperature is mainly sensitive to the opacity in the outer layers where bound - bound and bound - free transitions dominate , so that all zams models get cooler with increasing metallicities . the surface luminosity , on the other hand , is sensitive to the interior opacities . in low- and intermediate - mass stars , the main source of opacity is provided by bound - free transitions , and increases with metallicity . as a result , @xmath34 decreases with @xmath0 . in massive stars , however , electron - scattering is the main source of opacity . as this source is metallicity independent , the luminosity of low - metallicity massive stars is insensitive to @xmath0 . [ [ metal - rich - stars ] ] metal - rich stars : + + + + + + + + + + + + + + + + + as the metallicity increases above @xmath62 , the @xmath28-effect becomes competitive with the @xmath29-effect . in low- and intermediate - mass stars , the @xmath28-effect eventually exceeds the @xmath29-effect above @xmath63 , and a further increase in the metallicity results in _ hotter _ and _ more luminous_. zams models ( see sect . [ sect : mu - effect ] ) . in massive stars , @xmath34 already begins to increase as a function of @xmath0 at @xmath62 since the @xmath29-effect on @xmath34 is negligible in those stars . let us now consider the central temperature @xmath47 and density @xmath48 . those quantities are , contrarily to @xmath34 and @xmath33 , closely related to the nuclear energetic properties of the core material ( see sect . [ sect : epsilon - effect ] ) . in particular , they are sensitive to the mode of hydrogen burning . usually , the cno cycles are the main mode of core h burning for stars more massive than about ( depending on metallicity ) , while the pp chain operates at lower masses . at @xmath4 , however , all stars are found to burn their hydrogen through the cno cycles ( at least down to @xmath64 considered in our grids ) . when the cno cycle is the main mode of burning , @xmath48 and @xmath47 decrease with increasing @xmath0 ( see sect . [ sect : epsilon - effect ] ) . this is clearly visible in fig . [ fig : rot ] for the 3 and stars at @xmath65 . at @xmath4 , however , the higher surface luminosities of the models as compared to those at @xmath57 result in concomitant higher central temperatures . when the pp chain is the main mode of burning , on the other hand , the location in the @xmath66 is not very sensitive to @xmath0 . this is well verified for the models at @xmath67 ( fig . [ fig : rot ] ) . the mass of the convective core is mainly determined by the nuclear energy production , and thus by @xmath34 . at @xmath68 , it slightly decreases with increasing @xmath0 in low- and intermediate - mass stars , while it remains approximately constant in massive stars ( see fig . [ fig : zamsqcc ] ) . at @xmath69 , on the other hand , it increases with @xmath0 at all stellar masses - effect , and the core mass increases with increasing @xmath0 . ] . we recall furthermore that at @xmath4 all stars with @xmath70 as low as possess convective cores . the ms lifetimes @xmath71 are summarized in fig . [ fig : t ] for three different metallicities as a function of the initial stellar mass , and in fig . [ fig : th ] for several stellar masses as a function of metallicity . they can be understood in terms of two factors : the initial h abundance , which determines the quantity of available fuel , and the luminosity of the star , which fixes the rate at which this fuel burns . at @xmath21 , @xmath71 is mainly determined by @xmath34 , being shorter at higher luminosities . [ fig : th ] confirms , as expected from @xmath34 ( see fig . [ fig : hr ] ) , that @xmath71 increases with @xmath0 in low- and intermediate - mass stars , and is about independent of @xmath0 in massive stars . at @xmath72 , on the other hand , the initial h abundance decreases sharply with increasing @xmath0 ( the h depletion law is dictated by the adopted @xmath73 law ; for @xmath74 and @xmath75 , @xmath17 drops from 0.69 at @xmath5 to 0.42 at @xmath4 ) . the ms lifetimes are then mainly dictated by the amount of fuel available . this , combined with the higher luminosities at @xmath4 , leads to ms lifetimes which are about 60% shorter at @xmath4 than at @xmath5 . this result is independent of the stellar mass for @xmath76 . above this mass , however , the action of mass loss extends the ms lifetime , as can be seen from the curve in fig . [ fig : th ] . mass loss determines crucially the evolution of massive stars , especially at high metallicities . the stellar masses remaining at the end of the ms phase with the ` enhanced ' mass - loss prescription ( see sect . [ sect : models ] ) are shown in thick lines in fig . [ fig : mh ] ( the ones with ` standard ' mass - loss prescription is also shown in dotted lines for comparison ) . the most striking result at metallicities higher than twice solar is the rapid evaporation of massive stars with @xmath77 , and the consequent formation of wr stars during core h burning . at @xmath57 , stars more massive than @xmath78 leave the core hydrogen burning phase with less than . the same holds true at @xmath4 even though ` standard ' mass loss rates are used for that metallicity . the surface properties during the cheb phase are dictated by the same arguments developed in sect . [ sect : chemical composition ] and used to explain the ms properties ( see sect . [ sect : ms ] ) . at @xmath68 , both @xmath34 and @xmath33 decrease with increasing @xmath0 , due to the @xmath29-effect . at @xmath69 , on the other hand , both @xmath34 and @xmath33 increase with @xmath0 , due to the @xmath28-effect . these properties are well verified in figs . [ fig : hebhr3 ] and [ fig : hebhr20 ] which show the cheb tracks in the hr diagram of intermediate - mass and massive stars , respectively . low- and intermediate - mass stars of low metallicity are known to perform a blue loop in the hr diagram and cross the cepheid instability strip . the extent of that loop is a function of both the stellar mass ( more massive stars having more extended loops ) and the metallicity ( more metal - rich stars having less extended blue loops ) . for a star , the blue loop is suppressed at @xmath79 ( see fig . [ fig : hebhr3 ] ) , while it is suppressed at @xmath80 for a star . furthermore , no blue loop is obtained for any stellar mass at metallicities higher than @xmath81 . another interesting property is the effective temperature at which massive stars enter the cheb phase . low - metallicity massive star models ignite helium as blue supergiants , while solar metallicity models ignite helium as red supergiants . this location of helium ignition as a function of @xmath0 in the hr diagram is shown in fig . [ fig : hebhr20 ] . as a result , the number ratio of blue - to - red supergiants is a decreasing function of the metallicity . we finally notice , again , the departure of @xmath4 models to the general rule : they ignite helium at a surface temperature higher than those of solar metallicity models . this is due to the @xmath28-effect ( see sect . [ sect : mu - effect ] ) . the cheb lifetime @xmath82 is displayed in fig . [ fig : t ] for three different metallicities as a function of the initial stellar mass , and in fig . [ fig : the ] as a function of metallicity for several initial masses . it follows a pattern very similar to that of @xmath71 ( fig . [ fig : th ] ) for intermediate - mass stars , and can be understood on grounds of the same arguments developed in sect . [ sect : th ] . in massive stars , the cheb lifetime is extended as a result of the lower average luminosities due to mass loss . as a result , the cheb lifetime relative to its value at @xmath5 increases with the initial stellar mass . the ratio @xmath83 is displayed in fig . [ fig : theth ] . below @xmath81 , this ratio is a decreasing function of metallicity , for a given stellar mass , and an increasing function of stellar mass , for a given metallicity . above @xmath84 , this ratio converges to the value of @xmath85 , irrespective of the stellar mass . for stars more massive than @xmath78 , however , the high mass loss rates reduce further this ratio . the masses at the end of the cheb phase of massive stars are displayed in fig . [ fig : mhe ] for several metallicities . these are a decreasing function of increasing metallicity . the properties of the wr star models at metallicities between 0.001 and 0.04 have already been presented in maeder & meynet ( 1994 ) . here we shall focus our discussion on the expected characteristics of the wr stars in very metal - rich environments , i.e. at @xmath86 . the minimum initial mass for a single star to become a wr star is a decreasing function of metallicity . it goes from at @xmath5 to at @xmath57 ( see table 1 in maeder & meynet 1994 , @xmath87 case ) , and to approximately at @xmath4 . at @xmath4 , stars more massive than about 60 @xmath89 literally evaporate during the main sequence phase . they loose so much mass that they never reach the he - burning phase , and will likely end their nuclear life as helium white dwarfs . let us remark here that helium white dwarf stars are generally thought to be the end point evolution of very low mass stars typically with initial masses between 0.08 and 0.5 @xmath89 . their existence in our present universe is thus doubtful ( unless created through binary mass transfer ) , since the ms lifetimes of these very low mass stars are greater than the present age of the universe . in contrast , the above scenario ( already addressed by maeder & lequeux 1982 ; see also paper v ) , in which massive evaporating stars could end their stellar life as helium white dwarfs , allows the formation of such objects in young star forming regions at very high metallicities . stars more massive than about 50 @xmath89 never enter the wc phase at @xmath4 . due to very high mass loss rates acting already during the ms , the h - convective core rapidly decreases in mass , leaving behind a great portion of the star with cno - enriched material . as a consequence , the star goes through a very long wn phase . moreover , these stars enter the core he - burning phase , if they do enter this phase at all , with very low masses . this implies in particular so small mass loss rates , according to the @xmath90 relation proposed by langer ( 1989 ) , that their helium burning core never uncovers , preventing the star to become a wc wolf - rayet star . the number ratio of wr to o type stars expected in a constant star formation rate region at @xmath4 is estimated from these models to be around 0.45 . to compute this value , only the single star evolutionary channel was considered , we chose a salpeter initial mass function ( i.e. with a slope equal to 1.35 ) , and the upper mass limit was taken equal to 60 @xmath89 . from the above result , one expects to find at @xmath4 about 1 wr star per each 2 o - type stars detected ! this is a very large proportion compared to regions with solar and sub - solar metallicity ( in the solar neighborhood , the observed number ratio wr / o is 0.1 ) . this high proportion results mainly from the very high mass losses experienced by these metal - rich stars , which , as seen above , decrease the minimum initial mass for wr star formation , make the star enter the wr phase at an earlier stage of its evolution , and increase the core helium burning lifetime . moreover , the duration of the o - type phase of metal - rich stars is considerably reduced as compared to that at lower metallicities . for instance when z increases from 0.04 to 0.1 , the lifetime of the o - type phase decreases by about a factor 2 . this is essentially a consequence of the shorter duration of the h - burning phase at @xmath4 ( see fig . [ fig : th ] ) . let us now consider the final evolution of massive stars and estimate the relative fraction of supernovae with wr progenitors using some simplifying assumptions ( see also meynet et al . , 1998 ) . there is at present some uncertainties concerning the fate of wr stars , and to which type of supernovae they can give birth . as a first approximation , we make the hypothesis that all wr stars do lead to supernovae events ( see a discussion in maeder & lequeux 1982 ) . moreover , since the h - rich envelope is ejected through the wind during the wnl phase , we further assume that wne and wc stars result in supernovae of type i ( whether ib , ic , or another yet unknown subclassification ) . with these assumptions , and not considering type ia supernovae which have low - mass progenitors , one expects that metal - rich regions be characterized by a higher fraction of type i(b / c ) supernovae relative to the total number of supernovae ( maeder 1992 ) . as a numerical example , using the same imf as in sect . [ sect : wr - o ] and making the hypothesis of constant star formation rate , one obtains that , at @xmath4 , the fraction of supernovae with wne - wc progenitors amounts to 32% of the total number of supernovae having massive star progenitors . in other words , 1 supernova event out of 3 should be of type i(b / c ) in very metal - rich regions . for comparison , this proportion decreases to 1 wr explosion per @xmath91 supernovae at solar metallicity . this paper analyzes some properties of stellar models as a function of the metallicity , with particular attention to those at very high metallicities ( above twice or three times solar ) . the stellar properties as a function of @xmath0 are shown to result from the action of four ingredients : the mean molecular weight , an increase of which leads to more luminous and hotter stars ( the @xmath28-effect ) ; the opacity , an increase of which leads to less luminous and cooler stars ( the @xmath29-effect ) ; the nuclear energy production ( but to a lesser extent ) , an increase of which leads to less luminous and cooler stars with lower @xmath47 and @xmath48 ( the @xmath30-effect ; and mass loss , which acts indirectly on the stellar structure by reducing the stellar mass ( dramatically in the most massive metal - rich stars ) . the @xmath29-effect plays the main role at @xmath68 , while the @xmath28-effect becomes preponderant at @xmath26 . in the later case , the @xmath18 law is important . based on these considerations and on the geneva s grids of stellar models , it is shown in particular that very metal - rich stars exhibit properties which can not be extrapolated from model characteristics at lower metallicities . in particular , - very metal - rich stars ( such as at @xmath4 ) are more luminous and hotter than those at @xmath92 . this property contrasts with the usually known trend at lower metallicities ( whereby @xmath34 and @xmath33 decrease with increasing @xmath0 due to the @xmath29-effect ) ; - very metal - rich stars have much lower ms lifetimes than those at @xmath92 . for example the ms lifetime at @xmath4 is 2.5 times shorter than that at solar metallicity , independent of their initial mass ( at least up to ) ; - stellar population synthesis in metal - rich environments is significantly affected by the increased mass loss rates at high @xmath0 . in particular , a ) massive stars with @xmath4 enter the wnl wolf - rayet phase very early during their evolution , and those more massive than about probably never enter the wc phase ; b ) the most massive ( @xmath93 at @xmath4 ) high - metallicity stars loose the majority of their mass during their ms phase , and probably end their life as he white dwarfs ; and c ) the number of wr stars over o - type stars increases with metallicity , and metal - rich regions are expected to be characterized by a high fraction of type i ( b / c / other ? ) supernovae relative to the total number of supernovae having massive star progenitors ( up to 1 out of 3 core collapse supernovae events at @xmath4 ) . alexander , d. r. , ferguson , j. w. , 1994 , apj 437 , 879 audouze j. , 1987 , observational cosmology , iau symp . 124 , eds . a. hewitt et al . , reidel , p. 89 charbonnel , c. , meynet , g. , maeder , a. , schaller , g. , schaerer , d. , 1993 , a&a sup 101 , 415 ( paper iii ) cox , j.p , giuli , r.t . , 1969 , `` principles of stellar structure '' , gordon and breach , science publishers dorman , b. , rood , r.t . , oconnell , r.w . , 1993 , apj 419 , 596 fagotto , f. , bressan , a. , bertelli , g. , chiosi , c. , 1994 , a&a sup 105 , 39 iglesias , c. a. , rogers , f. j. , 1996 , apj 464 , 943 iglesias , c. a. , rogers , f. j. , wilson , b. g. , 1992 , apj 397 , 717 de jager , c. , nieuwenhuijzen , h. , van der hucht k. a. , 1988 , a&as 72 , 259 korista , k. , hamann , f. , ferguson , j. , ferland , g. , 1996 , apj 461 , 641 kudritzki , r.p . , pauldrach , a. , puls , j. , abbott , d. c. , 1989 , a&a 219 , 205 kurucz , r.l . , 1991 , in stellar atmospheres : beyond classical models , nato asi series c , vol . 341 langer n. , 1989 , a&a 220 , 135 langer n. , maeder , a. , 1995 , apj 450 , l320 maeder a. , 1992 , a&a 264 , 105 maeder a. , 1998 , in ` abundance profiles : diagnostic tools for galaxy history ' , eds . d. friedli , m.g . edmunds , c. robert , l. drissen , asp conf . , in press maeder a. , 1998 , in ` abundance profiles : diagnostic tool for galactic history ' , eds . d. friedly , m.g . edmunds , c. robert , l. drissen , asp conf . , in press maeder a. , conti p. , 1994 , ara&a 32 , 227 maeder a. , lequeux j. , 1982 , a&a 114 , 409 maeder a. , meynet g. , 1994 , a&a 287 , 803 meynet , g. , maeder , a. , schaller , g. , schaerer , d. , charbonnel , c. : 1994 , a&a sup 103 , 97 ( paper v ) meynet , g. , mowlavi , n. , schaerer , d. , pindao , m. : 1998 , iau symp . 184 , in press mowlavi , n. , 1995 , ph.d . thesis ( unpublished ) mowlavi , n. , schaerer , d. , meynet , g. , bernasconi , p. a. , charbonnel , c. , maeder , a. , in press ( paper vii ) peimbert m. , 1995 , in the light element abundances , ed . p. crane , springer - verlag , p. 165 reimers , d. , 1975 , mem . lige 6th ser . 8 , 369 rogers , f. j. , iglesias , c. a. , 1992 , apj sup . 79 , 507 schaerer , d. , charbonnel , c. , meynet , g. , maeder , a. , schaller , g. , 1993 , a&a sup 102 , 339 ( paper iv ) schaerer , d. , meynet , g. , maeder , a. , schaller , g. , 1993 , a&a sup 98 , 523 ( paper ii ) schaller , g. , schaerer , d. , meynet , g. , maeder , a. , 1992 , a&a sup 96 , 269 ( paper i ) simpson , j. p. , colgan , s. w. , rubin , r. h. , erickson , e. f. , haas , m. r. , 1995 , apj 444 , 721 mcwilliam , a. , rich , r. m. , 1994 , apjs 91 , 749 the luminosity and effective temperature on the zero age main sequence ( zams ) are given by the homology relations ( see , for example , cox and giuli 1969 , p 696 ) @xmath94 and @xmath95 where @xmath28 and @xmath70 are the mean molecular weight and the stellar mass , respectively , @xmath96 the temperature and density independent coefficient in the relation for the nuclear energy production @xmath97 ( see cox and giuli 1969 , p 692 ) , and @xmath98 the opacity coefficient in the kramers law @xmath99 . the numbers on the first line of the superscripts in eqs . [ eq : l ] and [ eq : teff ] apply when the nuclear energy production results from the @xmath100 chain , while the second line applies when the @xmath101 cycles provide the main source of nuclear energy . the variation of @xmath108 as a factor of @xmath102 is more difficult to evaluate . we can get an estimation of @xmath109 by comparing the opacities given by the rogers and iglesias ( 1992 ) tables at different metallicities . the opacities at @xmath5 , 0.06 and 0.1 ( and at @xmath110 given by eqs . [ eq : yz ] and [ eq : xyz ] ) are shown in fig . [ fig : kappa]_(b ) _ for temperature and density conditions representative of the interior of a , @xmath4 zams model [ fig . [ fig : kappa]_(a ) _ ; we assume , as a first approximation , that the same @xmath111 profile applies at those three metallicities ] . if the opacity profile were following a kramers law , then the derivative @xmath112 would be constant and equal to that of the kramers opacity coefficient @xmath113 ( @xmath114 and @xmath37 being held constant ) . inspection of fig . [ fig : kappa]_(c ) _ shows that this is obviously not the case . the derivative is seen to vary from 4 to 15 at @xmath5 and from 1.5 to 3.3 at @xmath115 . however in this order of magnitude approach , we consider mean values equal @xmath116 the variation of @xmath117 as a function of heavy element abundances @xmath102 can be estimated knowing that the cn cycle is the main energy producer in the star ( see e.g. mowlavi 1995 , fig . a.3 ) . the two reactions and contribute for more than 70% of the total nuclear energy at the center of the star . we can thus write that @xmath118 , which leads to , using eqs . [ eq : yz ] and [ eq : xyz ] , @xmath119 equations [ eq : deltal0 ] and [ eq : deltateff0 ] become with eqs . [ eq : deltamu ] , [ eq : deltaz02 ] , [ eq : deltaz10 ] and [ eq : deltaepsilon ] , and taking @xmath120 , @xmath121 for @xmath5 and @xmath122 , @xmath123 for @xmath4 , @xmath124 @xmath125 we note from eq . [ eq : deltalz ] the small effect of the nuclear production rate on the surface luminosity . these relations should not be considered very accurate given the simplifications made in their derivation ( mainly because of the departure of the opacity profile from the kramers law and the resulting uncertainty in eqs . [ eq : deltaz02 ] and [ eq : deltaz10 ] ) . they however enable to understand qualitatively the zams location of a star in the hr diagram as a function of the metallicity , and support the following conclusions found in fig . [ fig : hr ] : - the surface luminosity at zams is a decreasing function of @xmath0 at @xmath5 ( and lower metallicities ) and an increasing function of @xmath0 at @xmath4 . the metallicity at which the inversion operates can only be determined from complete stellar model calculations and is found to be @xmath63 . - the absolute shift @xmath126 resulting from an increase @xmath127 is much greater at @xmath4 than at 0.02 ( by a factor of about 14 according to eq . [ eq : deltalz ] ) . this result is qualitatively well verified by the zams model calculations displayed in fig . [ fig : hr ] .
an analysis of some properties of stellar models as a function of metallicity @xmath0 ( and helium content @xmath1 ) is presented , with special attention to those stars with metallicities higher than twice or three times solar . it is shown that the stellar properties as a function of @xmath0 are mainly determined by the effects of the opacities at sub - solar metallicities , and by the effects of the mean molecular weight and stellar mass loss at higher metallicities . as a result , very metal - rich stars ( @xmath2 ) exhibit properties that deviate from what is expected from the known characteristics at lower metallicities . in particular , they are more luminous and hotter than those at @xmath3 due to the effect of the mean molecular weight . they have main sequence lifetimes much shorter ( 60% shorter at @xmath4 than at @xmath5 ) than those at solar metallicity due to their lower initial hydrogen content . finally , the high mass loss rates at high metallicities affect significantly the population synthesis of massive stars in very metal - rich regions . an analysis of expected wolf - rayet and supernovae populations in such conditions is briefly presented . # 1@xmath6 # 1 _ # 1#2@xmath7 # 1#2 # 1#2#3#4#5#6@xmath8 # 1#2#3#4@xmath9 # 1#2#3#4@xmath10 # 1#2#3#4#5#6#7#8@xmath11 # 1#2#3#4#5#6#7#8@xmath12 # 1#2^#1_#2 # 1#2 c # 1 + # 2
emission lines arising from transitions in sulphur - like fexi have been detected in the spectra of a range of astrophysical sources , including the sun ( dere 1978 ) and other cool stars ( laming & drake 1999 ) , as well as active galaxies ( puchnarewicz et al . 1998 ) . the usefulness of these lines as electron density diagnostics for the emitting plasma was first shown by kastner & mason ( 1978 ) , who employed the atomic physics calculations of mason ( 1975 ) to generate theoretical emission - line ratios for this ion . since then , several authors have calculated atomic data and theoretical line ratios for fexi , particularly for applications to solar spectra ( see bhatia et al . 2002 and references therein ) . most work on the solar spectrum of fexi has focused on the @xmath0180200 wavelength range , containing 3s@xmath53p@xmath63s@xmath53p@xmath73d transitions ( see , for example , keenan et al . this is understandable , as this wavelength region contains numerous strong fexi lines ( including the most intense in the solar spectrum ) , many of which provide good electron density diagnostics . however , the @xmath0257407 region also contains several fexi lines , primarily arising from 3s@xmath53p@xmath63s3p@xmath8 transitions . in this paper we use the most recent atomic physics calculations for fexi to analyse solar spectra covering this wavelength region . specifically , we investigate observational data sets from two missions , namely the extreme - ultraviolet imaging spectrometer ( eis ) on board the _ hinode _ satellite , and the solar extreme - ultraviolet research telescope and spectrograph ( serts ) sounding rocket experiments . the eis long wavelength channel spectral region is 245291 ( culhane et al . 2007 ) , compared to @xmath0235449 for the serts first - order bandpass ( neupert et al . 1992 ) , and both have spectral resolutions of about 0.06 ( fwhm ) . although serts has a larger wavelength coverage , in principle allowing more lines from the same species to be detected , it has a limited dynamic range as most spectra have been recorded on photographic film . by contrast , eis spectra are obtained with ccds , and hence can reliably detect weaker lines . the combination of wavelength coverage and dynamic range provided by serts and eis allows us to undertake a detailed assessment of the fexi emission - line spectrum in the 257407wavelength region . the model ion adopted for fexi has been discussed in detail by keenan et al . briefly , it consisted of the 24 energetically lowest ls states , yielding a total of 48 fine - structure levels . experimental energy levels , where available , were obtained from shirai et al . ( 1990 ) and jupn et al . ( 1993 ) , with the theoretical results of aggarwal & keenan ( 2003a ) being used for the remainder . electron impact rates were taken from aggarwal & keenan ( 2003b ) , while for einstein a - coefficients the data of aggarwal & keenan ( 2003a ) were employed . finally , under solar plasma conditions proton impact excitation is only important for transitions within the 3s@xmath53p@xmath6 @xmath7p ground term , and we have used the calculations of landman ( 1980 ) for this atomic process . using the model ion outlined above , in conjunction with an updated version of the statistical equilibrium code of dufton ( 1977 ) , relative fexi emission - line strengths were calculated as a function of both electron temperature ( t@xmath1 ) and density ( n@xmath1 ) . details of the procedures involved and approximations made may be found in dufton ( 1977 ) and dufton et al . given errors in the adopted atomic data of typically @xmath910 per cent ( see the references above ) , we would expect the theoretical ratios to be accurate to better than @xmath920 per cent . the solar spectra analysed in the present paper are from two instruments , namely the eis on board the _ hinode _ satellite , and several serts rocket flights . our eis data sets are those of an active region and limb area originally presented and discussed by brown et al . ( 2008 ) , to which the reader is referred for details of the eis instrument and the data reduction and flux calibration procedures . these two features , denoted ar1 and limb by brown et al . , show the strongest fexi emission lines of the solar regions considered by these authors . the spectra are from the eis long wavelength channel and cover 245291 at a resolution of about 55m ( fwhm ) . our serts spectra are those of several quiet and active regions , plus an off - limb area , obtained during rocket flights in 1989 ( denoted serts89 ) , 1991 ( serts91 ) , 1993 ( serts93 ) and 1997 ( serts97 ) . details of these observations , including their reduction and flux calibration , may be found in thomas & neupert ( 1994 ) for serts89 , brosius et al . ( 1996 ) for serts91 and serts93 , and brosius et al . ( 2000 ) for serts97 . most of these data sets cover the 235 - 449 ( serts89 ) or 231445 ( serts91 and serts93 ) wavelength regions in first - order , at resolutions of between 5080m , with spectra recorded on photographic film . however , the intensity calibrations for the serts91 and serts93 spectra are highly uncertain for wavelengths shorter than 274 . by contrast , the serts97 spectrum spans 299353 in first - order , at a resolution of 115m , and was recorded on a ccd . we note that the version of serts flown in 1989 carried a standard gold - coated toroidal diffraction grating , while those in 1991 , 1993 and 1997 incorporated a multilayered - coated grating which enhanced the instrument sensitivity in the first - order wavelength range , by factors of up to 9 . in addition , the absolute intensity calibration of the serts89 spectrum has been re - evaluated since the publication of thomas & neupert , as a result of which emission - line intensities are uniformly a factor of 1.24 larger than those quoted in that paper ( see , for example , keenan et al . 2008 and references therein ) . we have searched for fexi emission lines in the eis and serts spectra using the detections of brown et al . ( 2008 ) and thomas & neupert ( 1994 ) , supplemented with those from other sources , including the national institute of standards and technology ( nist ) data base , the latest version ( v6.0 ) of the chianti data base ( dere et al . 1997 , 2009 ) , and the atomic line list of van hoof.peter/atomic/ ] in table 1 we list the fexi transitions identified in the data sets , along with their measured wavelengths . we also provide a note on each line , including whether it is detected in both eis and serts spectra , and if there are possible blending features . [ cols="^,<,<",options="header " , ] @xmath10determined from present line ratio calculations at t@xmath1 = 10@xmath11k ; n@xmath1 in @xmath4 . + @xmath12factor by which the theoretical line ratio varies between n@xmath1 = 10@xmath2 and 10@xmath3@xmath4 . an inspection of table 13 reveals that the densities derived from 308.54/352.67 are generally in very good agreement with those estimated for the serts solar features from line ratios in species formed at similar temperatures to fexi , which as noted in section 4.2 average about log n@xmath1 = 9.4 @xmath9 0.3 for active regions and log n@xmath1 = 9.1 @xmath9 0.3 for quiet sun and off - limb areas . hence we can state that , in general , the 308.54/352.67 ratio appears to provide a good n@xmath1diagnostic . the exception is the serts89 active region , where the derived density is more than an order of magnitude larger than expected . however , as already discussed in section 4.2 , there are problems with the 308.54 feature in the serts89 active region observations which make its measurement uncertain . the electron density derived for the serts97 active region from the 349.04/352.67 ratio is in good agreement with those found from other species , but in the case of the serts89 active region the density is much higher than expected . however , as noted by thomas & neupert ( 1994 ) , the 349.04 line in the serts89 observations is blended with the second - order fex 174.52 feature . we can estimate its contribution using the nearby fex 177.24 transition , which appears in first - order at 354.48 . the first - order intensity of this line is 3.6 erg@xmath13s@xmath14sr@xmath14 , and the 174.52 feature is predicted to be a factor of 1.8 stronger ( keenan et al . 2008 ) , yielding an intensity of 6.5 erg@xmath13s@xmath14sr@xmath14 . this in turn gives a corrected intensity for fexi 349.04 of 19.7 erg@xmath13s@xmath14sr@xmath14 , and revised experimental 349.04/352.67 ratio of 0.12 . from figure 5 , this implies n@xmath1 = 10@xmath15@xmath4 , in good agreement with other diagnostics . although the 308.54/352.67 ratio is probably the most useful fexi density diagnostic currently available in the 257407 wavelength range , 349.04/352.67 is superior in many respects . the lines are closer together , reducing the effects of possible errors in the instrument intensity calibration , and the ratio varies by a larger factor over the density range n@xmath1 = 10@xmath210@xmath3@xmath4 ( 14.7 c.f . 8.4 for 308.54/352.67 ) . however , the 349.04 feature is relatively weak , and is also blended with fex 174.52 , unless the first - order instrument response is enhanced , as for the serts97 observations . dbj and kma acknowledge financial support from stfc and epsrc , respectively . rom acknowledges support from the nasa postdoctoral program at the goddard space flight center , administered by oak ridge associated universities through a contract with nasa . _ hinode _ is a japanese mission developed and launched by isas / jaxa , collaborating with naoj as a domestic partner , nasa and stfc as international partners . scientific operation of the _ hinode _ mission is conducted by the _ hinode _ science team organised at isas / jaxa . support for the postlaunch operation is provided by jaxa and naoj , stfc , nasa , esa and nsc ( norway ) . the serts rocket programme is supported by rtop grants from the solar physics office of nasa s space physics division . jwb acknowledges additional nasa support under grant nag513321 . fpk is grateful to awe aldermaston for the award of a william penney fellowship . the authors thank peter van hoof for the use of his atomic line list . chianti is a collaborative project involving the naval research laboratory ( usa ) , rutherford appleton laboratory ( uk ) , and the universities of florence ( italy ) and cambridge ( uk ) . we are very grateful to the referee , peter young , for his comments on an earlier version of the paper , in particular regarding the analysis of the eis spectra .
theoretical emission - line ratios involving fexi transitions in the 257407 wavelength range are derived using fully relativistic calculations of radiative rates and electron impact excitation cross sections . these are subsequently compared with both long wavelength channel extreme - ultraviolet imaging spectrometer ( eis ) spectra from the _ hinode _ satellite ( covering 245291 ) , and first - order observations ( @xmath0235449 ) obtained by the solar extreme - ultraviolet research telescope and spectrograph ( serts ) . the 266.39 , 266.60 and 276.36 lines of fexi are detected in two eis spectra , confirming earlier identifications of these features , and 276.36 is found to provide an electron density ( n@xmath1 ) diagnostic when ratioed against the 257.55 transition . agreement between theory and observation is found to be generally good for the serts data sets , with discrepancies normally being due to known line blends , while the 257.55 feature is detected for the first time in serts spectra . the most useful fexi electron density diagnostic is found to be the 308.54/352.67 intensity ratio , which varies by a factor of 8.4 between n@xmath1 = 10@xmath2 and 10@xmath3 @xmath4 , while showing little temperature sensitivity . however , the 349.04/352.67 ratio potentially provides a superior diagnostic , as it involves lines which are closer in wavelength , and varies by a factor of 14.7 between n@xmath1 = 10@xmath2 and 10@xmath3 @xmath4 . unfortunately , the 349.04 line is relatively weak , and also blended with the second - order fex 174.52 feature , unless the first - order instrument response is enhanced . [ firstpage ] atomic data sun : activity sun : corona sun : ultraviolet .
the pamela ( a payload for antimatter matter exploration and light - nuclei astrophysics ) experiment is a satellite - borne apparatus designed to study charged particles in the cosmic radiation with a particular focus on antiparticles ( antiprotons and positrons ) . pamela is installed inside a pressurized container attached to a russian resurs dk1 earth - observation satellite that was launched into space by a soyuz - u rocket on june 15@xmath0 2006 from the baikonur cosmodrome in kazakhstan . the satellite orbit is elliptical and semi - polar , with an altitude varying between 350 km and 600 km , at an inclination of 70@xmath1 . the mission is foreseen to last for at least three years . the pamela mission is devoted to the investigation of dark matter , the baryon asymmetry in the universe , cosmic ray generation and propagation in our galaxy and the solar system , and studies of solar modulation and the interaction of cosmic rays with the earth s magnetosphere . the primary scientific goal is the study of the antimatter component of the cosmic radiation , * in order to search for evidence of dark matter particle ( e.g. non - hadronic particles outside the standard model ) annihilations by precisely measuring the antiparticle ( antiproton and positron ) energy spectra ; * in order to search for antinuclei ( in particular , anti - helium ) ; * in order to test cosmic - ray propagation models through precise measurements of the antiparticle energy spectrum and precision studies of light nuclei and their isotopes . concomitant goals include , * a study of solar physics and solar modulation during the 24@xmath0 solar minimum by investigating low energy particles in the cosmic radiation ; * reconstructing the cosmic ray electron energy spectrum up to several tev thereby allowing a possible contribution from local sources to be studied . table [ t : science ] shows the design goals for pamela performance . the various cosmic - ray components and energy ranges over which pamela will provide new results are presented . .[t : science ] design goals for pamela performance . [ cols="^,^",options="header " , ] pamela is built around a 0.43 t permanent magnet spectrometer equipped with 6 planes of double - sided silicon detectors allowing the sign , absolute value of charge and momentum of traversing charged particles to be determined . the acceptance of the spectrometer ( which also defines the overall acceptance of the pamela experiment ) is 21.5 @xmath2sr and the maximum detectable rigidity is @xmath31 tv . spillover effects limit the upper detectable antiparticle momentum to @xmath3190 gev / c ( @xmath3270 gev / c ) for antiprotons ( positrons ) . the spectrometer is surrounded by a plastic scintillator veto shield . an electromagnetic calorimeter mounted below the spectrometer measures the energy of incident electrons and allows topological discrimination between electromagnetic and hadronic showers ( or non - interacting particles ) . planes of plastic scintillator mounted above and below the spectrometer form a time - of - flight system which also provides the primary experimental trigger . the timing resolution of the time - of - flight system allows albedo particles to be identified and proton - electron separation is also possible below @xmath31 gev / c . ionising energy loss measurements in the time - of - flight scintillator planes and the silicon planes of the magnetic spectrometer allow the absolute charge of traversing particles to be determined . the volume between the upper two time - of - flight planes is bounded by an additional plastic scintillator anticoincidence system . a plastic scintillator system mounted beneath the calorimeter aids in the identification of high energy electrons and is followed by a neutron detection system for the selection of very high energy electrons ( up to 2 tev ) which shower in the calorimeter but do not necessarily pass through the spectrometer . the pamela subdetectors are read out and controlled by a data acquisition system based around actel ( 54sx series ) field programmable gate arrays ( fpga ) @xcite and analog devices ( adsp-2187l ) digital signal processors ( dsp ) @xcite . connections between different systems are realised with redundant data - strobe @xcite low voltage differential signaling ( lvds ) links . each subdetector is also connected to a global trigger system and can issue alarm conditions ( e.g. over - temperature , data corruption ) to a housekeeping system . all the data acquisition boards ( except for the calorimeter ) are housed in a custom crate secured to the pamela superstructure , as shown in figure [ pam ] ( bottom ) . in order to promote reliability , common design rules have been followed for all electronics systems in pamela , e.g. over - current protection on all electronics boards , redundant data links , redundant power connections and the use of radiation qualified components . the tof system @xcite comprises 6 layers of fast plastic scintillators ( bicron bc-404 @xcite ) arranged in three planes ( s1 , s2 and s3 ) , with alternate layers placed orthogonal to each other , as shown in figure [ fig : tof ] . the distance between s1 and s3 is 77.3 cm . time - of - flight information for charged particles passing between planes s1 and s3 is combined with track length information derived from the magnetic spectrometer ( see section [ sec : spectrometer ] ) to determine particle velocities and reject albedo particles . ionisation ( de / dx ) measurements in the scintillator layers allow the particle charge to be determined at least up to z@xmath48 . coincidental energy deposits in combinations of planes provide the main trigger for the experiment , as described in section [ sec : trigger ] . the segmentation of each plane allows redundant studies of the trigger efficiency . the sensitive area of each of the two s1 layers is ( [email protected] ) @xmath6 with the first layer divided into 8 bars and the second layer divided into 6 bars . the total sensitive area of the s2 and s3 planes is ( 15@xmath518 ) @xmath6 segmented into 2@xmath52 and 3@xmath53 orthogonal bars , respectively . the s1 and s3 layers are 7 mm thick while the s2 layers are 5 mm thick . there are 24 scintillator bars in total . both ends of each scintillator bar are glued to a plastic light guide which is mechanically coupled to a hamamatsu r5900u photomultiplier ( pmt ) by means of silicone pads of thickness 3 mm ( s1 and s2 ) and 6 mm ( s3 ) . the differences in thickness reflect the different vibrational spectra expected during launch . the scintillators and light - guides are wrapped in 2 layers of 25 @xmath7 m thick mylar foil . the s3 plane is mounted directly on the base plate of pamela , while the other two planes are enclosed in light - proof boxes suspended off the pamela structure . a high - voltage divider circuit is mounted directly behind each pmt . the high - voltage and discrimination threshold for each pmt is chosen to optimize the performance of a given tof bar . the tof electronics system converts the 48 pmt pulses into time- and charge - based measurements . in the timing section , a capacitor is linearly charged during a time interval defined by the passage of a particle through the tof system . in the charge section , a capacitor is charged with the pmt pulse charge . in both cases , during read out the capacitor is linearly discharged into a time - to - digital converter . the tof electronics system comprises a nine board electronics system based around the pamela - standard fpgas and dsps . a separate trigger board processes signals @xcite from the 48 pmts as well as trigger signals from the calorimeter and bottom scintillator ( see section [ sec : trigger ] ) . rate counters , dead-/live - time counters and the logic to generate calibration pulse sequences for different subsystems are also implemented . control masks select trigger types ( see section 3.2 ) and allow noisy or dead pmt channels to be vetoed and the pmt hit pattern to be recorded for each trigger . figure [ tof1 ] shows the velocity of particles ( in units of speed of light , @xmath8 ) measured by the tof system as a function of their rigidity for data recorded at ground . most of the events are relativistic muons . a small proton component is visible at low rigidity ( the solid line indicates the theoretical @xmath8 for protons ) . the measured time - of - flight resolution of @xmath3250 ps will allow electrons ( positrons ) to be separated from antiprotons ( protons ) up to @xmath31 gev / c . albedo particles can also be rejected with a significance of 60 standard deviations . in addition , the measurement of ionization losses in the tof scintillators will allow the determination of the absolute charge of the particles , as shown in figure [ gsi ] . these data were collected during a beam test performed at the gsi facility in darmstadt . prototype versions of the s1 , s2 and s3 tof paddles were exposed to @xmath9c beams . targets of aluminium and polyethylene were used to generate a variety of fragmentation products . during this test , the s1 and s2 layers were used to clean the data sample , and the particle charge was subsequently measured using the s3 layer . data taken during this test also allowed the timing resolution for carbon to be determined as 70 ps . this improvement is reasonable ( compared to the 250 ps quoted above ) , since the timing resolution improves with the inverse square root of the number of photons created in the scintillator . simulations have shown that the majority ( @xmath3 75% ) of triggers in orbit are false " triggers @xcite , i.e. where the coincidental energy deposits in the time of flight scintillators are generated by secondary particles produced in the mechanical structure of the experiment , as shown in figure [ ints ] . the aim of the anticoincidence systems is to identify these events during offline data analysis , or through the use of a second - level trigger in - orbit ( see section [ sec : trigger ] ) . the pamela experiment contains two anticoincidence ( ac ) systems @xcite . the primary ac system @xcite consists of 4 plastic scintillators ( cas ) surrounding the sides of the magnet and one covering the top ( cat ) , as shown in figure [ fig : ac ] . a secondary ac system consists of 4 plastic scintillators ( card ) that surrounds the volume between the first two time - of - flight planes . the card detectors are scaled - down versions of cas . the ac systems use 8 mm thick plastic scintillators ( bicron bc-448 m @xcite ) read out by hamamatsu @xcite r5900u pmts . each scintillator is covered in two layers of reflective tyvek @xcite material and coupled via a 7 mm thick silicone pad to the pmts . each cas and card detector is read out by two identical pmts in order to decrease the possibility of single point failure . also for this reason , and to cover the irregularly shaped area , the cat detector is read out by 8 pmts . a high - voltage divider is mounted directly behind each pmt and operated at a fixed voltage of -800 v. the scintillators and pmts are housed in aluminium containers which provide light - tightness , allow fixation to the pamela superstructure and ensure that a reliable scintillator - pmt coupling is maintained . the small fringe field from the magnetic spectrometer at the position of the pmts means that additional magnetic shielding is not required . the signals from the 24 pmts are divided between two independent data acquisition boards with signals from pmts for a given cas or card detector or cat quadrant routed to different boards . only binary hit information is stored from each pmt indicating whether the deposited energy exceeds 0.5 mip ( where 1 mip is the most probable energy deposited by a normally incident minimum ionising particle ) . on each board , an analogue front - end electronics system comprising an integration / amplification and discrimination stage processes the pmt signals before they are fed into a fpga . the core of this digital system is a 16 bit shift register allowing hit information to be recorded in a time window of length 1.28 @xmath7s centered on the trigger time . within this window the hit can be located with an accuracy of 80 ns . the fpga also allows the pmt singles rates to be monitored and controls the data acquisition system . a dsp controls a monitoring system which is based around 640 nm miniature leds glued directly to the scintillator material . the efficiency of the large area cas detectors has been studied using an external drift chamber to map the spatial distribution of incident cosmic ray muons . a detection efficiency for mips of ( [email protected])% was observed @xcite . the ac system has also been tested by studying the backscattering of particles ( see figure [ ints ] ) from the calorimeter during tests with high energy particle beams @xcite . the robustness of the ac system has been determined by studying the stability of the scintillator - pmt coupling to variations in temperature @xcite and the vibration spectra expected during launch @xcite . the central part of the pamela apparatus is a magnetic spectrometer @xcite consisting of a permanent magnet and a silicon tracker . the magnetic spectrometer is used to determine the sign of charge and the rigidity of particles up to @xmath31 tv / c . ionisation loss measurements are also made in the silicon planes , allowing absolute particle charge to be determined up to at least z=6 . the magnet is composed of five modules forming a tower 44.5 cm high . each module comprises twelve magnetic blocks , made of a nd - fe - b alloy with a residual magnetisation of 1.3 t. the blocks are configured to provide an almost uniform magnetic field oriented along the y - direction inside a cavity of dimensions ( [email protected] ) @xmath2 . the dimensions of the permanent magnet define the geometrical factor of the pamela experiment to be 21.5 @xmath2sr . to allow precise rigidity measurements to be obtained from the reconstructed particle trajectory , the magnetic field has been precisely measured with a hall probe through - out the cavity volume and the surrounding regions . figure [ ybz0 ] shows the y - component of the magnetic field measured in the z=0 plane as a function of x and y and the y - component as measured along the z - axis . the mean magnetic field inside the cavity is 0.43 t with a value of 0.48 t measured at the centre . any stray magnetic field outside of the cavity can potentially interfere with the satellite instruments and navigation systems . in order to attenuate the stray field , the magnet is enclosed by ferromagnetic shielding . six equidistant 300 @xmath7 m thick silicon detector planes are inserted inside the magnetic cavity . the double - sided silicon sensors provide two independent impact coordinates on each plane . the basic detecting unit is the ladder which comprises two sensors , ( [email protected] ) @xmath2 , assembled with a front - end hybrid circuit , as shown in figure [ hybrid ] . each plane is built from three ladders that are inserted inside an aluminium frame which connects to the magnet canister . in order to limit multiple scattering in dead layers , no additional supporting structure is present above or below the planes . each high resistivity n - type silicon detector is segmented into micro - strips on both sides with p@xmath11 strips implanted on the junction side ( bending- , x - view ) and n@xmath11 strips on the ohmic side ( non - bending , y - view ) . in the x - view , the implantation pitch is 25 @xmath7 m and the read - out pitch is 50 @xmath7 m . in the y - view , the read - out pitch is 67 @xmath7 m with the strips orthogonal to those in the x - view . the mip efficiency for a single plane ( including dead regions ) exceeds 90% . the front - end electronics system is based around va1 application specific integrated circuits ( asics ) @xcite which contain 128 charge sensitive preamplifiers connected to shapers and a sample and hold circuit . the signals from the va1 chips are sent over 5 cm long kapton cables to be digitised by analog - to - digital ( adc ) boards mounted on the magnet canisters . the digitised data are transferred by serial links to dsp - based read - out boards where they are compressed using a zero order predictor ( zop ) algorithm . the compression factor is estimated at 95% . the main task of the spectrometer is to measure the rigidity , @xmath12 , of charged particles . the momentum of the particle and the sign of its electric charge can then be derived from the relation @xmath13 , where @xmath14 is the electron charge , @xmath15 the momentum , @xmath16 the speed of light and @xmath17 is the absolute charge . as discussed in section [ tof ] , the tof system can be used to identify physical albedo activity . instrumental albedo , e.g. due to particles backscattered from the calorimeter , can be identified a combination of the tof system , the anticoincidence systems , and the tracking capabilities of the calorimeter ( see section [ calorimeter ] ) . the magnetic spectrometer measures the deflection of a particle , which is defined as the inverse of the rigidity . the resolution in the deflection measurement depends on the geometrical configuration of the spectrometer , on the intensity of the magnetic field and on the spatial resolution of the position measuring system - the silicon sensors in this case . this spatial resolution depends on the particle incidence angle . for normally incident tracks , tests with particle beams show a spatial resolution of ( [email protected] ) @xmath7 m and ( [email protected] ) @xmath7 m in the bending and non - bending views , respectively . the spatial resolution in the bending view is shown in figure [ res ] ( left ) . figure [ res ] ( right ) shows the resulting deflection error as a function of rigidity obtained with proton beams . from this plot a maximum detectable rigidity ( mdr ) of @xmath18 tv can be inferred . note that this exceeds the design goal presented in table [ t : science ] . in flight , the deflection measurement of the tracking system will be cross - checked with the energy measurement of the calorimeter for high - energy electrons . protons and electrons dominate the positively and negatively charged components of the cosmic radiation , respectively . the main task of the calorimeter is to select positrons and antiprotons from like - charged backgrounds which are significantly more abundant . positrons must be identified from a background of protons that increases from about 10@xmath19 times the positron component at 1 gev / c to @xmath35@xmath510@xmath19 at 10 gev / c , and antiprotons from a background of electrons that decreases from @xmath35@xmath510@xmath19 times the antiproton component at 1 gev / c to less than 10@xmath20 times above 10 gev / c . this means that the pamela system must separate electrons from hadrons at a level of 10@xmath21 - 10@xmath22 . much of this separation must be provided by the calorimeter , i.e. electrons must be selected with an acceptable efficiency and with as small a hadron contamination as possible . the sampling electromagnetic calorimeter comprises 44 single - sided silicon sensor planes ( 380 @xmath7 m thick ) interleaved with 22 plates of tungsten absorber @xcite . each tungsten layer has a thickness of 0.26 cm , which corresponds to 0.74 x@xmath23 ( radiation lengths ) , giving a total depth of 16.3 x@xmath23 ( @xmath30.6 nuclear interaction lengths ) . each tungsten plate is sandwiched between two printed circuit boards upon which the silicon detectors , front - end electronics and adcs are mounted . the ( 8@xmath58 ) @xmath2 silicon detectors are segmented into 32 read - out strips with a pitch of 2.4 mm . the silicon detectors are arranged in a 3@xmath53 matrix and each of the 32 strips is bonded to the corresponding strip on the other two detectors in the same row ( or column ) , thereby forming 24 cm long read - out strips . the orientation of the strips of two consecutive layers is orthogonal and therefore provides two - dimensional spatial information ( views " ) . figure [ calo ] shows the calorimeter prior to integration with the other pamela detectors . the calorimeter front - end electronics is based around the cr1.4p asic @xcite which provides 16 channels containing a charge - sensitive preamplifier , a cr - rc shaper , a track - and - hold circuit and an output multiplexer . a charge - injection calibration system is also implemented . six cr1.4p chips are used per plane with the outputs multiplexed into a single 16-bit adc . data from all 44 adcs are processed by 4 dsp - based read - out boards mounted within the calorimeter housing before being sent over serial links to the main pamela data acquisition system . the read - out is divided into 4 independent sections , corresponding to the x - even , y - even , x - odd and y - odd planes . the longitudinal and transverse segmentation of the calorimeter , combined with the measurement of the particle energy loss in each silicon strip , allows a high identification ( or rejection ) power for electromagnetic showers . electromagnetic and hadronic showers differ in their spatial development and energy distribution in a way that can be distinguished by the calorimeter . this is demonstrated in figure [ topo ] which shows examples of an electromagnetic shower induced by an electron ( left ) and an interacting proton ( right ) , recorded during tests with particle beams at the cern sps facility . all incident particles have a momentum of 50 gev / c . the electron - hadron separation performance of the calorimeter has been extensively studied @xcite and the calorimeter is found to have sufficient performance to reach the primary scientific objectives of pamela , providing a proton rejection factor of about 10@xmath21 while keeping about 90% efficiency in selecting electrons and positrons . from simulations , an electron rejection factor of about 10@xmath21 in antiproton measurements ( about 90% antiproton identification efficiency ) is demonstrated . the calorimeter will also be used to reconstruct the energy of the electromagnetic showers . this will provide a measurement of the energy of the incident electrons independent from the magnetic spectrometer , thus allowing a cross - calibration of the two energy determinations . as shown in figure [ eres ] , the constant term for the calorimeter energy resolution has been measured as @xmath24 for electromagnetic showers generated by particles entering the calorimeter within the acceptance of the tracking system up to an energy of several hundred gev . the calorimeter is also equipped with a self - trigger capability , as discussed in section [ sec : trigger ] . the shower tail catcher scintillator ( s4 ) improves the pamela electron - hadron separation performance by measuring shower leakage from the calorimeter . it also provides a high - energy trigger for the neutron detector ( described in the next section ) . this scintillator is placed directly beneath the calorimeter . it consists of a single square piece of 1 cm thick scintillator of dimensions ( 48@xmath548 ) @xmath2 which is read out by six pmts , as shown in figure [ nd_s4 ] . the neutron detector complements the electron - proton discrimination capabilities of the calorimeter . the evaporated neutron yield in a hadronic shower is 1020 times larger than expected from an electromagnetic shower . the neutron detector is sensitive to evaporated neutrons which are thermalised in the calorimeter . joint analysis of the calorimeter and neutron detector information are expected to allow allow primary electron energies to be determined up to several tev . the neutron detector @xcite is located below the s4 scintillator and consists of 36 proportional counters , filled with @xmath19he and surrounded by a polyethylene moderator enveloped in a thin cadmium layer to prevent thermal neutrons entering the detector from the sides and from below . the counters are stacked in two planes of 18 counters , oriented along the y - axis of the instrument . the size of the neutron detector is ( @xmath25 ) cm@xmath19 and is shown in figure [ nd_s4 ] . a schematic overview of the pamela data acquisition ( daq ) system is shown in figure [ daq ] . the pscu ( pamela storage and control unit ) handles all slow controls , communication with the satellite , data acquisition , storage and downlink tasks . the pscu contains 4 subsystems : * a processor module built around a cpu based on a erc-32 architecture ( sparc v7 implementation ) running the rtems real time operating system at 24 mhz . the cpu is custom built by laben and is fully space qualified . there is no redundant back - up . communication with the resurs satellite is realised via a standard 1553b data - bus ; * two redundant 2 gbyte mass memory modules . the modules include latch - up detection , allowing operation to be transparently switched to the safe module when a latch - up is detected ; * a pif ( pamela interface board ) that performs three main tasks : communication with the idaq ( intermediate daq ) system through a dma ( dynamic memory access ) controller , handling the interface with the mass memory , and providing the interface with the vrl ( very high - speed radio link ) module of the satellite ; * a tmtc ( telemetry and control ) board that handles the housekeeping operations of pamela , such as alarm , temperature and voltage monitoring ( once per second ) . such monitoring is performed both directly ( adc inputs and contact closure telemetries ) and through a dedicated housekeeping board that communicates through serial data links with the subdetector read - out boards , with the idaq board and with the power supply control boards . data acquisition from the subdetectors is managed by the idaq system at a rate of 2 mbyte / s . upon receipt of a trigger , the pscu initiates the idaq procedure to read out data from the subdetectors in sequence . the resulting data are stored in the pscu mass memory . several times a day , the data are transferred to the satellite on - board memory via the 12 mbyte / s vrl bus where it is stored prior to downlinking to earth . approximately 15 gbytes are transferred to ground per day during 2 - 3 downlink sessions . the pscu automatically handles the flow of pamela physics tasks and continuously checks for proper operation of the apparatus . at boot , the pscu manages the operation of the power supply system to power up all subsystems , initializes all detectors and starts the data acquisition cycle . in parallel , once per second the pscu checks the tmtc information on voltages and alarms . in case of abnormal conditions the pscu can perform a hardware reset of the whole system or , if insufficient to solve the problem ( e.g. in case of electronics latch - up ) , powers down and then up pamela . the pscu also checks the temperature environment by reading dedicated temperature sensors distributed in various locations around the instrument . if the readings exceed predefined values ( set with dedicated commands from ground ) the pscu powers down pamela until acceptable working conditions are reached . the pscu also handles communication with the resurs satellite cpu and vrl system . data is downloaded to the vrl upon receipt of a dedicated command from the resurs cpu . the scheduling of data downloads from the pamela mass memory to the vrl hard disk system is defined from ground on a daily basis . the pscu organizes the data acquisition cycle in runs " . a run is defined as a continuous period of data taking in which the trigger and detector configurations are constant . these configurations are defined by the pscu according to information stored in on - board memory or received from ground . the duration of a run is determined by the pscu according to the orbital position ( e.g. inside radiation belts or south atlantic anomaly saa or outside these areas ) . the orbital position also dictates the trigger configuration , as described in the following section . the orbital position is derived from the ascending node " notification issued by the resurs cpu when the satellite crosses the equator from the southern hemisphere to the northern hemisphere . from this position information , the cpu extrapolates the entry time into high radiation environments . this can be performed in three ways , chosen from ground : * when the counting rate of the s1 scintillator exceeds a given threshold ( changeable from ground with dedicated command ) ; * according to fixed time periods conservatively chosen and modifiable from ground ; * according to a table with crossing times in absolute moscow time 1 s. ] provided on a bi - weekly basis from ground with a dedicated command . additionally , the pscu can interrupt and close a run if anomalous conditions that require action upon the subsystems ( e.g. hardware resets , etc . ) are detected . periodically the pscu calibrates the detectors , namely the anticounter system , the tracker , the calorimeter and the s4 scintillator . by default , the calibration is performed at the point of lowest cosmic - ray trigger rate , i.e. the equator , upon receiving an ascending node " notification from the resurs cpu . the frequency of calibrations can be modified from ground . the pamela trigger condition is defined by coincident energy deposits in the scintillator tof layers . various configurations can be selected . the default ones ( the subscripts 1 and 2 refer to the upper and lower layers in each tof plane ) used outside and inside radiation environments are : * outside radiation belts and saa ; * inside radiation belts and saa ; since , according to simulation , the radiation environment will saturate the s1 counting rate but will not affect significantly the s2 and s3 scintillators since they are more shielded . these trigger configurations can be changed from ground with dedicated commands to the pscu . a total of 29 configurations have been implemented on the trigger board . various combination of _ and _ or _ or _ of the scintillators layers with or without the calorimeter self - trigger and s4 trigger ( described below ) are implemented . the pmts can be masked on the trigger board by the pscu . the calorimeter is equipped with a self - trigger capability . a trigger signal is generated when a specific energy distribution is detected in predetermined planes within the lower half of the calorimeter . the sets of planes used in this configuration can be changed with a dedicated command from ground . this allows pamela to measure very high - energy ( @xmath3300 gev to @xmath271 tev ) electrons in the cosmic radiation . at present , very few measurements have covered this energy range @xcite . since these events are rare , it is important to have a large geometrical factor . by requiring that triggering particles enter through one of the first four planes and cross at least 10 radiation lengths , the geometrical factor is @xmath3600 @xmath6sr , i.e. about a factor of 30 larger than the default pamela acceptance defined by the magnetic spectrometer . the behaviour of the calorimeter in self - trigger mode has been studied by means of simulations @xcite . the simulated energy resolution of the calorimeter in self - trigger mode is approximately constant ( @xmath312% ) up to about 800 gev , as shown in figure [ eres ] . at higher energies the resolution decreases because of increasing longitudinal leakage and saturation of the signal from the strips ( about 1000 mip ) . the choice of energy loss and activated planes implemented in the calorimeter electronics to generate a trigger signal has been taken to have the highest proton rejection while keeping a trigger efficiency of better than 90% for electrons of energies higher than 300 gev @xcite . combined with the neutron detector information , the apparatus will be able to cleanly identify very high - energy electrons . the neutron detector can also be triggered when an energy deposit exceeding 10 mip is detected in the s4 scintillator . the trigger rate observed during typical orbits is shown in figure [ trig_rate ] . the maxima at @xmath32200 events per minute ( @xmath335 hz ) correspond to passages over the polar regions ( north pole , np and south pole , sp ) while the minima ( @xmath315 hz ) correspond to equatorial regions ( e ) . the contribution from the south atlantic anomaly ( saa ) is clearly visible ( @xmath370 hz , maximum ) . note that data is taken in the saa using the second default trigger configuration . the missing acquisition time after the peaks of the saa corresponds to the detector calibrations upon crossing the equator ( about 1 minute in duration ) . dead and live times are monitored by two clocks that count the time during which the data acquisition system is busy or is waiting for a trigger , respectively . the dead time varies significantly over an orbit , due to the significant changes in trigger rate shown in figure [ trig_rate ] . furthermore , if the satellite crosses the saa the dead time increases . the dead time also depends on the trigger configuration . for an orbit not crossing the saa the fractional dead time is approximately 26% , i.e. the fractional live time is about 74% . as discussed previously , an automatic procedure changes the trigger configuration when entering radiation environments thus reducing the trigger rate and , consequently , the dead time . complete details of in - flight performance will be detailed in a future publication . large solar particle events ( spe ) can lead to a high rate of particles hitting the top scintillator ( s1 in figure [ pam ] , top ) . although very large events such as the one observed on 24@xmath0 october 1989 could result in rates of @xmath37 mhz on s11 , most spe will occur at solar minimum ( @xmath310 spe are expected during a 3 year pamela mission ) and will be of much smaller intensity . for example , a coronal mass ejection such as that of 24@xmath0 september 1997 would result in a s1 rate of less than 100 hz , much less than that encountered in the saa ( @xmath3 1 khz ) . for large events the automatic trigger selection procedure would switch to a configuration without the s1 detector , as currently happens during passages through the saa . after this , a specific trigger configuration suited to the size and expected temporal evolution of the event can be selected from ground . if the amount of event data exceeds the storage dedicated to pamela on - board the resurs satellite or the daily downlink limit , an on - line event selection is provided by a second level trigger . the second level trigger is not normally activated and must be activated via an uplinked command from ground . information from the cas anticoincidence system is used to reject false " triggers ( see section 2.3 ) and information from the calorimeter is used to reduce the impact of particles backscattered from the calorimeter . the second level trigger is described in detail elsewhere @xcite . the resurs dk1 satellite is manufactured by the russian space company tsskb progress to perform multi - spectral remote sensing of the earth s surface and acquire high - quality images in near real - time . data delivery to ground is realised via a high - speed radio link . the satellite is presented in figure [ resurs - pc ] , has a mass of @xmath36.7 tonnes and a height of 7.4 m. the solar array span is @xmath314 m. the satellite is three - axis stabilized with an axis orientation accuracy of 0.2 arcmin and an angular velocity stabilization accuracy of 0.005@xmath28/s . the orbital altitude varies between 350 km and 600 km at an inclination of 70@xmath28 . the design lifetime is three years . pamela is mounted in a dedicated pressurized container ( pc ) attached to the resurs dk1 satellite . during launch and orbital manoeuvres , the pc is secured against the body of the satellite . during data - taking it is swung up to give pamela a clear view into space . the container is cylindrical in shape and has an inside diameter of about 105 cm , a semi - spherical bottom and a conical top . it is made of an aluminium alloy , with a thickness of 2 mm in the acceptance of pamela . figure [ pc_test ] shows tests of the pc tilting mechanism performed in may 2002 at the tsskb progress facility in samara . the movement of the pc from the parked to the data - taking position was tested in simulated weightless conditions . the ground segment of the resurs dk1 system is located at the research center for earth operative monitoring ( nts omz ) in moscow , russia @xcite . this forms part of the russian space agency ( roskosmos ) ground segment designed for acquiring , recording , processing and distributing data from remote sensing systems in space . the reception antenna at nts omz is a parabolic reflector of 7 m diameter , equipped with an azimuth - elevation rotation mechanism , and has two frequency multiplexed radio channels . the resurs dk1 radio link towards nts omz is active 2 - 3 times a day . the average volume of data transmitted during a single downlink is currently @xmath36 gbytes , giving a total of 15 gbytes / day . data received from pamela are collected by a data - set archive server . the server calculates the downlink session quality ( the error probability per bit ) and faulty downlink sessions can be assigned for retransmission up to several days after the initial downlink . the downlinked data are transmitted to a server dedicated to data processing for instrument monitoring and control , and is also written to magnetic tape for long - term storage . all such operations are automatized to minimize the time delay between the data reception and the extraction of monitoring information . after this first level of data analysis , both raw and preliminary processed data are moved through a normal internet line to the main storage centre in eastern europe , which is located at mephi ( moscow , russia ) . from here , grid infrastructure is used to move raw and first level processed data to the main storage and analysis centre of the pamela collaboration , located at cnaf ( bologna , italy ) , a specialized computing centre of infn . here data are accessible to all various institutions within the pamela collaboration . space - borne apparatus must maintain a high level of performance and stability throughout the mission duration in the harsh environment of space . the mechanical design must be such that the payload and satellite withstand the significant shocks and vibrations of the launch . the extremes of temperature that may be encountered in space require that the thermal and mechanical designs be such that the sensitive components maintain excellent stability over a broad range of temperatures . the radiation environment in space is a major consideration in the design of electronic circuitry . all chosen components must be tested for radiation tolerance prior to use . electromagnetic interference ( emi ) from electronic devices must be minimized by the use of different types of filters and shielded cables . in this section the steps taken to qualify pamela for operation in space are reviewed . in orbit all on - board electronic devices will be subject to the passage of ionizing particles , which can degrade their performance and eventually lead to their permanent damage or loss of functionality . since malfunctioning components can not be replaced once the instrument is in orbit , all critical devices must either be already space qualified , or tested for radiation tolerance before use . for economic , performance and power consumption reasons , most of the pamela electronic components are off - the - shelf " commercial products . radiation tolerance tests therefore had to be carried out before their integration into electronic boards . a selection of electronic components have been tested under gamma and heavy ion beams during the construction phase of the pamela subsystems . as an example , the dsp and fpga chips used through - out the pamela data acquisition system were extensively tested in the period 2000 - 2002 , using heavy - ion beams . the tests were performed at gsi in darmstadt ( germany ) , and jinr in dubna ( russia ) . at gsi the devices were exposed to beams of @xmath29xe and @xmath30u , in the energy range 100 - 800 mev / n . different incidence angles allowed different doses to be achieved . at jinr slow beams of @xmath31 mg at 150 mev / n were used , in order to maximize the energy transfer to the components under test . test results have been published elsewhere @xcite . it is not expected that solar particle events pose a hazard to pamela . however , in case of very large events pamela will be switched off in the time occurring between observation of the flare at the sun and the particles reaching earth . the mechanical and thermal space qualification tests of the pamela instrument were performed in the years 2002 - 2003 . in order to perform such tests , a mock - up of the entire instrument , mass - dimensional and thermal model ( mdtm ) , was manufactured . the mdtm reproduces the geometrical characteristics of pamela ( e.g. dimensions , total mass , center of gravity , inertial moments ) and the basic thermal behaviour . all particle detectors in the mdtm were simulated by dummy aluminium boxes . the electronics systems were non - functional and only reproduced the power consumption of each subsystem . in order to ensure that no damage occur to pamela or the spacecraft during any of the different operational phases of the mission ( transport , launch , orbital operations , unlocking of the pressurized container , flight ) , the mdtm was exposed to vibration spectra at mechanical loads exceeding those expected during the mission . the mdtm vibration tests were performed at iabg laboratories ( munich , germany ) in august 2002 , as shown in figure [ iabg ] . during the test it was verified that structural integrity was maintained and that there was no change in the dynamic behaviour of mdtm ( using resonance searches ) . the mdtm structure was subjected to the required vibration loads along three orthogonal axes . additional transport , vibration and shock tests of the mdtm whilst integrated into the pressurized container were performed at the tsskb - progress testing center in may 2003 . additional information about pamela mechanical space qualification can be found in @xcite . the pamela thermal cooling system consists of a 8.6 m long pipe that joins 4 radiators and 8 flanges connected throughout the pamela detector system . the task of this system is to dissipate the heat produced by the pamela subsystems and transfer it into the spacecraft , where a custom designed thermal control system is located . this transfer is performed by means of a heat - transfer fluid pumped by resurs satellite through the pamela pipelines . the total heat release of pamela can not exceed 360 w. thermal and vacuum tests of the pamela mdtm were performed in the laboratories of tsskb - progress in april 2003 . six thermal modes of operation were implemented , where the three relevant parameters which regulate the instrument thermal behaviour ( pamela power consumption , external heat flows and heat - transfer fluid temperature and flow rate ) were varied between the design extrema to simulate in - flight operations . each mode persisted until a steady state condition was reached . as an example , a test simulating an interruption in the flow of the heat - transfer fluid due to a malfunction was interrupted after 3 hours when the pamela mdtm reached a temperature of @xmath360@xmath28c . the qualification test of the pamela thermal system showed that all parameters of the system stayed within the design limits ( 5@xmath28c - 40@xmath28c ) . the test shows that during the resurs dk1 orbit the expected operating temperature range of pamela will vary between 7@xmath28c for the coldest systems and 38@xmath28c for the warmest ones , as shown in figure [ dallas ] . additional information about pamela thermal space qualification can be found in @xcite . to perform tests of the electrical interface between pamela and the spacecraft , a second mock - up of the pamela instrument was assembled . this technological model " was an exact copy of the flight model from the point of view of electrical connections to the satellite and for the readout electronics boards , with the particle detectors substituted by dummies . the technological model was shipped to tsskb - progress in april 2004 ( see figure [ techno ] ) . the task of the technological model was to thoroughly test the electrical interface to the resurs dk1 satellite . in addition , it was used to check that the residual magnetic field from the pamela spectrometer did not interfere with the resurs instrumentation . these complex tests proceeded in phases . a first test was performed in rome in december 2003 , with the satellite emulated by a ground support equipment ( egse ) system . a second test started in may 2004 at tsskb - progress and verified the powering procedures . in october 2004 the pamela technological model was fully integrated into the resurs dk1 to complete all remaining tests . between july 2000 and september 2003 , the pamela subsystems were periodically exposed to particle beams at the cern ps and sps facilities . electron and proton beams were used with energies in the 10 s - 100 s gev range . results from these tests are described in section [ sec : pamela ] . prior to delivery to russia , the pamela instrument was assembled at the infn laboratories of roma tor vergata , rome , italy . the system was tested with cosmic rays over a period of several months . figures [ muon ] and [ prot ] show two cosmic ray events recorded in rome . the first is a 1.5 gev/@xmath16 negatively charged particle , with high probability of being a @xmath32 considering the clean non - interacting pattern in the calorimeter . the second is a 67 gev/@xmath16 particle with a hadronic interaction in the calorimeter , consistent with a proton . all pamela detectors are shown in the figures along with the signals produced by the particles in the detectors and derived information . highly detailed information is provided for each cosmic - ray event . the solid lines indicate the tracks reconstructed by the fitting procedure @xcite of the magnetic spectrometer . the figures show also the ghost " hits due to the common readout of the 2 silicon sensors of the same ladder in the non - bending projection . this ambiguity is solved with the help of track fitting procedure and with a consistency check with the other pamela subdetectors . pamela was successfully launched on june 15@xmath0 2006 and was first switched - on on the 20@xmath0 of june . after a brief period of commissioning pamela has been in a continuous data - taking mode since july 11@xmath0 . data downlinked to ground show that the entire instrument is working as expected . figure [ flight1 ] shows a 3 gv non - interacting proton recorded in - orbit while figure [ flight2 ] shows a 13 gv helium nucleus interacting in the calorimeter . the in - orbit performance of pamela will be discussed in future publications . pamela is a multi - purpose satellite - borne apparatus designed to study charged particles in the cosmic radiation with a particular focus on antiparticles ( antiprotons and positrons ) . the energy range over which observations are made and the foreseen statistics mean that pamela stands to deliver results of great scientific relevance in several fields of cosmic ray research . to ensure reliable operation in space extensive space qualification tests of the pamela detector systems , electronics and mechanical structures were performed prior to launch . the performance of the individual detector components and the pamela system as a whole were also verified at particle beam facilities . the pamela instrument was launched into orbit from the baikonur cosmodrome on - board a resurs dk1 satellite on june 15@xmath0 2006 . all systems have been observed to operate as expected and scientific data analysis is now on - going . the pamela mission is sponsored by the italian national institute of nuclear physics ( infn ) , the italian space agency ( asi ) , the russian space agency ( roskosmos ) , the russian academy of science , the german space agency ( dlr ) , the swedish national space board ( snsb ) and the swedish research council ( vr ) . pamela is the result of a collaborative work that has lasted for several years . we gratefully acknowledge the contributions from the italian companies carlo gavazzi space , laben , caen , kayser , forestal and aerostudi . we also would like to thank the scientific laboratories and test facilities that assisted the pamela team during the qualification phases : cern ( geneva , switzerland ) , jinr ( dubna , russia ) , gsi ( darmstadt , germany ) , iabg ( munich , germany ) , enea - casaccia ( rome , italy ) and galileo ( florence , italy ) . we also thank the following engineers and technicians for their valuable contribution to the project : l. andreanelli , e. barbarito , a. bazarov , f. ceglie , s. ciano , c. fiorello , m. franco , a. gabbanini , e. gaspari , m. grandi , m. lundin , g. mazzenga , m. mongelli , r. mllerberg , o. panova , p. parascandolo , g. passeggio , g. pontoriere , e. reali , r. rocco , s. rydstrm , a. sedov , b. talalaev , m. tesi , e. vanzanella , and v. zotov . we thank the data reception centre , nts omz , for its first class preparation of the pamela data receiving station in moscow . finally , we would like to express our gratitude to tsskb - progress , who carefully followed the pamela mission during all its phases and made the pamela launch a reality . h. aizu et al . , phys . rev . 121 ( 1961 ) 1206 ; g.f . smoot et al . , phys . 35 ( 1975 ) 258 ; g.d . badhwar et al . , nature 274 ( 1978 ) 137 ; a. buffington et al . , astrophys . j. 248 ( 1981 ) 1179 ; r.l . golden et al . , astrophys . j. 479 ( 1997 ) 992 ; j.f . ormes et al . , astrophys . j. lett . 482 ( 1997 ) l187 ; t. saeki et al . b 422 ( 1998 ) 319 ; j. alcaraz et al . , phys . lett . b 461 ( 1999 ) 387 ; m. sasaki et al . , nucl . phys . ( proc . suppl . ) b 113 ( 2002 ) 202 .
the pamela experiment is a satellite - borne apparatus designed to study charged particles in the cosmic radiation with a particular focus on antiparticles . pamela is mounted on the resurs dk1 satellite that was launched from the baikonur cosmodrome on june 15@xmath0 2006 . the pamela apparatus comprises a time - of - flight system , a magnetic spectrometer , a silicon - tungsten electromagnetic calorimeter , an anticoincidence system , a shower tail catcher scintillator and a neutron detector . the combination of these devices allows antiparticles to be reliably identified from a large background of other charged particles . this paper reviews the design , space qualification and on - ground performance of pamela . the in - orbit performance will be discussed in future publications . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
barium titanate ( batio@xmath0 ) is well known to exhibit three ferroelectric phase transitions : stable at high temperature in a perovskite cubic phase , its structure becomes successively tetragonal , orthorhombic and finally rhombohedral as the temperature goes down . many works investigated and discussed the origin of its ferroelectric phase transitions @xcite . amongst them , the most gratifying contribution was probably due to cochran @xcite who associated the ferroelectric instability with the softening of a transverse optic phonon , and emphazised the connection between the structural instability and the lattice dynamics . consequently to cochran s work , a large amount of experiments have been performed in order to confirm the existence of a soft ferroelectric mode in batio@xmath0 . they include infra - red @xcite and raman @xcite measurements of the @xmath3 phonon modes as well as various neutron diffraction data @xcite . these experiments focused on the temperature behaviour of the soft phonon and were mainly concerned by the low frequency modes : only a few results are available for the higher energy vibrations . simultaneously , theoretical phonon dispersion curves were deduced from fits to the experimental data using different shell models . let us mention the pseudo - ionic model developped by gnininvi and bouillot @xcite or the rigid - shell model used by jannot _ et al . _ these models were however not particularly suited to describe abo@xmath0 crystals , like batio@xmath0 . migoni , bilz and buerle @xcite pointed out that the behaviour of the ferroelectric soft mode in the oxidic perovskites originates from the unusual anisotropic polarizability of the oxygen that , in turn , may be connected to hybridization between o-2p and b - d states . consequently , a more sophisticated `` polarizability model '' @xcite was introduced in order to include the specific physical features of abo@xmath0 compounds . the application of this model to batio@xmath0 was reported by khatib _ et al . _ @xcite . in their work , they obtained a full phonon band structure and investigated the temperature behaviour of the ferroelectric soft mode . however , their interesting results still remained at a semi - empirical level . since a few years , theoretical advances have enabled one to determine the phonon dispersion curves of solids within a truly first - principle approach . recently , _ ab initio _ results have been reported for knbo@xmath0 @xcite and srtio@xmath0 @xcite . similarly , in this paper , we investigate the lattice dynamics of batio@xmath0 . our phonon frequencies compare well with the experimental data available . moreover , our dispersion curves are very similar to those reported for knbo@xmath0 @xcite , a perovskite material presenting the same sequence of phase transition as batio@xmath0 . in particular , the wave vector of the unstable phonon modes remains located in three @xmath4 slab region of the brillouin zone , emphasizing a chain - structure instability in real space @xcite . this behaviour is illustrated with a simplified model . the range and anisotropy of the interatomic force constants is also discussed . our calculations have been performed in the general framework of the density functional formalism @xcite . the exchange - correlation energy functional was evaluated within the local density approximation , using a polynomial parametrization of ceperley - alder @xcite homogeneous electron gas data . we used a plane - wave pseudopotential approach . the all - electron potentials were replaced by the same extended norm conserving highly transferable pseudopotentials as in ref . we have considered 5s , 5p and 6s as valence states to build the ba pseudopotential , 3s , 3p , 3d and 4s valence states for the ti pseudopotential and 2s and 2p valence states for the o pseudopotential . the electronic wave function was expanded in plane waves up to a kinetic energy cutoff of 45 hartree ( about 6200 plane waves ) . integrals over the brillouin zone were replaced by a sum on a mesh of 6x6x6 special @xmath5-points @xcite . the dynamical matrix , born effective charges and dielectric tensor were computed within a variational formulation @xcite of the density functional perturbation theory @xcite . first , calculations were carried out to determine the dynamical matrix on different meshes of @xmath6-points . then , an interpolation was performed following the scheme proposed in ref . this technique takes properly into account the long - range behaviour of the dipole - dipole interaction which is separated from the remaining short - range forces owing to the knowledge of the born effective charges and the optical dielectric tensor . an insight into the convergence reached on the phonon dispersion curves is reported in fig . [ fig.gr ] . the frequencies deduced from the dynamical matrix at @xmath7 and @xmath8 are compared to those extrapolated from two different meshes of @xmath6-points : the first mesh ( m1 ) includes @xmath3 ( .0 , .0 , .0 ) , x ( .5 , .0 , .0 ) , m ( .5 , .5 , .0 ) and r ( .5 , .5 , .5 ) points ; the second mesh ( m2 ) is the cubic mesh m1 to which the @xmath9 ( .25 , .25 , .25 ) point was added . we obtain a very good convergence with the m2 mesh . our calculations are performed at the experimental lattice parameter of 4.00 . this choice facilitates the comparison with the experimental data . some indications on the volume dependence of the phonon frequencies can be found in ref . @xcite , where the frequencies of the @xmath3 phonons at different lattice constants have been compared . first , we present in table [ table.z ] the born effective charges and the optical dielectric constant @xcite , important ingredients of the present study since their knowledge allows to identify the long - range part of the interatomic force constants and makes the interpolation of phonon frequencies tractable . for ba and ti atoms , the effective charge tensor is isotropic . for o , the two independent components of the tensor correspond respectively to a displacement of the atom parallel ( @xmath10 ) and perpendicular ( @xmath11 ) to the ti - o bond . @xmath12 ( @xmath13 ) and @xmath10 ( @xmath14 ) are anomalously large @xcite with respect to the nominal ionic charges ( + 4 for ti and @xmath15 for o ) . this surprising phenomenon was explained recently in connection within dynamic changes of hybridization between o-2p and ti-3d orbitals @xcite . this specific feature is at the origin of a large destabilizing dipole - dipole interaction , connected to the anomalous mode effective charge @xcite and the instability of the ferroelectric mode @xcite . our computed optical dielectric constant ( 6.75 ) largely overestimates the experimental value ( 5.40 ) @xcite , as usual within the local density approximation . this discrepancy will essentially affect the position of the highest longitudinal optic mode : when replacing the theoretical dielectric constant by the experimental value , its frequency at the @xmath3 point changes from 631 to 696 @xmath16 . at the opposite , the frequencies of the two other longitudinal modes at the @xmath3 point are affected by less than 2 @xmath16 . the calculated phonon dispersion curves are plotted along high symmetry directions in fig . [ fig.bs ] . the @xmath3-x , @xmath3-m and @xmath3-r lines are along the @xmath1100@xmath2 , @xmath1110@xmath2 and @xmath1111@xmath2 directions , respectively . the unstable modes associated to a negative curvature of the energy hypersurface have imaginary phonon frequencies . the frequencies at the high symmetry points are reported in table [ table.w ] . our result can be compared to the experimental data @xcite . however , a complication arises from the fact that all the experimentally observed vibrational excitations have a real frequency while the computed unstable modes are obtained with an imaginary frequency . as the soft mode can be clearly identified by its symmetry , the associated experimental frequencies were removed from the comparison , for clarity . in the low frequency region , the presence of this additional soft mode may have slightly modified the frequency of the other modes . in spite of these difficulties we observe a good correspondence between our theoretical frequencies and the experimental data , specially for the acoustic modes for which a large variety of data are available . the ferroelectric phase transitions are driven by the unstable phonon modes . we are therefore mainly concerned by the analysis of these specific phonons . two transverse optic modes are unstable at the @xmath3 point : they correspond to a displacement of the ti atom against the oxygen cage . the associated displacement eigenvector is equal to [ @xmath17 , @xmath18 , @xmath19 , @xmath20 @xcite . these two modes remain unstable all along the @xmath3-x line , with very little dispersion . one of them stabilizes along the @xmath3-m and x - m lines . examination of the eigenvectors reveals that the unstable mode at the m ( .5 , .5 , .0 ) point is polarized along the @xmath21-direction : its displacement eigenvector is equal to [ @xmath22 , @xmath23 @xcite . both of the unstable modes become stable when deviating from the three @xmath3-x - m planes to the r - point . these features were also observed for knbo@xmath0 @xcite and point out a marked 2d character of the instability in the brillouin zone . this behaviour is more easily visualized in fig . [ fig.im ] where we show the frequency isosurface of the lowest unstable phonon branch corresponding to @xmath24 . the region of instability , @xmath25 , lies between three pairs of flat surfaces , that are parallel to the faces of the brillouin zone cube . in other words , the unstable modes are contained in three perpendicular interpenetrating slab - like regions of finite thickness containing the @xmath3 point . as highlighted by yu and krakauer @xcite , this behaviour corresponds to chain instabilities in real space . at the m - point , we have seen that there is a single unstable mode polarized along the @xmath21-axis and dominated by the ti@xmath26 and o1@xmath26 displacements . at this wave vector ( @xmath27 ) , the ti and o@xmath28 atoms will be coherently displaced all along an _ infinite _ @xmath1001@xmath2 chain . going now from m to the r - point , the coherency of the displacement will gradually disappear and a _ finite _ length of correlation will be reached for which the phonon becomes stable . the finite thickness of the slab region of instability therefore corresponds to a minimum correlation length of the displacement required to observe an unstable phonon mode . from fig . [ fig.im ] , the length of the shortest unstable chain can be estimated to @xmath29 16 @xcite . note finally , the small dispersion of the unstable mode in the @xmath3-x - m plane suggests a small correlation of the displacements between the different ti o chains . in cubic batio@xmath0 , we will see that the single displacement of a particular atom never leads to an instability : when one atom is displaced , a force is induced and brings it back in its initial position ( see tables [ table.ifc ] - [ table.o - o ] : the self - force on ti and o is positive ) . however , its atomic displacement will simultaneously induce forces on the other atoms . it is only the additional displacement of some other atoms in this force field that can lower the total energy and produce an instability . the amplitude and the range of the interatomic force constants ( ifc ) associated to this mechanism can be analysed in order to clarify the chain instability pointed out in the previous section . moreover , the specific role of the dipole - dipole interaction ( dd ) can be separated from that of the short - range forces ( sr ) following the scheme proposed in ref . our conventions on the interatomic force constants @xmath30 are such that the force @xmath31 induced on atom @xmath32 by the displacement @xmath33 of atom @xmath34 is given by : @xmath35 . let us first investigate the ifc with respect to a reference ti atom along a ti - o chain ( table [ table.ifc ] ) . as previously mentionned , we note that the self - force on the ti atom is large and positive ( + 0.15215 ha / bohr@xmath36 ) . we observe also that the longitudinal ifc with the first neighbour o atom is surprisingly small ( + 0.00937 ha / bohr@xmath36 ) ; moreover , it is positive . the analysis of the dd and sr contributions points out that these characteristics are the result of a destabilizing dd interaction , sufficiently large to compensate the sr forces . it is this close compensation which allows the displacement of ti against the o atoms . another insight on this balance of forces was already reported in ref . consequently to the very small total ifc , the ti and o displacements might be relatively decoupled . at the opposite , the dd forces induced on the next ti atom are negative : they will combine with the sr forces in order to produce sizeable coupling ( @xmath370.06721 ha / bohr@xmath36 ) . this mechanism is at the origin of the chain correlation of the ti atomic displacements . by contrast , the _ transverse _ force on the first ti neighbour is very small and confirms the small correlation of the displacements from chain to chain . the decay of the ti ti and o o longitudinal ifc with the interatomic distance can also be investigated . the results are reported in table [ table.ti-ti ] and [ table.o - o ] . it is seen that the longitudinal ifc are anisotropic : they propagate essentially along the ti o chain . this appears clearly for the sr part . for o , the dd contribution is also highly anisotropic due to the anisotropy of the born effective charges . the anisotropy of the ifc is inherent to the chain correlation previously mentionned in this paper . from the knowledge of the ifcs previously reported , we can also investigate the energy surface of batio@xmath0 and illustrate the chain correlation highlighted in section iii . let us consider that we have a bulk cubic crystal with the atoms frozen at their equilibrium position @xmath38 . then , we allow displacements of ti and o atoms belonging to a [ 100 ] single ti o chain of finite but increasing size . the total energy of this system will be given by : @xmath39 where @xmath40 is the interatomic force constant matrix and the sum on @xmath32 and @xmath34 is restricted to the ti and o atoms that are allowed to move . with the help of this equation , we can track the appearance of an instability in terms of the length of the chain of displaced atoms . an instability will correspond to a specific displacement pattern that lowers the total energy of the system : it will be associated to a negative eigenvalue of the restricted force constant matrix . in fig . [ fig.chain ] , we report the evolution of the lowest eigenvalue of the force constant matrix with respect to the length of the chain of moving atoms . displacing only a single atom , the force induced on the ti is larger than that on the o atom . with 3 atoms , we observe , at the opposite , that the ti - terminated chain ( ti o ti ) is more stable than the o - terminated one ( o ti o ) : it points out the important role of the ti ti interaction . the difference between ti and o terminated chains will disappear progressively with the chain length . it is seen that an instability takes place for a chain longer than 10 atoms ( 5 unit cells ) . this is in close agreement with the correlation length estimated in the previous section . it suggests that the behaviour of batio@xmath0 is already well reproduced when considering an isolated ti o chain of displacements . it confirms that the correlation between the different chains may play a minor role . going further and frozing all the o atoms in such a way that only the ti atoms are allowed to move along the chain , we can repeat the previous investigations . for this case , however , we do not observe any instability even for an infinite chain of correlated ti displacements . this result aims to prove that the relatively weak coupling between ti and o displacements still remains an important feature in the appearance of the structural instability . in summary , we have reported first - principles linear response calculations of the phonon dispersion curves of cubic batio@xmath0 . our results are in good agreement with the experimental data . moreover , they are very similar to those reported for knbo@xmath0 , a cubic perovskite material presenting a sequence of phase transitions analogous to batio@xmath0 . in particular it was seen that a ferroelectric instability takes place when correlated atomic displacements are allowed along @xmath1100@xmath2 chains of finite length . this has been investigated with a simplified model and the length of the shortest unstable chain estimated to a dozen of atoms . the interatomic force constants are anisotropic and propagate essentially along the ti - o chains . their analysis has emphasized the correlation of the ti displacements . it has shown that the ti and o atomic displacements are only weakly coupled . this small coupling remains however an important ingredient to reproduce the ferroelectric instability . we thank j .- m . beuken for permanent computer assistance . x.g . is grateful to the fnrs - belgium for financial support . this paper presents research results of the belgian program on interuniversity attraction poles initiated by the belgian state - prime minister s office science policy programming . we aknowledge the use of the rs 6000 work stations and the namur scientific computing facility ( namur - scf ) , which are common projects between ibm belgium and respectively , the catholic university of louvain ( ucl - pcpm ) and the facults universitaires notre dame de la paix ( fundp - lpmps ) . m. e. lines and a. m. glass `` principles and applications of ferroelectrics and related materials '' , clarendon press , oxford , 1977 . w. cochran , adv . * 9 * , 387 ( 1960 ) . y. luspin , j. l. servoin and f. gervais , j. phys . c * 13 * , 3761 ( 1980 ) . servoin , f. gervais , a. m. quittet and y. luspin , phys . b * 21 * , 2038 ( 1980 ) ; j .- servoin , y. luspin and f. gervais , ferroelectrics * 37 * , 523 ( 1981 ) ; k. a. mller , y. luspin , j .- l . servoin , f. gervais , j. physique * 43 * , l-537 ( 1982 ) . c. h. perry and d. b. hall , phys . lett . * 15 * , 700 ( 1965 ) ; a. scalabrin , a. s. chaves , d. s. shim and s. p. s. porto , phys . ( b ) * 79 * , 731 ( 1977 ) ; b. jannot , l. gnininvi and g. godefroy , ferroelectrics * 37 * , 669 ( 1981 ) ; g. godefroy , b. jannot , a. levanyuk and a. sigov , ferroelectrics * 54 * , 297 ( 1984 ) ; k. laabidi , m. d. fontana and b. jannot , solid state comm . * 76 * , 765 ( 1990 ) ; t. p. dougherty _ _ , science * 258 * , 770 ( 1992 ) ; t. nakamura , ferroelectrics * 137 * , 65 ( 1992 ) ; m. d. fontana , k. laabidi and b. jannot , j. phys . * 6 * , 8923 ( 1994 ) . g. shirane , b. c. frazer , v. j. minkiewicz , j. a. leake and a. linz , phys . * 19 * , 234 ( 1967 ) . y. yamada , g. shirane and a. linz , phys . rev . * 177 * , 848 ( 1969 ) . g. shirane , j. d. axe , j. harada and a. linz , phys . b * 2 * , 3651 ( 1970 ) . j. harada , j. d. axe and g. shirane , phys . b * 4 * , 155 ( 1971 ) . j. bouillot , c. escribe , w. j. fitzgerald and l. gnininvi , solid state comm . * 30 * , 521 ( 1979 ) . b. jannot , c. escribe - filippini and j. bouillot , j. phys . c * 17 * , 1329 ( 1984 ) . l. gnininvi and j. bouillot , ferroelectrics * 15 * , 173 ( 1977 ) . r. migoni , h. bilz and d. buerle , phys . rev * 37 * , 1155 ( 1976 ) . h. bilz , g. benedek and a. bussmann - holder , phys . b * 35 * , 4840 ( 1987 ) ; a. bussmann - holder , h. bilz and g. benedek , phys . b * 39 * , 9214 ( 1989 ) . d. khatib , r. migoni , g. e. kugel and l. godefroy , j. phys . : condens . matter * 1 * , 9811 ( 1989 ) . r. yu and h. krakauer , phys . lett . * 74 * , 4067 ( 1995 ) . c. lasota , c .- z . wang , r. yu and h. krakauer , ferroelectrics * 189 * , xxx ( 1996 ) . w. e. pickett , comput . phys . rep . * 9 * , 115 ( 1989 ) ; m. c. payne , m.p . teter , d.c . allan , t.a . arias and j.d . joannopoulos , rev . * 64 * , 1045 ( 1992 ) . d. m. ceperley and b. j. alder , phys . * 45 * , 566 ( 1980 ) . ghosez , x. gonze and j .- p . michenaud , ferroelectrics * 153 * , 91 ( 1994 ) . h. j. monkhorst and j. d. pack , phys . rev . b * 13 * , 5188 ( 1976 ) ; ibid . , phys . rev . b * 16 * , 1748 ( 1977 ) . x. gonze , d. c. allan and m. p. teter , phys . lett . * 68 * , 3603 ( 1992 ) . n. e. zein , sov . solid state * 26 * , 1825 ( 1984 ) ; s. baroni , p. giannozzi and a. testa , phys . * 58 * , 1861 ( 1987 ) . p. giannozzi , s. de gironcoli , p. pavone and s. baroni , phys . b * 43 * , 7231 ( 1991 ) . x. gonze , j .- c . charlier , d.c . allan and m.p . teter , phys . b * 50 * , 13035 ( 1994 ) . ghosez , x. gonze and j .- p . michenaud , europhys . * 33 * , 713 ( 1996 ) ; ph . ghosez , x. gonze and j .- p . michenaud , ferroelectrics * 189 * , xxx ( 1996 ) . the values presented here are slightly different from those reported by us elsewhere @xcite due to the different convergence reached in the calculation . j. d. axe , phys . rev . * 157 * , 429 ( 1967 ) . w. zhong , r. d. king - smith and d. vanderbilt , phys . lett . * 72 * , 3618 ( 1994 ) . ghosez , x. gonze , ph . lambin , and j .- michenaud , phys . b * 51 * , 6765 ( 1995 ) ; ph . ghosez , x. gonze and j .- p . michenaud , ferroelectrics * 164 * , 113 ( 1995 ) . m. posternak , r. resta and a. baldereschi , phys . b * 50 * , 8911 ( 1994 ) . @xmath41 was obtained by extrapolating to zero frequency index of refraction measurements [ g. burns and f. h. dacol , solid state comm . * 42 * , 9 ( 1982 ) ] at different wavelengths using a one - oscillator sellmeier equation . r. a. cowley , phys . rev . * 134 * , a981 ( 1964 ) . the eigendisplacement vector @xmath42 was normalized such that @xmath43 , where m is such that @xmath44 and @xmath45 is the mass of atom @xmath32 in atomic mass units . the length of the shortest unstable chain is slightly different from that reported for knbo@xmath0 . changes in material properties could explain this difference although part of it could be due to the different @xmath5-point and @xmath6-point convergence achieved in ref . @xcite : as observed in fig . [ fig.gr ] , the use of a finer mesh of @xmath6-points could still slightly decrease the size of the zone of instability of batio@xmath0 .
the phonon dispersion curves of cubic batio@xmath0 have been computed within a first - principles approach and the results compared to the experimental data . the curves obtained are very similar to those reported for knbo@xmath0 by yu and krakauer [ phys . rev . lett . * 74 * , 4067 ( 1995 ) ] . they reveal that correlated atomic displacements along @xmath1100@xmath2 chains are at the origin of the ferroelectric instability . a simplified model illustrates that spontaneous collective displacements will occur when a dozen of aligned atoms are coupled . the longitudinal interatomic force constant between nearest neighbour ti and o atoms is relatively weak in comparison to that between ti atoms in adjacent cells . the small coupling between ti and o displacements seems however necessary to reproduce a ferroelectric instability . + + _ keywords : batio@xmath0 , phonon dispersions curves , ferroelectric instability , density functional theory , linear response . _
the unification of gauge coupling constants is an attractive proposal under the name of grand unification@xcite , which can not be understood in the standard model(sm ) . at the unification scale @xmath7 the strong , weak and electromagnetic coupling constants are the same since they are described by a simple or semi - simple group @xmath8 for grand unification . below the unification scale , this grand unification(gut ) group @xmath8 is broken down to the standard model group @xmath9 and the difference between the sm couplings is generated@xcite . since the gut unifies the interactions and some quarks and leptons are assigned in a same @xmath8 multiplet , the quark and lepton transition is possible in guts , triggering proton to decay . the lepto - quark gauge bosons and colored scalars are responsible for proton decay . the lepto - quark gauge boson mass is the unification scale @xmath7 . but the higgs multiplet containing the sm higgs doublet must contain a light spectrum for the doublet . the colored partner of the doublet must be superheavy , or proton lifetime is absurdly short , @xmath10 seconds , for o(1 ) couplings . for example , the @xmath11 of @xmath0 contains a higgs doublet field @xmath12 and color triplet field @xmath13 , where it is assumed that @xmath4 is light and @xmath14 is superheavy . there exists the difficulty in splitting the doublet - triplet masses , which is the split multiplet problem . recently , it was pointed out that the split multiplet problem can be understood in 5d theories with the @xmath15 orbifold compactification@xcite . it is because of the geometric twist of the gauge group such that some fields are projected out from the massless spectrum . indeed , the orbifold compactification in string models@xcite has shown already some models without colored scalars , realizing the split multiplet . thus , orbifold compactification in higher dimensional theories may be the underlying reason for the split multiplet@xcite . in the context of this orbifold breaking of the gut groups , some issues can be reconsidered as for the gauge coupling unification@xcite , the larger gut groups@xcite , and the flavor unification@xcite in this paper , we try to understand _ geometrically _ the top - bottom mass hierarchy and the @xmath6 puzzle@xcite . the fifth dimensional coordinate @xmath16 is compactified to a torus @xmath17 . furthermore , the point @xmath18 is identified to @xmath19 ( @xmath20 symmetry ) and the point @xmath21 is identified to @xmath22 ( @xmath23 symmetry ) . this modding introduces a fundamental region @xmath24 $ ] and there arise two fixed points , @xmath25 and @xmath26 . this geometry is used to twist the gut multiplet . in particular the gut multiplet @xmath11 living in the bulk is twisted , the twisting being represented by @xmath27 and @xmath28 . obviously , the twisting breaks @xmath0 . but in the bulk the @xmath0 symmetry is manifest above the unification scale and the gauge coupling unification is assumed above @xmath7 . the bulk fields are split into four different kaluza - klein(kk ) categories @xmath29 with the @xmath2 quantum numbers @xmath30 , @xmath31 @xmath32 @xmath33 @xmath34 where @xmath35 is the 4d spacetime coordinate , @xmath36 and @xmath37 for @xmath38 . the massless field is @xmath39 for @xmath40 . in this way , the massless 4d higgs doublet is obtainable from 5d while color triplets are all heavy . now let us extend the study to include @xmath41 supersymmetry . in 5d , there exists an @xmath42 supersymmetry . one @xmath20 breaks down the @xmath42 down to @xmath41 and the other @xmath20 breaks @xmath8 down to the sm . two 4d spinors(e.g . two weyl spinors ) make up one 5d spinor . thus , a 5d field thery is not anomalous . we can introduce only one @xmath0 fermion multiplet in the bulk without worrying about the anomaly , say a hypermultiplet @xmath43 . upon compactification , the @xmath41 supermultiplets are @xmath44,\ { \rm mass}=2n / r\\ & h&_t^{(2n+1)}[(+-);(3,1,\frac{1}{3})],\ { \rm mass}=(2n+1)/r\\ & \hat h&_t^{(2n+1)}[(-+);(\bar 3,1,-\frac{1}{3})],\ { \rm mass}=(2n+1)/r\\ & \hat h&_d^{(2n+2)}[(--);(1,2,\frac{1}{2})],\ { \rm mass}=(2n+2)/r\end{aligned}\ ] ] where the brackets [ ] contain the quantum numbers of @xmath45 . the original 5d @xmath0 theory with one anti - quintet is anomaly free . but the orbifolding introduces one massless fermion doublet only , @xmath46 . the other massive fields in the bulk pair up to form massive kk towers of mass @xmath47 where @xmath48 . since the low energy theory should be anomaly free , we are dictated to introduce brane fermions . so at one fixed point we introduce a 4d @xmath41 sm higgs supermultiplet @xmath5 with the quantum number @xmath49 under @xmath9 . at this field theory level , the introduction of anomaly cancelling fermions at the fixed points is arbitrary . it is needed from the renormalizability of the low energy effective theory . however , the orbifold compactification in string models introduces fixed point fermions definitely once the bulk fermions carry anomaly@xcite . under the framework of the preceding paragraph , we will consider two models in sec . ii model ( i ) one @xmath11 in the bulk model ( ii ) one @xmath11 plus @xmath50 in the bulk where @xmath51 denotes a higgs field and @xmath52 denotes some fermions of the sm . in sec . iii , we try to understand the @xmath6 puzzle geometrically along the line of the split multiplet in the bulk , and present model ( iii ) for an explicit presentation . since there appears the kk tower of the split multiplet in the bulk we expect a correction to @xmath53 from the usual susy gut prediction , @xmath54 which is @xmath55@xcite . the superscript 0 denotes no threshold correction . we will show that in models ( i ) , ( ii ) and ( iii ) the kaluza - klein mode corrections are in the favorable direction toward the experimental data . at the minimal supersymmetric standard model(mssm ) level , @xmath4(coupling to @xmath56 quark ) and @xmath5(coupling to @xmath57 quark ) are not distinguished except for their gauge quantum numbers . thus , the apparent disparity of the top bottom masses is not understood . it is fixed either by a large top yukawa coupling and a small bottom coupling with @xmath58 or by comparable yukawa couplings and a large @xmath59 . in this section , we explore a possibility that the couplings and vacuum expectation values are comparable , but the mass hierarchy is understood from a geometric origin@xcite . namely , the origin of @xmath4 and @xmath5 are different in a higher dimensional theory . and @xmath5 in the bulk and brane was considered before@xcite . however , in our gut theory , assigning @xmath5 at a brane is needed to explain the difference of @xmath60 mass scales in the low energy effective theory . ] to concentrate on the @xmath60 disparity , we restrict our discussion to the third family only . as the simplest model of the field theoretic orbifold compactification , let us introduce a @xmath11 in the 5d bulk . the compactification is @xmath15 as shown in introduction . because of the unification in the 5d bulk the gauge coupling is unified above the gut scale @xmath7 which can be a string scale in a theory from string compactification . the @xmath15 compactification produces one massless supermultiplet @xmath4 containing one higgs doublet . the compactification is schematically drawn in fig . 1 where two fixed points(3-branes ) @xmath61 and @xmath62 are shown and the thick line is the fundamental region(=the bulk ) in 5d . in the bulk @xmath0 gauge fields and @xmath11 live . at the 3-brane @xmath62 we locate the missing higgs doublet @xmath5 and the sm fields(including three copies of supermultiplets of 15 chiral fields ) . the 5d lagrangian contains @xmath63\\ & = \int d^4x\left [ \partial^\mu h_1^{(0)\dagger}(x)\partial_\mu h_1^{(0)}(x ) + y_bh_1^{(0)}q_3d_3^c + y_th_2q_3u_3^c\right]\end{aligned}\ ] ] where @xmath64 and @xmath65 . thus , we obtain hierarchic masses @xmath66 note that the geometric suppression is the square root of @xmath67 , which may not be large enough . therefore , to enhance the suppression we consider the following model . in model ( i ) we inserted only @xmath43 in the bulk . here , we introduce @xmath68 in the bulk also . the bulk field must be an @xmath0 multiplet . for no mass hierarchy between @xmath56 and @xmath69 masses , we need a complete multiplet . but an @xmath0 multiplet field in the bulk allows only a split massless field . for a complete multiplet @xmath70 to be massless , we have to introduce two @xmath71 s so that an anti - quark singlet from one @xmath71 and a lepton doublet from the other @xmath71 survives as a massless field by appropriately twisting the bulk fields . in fig . 2 , we assign the fields in the bulk and at the 3-brane @xmath62 . except the quintets containing @xmath72 , all the sm fermions are located at @xmath62 . of course , we locate @xmath5 at @xmath62 to cancel the gauge anomaly . the relevant 5d lagrangian is @xmath73\\ & = \int d^4x\left[\cdots+y_bh_1^{(0)}q_3d_3^{c(0 ) } + y_th_2q_3u_3^c+y_\tau h_1^{(0)}l_3^{\prime(0)}e_3^c\right]\nonumber\end{aligned}\ ] ] from which we obtain a linear relation in @xmath67 , @xmath74 . note that @xmath75 , and @xmath76 . the mass scales of interest in our scenario are the electroweak scale , the unification scale(or the string scale ) @xmath7 , and the inverse compactification length @xmath77 . we assume that the compactification mass is smaller than the unification mass so that the running of the kaluza - klein(kk ) towers between @xmath7 and @xmath78 helps toward the unification condition . let us define the ratio of these two scales as @xmath79 @xmath80 the masses of the kk modes are @xmath81 where the columns show @xmath82 quantum numbers , kk masses , and the @xmath83 function coefficients . the tower of kk excitations up to @xmath7 contributes to the running of gauge couplings at their thresholds@xcite . at the scale @xmath84 below the compactification scale , the gauge coupling constant is @xmath85 up to a threshold correction @xmath86 . @xmath87 is the unification coupling . stirling s formula gives @xmath88 and @xmath89 thus , the low energy mssm couplings become @xmath90+\tilde b_i\ln\frac{m_u}{m_c'}\label{master}\ ] ] where @xmath91 , and @xmath92 0.5 cm * model ( i ) : * in model ( i ) , from the fields @xmath93 in the bulk we obtain @xmath94 and from the field @xmath5 in the brane @xmath95 from the vector multiplet in the bulk , @xmath96 , @xmath97 the sum of the brane higgs and the bulk vector contributions define the total value , @xmath98 . therefore , from eq . ( [ master ] ) , we obtain a relation between couplings at the electroweak scale @xmath99 where @xmath100 . strong coupling unification considered in ref.@xcite is due to the duplication of matter fields appearing in the extension of chiral multiplets to hypermultiplets which make the gauge coupling be strong at high energy scales . but , in all models considered here , most matter fields are living at the brane and the 5-d gauge theory becomes asymptotically free : @xmath101 in eq . ( [ master ] ) for models i , ii , and iii , respectively . in the unification models , such as in susy @xmath0 , one can determine the unification mass and gauge coupling constant @xmath102 at the unification scale . namely , if @xmath7 and @xmath102 are given , one can predict @xmath103 at the electroweak scale . these coupling constants satisfies the relation given in eq . ( [ unify ] ) , and the experimental values at @xmath104 may not satisfy eq . ( [ unify ] ) . in our models , the onset of the kk modes introduces another parameter @xmath78 . thus , we can satisfy the condition ( [ unify ] ) by appropriately choosing @xmath78 . however , in our split multiplet models the unified gauge coupling constant is extremely small due to the asymptotic freedom . namely , @xmath78 turns out to be far below the string scale , @xmath105 gev with @xmath106 gev . however , it may be a better treatment of the problem if we satisfy one condition . we choose the condition as the ratio @xmath107 in view of the experimental errors including the error in @xmath108 and other effects ( running effects due to yukawa couplings and two loop runnings ) . thus , the unification condition ( [ unify ] ) is satisfied approximately but not exactly due to the error bars allowed and hence we will study just the modification of @xmath108 . then , @xmath109 where @xmath110 is the kk mode corrected value in @xcite with reasonable choices of the ratio @xmath107 . for @xmath111 , we obtain @xmath112 , respectively . then , @xmath113 which correspond to @xmath114 away from the experimental data . in all cases considered above the additional logarithmic running reduces the discrepancy between experimental value and the prediction of susy gut even with no threshold correction . the coupling at the unification scale becomes @xmath115 and @xmath116 , respectively . we used the conventional value @xmath117 at @xmath118 . at @xmath7 we also cutoff the power running and there appears an o(1 ) uncertainty in @xmath119 , which however does not affect the unification condition . 0.5 cm * model ( ii ) : * we can repeat the same calculation for model ( ii ) . but note that the additional fields @xmath120 and @xmath121 have the following kk modes @xmath122 so that the zero modes are @xmath123 which mimicks a gut multiplet . another massive gut - like multiplets are the even kk modes @xmath124 and @xmath125 which contribute to the log running . the odd kk modes contributes only to the power running . thus , the difference of gauge couplings and the ratio @xmath119 are not changed , viz . eq.([unify ] ) . for @xmath126 the unification coupling is changed to @xmath127 for the exact unification . a similar analysis as in the study of model ( i ) for @xmath128 and @xmath129 , the coupling constant at the unification scale becomes @xmath130 and @xmath131 , respectively . the kk mode correction to @xmath132 is the same as in model ( i ) . as discussed in the preceding section , there are a lot of possibilities for obtaining hierarchies of couplings by locating some fields in the bulk and some fields at a brane . in this section , we explore one more possibility for geometrically generating hierarchical coupling structure . one of the puzzles in the @xmath0 gut is that in the second family the quark yukawa coupling is too small ( by a factor of 3 ) compared to the lepton coupling , which is the @xmath6 puzzle . to obtain a desired suppression for the @xmath133 quark coupling , georgi and jarlskog introduced @xmath134 in addition to the usual @xmath135@xcite . in our scenario of keeping a split part of an @xmath0 multiplet as a massless spectrum in the bulk , there is a possibility of geometrically understanding the @xmath6 puzzle . for simplicity , we modify the simplest example , model ( i ) of the previous section , and comment on another possibility after the discussion on model ( iii ) . the higgs fields , the first and the third family members are the same as in model ( i ) . we only change the members of the second family . * model ( iii ) * among the second family members , some fields are put in the bulk . it is a split multiplet from @xmath136 . in the bulk , the members of @xmath137 $ ] under @xmath138 are assigned the @xmath2 parity as @xmath139 and @xmath140 , respectively . thus , only the quark doublet @xmath141 has a zero mode spectrum @xmath142 which we interpret as the second family quark doublet . at low energy , the theory must be anomaly - free and hence we locate the rest members of the second family , @xmath143 , at the brane , which is shown in fig . 3 . the 5d lagrangian contains @xmath144\\ & = \int d^4x\left[y_sh_1^{(0)}q_2^{(0)}d_2^c+y_\mu h_1^{(0)}l_2e_2^c+\cdots\right]\nonumber\end{aligned}\ ] ] note that @xmath145 , implying @xmath146 thus , the strange quark yukawa coupling is suppressed compared to the muon yukawa coupling . from the bulk zero mode @xmath142 , @xmath147 . from the brane fields , @xmath148 and @xmath149 , @xmath150 for @xmath151 . from the brane higgs @xmath5 , @xmath152 . from the bulk higgs @xmath153 , @xmath154 . from the vector multiplet , @xmath155 . thus , we obtain @xmath156 therefore , @xmath157 model ( iii ) is interesting since it turns out that @xmath158 is very large @xmath159 and @xmath7 is the usual unification scale @xmath160 gev in order to satisfy the eq . ( [ unify ] ) . the unification coupling constant is also close to the sgut value @xmath161 . due to the large @xmath162 compared to the other models , model ( iii ) allows the perfect unification with the logarithmic running between a small scale difference of @xmath163 . if we consider model ( iii ) with the backbone of model ( ii ) instead the backbone of model ( i ) , we obtain a better relation between the top - bottom mass hierarchy since the suppression will be linear in @xmath164 . assuming all the couplings to be order 1 , we obtain @xmath165 at the unification scale , implying @xmath166 . @xmath132 is the same as in model ( iii ) . in this paper , we studied a new possibility for a field theoretic orbifold compactification possessing supersymmetry , which was applied to the top - bottom mass hierarchy and the @xmath6 puzzle . this possibility relies on the missing massless spectrum in the bulk . the 5d bulk theory with any fermion representation is anomaly free , but the orbifold compactification may project out split multiplet in the bulk . this situation has been observed in orbifold compactifications in string models@xcite . because of the split multiplet , the anomaly - free condition dictates to put some massless fermions at the brane so that the resulting 4d theory is anomaly free . in string examples , the field content and assignment of the fields at the fixed points are determined uniquely by the modular invariance requirement . but in our field theoretic example , the field content and the location are arbitrary . in this paper , we chose the simplest possibility . in our examples , we put the @xmath43(containing the @xmath4 higgs doublet ) in the bulk . because of the twisting , only @xmath4 from the @xmath11 remains massless in the bulk . thus , the needed @xmath5 is put at a brane where the sm fields are located . thus , the bottom and the top quarks have geometrically different factors for the effective 4d yukawa couplings , rendering a top - bottom mass hierarchy . to enhance the hierarchical factor some sm fermions are put in the bulk in our second example . similarly , the @xmath6 puzzle is understood geometrically by putting the strange quark doublet in the bulk , thus reducing the strange quark yukawa coupling compared to the muon yukawa coupling . there are other applications along this line , e.g. reducing the up quark mass compared to the down quark mass . in all these examples we considered , the corrections to the strong coupling constant are in the right direction , making the low energy effective mssm predictions closer to the experimental value . one tempting question to ask in this scenario might be the @xmath84 problem@xcite . certainly , one can not write @xmath167 in the bulk . it can be written only at the brane a. but the need to introduce @xmath5 at a is below the compactification scale @xmath77 . therefore , writing the dimensional parameter such as @xmath84 must have a suppression factor , certainly less than @xmath78 . but at this moment , we do not understand geometrically how large the suppression factor is . we may need an additional discrete or global symmetry to sufficiently suppress @xmath84 . in any case , these extra symmetries are needed for proton longevity . the field theoretic orbifolding considered recently is very simple compared to the string theory orbifolding . however , it seems to be arbitrary in choosing and assigning the fields , and we expect that some string compactification in the future may lead to the above types of field theoretic orbifolding so that the assignment of the @xmath5 and the sm fields at the brane is no longer arbitrary . 0.5 cm * note added : * recently , there appeared an argument@xcite that it would not be possible to have a consistent susy field theory on the @xmath168 orbifold with a single bulk higgs multiplet , since there are gauge anomalies localized at the orbifold fixed points@xcite . in our case , however , the local gauge anomalies can be cancelled by introducing a brane higgs field and 5d chern - simons terms in the bulk@xcite . the models considered in the present paper are consistent up to introducing the chern - simons terms . # 1#2#3astrophys . j. * # 1 * ( # 2 ) # 3 # 1#2#3int . j. mod . # 1 * ( # 2 ) # 3 # 1#2#3mod . phys . # 1 * ( # 2 ) # 3 # 1#2#3nature * # 1 * ( # 2 ) # 3 # 1#2#3nucl . * b#1 * ( # 2 ) # 3 # 1#2#3phys . * b#1 * ( # 2 ) # 3 # 1#2#3phys . * d#1 * ( # 2 ) # 3 # 1#2#3phys . lett . * # 1 * ( # 2 ) # 3 # 1#2#3phys . rep . * # 1 * ( # 2 ) # 3 # 1#2#3sov . j. nucl . # 1 * ( # 2 ) # 3 # 1#2#3z . # 1 * ( # 2 ) # 3 # 1#2#3jhep * # 1 * ( # 2 ) # 3 n. arkani - hamed and m. schmaltz , [ hep - ph/9903417 ] ; n. arkani - hamed , l. hall , d. smith , and n. weiner , [ hep - ph/9909326 ] ; e. a. mirabelli and m. schmaltz , [ hep - ph/9912265 ] ; g. dvali and m. shifman , [ hep - ph/0001072 ] ; d. e. kaplan and t. m. tait , [ hep - ph/0004200 ] ; d. e. kaplan and t. m. tait , [ hep - ph/0110126 ] ; m. kakizaki and m. yamaguchi , hep - ph/0110266 .
a supersymmetric 5d @xmath0 grand unification is considered . the @xmath0 is broken down to @xmath1 by the @xmath2 assignment of the bulk field(s ) . the matter fields are located at the fixed point(s ) . in the bulk , a higgs multiplet @xmath3(containing the bottom doublet @xmath4 ) and the @xmath0 gauge multiplet are located . at one fixed point , @xmath5(the top doublet ) and the standard model matter multiplets are presented . because of the difference of the locations of @xmath4 and @xmath5 , one can obtain a hierarchy between top and bottom yukawa couplings . we also present a possibility to understand the @xmath6 mass puzzle in this framework of the split multiplet . + 0.5cm[key words : split multiplet , top - bottom mass hierarchy , @xmath6 puzzle ]
determining the energy levels of a quantum system was a desperate enterprise before the advent of the bohr sommerfeld quantization conditions . max planck had emphasized that the energy of the harmonic oscillator is quantized in units of a fundamental energy given by the product of what we now call dirac s constant and the frequency of the oscillator . however , this rule failed miserably when used in other quantum systems . paul ehrenfest s adiabatic principle applied to the pendulum @xcite whose length slowly changes as a function of time , made clear that it is not the energy , but the action that is quantized , but why ? when we change the length of the pendulum adiabatically , the amplitude of oscillation does not stay constant , neither does the frequency nor the energy . what stays constant is the action , that is , the area in phase space . for the founding fathers of quantum mechanics it must have been a miracle , an amazing fact symbolizing in post schrdinger language that the number of nodes in an energy wave function stays constant under adiabatic changes . the classical dynamics of the pendulum serves as an excellent guide for many quantum phenomena . for example , the upside down pendulum yields insight into the energy wave function of a periodic potential @xcite . here we do not focus on the familiar bloch states which appear when the potential enjoys a strict periodicity over the whole space . our states occur when the modulation of the potential extends only over a finite domain of space . due to the shape of this potential we refer to it as the _ accordion potential_. extensions of these one dimensional considerations to two and three dimensions lead us to the effect of the quantum anti centrifugal force @xcite . in the present paper we take seriously the joke ` physics takes mathematics and makes it understandable ' , therefore we shall suppress all the mathematics and highlight the essential ideas . this approach is justified by the fact that we are not inventing or applying new mathematics , but attempt to draw together phenomena of different fields exposing a common thread . we shall focus on a general point of view but emphasize that the modern tools of cold atoms in a standing electromagnetic wave can demonstrate these unusual bound or localized states predicted in this paper . we follow one ` leitmotif ' , that is , a theme common to all phenomena discussed in this article : the classical physics of the upside down pendulum . moreover , we develop a little side theme summarized by the phrase `` wave mechanics in quantum potentials '' . our paper is organized as follows : in sec.2 we briefly summarize the classical physics of the upside down pendulum and then turn in sec.3 to related phenomena , such as the paul trap , the helium atom and field induced multipoles . we devote sec.4 to the transition to quantum mechanics and in particular , discuss the shape of the ground state wave function in the accordion potential . in sec.5 we show that in three dimensions a localized ground state can exist even at zero energy . the quantum anti centrifugal potential emerges when in sec.6 we focus on the schrdinger equation in two dimensions . we conclude in sec.7 with an outlook and present extensions of the concepts discussed in this paper . an appendix summarizes the mathematical aspects of the upside down pendulum and the schrdinger equation of the particle in a periodic potential . in the mid - forties it was recognized @xcite that the cyclotron would never be able to accelerate particles to energies above a critical value . the problem lies in the fact that in free space static electric or magnetic or both fields together can not focus a beam of charged particles in all planes through the axis of the beam . fortunately , this difficulty , which applies to linear accelerators , can be circumvented to some degree in circular systems . however , the degree of focusing permitted by the effect of the centrifugal force is extremely weak . these considerations suggested that energies higher than 10 mev are not attainable without unreasonable expenditures of money and materials . however , in 1949 nick christofilos from athens , who at that time was an elevator operator , dreamt up the principle of strong focusing . although he got a patent , his work was unpublished and the principle was re discovered @xcite in 1952 . the central idea of the principle of strong focusing is to use a sequence of strong focusing and defocusing fields to achieve a net focusing effect . this idea is best illustrated using the principle of the upside down pendulum @xcite . consider the motion of a point mass on one end of a massless rod . the opposite end of the rod is connected via a hinge to another rod that is fixed to a support . this pendulum experiences a constant gravitational force pulling downwards . when the point of suspension is such that the pendulum hangs down and we concentrate on the limit of small displacements from the stable hanging position we can approximate the motion of the mass by that of a harmonic oscillator . however , when the pendulum is upside down and stands straight against the gravitational force , the motion is unstable and the mass tends to fall down . in this case we face an inverted harmonic oscillator of negative steepness . in order to stabilize it , pyotr l. kapitza @xcite in 1951 suggested a rapid vertical modulation of the foundation of the pendulum . when the modulation frequency is above a critical value determined by the length of the pendulum , the gravitational acceleration and the modulation amplitude , the motion of the pendulum becomes stable . why is this so ? many mathematical arguments offer themselves . they range from the method of averaging @xcite via the secular growth theory @xcite to the floquet theorem @xcite . however , none of these mathematical techniques provide deeper insight into the physics of the stabilizing mechanism . for the sake of argument and to illustrate this lack of insight , we follow the floquet reasoning . the equation of motion for the phase angle @xmath0 of the pendulum , measured from its upside down unstable position , is the mathieu equation @xcite , as discussed in the appendix . the floquet theorem @xcite determines the domains of stability and instability of this equation . hence , it is the stability chart @xcite of the mathieu equation that governs the parameter regime for which the upside down pendulum is stable . what is the mechanism of this stabilization ? when the foundation accelerates the pendulum upwards , the gravitational force which pulls the pendulum down and makes its motion unstable , is reduced . a push up therefore corresponds to a stabilizing force . on the other hand , a motion of the foundation downwards increases the effective gravitational force and thereby enhances the instability of the pendulum : a pull down corresponds to a de stabilizing force . on the first sight one might think that during a complete cycle of modulation , the two effects average out , however , during a majority of time there is a pushing up effect . why is this ? in the stable mode the pendulum performs two types of motion , the secular and the micro motion . the secular motion corresponds to a slow oscillation of the mass between two extremes of phase angles . in addition , the mass undergoes a rapid motion with a frequency identical to the modulation frequency . it is this micro motion , superimposed on the secular motion , which makes the pendulum stable and which breaks the symmetry between the upward and the downward push . indeed , due to the geometry of the pendulum , the angular acceleration is not only determined by the acceleration due to the modulation , but by the product of this acceleration and the instantaneous phase angle . the corresponding force is therefore a tidal force : no displacement of the oscillator no force ; large displacement large force . the modulation of the foundation translates into a modulation of the phase angle . since the effective acceleration is the product of the acceleration due to the modulation and the phase angle we find the product of two oscillatory functions with the same phase . the time average of such a quantity is a positive constant providing a harmonic oscillator potential of positive steepness for the slow motion . many other applications of the upside down pendulum come to mind , for example the paul trap @xcite that allows us to store and manipulate single ions in a controlled way . it is worth mentioning that paul traps play an important role in the recent proposals for a quantum computer . in the paul trap the need for a time dependent force is dictated by the poisson equation of electrostatics making it impossible to create binding forces in all three dimensions of space . indeed , the poisson equation enforces the feature that even when two spatial directions enjoy a binding potential the third one is anti binding . to overcome the instability of the trap in the third direction two possibilities offer themselves : apply a time independent homogeneous magnetic field that confines the ion in the third direction or apply an alternating voltage to the trap . the first approach corresponds to the penning trap , the second to the paul trap . in the most elementary case the paul trap consists of a quadrupole field , giving rise to harmonic oscillator potentials . the alternating voltage applied to the electrodes creates a dynamical binding in three dimensions . in each dimension the physics of this binding phenomenon is identical to the upside down pendulum . the helium atom represents another interesting application of this concept of dynamical binding . for more than one hundred years physicists have tried to understand the motion of two electrons around the nucleus . a one dimensional model in which the electrons move along a line with the nucleus at the origin brings out the essential physics . on first sight , a situation in which the electrons are on different sides of the nucleus seems to be preferable since in this way the electrons can avoid each other in a most effective way . however , an extremely interesting situation @xcite occurs when both electrons are on the same side . indeed , here one electron is close to the nucleus and oscillates rapidly between two equilibrium points . the second electron is far away from the nucleus and moves slowly feeling the attractive force of the nucleus and the repulsion from the inner electron . indeed , the fast motion of the inner electron creates a time averaged repulsive potential , which superposes with the attractive coulomb potential of the nucleus , giving rise to a local potential minimum for the outer electron . in this case , the electron electron repulsion , together with the rapid motion of the inner electron , forms a time dependent barrier to stop the unstable motion of the outer electron caused by the attraction towards the nucleus . this phenomenon is the upside down pendulum in disguise @xcite . our last illustration of effective potentials , arising from time averaging , is the effect of an induced dipole or multipole . it helps us to make the transition to unusual bound states in quantum mechanics . two examples offer themselves , ( _ i _ ) the binding of an electron in a negative ion and ( _ ii _ ) the atom in a standing electromagnetic wave . the field of an electron can polarize a neutral atom even when the electron is at a distance large compared with the atomic dimensions . this interaction between the electron and the atom leads to a force of attraction . this attraction is the reason why some atoms , such as hydrogen or halogen , are able to form negative ions by an attachment of an electron . we now consider the motion of an atom in a standing electromagnetic wave . indeed , in quantum optics many such experiments have been and are being performed . the light induces a dipole moment in the atom and this dipole again interacts with the light field . since the interaction energy is the product of the induced dipole and the field , it enters quadratically . consequently , the position dependence of the interaction energy follows from the position dependence of the square of the field that is , from the square of the mode function thereby creating the effective potential for the motion of the atom . all examples in the previous sections illustrate how dynamical binding can occur in classical physics . we now turn our attention from newton s equation of motion @xmath1\varphi(t)=0\ ] ] for a harmonic oscillator with time dependent steepness @xmath2 to the time independent schrdinger equation @xmath3 \ u(x)= 0 \label{schrodinger}\ ] ] for a point particle of mass @xmath4 in a position dependent potential @xmath5 ; both equations are identical in form . the role of the angle coordinate of the classical upside - down pendulum is now played by the energy wave function @xmath6 . moreover , the time variable @xmath7 is replaced by the coordinate variable @xmath8 . indeed , newton s equation of motion is an equation of second order in time for the position and the schrdinger equation is an equation of second order in position for the wave function . we emphasize that this analogy between the two equations only holds true for the classical harmonic oscillator . whereas the schrdinger equation is always linear , a feature independent of the form of the potential , the form of newton s equation strongly depends on the shape of the potential and in general is nonlinear . we focus on a periodic potential in space . the particular shape of this potential is of no importance as long as it is averages out to zero over space ; it is as often positive as it is negative . for our analysis it is crucial that the oscillations of the potential start at a given point in space and end thereby connecting a flat space through a domain of wiggles back to a flat space . we measure energy relative to flat space . fig.1 shows an example of such a potential . it is not surprising that under these circumstances we find bound states of negative energy , however , under appropriate conditions , this potential can also display a localized state of zero energy . thus , it is again the physics of the upside down pendulum that allows us to understand this phenomenon . we first concentrate on the one dimensional case . guided by the analogy between the newton and the schrdinger equation we expect the corresponding ground state energy wave function to exhibit a slowly varying envelope that is modulated with the period of the periodic potential ; but where does the binding come from ? the origin of this bound state lies in the modulation of the envelope without the modulation there is no binding . the modulation can be easily understood : the particle is more likely to be in a stable minimum of the potential rather than in an unstable maximum . the curvature of the wave function at a given position , expressed by its second derivative , is determined , not only by the potential , but also by the wave function at that point . indeed , according to the schrdinger equation ( [ schrodinger ] ) , it is the product of the potential and the wave function that governs the curvature . this quantity contains , apart from other terms , the product of the oscillatory potential and the modulations of the wave function . the simplest case of a cosine modulation gives rise to the square of the cosine , thus creating a constant potential and a part that varies in space with twice the modulation frequency . we focus on the constant part and note that this constant is always negative . in order to explain this feature we recall that the modulation of the envelope of the wave function is such that at the minima of the potential we have local maxima of the wave function and vice versa . hence , the modulation and the potential defer in their sign and their product creating the potential well is negative . the fact that the potential only wiggles over a finite domain makes this constant negative potential into an attractive potential well whose width is determined by the range of the oscillations . the depth of the well involves , apart from other parameters , the square of the height of the periodic potential . for more mathematical treatment and explicit expressions for the well , we refer to the appendix . there is a close analogy between the ground state in the periodic potential and the problem of the upside down pendulum . in both cases the binding effect is a consequence of the average of the product of two oscillatory functions : for the pendulum it is the time dependent modulation of the foundation and the time dependent angle of oscillation , which creates a harmonic oscillator of positive steepness . when considering the particle in the periodic potential , it is the product of the position dependent potential and the wave function modulated in space that gives rise to a well . however , there is a dramatic difference : whereas the modulation of the pendulum is on for all times , the spatial modulation is confined to a certain domain of space . it is this confinement in space , together with the creation of an effective potential , which defines the bound state . the spatial domain , over which the wiggles are present , and the depth of the potential well , determine the ground state energy . is it possible to choose these parameters so as to achieve a ground state of zero energy ? the answer is : it depends ! in one dimension the answer is a flat no , but in three dimensions it is possible , as we shall now show . we start by considering the situation in one dimension . as we decrease the depth of the well and the energy of the ground state approaches zero , the tails of the wave function reach more and more into the forbidden region outside of the well . this feature is a consequence of the fact that the decay is governed by the square root of the absolute value of the energy . in the limit of zero energy the wave function is no longer localized ; no mechanism prevents the wave from flooding all space . consequently , there is no ground state of zero energy in one dimension . is it possible to construct such a zero - energy bound state in higher dimensions ? in order to answer this question , we first concentrate on the three - dimensional case and consider a potential @xmath9 that wiggles along the radial direction . beyond a final radius @xmath10 , the potential vanishes . the time independent schrdinger equation @xmath11 \ u_{0}(r)=0 \label{radial}\ ] ] for the radial energy wave function @xmath12 corresponding to vanishing angular momentum is then identical to the one - dimensional case . due to the product of the oscillatory potential and the wave function , and the finite domain of oscillations , we find an attractive potential well of radius @xmath10 . there is however , a subtle difference to the one - dimensional case : it is the origin where the radial wave function has to vanish @xcite corresponding to an infinitely high potential wall at the origin . this feature is a consequence of the spherical symmetry . therefore , the infinite wall at the origin provides one side of the effective well creating even a node of the wave function . the other wall of finite height is due to the effective potential together with the finite range of the oscillatory domain . for appropriate parameters such as the final radius and the depth of the modulation of the potential we can fit a ground state of negative energy into this well , as shown in fig.2 . the corresponding wave function displays a node at the origin and an exponential decay in the classically forbidden domain beyond the critical radius , with a single maximum lying in between . can we now take the limit of zero energy and thus realize a zero - energy ground state ? the answer is yes ! indeed , we can arrange the depth of the well such that exactly one quarter of a period of a sine oscillation fits into the potential well . why a quarter of a period ? since the decay of the wave function outside of the potential is governed by the square root of the energy we find that for zero energy the wave function is constant . due to the condition of continuity on the wave function it must take on the value at the wall . the radial wave function displays a node at the origin and increases like a sine function to a constant value . however , the sine function must merge smoothly into the constant , which is only possible at a point where the sine function has an extremum . since we want to have the ground state without a node , except the one at the origin , the only possibility for such a merger appears at one quarter of a period @xcite of the sine function . obviously , this zero energy ground state radial wave function does not display any localization . the localization becomes apparent when we recall that we have to divide the radial wave function by the radial variable in order to find the total wave function , that is @xmath13 indeed , now the total wave function enjoys a maximum at the origin and decays with the inverse power of the radius . we emphasize however , that the total wave function is not square integrable since the volume element @xmath14 brings in the square of the radius . this contribution cancels exactly the factor creating the localization in the probability density , that is , in the absolute value squared of the wave function . for this reason we refer to this state as localized state rather than bound state . we conclude this discussion by briefly addressing the case of positive energies . here , the exponential decay into the classically forbidden region of a wave function corresponding to negative energy turns into right and left running plane waves of positive energy . obviously , plane waves are not localized wave functions , nevertheless , the total wave function is concentrated at the origin due to the inverse power of the radius ; in this sense we have localized wave functions in the continuum . the physics of a bound state in two dimensions is very different from the one in other dimensions ; and this for many reasons . in @xcite we have studied this case in more detail . in the present paper we only highlight and motivate the results of @xcite . for detailed derivations we refer to that article . to bring out the peculiarities of the two dimensional case we first consider the case of no external potential . in two dimensions a vanishing angular momentum does not imply @xcite a vanishing centrifugal potential as in three dimensions . indeed here the radial schrdinger equation reads @xmath15 \ u_{0}(r)=0 . \label{polar}\ ] ] we can trace the origin of the new potential @xmath16 back to the non - vanishing commutation relation between the operators of momentum and the radial unit vector . this potential is , therefore , a true quantum potential . since , in addition , it is attractive rather than repulsive we have named it the quantum anti - centrifugal potential @xcite . moreover , the potential is attractive only in a two dimensional world . for one or three dimensions it vanishes , but for higher than three dimensions it becomes repulsive . the quantum anti - centrifugal potential has brought us into a rather unusual situation . usually , we perform wave mechanics in a classical external potential . in the present context however , we have a quantum potential . this feature stems from the reduction of space from three to two dimensions , reminiscent of the born - oppenheimer approximation in molecular physics . a molecule is a quantum system consisting of several degrees of freedom . in the framework of oppenheimer approximation we solve the schrdinger equation for the electrons for a fixed position of the nuclei and then use the resulting potential curves , that is , the electronic energies as a function of the separation of the nuclei to determine their relative motion . in this sense we indeed perform wave mechanics on potentials that are not due to external forces but due to quantum mechanics . in the example of the nuclear motion in a molecule , the appearance of wave mechanics in a quantum potential is a consequence of the reduction of the degrees of freedom @xcite . the attractive quantum anti - centrifugal potential manifests itself even in the energy wave function of a free particle , that is in the absence of any other external potential . the ordinary bessel functions @xmath17 and @xmath18 are two independent solutions of the time independent schrdinger equation for positive energy . both show a bunching of nodes towards the origin , as suggested by an attractive potential leading to an acceleration of the particle towards the origin and demonstrated for the example of @xmath17 in fig.3 . we have to compare and contrast this situation with the case of a particle with one unit of angular momentum , where the centrifugal potential is indeed repulsive . it is worth noting that the quantum anti - centrifugal potential reduces the strength of repulsion of the centrifugal potential and the effective potential is not as repulsive as in the classical case . in the case of one unit of angular momentum , the solutions of the time independent schrdinger equation are the ordinary bessel functions @xmath19 and @xmath20 . as expected by the repulsive potential enforcing a deceleration of the particle as it approaches the origin , the nodes of the bessel functions show anti - bunching towards the origin . one might wonder if the quantum anti - centrifugal potential is strong enough to support a bound state with negative energy . in the transition from positive to negative energies real wave numbers transform into purely imaginary ones as a consequence of the quadratic dispersion relation between energy and wave number . therefore , the ordinary bessel functions @xmath17 and @xmath18 turn into modified bessel functions @xmath21 and @xmath22 . since @xmath21 explodes for large arguments @xcite , that is , large radial distances , this solution does not satisfy the requirement of exponential decay enforced by the attractive quantum anti - centrifugal potential . only the function @xmath22 achieves this goal , but it has another disease : it explodes at the origin like a logarithm . this singularity at the origin indicates that the solution @xmath23 of the schrdinger equation , shown in fig.4 , does not describe a truly free particle but that there is a delta function potential present . the strength of this external potential determines @xcite the energy eigenvalue of this bound state . moreover , due to the quantum anti - centrifugal potential the probability of finding the particle within a given area in space is concentrated along a band around the location of the delta function ; a remarkable feature unique to two dimensions . in a similar arrangement in one or three dimensions the maximum of the probability is at the location of the potential minimum . this property might have interesting applications for guiding atoms along wires @xcite or electro - magnetic waves in wave - guides @xcite . we now return briefly to the case of an energy eigenstate in a two dimensional oscillatory potential of finite range . as emphasized in the previous paragraph , the case of vanishing angular momentum contains the most quantum effects . here , the ground state wave function of negative energy feels the combination of the quantum anti - centrifugal potential and the potential well created by the product of the wave function and the potential . the examples discussed in the preceding sections represent but only a small subsection of the class of unusual bound states . they have been selected because they follow or are motivated by the physics of the upside - down pendulum . however , there are many more intriguing bound states in atomic and molecular physics @xcite . we conclude our paper by briefly summarizing three examples starting with one that is still closely related to our main theme of the upside - down pendulum and gradually moving away towards new frontiers . consider a diatomic molecule with various electronic states . within oppenheimer approximation the electronic states provide the potentials for the relative motion of the two nuclei . a strong external time dependent laser field interacts with the electrons and couples the individual electronic potential surfaces . the wave functions for the vibratory motion of the nuclei are therefore coupled through a time dependent periodic interaction hamiltonian . the product of the interaction hamiltonian and the nuclear wave functions govern the time evolution of the wave functions ; a feature that is again reminiscent of the upside - down pendulum . in the latter the product consists of the time dependent steepness of the oscillator and the phase angle . the only difference lies in the fact that in the classical case the newton equation is of second order , whereas the schrdinger equation is a first order . however , this distinction is superficial since we are dealing with the nuclear wave functions corresponding to different electronic states and hence , with a system of differential equations of first order . due to this product of two periodic functions with the same period ( interaction hamiltonian and wave function ) we find again a constant which lifts up the electronic potential by integer multiples of the energy of the absorbed photon . in mathematical terms this energy elevation is a consequence of floquet states . however , it can be viewed as just one more illustration of the upside - down pendulum with counter - intuitive consequences . indeed , when we consider a binding and a repulsive potential this constant shift in energy can result in a crossing of the potential curves . since in a diatomic molecule there exists a non - crossing rule we obtain an avoided crossing forming new potentials ; one is binding and one is repulsive . however , these potentials are rather peculiar because each of them consists of parts of the two original potentials . the formation of new bound states @xcite in this light induced molecule has been observed experimentally @xcite . trojan asteroids are celestial bodies that are held in their positions by the gravitational forces of the sun and jupiter . a similar phenomenon can occur @xcite in an atom where the electron plays the role of a trojan asteroid , and the nucleus substitutes for one of the two planets . a circularly polarized electromagnetic field simulates the effect of the second planet . indeed , the electromagnetic interaction of the electron with the field and the coulomb attraction between the electron and the nucleus replace the gravitational attraction of the celestial bodies . in a frame rotating with the electromagnetic field , the motion of the electron is in a plane and governed by three potentials which depend on the two - dimensional radial direction : the coulomb attraction of the nucleus , the centrifugal potential of the circular motion and the linear potential arising from the interaction of the electron with the electromagnetic field . for short distances the coulomb potential dominates , whereas for large distances the repulsive potential of the circular motion prevails . consequently , for intermediate distances , an unstable potential maximum occurs . this feature is in complete analogy with the upside down pendulum . but where does the stabilizing drive come from ? the answer to this question lies in the fact that the radial motion is coupled to the angular motion . the latter is locked to the rotating electromagnetic field and the electron performs angular vibrations around this stable point of equilibrium . these vibrations couple into the radial motion creating a dynamically binding potential . this classical picture also holds true for the quantum case as verified by extensive studies of wave packets propagating in this minimum . in our last example of unusual bound states it is the balance of static forces that is responsible for the formation of a bound state of a lone electron in a rydberg atom in crossed electric and magnetic fields @xcite . the crossed - field situation is of particular appeal in the field of quantum chaology since in this system energy is the only conserved quantity . the classical system displays chaos and is therefore of interest in the search for fingerprints of chaos in the corresponding quantum system . apart from these questions of quantum chaos there is another quite interesting aspect of this system . the electron in the atom experiences , not only the attractive coulomb potential of the nucleus , but also a linear potential due to a constant electric field and a binding harmonic oscillator potential due to the magnetic field . the superposition of all three potentials creates a local potential minimum far away from the nucleus . an electron bound in this minimum displays a large dipole moment that can be observed by sending the atom through an inhomogeneous electric field . indeed , the experiments @xcite have confirmed the existence of this far outside lying minimum . we have come a long way on our journey into unusual bound states . starting from the classical physics of the upside - down pendulum we have been led to the phenomenon of a zero - energy ground state in a periodic potential in three dimensions . the two - dimensional world has even more surprises in store : the attractive quantum anti - centrifugal force that manifests itself in a free particle through the bunching of the nodes or even in a bound state . here the probability of finding the particle is concentrated in the domain where no force is acting . this feature resulting from the action of the quantum anti - centrifugal potential could be useful in guiding atoms along wires or electromagnetic waves along fibers . experiments on trapping cold atoms by single photons @xcite or along a whispering gallery mode of a glass sphere resonator @xcite are yet more realizations of the physics of the upside - down pendulum and give us confidence that our predictions of unusual bound and localized states can be verified experimentally in the near future . this work started when i ( wps ) had the great privilege to be a postdoc with john a. wheeler more than a decade ago . since that time , john and i have frequently returned to the topics addressed in the present paper . these discussions took place in trains , planes and automobiles and at various locations such as hightstown , high island , and ulm . i thank john for this wonderful time , his outstanding hospitality at hightstown and high island and especially for the unique experience to work with him . the present paper is partially based on notes and sketches of figures prepared jointly . i am grateful for many fruitful discussions and john s deep insights and , in particular , for allowing us to use the material that had originally been obtained in close collaboration with him . the proceedings of the lake garda conference are a most welcome opportunity to finally summarize this project started many years ago . moreover , we are grateful to i. biaynicki - birula for a critical reading of the manuscript . two of us ( mac and wps ) thank r. bonifacio for his hospitality and for organizing a most interesting conference in the splendid surroundings of lake garda . the work of wps is partially supported by the deutsche forschungsgemeinschaft , moreover , he gratefully acknowledges a travel grant from the universitt ulm which made part of this research possible . in contrast to the main body of the paper , we pursue in the present appendix a more mathematical approach . in particular , we emphasize the similarities between the classical equation of motion for the upside down pendulum and the schrdinger equation for a nonrelativistic particle in a periodic potential . for this purpose , we first cast the corresponding equations into a dimensionless form . this approach allows us to simultaneously derive equations for the dynamics contained in the macro and micro motion of the pendulum or the envelope function and the modulation of the ground state wave function . we start by summarizing the classical equation of motion for the upside down pendulum . for the sake of simplicity we consider the limit of small angles @xmath24 measured relative to the vertical position . a vertical acceleration @xmath25 of the foundation translates into a vertical acceleration of the mass @xmath4 of the pendulum and adds to the gravitational acceleration @xmath26 pointing downwards . here , the sign of @xmath25 is crucial : when @xmath25 is positive , that is , when the pendulum is accelerated upwards , the gravitational acceleration pointing downwards is reduced . likewise , when @xmath25 is negative , that is , when the pendulum is accelerated downwards , the gravitational acceleration is increased . consequently , the total acceleration is @xmath27 . hence for small angles @xmath24 the force tangentially to the rod of the pendulum of length @xmath28 reads it is reasonable to assume that the modulation @xmath32 is a periodic function of period @xmath33 . this time scale allows us to introduce a dimensionless variable @xmath34 and the differential equation of the upside down pendulum takes the form here we have introduced the abbreviations @xmath36 and @xmath37 for the steepness of the inverted harmonic oscillator and the acceleration due to the modulation , respectively . moreover , prime denotes differentiation with respect to @xmath38 . the two dimensionless equations of the driven pendulum ( [ eight ] ) and the quantum particle in the periodic potential ( [ ten ] ) have identical structure . in secs.2 and 4 we have presented qualitative arguments to explain the motion of the pendulum as a superposition of a slow motion ( macromotion ) and a rapid motion ( micromotion ) or , in the language of wave functions , to decompose the energy wave function into an envelope and a modulation . we now take advantage of the analogy between the two systems to support these qualitative arguments by rigorous mathematics . for this purpose we start from the equation again , we emphasize that @xmath6 can either be the phase angle of the pendulum or the wave function . likewise , @xmath38 represents time or position and the drive @xmath47 is either the acceleration of the foundation or the potential . @xmath50{\cal a } \nonumber \\ & & -\left [ { \cal a}''+\left ( \eta + \frac{1}{\epsilon}+ \frac{v''}{v}\right ) { \cal a}\right ] \epsilon v \nonumber \\ & & -2\epsilon { \cal a } ' v'=0 . \label{star}\end{aligned}\ ] ] so far , our analysis is exact . we now solve the equation ( [ star ] ) in an approximate way . for this purpose , we first neglect the last contribution in ( [ star ] ) involving the product of the first derivatives of the envelope function and the potential . this approximation is justified since @xmath51 is out of phase with @xmath52 and can not therefore lead to a significant contribution . moreover , we choose @xmath53 such that we emphasize that this is not possible in a strict sense , since @xmath55 is not necessarily a constant . however , in the oscillatory domain of the potential the most elementary model @xmath56 suggests the estimate @xmath57 and hence we find for the case of the upside down pendulum , the parameter @xmath61 is negative . in the absence of the drive , the negative value of @xmath61 leads to an exponential growth of @xmath59 . however , the square of the drive is always positive and can create an overall positive coefficient in front of @xmath59 , providing an oscillation rather that an exponential explosion . in this case , the envelope @xmath59 is an energy wave function of this potential well and the parameter @xmath61 is the dimensionless energy eigenvalue . hence , @xmath61 is not determined from the outside , but rather by the well ( [ well ] ) itself . if the well is deep enough , it can support , apart from the ground state , also excited states with negative energy . p. ehrenfest , naturwiss . * 11 * , 543 ( 1923 ) ; for an excellent introduction into and comprehensive summary of the early quantum mechanics we refer to d. ter haar , the old quantum theory , pergamon press , oxford , 1967 , p. 44 . a classic in this field is m. born , atommechanik , springer , heidelberg , 1925 ; see also j. duck and e. c. g. sudarshan , 100 years of planck s quantum , world scientific , singapore , 2000 . for a summary of the problem of acceleration of charged particles to high energies see the article by j. p. blewett , in e. u. condon and h. odishaw ( eds ) , handbook of physics , mcgraw - hill , new york , 1958 , chapter 9 , p. 153 . j. a. wheeler in many discussions with one of us ( wps ) repeatedly credited h. bethe with the discovery of the problems leading to the principle of strong focusing . unfortunately , we have not been able to locate the appropriate reference . we have talked to prof . bethe , but also he was not able to point out the article . m. abramowitz and i. a. stegun , handbook of mathematical functions , dover publications , new york , 1965 ; for a particularly illuminating method to motivate the stability chart of the mathieu equation using wkb wave functions , see m. j. richardson , am . j. phys . * 39 * , 560 ( 1971 ) . w. paul , rev . phys . * 62 * , 531 ( 1990 ) . a pedagogical introduction into the physics of paul traps emphasizing the concept of the effective potential is given by p. e. toschek , in g. grynberg and r. stora ( eds ) , new trends in atomic physics , north - holland , amsterdam , 1984 , p. 390 . for a summary of current activities in paul traps see h. walther , in b. bederson and h. walther ( eds ) , advances in atomic , molecular and optical physics , academic press , boston , 1995 . k. richter and d. wintgen , phys . * 65 * , 1965 ( 1990 ) ; j. phys . b * 24 * , l565 ( 1991 ) : for an overview of the full classical and quantum dynamics of two - electron systems see k. richter , g. tanner , and d. wintgen , phys . a * 48 * , 4182 ( 1993 ) . we can also interpret the system of the two electrons in the helium atom as one realization of the fermi accelerator . in the most elementary version of this device an oscillating wall confines the unbounded motion of a particle in a linear potential . for a review of the fermi accelerator see f. saif , i. biaynicki - birula , m. fortunato and w. p. schleich , physics reports , to be published . l. d. landau and e. m. lifshitz , quantum mechanics , non - relativistic theory , pergamon press , oxford , 1965 , p. 267 ; there exists a large amount of literature on the physics of negative ions . see for example the classic book h. massey , negative ions , cambridge university press , cambridge , 1976 . for the most recent activities in this field see the springer series , production and neutralization of negative ions and beams . a similar effect appears also for an electron in the field of a super - heavy nuclei , see for example f. g. werner and j. a. wheeler , phys . * 109 * , 126 ( 1958 ) . for summary see c. n. cohen - tannoudji , in j. dalibard et al . ( eds ) , fundamental systems in quantum optics , elsevier science , amsterdam , 1992 ; see also the nobel lectures by s. chu , c.n . cohen - tannoudji , and w. d. phillips , rev . phys . * 70 * , 685 ( 1998 ) . for a most striking and counter - intuitive bound state resulting from the peculiar behavior of the radial wave function at the origin of three - dimensional space we refer to j. rauch and m. reed , comm . * 29 * , 105 ( 1973 ) ; and m. reed , and b. simon , methods of modern mathematical analysis ii , academic press , new york , 1975 . in this example the potential is a sequence of appropriately constructed steps that lead continuously downwards as the radial variable increases . classically a particle of given energy has to fall down the steps . however , the reflections of the quantum wave at the individual steps interfere in a way as to localize the particle . likewise , the same authors discuss a potential consisting of an infinite sequence of potential spikes that classically would keep a particle trapped , however , due to the tunneling effect the quantum particle escapes . a similar reasoning appears @xcite in general relativity , that is in geometrodynamics , when we determine the metric coefficients of a system . they follow from brill s equation @xcite , which is similar to the time independent schrdinger equation for zero energy . in contrast to quantum mechanics where the schrdinger equation determines the energy eigenvalues we now have to solve the equation under the constraints that the wave is not allowed to have nodes and corresponds to zero energy . this observation is in complete accordance with a recent paper arguing that the time dependent schrdinger equation is an approximation of the time independent schrdinger equation resulting from the elimination of degrees of freedom . see for example j. s. briggs and j. m. rost , epjd * 10 * , 311 ( 2000 ) . two examples illustrate the unusual bound states that originate from the mutual interaction of the electrons in a heavy atom : the thomas - fermi potential together with the centrifugal potential can form a second potential minimum , which is very deep and located close to the nucleus . this effect occurs provided the atomic number is larger than 57 and we are dealing with an energy eigenstate corresponding to the angular momentum quantum number @xmath68 , see for example m. goeppert - mayer , phys . rev . * 60 * , 184 ( 1941 ) . since a. sommerfeld we associate the motion of an electron in an atom with an ellipse or a circle . however , the electron at the top of the sea of filled atomic states moves in an effective screened potential giving rise to a necklace orbit as pointed out by j. a. wheeler , in e. h. lieb et al . ( eds ) , studies in mathematical physics , princeton university press , princeton , 1976 , p. 383
we summarize unusual bound or localized states in quantum mechanics . our guide through these intriguing phenomena is the classical physics of the upside down pendulum , taking advantage of the analogy between the corresponding newton s equation of motion and the time independent schrdinger equation . we discuss the zero energy ground state in a three dimensional , spatially oscillating , potential . moreover , we focus on the effect of the attractive quantum anti centrifugal potential that only occurs in a two dimensional situation . quantum mechanics ; bound states ; parametric oscillator ; periodic potential .
in lattice qcd calculation , evaluating the fermion determinant is computationally expensive and as a consequence @xmath2 is often not evaluated and is simply set equal to a constant . this is called the _ quenched approximation _ and amounts to neglecting internal fermion loops . the goal of this work is to study the systematic error introduced by quenching and to compensate for quenching errors by constructing an expansion in quark - loop count and using it to improve monte carlo estimators of wilson loops measured on quenched gauge field configurations . we examine a source of statistical error and a method for its partial alleviation . with the full qcd action @xmath3 = s^{wilson}_{fermion}[u ] + s_{gauge}[u,\psi,\bar\psi ] \ ; , \label{eq : qcd_lattice_action}\ ] ] the expectation value of some operator , @xmath4 is given by @xmath5[d\psi][d\bar\psi ] { \cal o}e^{-s[u,\psi,\bar\psi ] } \nonumber \\ & = & \frac{1}{z } \int[du ] { \cal o}det{\bf m}[u]e^{-s_{gauge}[u ] } \ ; , \label{eq : o_expectation}\end{aligned}\ ] ] where @xmath6 det{\bf m}[u]e^{-s_{gauge}[u ] } \;. \label{eq : z_qcd}\end{aligned}\ ] ] by rewriting @xmath7 $ ] as @xmath8 & = & n_f \log{\det{{\bf m}[u ] } } \nonumber\\ & = & n_f \mbox{tr}\log{\bf m}[u ] \nonumber\\ & = & \delta s[u ] \ ; , \label{eq : delta_s_log}\end{aligned}\ ] ] and dividing through by @xmath9 , ( [ eq : o_expectation ] ) becomes @xmath10 { \cal o}e^{-s_{\rm gauge}[u ] } e^{(\delta s[u]-\left\langle \delta s\right\rangle ) } } { \int[du ] e^{-s_{\rm gauge}[u ] } e^{(\delta s[u]-\left\langle \delta s\right\rangle ) } } \;. \label{eq : o_reweighted}\end{aligned}\ ] ] expanding the exponential in the fluctuations in @xmath11 leads to @xmath12 here , the subscript on @xmath13 signifies the quenched expectation value of @xmath4 . in order to make use of ( [ eq : o_reweighted_first_order ] ) one needs to be able to find @xmath11 . we use the pad z@xmath0 method to find @xmath14 as per @xcite . the pad approximant for @xmath14 can be written as @xmath15 @xmath16 can be estimated using noise z@xmath0 vectors , @xmath17 . @xmath18 where @xmath19 are the solutions of @xmath20 where @xmath21 and @xmath22 . the variance of these estimators can be greatly reduced by subtracting suitably chosen traceless matrices @xmath23 where the @xmath24 are chosen to minimise the variance of the estimator . the @xmath25 used are @xmath26 where @xmath27 . the odd powers of @xmath28 are traceless , but the even powers greater than two are not . @xmath29 and @xmath30 are calculated from @xmath31 and @xmath32 link loops respectively . for this work we expand to order @xmath33 , using @xmath34 noise vectors and @xmath35 traceless subtraction matrices . the variational procedure to set @xmath24 was found not to be necessary and so we set them equal to unity . to solve ( [ eq : original_ax_equals_b ] ) we used the mr@xmath36 algorithm from @xcite and the odd even preconditioner in from @xcite . we apply the preceding numerical techniques to finding improved estimates of wilson loops . measurements were made on two sets of @xmath37 gauge field configurations ; a set of @xmath38 quenched qcd configurations and a set of @xmath39 full qcd configurations , generated using the milc collaboration software . both have @xmath40 , and the full qcd configurations have @xmath41 and @xmath42 . fig . [ fig : nodeltabetawilson ] shows the results for various wilson loops with corrections done to first order as per ( [ eq : o_reweighted_first_order ] ) in all of figs . [ fig : nodeltabetawilson ] [ fig : yesdeltabetawilsonrelativelimitedsum ] the loop number on the x axis labels @xmath43 from left to right in the following order : @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 . in figs . [ fig : nodeltabetawilson ] , [ fig : nodeltabetawilsonrelative ] and [ fig : yesdeltabetawilsonrelative ] , points have been offset slightly for clarity . a method for controlling the statistical error in ( [ eq : o_reweighted_first_order ] ) is suggested by looking at the form of the correlator between the diagonal elements of @xmath56 and the operator , @xmath4 , being measured . given that @xmath4 and @xmath11 can be written @xmath57 then the first order correction to @xmath4 can be written as @xmath58 if we write @xmath59 , where @xmath60 , then @xmath61 fig . [ fig : nodeltabetawilsonlogcorr3 ] shows @xmath62 and @xmath63 as a function of @xmath64 for the case @xmath65 . the former shows that the two operators are correlated , as would be expected . the latter shows that most of the signal for @xmath66 is at small @xmath67 and but most of the noise is at large @xmath67 , and that for large lattices this has the potential to swamp the signal . one way to address the problem is to cut the integral off at some @xmath68 . @xmath69 this has been done using @xmath70 , and the results are shown in fig . [ fig : nodeltabetawilsonrelative ] . the cut off scheme reduces the size of the statistical error but also looses some signal . the monte carlo estimator for @xmath71 is @xmath72 an efficient way to do this is with fourier convolution . given a windowing function @xmath73 , where @xmath74 then ( [ eq : o_reweighted_first_order_limited_sum_estimator ] ) becomes @xmath75_x \nonumber \label{eq : o_reweighted_first_order_limited_sum_estimator_fourier}\end{aligned}\ ] ] where @xmath76 and @xmath77 are defined in the usual way and implemented as fast fourier transforms . using this method to compute @xmath78 also makes it easy to choose a more elaborate windowing function should this be desired . following @xcite and identifying @xmath79 we again look at corrections to wilson loops . these are measured on a set of @xmath80 @xmath37 full qcd configurations with @xmath81 , @xmath41 and @xmath42 and also the set of quenched qcd configurations used earlier . the results are shown in fig . [ fig : yesdeltabetawilsonrelative ] and are in accordance with @xcite . a notable difference between the two is the computational savings afforded by the choice of algorithm for @xmath14 . use of unbiased subtractors means that only @xmath82 noise vectors are needed . in addition to this we do not have to fix the configurations to landau gauge . the previously mentioned method of cutting the integral off to reduce the statistical error turns out not to work for @xmath11 defined in ( [ eq : lw_delta_s ] ) . the contributions from @xmath14 and @xmath83 are nearly equal but are opposite in sign and so the total correction is small as is shown in fig . [ fig : yesdeltabetawilsonrelativenolimitedsum ] introducing a cutoff reduces the contributions from @xmath14 and @xmath83 by different amounts which produces a large variation in the total correction . this is shown in fig . [ fig : yesdeltabetawilsonrelativelimitedsum ] . we have also looked at corrections to two point functions but the noise is such that this method can not be used _ as is _ to provide improved mass measurements . a method of improving estimates of observables measured on quenched qcd configurations was tested on some sets of small ( @xmath37 ) configurations . expanding to first order in @xmath84 gave moderate improvement to estimators for wilson loops . analysis of the source of the statistical error hinted at problems that the method would have on larger lattices and with two point functions , and also suggested a technique for its partial alleviation . the technique does have the potential to address these problems but at the price of reducing the size of the correction . it was shown in @xcite that combining the above method with an appropriate shift in the gauge coupling can provide more accurate corrections to wilson loop values . we obtained the same results using a more economical method of calculating @xmath14 . we would like to thank keh - fei liu , john sloan and don weingarten for useful discussions and in particular keh - fei liu for his support of john markham during his visit to the university of kentucky . c. thron , s.j . dong , k.f . liu and h.p . ying , phys . rev . * d57 * ( 1998 ) 1642 ( hep - lat/9707001 ) . u. glassner , s. gusken , t. lippert , g. ritzenhofer , k. schilling and a. frommer , hep - lat/9605008 . a. frommer , b. nockel , s. gusken , t. lippert and k. schilling , int . c6 * ( 1995 ) 627 ( hep - lat/9504020 ) .
we use the pad z@xmath0 stochastic estimator for calculating @xmath1 to compute quark loop corrections to quenched qcd . we examine the main source of error in this technique and look at a way of controlling it .
within the standard model ( sm ) the neutrinos are massless and hence there is no cp violation in the lepton ( l ) sector . the current evidence @xcite from the neutrino oscillation experiments , on the other hand , suggest that neutrinos are massive , however small , and they mix up . the goal of the present neutrino oscillation experiments is to determine the nine degrees of freedom in the low energy neutrino mass matrix . they are parametrized by three masses , three mixing angles and three cp violating phases out of which two are majorana and one is dirac . at present the neutrino oscillation experiments able to measure the two mass square differences , the solar and the atmospheric , and three mixing angles with varying degrees of precision , while there is no information about the phases . assuming that the neutrinos are of majorana type the small masses of the physical left handed neutrinos can be explained by the elegant seesaw mechanism @xcite which involves singlet right - handed neutrinos ( type - i seesaw ) or triplet higgs ( type - ii seesaw ) or can be both ( hybrid seesaw ) . in the present article we limit ourselves to the case of type - i seesaw models . although we call them right - handed neutrinos , in the extensions of the sm they are just singlet fermions that transform trivially under the sm gauge group . so , there is no apparent reasons for the number of heavy singlet neutrinos to be same as the number of left - handed neutrinos . so , for the main part of our discussions we restrict ourselves to only two right - handed neutrinos . these results will also be true when there are three right - handed neutrinos , but the third right - handed neutrino do not mix with the other two neutrinos . we start with three right - handed neutrinos and after some general comments work mostly with two right - handed neutrinos . while there is no information about the absolute mass scales of the physical neutrinos , the currently discovered tiny mass scales ; the atmospheric neutrino mass ( @xmath0 ) in the @xmath1 oscillation and the solar neutrino mass ( @xmath2 ) in the @xmath3 oscillation , can be explained by adding at least two right handed neutrinos to the sm lagrangian . however , with two right - handed neutrinos the seesaw mechanism predicts one of the physical light neutrino mass to be exactly zero which is permissible within the current knowledge of neutrino masses and mixings . the majorana mass of the right - handed neutrino violates @xmath4-number and hence is a natural source of l - asymmetry in the early universe @xcite . a partial l - asymmetry is then converted to baryon ( b ) asymmetry through the non - perturbative sphaleron processes , unsuppressed above the electroweak phase transition . currently the @xmath5-asymmetry has been measured precisely by the wilkinson microwave anisotropy probe ( wmap)@xcite and is given by @xmath6 it is legitimate to ask if there are any connecting links between leptogenesis and the cp violation in the low energy leptonic sector , in particular neutrino oscillation and neutrinoless double beta decay . in the context of three right - handed neutrino models several attempts have been taken in the literature to connect the cp violation in leptogenesis and neutrino oscillations @xcite . it is found that there are almost no links between these two phenomena unless one considers special assumptions @xcite . in fact it is shown that leptogenesis can be possible irrespective of the cp violation at low energy @xcite . on the other hand , in the two right - handed neutrino models there is a ray of hope connecting leptogenesis with the cp violation in neutrino oscillation @xcite and neutrinoless double beta decay processes . while the magnitude of cp violation is fairly known in the quark sector , it is completely shaded in the leptonic sector of the sm . therefore , searching for cp violation in the leptonic sector is of great interest in the present days . it has been pointed out that the dirac phase , being involved in the l - number conserving processes , can be measured in the long baseline neutrino oscillation experiments @xcite , while the majorana phase , being involved in the l - number violating processes , can be investigated in the neutrinoless double beta decay @xcite processes . at present the magnitude of b - asymmetry is precisely known , while the sign of this asymmetry is not known yet . however , by knowing the cp violating phases in the leptonic mixing matrix one can determine the sign of the b - asymmetry . this is the study taken up in this work . we consider a minimal extension of the sm by including two singlet right - handed neutrinos which are sufficient to explain the present knowledge of neutrino masses and mixings . we adopt a general parameterization of the neutrino dirac yukawa coupling and give the possible links between the cp violation in leptogenesis and neutrino oscillation , cp violation in neutrinoless double beta decay and leptogenesis . it is shown that the knowledge of low energy cp violating re - phasing invariants can indeed determine the sign of the b - asymmetry since the size of this asymmetry is known precisely . rest of the manuscript is arranged as follows . in section ii we elucidate the canonical seesaw in the framework of three right handed neutrinos . we then display the possible links between leptogenesis and the low energy cp - violating phases appearing in the leptonic mixing matrix in certain special circumstances . it is found that there are almost no links between these two phenomena occurring at two different energy scales . therefore , in section iii we give a parameterization of @xmath7 in the two right - handed neutrino models . in section iv we calculate the neutrino masses and mixings by using the parameterization of @xmath7 given in section iii . in section v we estimate the cp violation in leptogenesis . in section vi we consider the re - phasing invariant formalism to study the possible links between the cp violating phases responsible for leptogenesis and the cp violation at low energy phenomena . first we calculate th cp violation in neutrino oscillation and then elucidate its link to leptogenesis . after that we calculate the cp violation in low energy lepton number violating process , i.e. , the neutrinoless double beta decay , and then elucidate its link to leptogenesis . we conclude in section vii . to account for the small neutrino masses we extend the sm by including right - handed neutrinos . the corresponding leptonic lagrangian is given by @xmath8 where @xmath9 and @xmath10 runs from 1 to 3 , representing the left - handed fields . @xmath11 represent the right handed neutrino indices . @xmath12 represents the @xmath13 doublets , @xmath14 and @xmath15 are right - handed singlets of the theory . after the electroweak symmetry breaking the canonical seesaw @xcite gives the effective neutrino mass matrix @xmath16 where @xmath17 is the dirac mass matrix of the neutrinos with @xmath18 is the vev of sm higgs and that of @xmath19 is the mass matrix of right handed neutrinos . without loss of generality we consider @xmath19 to be diagonal and in this basis @xmath7 contains rest of the physical parameters that appears in @xmath20 . the diagonalization of @xmath21 , through the lepton flavor mixing matrix @xmath22 @xcite , gives us three masses of the physical neutrinos . its eigenvalues are given by @xmath23 where the masses @xmath24 are real and positive . the standard pdg parametrization @xcite of the pmns matrix reads : @xmath25 where @xmath26 and @xmath27 , @xmath28 stands for @xmath29 and @xmath30 respectively . the two physical phases @xmath31 and @xmath32 associated with the majorana character of neutrinos are not relevant for neutrino oscillations . thus we see that there are three phases in the low energy effective theory responsible for cp violation . however , these phases may not give rise to cp violation at high energy regime , in particular , leptogenesis to our interest . in the following we study this in the framework of three and than two right - handed neutrino models . in general if @xmath33 and @xmath34 are the number of generations of the left- and right - handed neutrinos that take part in the seesaw then the total number of parameters in the effective theory is estimated to be @xcite @xmath35 for @xmath36 and @xmath37 , @xmath38 and @xmath39 , which in the effective theory manifests as three masses of charged leptons , three masses of right - handed neutrinos and remaining 15 parameters including nine moduli and six phases in the dirac mass matrix @xmath7 in a basis where the charged lepton mass matrix is real and diagonal . in the bi - unitary parameterization the mass matrix @xmath7 can be given as @xmath40 where @xmath41 and @xmath42 are @xmath43 unitary matrices . @xmath41 diagonalizes the left - handed sector while @xmath42 is the diagonalizing matrix of @xmath44 . any arbitrary @xmath45 unitary matrix @xmath46 can be written as @xmath47 where @xmath48 is an overall phase and @xmath49 are phase matrices . @xmath50 is a ckm like matrix parametrized by three angles and one embedded phase . now using eq . ( [ udef ] ) in eq . ( [ para_step1 ] ) we get m_d = e^i(-_l+_r)p_2l^^ p_1l^m_d^diag p_1ru_rp_2r . [ para_step2 ] without loss of generality three of the left phases can be absorbed in the redefinition of charged lepton fields . as a result the effective dirac mass matrix turns out to be m_d = u_l^p_3 m_d^diag u_rp_2r , [ para_step3 ] where @xmath51 is an effective phase matrix . thus in the models with three right - handed neutrinos @xmath7 contains 15 parameters . in leptogenesis , the cp asymmetry comes in a form @xmath44 , which contains @xmath52 and @xmath53 , _ i.e. _ , m_d^m_d = p_2r^^(m_d^diag)^2 u_rp_2r , and hence is independent of @xmath54 and @xmath55 . although it would be good to know the exact relationship of the phases in @xmath52 and @xmath53 with the phases appearing in the @xmath22 matrix but that is not possible so , we try with some special cases . * case - i : * let us first consider the case , when @xmath53 is a diagonal matrix . this is the case when the right - handed neutrino majorana mass matrix is diagonal to start with . the mass matrix can still contain majorana phases . in that case , @xmath53 and @xmath56 will commute and hence @xmath44 will be real and there will not be any leptogenesis . this already tells us that the phases in leptogenesis crucially depends on the mixing of the right - handed physical neutrinos . even in this case there will be cp violation at low energy as we shall see below . the light neutrino mass matrix is given by @xmath57 so that the pmns matrix will become @xmath58 thus both the dirac and majorana phases at low energy are non - vanishing . * case - ii : * we shall now consider another special case when there is no leptogenesis . if the diagonal dirac neutrino mass matrix is proportional to a unit matrix , i.e. , @xmath59 ( @xmath60 is the identity matrix ) , again there is no leptogenesis , @xmath61 in this case the light neutrino mass matrix becomes @xmath62 so that the pmns matrix can be read off to be @xmath63 even in this case the dirac and majorana phases are present . thus in both these examples , even if cp violation is observed at low energy neutrino experiments , this cp violation may not be related to leptogenesis . since it is not possible to make any further progress with three heavy neutrinos , we shall now restrict ourselves to models with two heavy neutrinos . from now on we shall work with only two right - handed ( 2rh ) neutrinos . this result will be applicable when there are only two heavy neutrinos or when there are three heavy neutrinos but one of them do not mix with others and heavier than the other right - handed neutrinos and hence its contribution to the light neutrinos is also negligible . in the present case where we have @xmath36 and @xmath64 , from eq . ( [ mod ] ) and ( [ phs ] ) , we get @xmath65 and @xmath66 . the 14 parameters in the effective theory manifest them as three masses of charged leptons , two masses of right handed neutrinos and remaining nine parameters including six moduli and three phases appear in the dirac mass matrix @xmath7 . there are various textures and their phenomenological implications of @xmath7 in the 2rh neutrino models that have been considered in the literature @xcite . in this article a general parametrization of the @xmath67 mass matrix of the dirac neutrinos is considered . this is given by @xmath68 where @xmath46 is an arbitrary unitary matrix and the yukawa coupling of the two rh neutrino model is given as @xmath69 a derivation of eq . ( [ tri1 ] ) is given in the appendix [ appa ] . however , we declare that the texture of @xmath70 is not unique . by choosing appropriately the @xmath46 matrix one can place @xmath71 at different positions so as to get the different textures of @xmath70 as shown in appendix [ appb ] . using ( [ udef ] ) in eq . ( [ trip ] ) we get @xmath72 where @xmath50 contains four parameters including three moduli and one phase , @xmath73 contains two phases and @xmath70 contains four parameters including three moduli and one phase which all together makes ten parameters in @xmath7 . however , by multiplying the phase matrix @xmath73 with @xmath70 one can see that one of the phases in the phase matrix @xmath73 , i.e. , @xmath74 becomes redundant and can be dropped without loss of generality . as a result the total number of effective parameters is actually nine and hence consistent with our counting . substituting @xmath7 , given by eq . ( [ emd ] ) , in eq . ( [ seesaw ] ) we can calculate the effective neutrino mass matrix , @xmath20 . the diagonalization of @xmath21 , through the lepton flavor mixing matrix @xmath22 @xcite , then gives us two non - zero masses of the physical neutrinos while setting one of the mass to be exactly zero as shown in the following section . the unitary matrix @xmath50 , appearing in eq . ( [ emd ] ) , can be parameterized as . ] @xmath75 it turns out that this parameterization is useful in determining the leptonic mixing matrix in 2rh neutrino models . now from eqs . ( [ seesaw ] ) and ( [ emd ] ) we get the effective neutrino mass matrix to be @xmath76 where @xmath77 for simplicity of the calculation let us take @xmath78 common from 2nd row of @xmath70 matrix given by eq . ( [ tri1 ] ) and absorb it in @xmath79 by redefining @xmath80 as @xmath81 . as a result opposite phase will reappear with @xmath82 . then the matrix @xmath70 turns out to be @xmath83 using eq . ( [ tri2 ] ) in the above eq . ( [ xdef ] ) we get @xmath84 in writing the above equation we have used a diagonal basis of the rh neutrinos where @xmath85 . for simplicity , we absorb @xmath86 and @xmath87 in @xmath88 and @xmath82 as @xmath89 @xmath90 and @xmath91 . so @xmath92 can be rewritten as : @xmath93 looking to the effective neutrino mass matrix as given by eq . ( [ ss1 ] ) we can guess that the diagonalizing matrix would be of the form @xmath94 where @xmath95 is an unitary matrix . using eqs . ( [ diag ] ) and ( [ uk ] ) in eq . ( [ ss1 ] ) we see that @xmath96 which implies that @xmath95 would diagonalize the matrix @xmath97 . from the structure of @xmath92 it is clear that one of the light physical neutrinos must be massless . the matrix @xmath95 can be parameterized as @xmath98 where @xmath99 and @xmath100 with @xmath101\ , , \label{omeg}\end{aligned}\ ] ] and @xmath102\ , . \label{ph}\end{aligned}\ ] ] since @xmath103 diagonalizes the matrix @xmath92 the resulting diagonal matrix will have complex eigenvalues in general . however , by choosing appropriately the phases of @xmath104 one can make the eigenvalues of @xmath92 real . using eqs . ( [ omeg ] ) and ( [ ph ] ) we get the eigenvalues @xmath105 of @xmath92 to be @xmath106 the absolute masses of the physical neutrinos are then given by @xmath107 . the msw effect in the solar neutrino oscillation experiments indicates that @xmath108 . the corresponding mass scale , giving rise to the @xmath109 oscillation , is given by @xmath110 using eq . ( [ lambda ] ) in the above equation we get the solar neutrino mass scale to be @xmath111 ^ 2 - 4 a^4 c^4 \right\}^{1/2 } \nonumber \\ & \simeq & 8\times 10^{-5 } ev^2\ , . \label{sol - mass}\end{aligned}\ ] ] the atmospheric mass scale , on the other hand , is given by @xmath112 now using eq . ( [ lambda ] ) in the above equation we get the atmospheric mass scale to be @xmath113 these equations may be inverted to obtain @xmath114 now using eqs . ( [ p2 ] ) and ( [ rpar ] ) in eq . ( [ kpar ] ) we can rewrite the matrix @xmath95 as @xmath115 thus using eqs . ( [ krlab ] ) and ( [ up ] ) in eq . ( [ uk ] ) the pmns matrix @xmath22 is given as @xmath116 where @xmath117 in the above equation we have @xmath118\ , , \label{2 - 1}\\ \sin(\rho_{1}+\rho_{2})&=&\frac{\sin 2\omega \sin(\phi+\beta_{1 } ) } { \sin 2\theta_{12}'}\ , . \label{2 + 1}\end{aligned}\ ] ] for further simplification of the pmns matrix ( [ eff - pmns ] ) we now compute the matrix product @xmath119 which is given as @xmath120 now using eq . ( [ rkmult ] ) in eq . ( [ eff - pmns ] ) the @xmath22 matrix can be rewritten as : @xmath121 where @xmath122 is the ckm like matrix and @xmath123 is the majorana phase matrix . the effective @xmath124 violating phase in the @xmath122 matrix is given by @xmath125 note that in writing eq . ( [ rdfu ] ) the overall phase @xmath126 has been taken out . moreover , we absorb the unphysical phase matrix @xmath127 into the redefinition of charged lepton fields . from eqs . ( [ mns - matrix ] ) , ( [ up ] ) and ( [ dcp ] ) we see that , for the chosen parameterization of @xmath70 , two of the mixing angles @xmath128 and @xmath129 remains same as of the @xmath130 and @xmath131 mixing angles in pdg parameterization of the leptonic mixing matrix . thus we can write @xmath132 and @xmath133 . while @xmath134 gets modified to @xmath135 and is given by eq . ( [ thetap ] ) , the modified cp violating phase @xmath136 is given by eq . ( [ dcp ] ) . at present the best fit value of @xmath128 is given to be @xmath137 , while the best fit value with @xmath138 error the value of @xmath135 is given to be @xmath139 @xcite . the chooz experiment gives a bound on @xmath129 . currently the most conservative upper bound on @xmath129 at the @xmath140 confidence level is given to be @xcite @xmath141 which gives @xmath142 . the majorana mass of the rh neutrino violates l - number and hence is considered to be a natural source of l - asymmetry in the early universe @xcite provided its decay violates cp symmetry , a necessary criteria of sakharov @xcite . in a mass basis where the rh neutrinos are real and diagonal the majorana neutrinos are defined as @xmath143 . in this basis the cp asymmetry is given by @xmath144 where @xmath145 is the decay rate of @xmath146 . if we assume a normal mass hierarchy ( @xmath147 ) in the rh neutrino sector then the final l - asymmetry is given by the decay of the lighter rh neutrino , @xmath148 . the cp asymmetry parameter , arising from the decay of @xmath148 , is then given by @xmath149 ^ 2 } { ( m_d^\dagger m_d)_{11 } } \ , . \label{cpasym}\ ] ] using eqs . ( [ emd ] ) and ( [ tri1 ] ) in the above eq . ( [ cpasym ] ) we get @xmath150 from the above eq . ( [ cpasym-1 ] ) it is clear that if @xmath151 then there is no cp violation in leptogenesis . therefore , @xmath152 can be thought of the phase associated with @xmath153 in a basis where @xmath153 s are complex . moreover , @xmath152 always hangs around @xmath154 . so @xmath155 implies no leptogenesis . we will discuss more about it in sec.vi while we compare the cp violation in leptogenesis , neutrino oscillation and neutrinoless double beta decay processes . we now estimate the magnitude of @xmath4-asymmetry . a net l - asymmetry arises when @xmath156 fails to compete with the hubble expansion parameter , @xmath157 , where @xmath158 is the number of relativistic degrees of freedom at the epoch of temperature @xmath159 . in a comoving volume the @xmath4-asymmetry is defined as @xmath160 where @xmath161 is the dilution factor arises due to the competitions between @xmath156 and @xmath162 at @xmath163 . now using eq . ( [ cpasym-1 ] ) in the above eq . ( [ yl - def ] ) we get @xmath164 a part of the l - asymmetry is then transferred to the b - asymmetry via the sphaleron processes which are unsuppressed above the electroweak phase transition . taking into account the particle content in the @xmath165 the b- and l - asymmetries are related as @xmath166 where @xmath167 appropriate for the particle content in the @xmath165 . as a result we get the net b - asymmetry per comoving volume to be @xmath168 the observed b - asymmetry , on the other hand , is given by @xmath169 comparing the above eq . ( [ final - asy ] ) with the observed matter antimatter asymmetry , given by eq.([asymmetry ] ) , we get @xmath170 we have shown the allowed values of @xmath154 in fig . ( [ ytheta ] ) , using @xmath171 , for hierarchical rh neutrinos in the @xmath172 plane . it is shown in fig . [ ytheta](a ) that for @xmath173 the minimum allowed value of @xmath154 is @xmath174 . however , this value is lifted up to @xmath175 for @xmath176 as shown in fig . [ ytheta](b ) . it is convenient to study cp violation in a re - phasing invariant formalism . in particular , for the cp violation in the leptonic sector the latter makes it very interesting . the cp violation in any lepton number conserving processes comes out to be of the form @xcite @xmath177,\ ] ] where @xmath122 is the ckm like matrix in the lepton sector . on the other hand , cp violation in any lepton number violating processes will be of the form @xcite @xmath178\,.\ ] ] now one can have as many independent re - phasing invariant measures @xmath179 as many independent majorana cp phases . for three generations there are two independent @xmath179 s ( denoted as @xmath180 and @xmath181 ) and one j ( denoted as @xmath182 ) . for example , in the neutrinoless double beta decay the following re - phasing invariant will appear @xmath183 \sim t_{aij } t^\ast_{bij}.\ ] ] it has been shown that the re - phasing invariant cp violating quantity @xmath182 only appears in the neutrino oscillations and that of @xmath180 , @xmath181 appears in the neutrinoless double beta decay processes which may be observed in the next generation experiments . it has been pointed out that the dirac phase @xmath136 can be measured in the long baseline neutrino oscillation experiments @xcite . in that case the cp violation arises from the difference of transition probability @xmath184 . it can be shown that the transition probability @xmath185 is proportional to the leptonic jarlskog invariant @xmath186\ , . \label{jcp}\ ] ] using eq . ( [ rdfu ] ) the re - phasing invariant @xmath182 can be rewritten as @xmath187 now using eqs . ( [ omeg ] ) , ( [ ph ] ) , ( [ thetap ] ) and ( [ 2 + 1 ] ) in the above eq . ( [ jcp-1 ] ) we get @xmath188 ^ 2 - 4a^4c^4}}\nonumber\\ & \times & \left [ 2ab \cos\delta'_{13}\ { -c^2 \sin 2\theta \cos\beta_1+(a^2+b^2+c^2\cos2\theta)\sin\beta_1\}\right.\nonumber\\ & + & \left . 2ab\cos 2\theta_{12}\sin\delta'_{13}\ { ( a^2+b^2+c^2\cos 2\theta)\cos\beta_1+c^2\sin 2\theta \sin\beta_1\}\right.\nonumber\\ & + & \left.\sin \delta'_{13}\sin 2\theta_{12}(-a^4+b^4+c^4 + 2b^2 c^2 \cos 2\theta ) \right]\ , . \label{jcp-2}\end{aligned}\ ] ] from the above eq . ( [ jcp-2 ] ) it is obvious that @xmath189 only if both @xmath190 and @xmath191 , while only @xmath191 ( equivalently @xmath155 ) implies the condition for no leptogenesis " . this indicates that there is no one - to - one correspondence between the cp violation in neutrino oscillation and the cp violation in leptogenesis , even in the 2rh neutrino models . however , it is interesting to see the common regions in the plane of @xmath192 versus @xmath182 . this is shown in fig . ( [ jcp - fig ] ) by taking a typical set of parameters . the main aim is to illustrate the maximal contrast between the positive and negative values of @xmath193 for a given set of values of @xmath182 . this helps us in determining the sign of the asymmetry by knowing the size of the asymmetry . from the fig . ( [ jcp - fig ] ) it is obvious that for the given set of parameters the positive sign of the asymmetry allows the values of @xmath182 in the range @xmath194 for @xmath173 . however , this range is significantly reduced for @xmath176 . on the other hand , the negative sign of the asymmetry allows the values of @xmath182 in the range @xmath195 for @xmath173 which is further reduced for @xmath176 . in this figure the value of @xmath134 is used from fig . ( [ theta12-theta ] ) where we have shown the allowed values of @xmath134 as @xmath152 varies from @xmath196 to @xmath197 . note that the above results are true for a non - zero @xmath129 . consequently the allowed range of values of @xmath182 may vary depending on the values of @xmath129 . thus we anticipate that in the 2rh neutrino models a knowledge of @xmath182 can predict the sign of matter antimatter asymmetry of the universe . we should note that the predictive power of the model depends on the cp violating phases @xmath198 and @xmath199 . this can be visible from fig . ( [ jcpvalues - fig ] ) where we have shown the variation of @xmath193 with @xmath182 for different values of @xmath198 and @xmath199 . in particular , for the choice ( @xmath200 , @xmath201 ) and ( @xmath202 , @xmath203 ) , the contrast between the positive and negative values of @xmath193 is almost zero for a given set of values of @xmath182 . on the other hand , for the choice ( @xmath200 , @xmath203 ) and ( @xmath202 , @xmath201 ) , the contrast between the positive and negative values of @xmath193 is maximal and can be chosen for the present purpose . the observation of the neutrinoless double beta decay would provide direct evidence for the violation of total @xmath4-number in the low energy effective theory and hence probing the left - handed physical neutrinos to be majorana type . note that the @xmath4-number violation at high energy scale is a necessary criteria for leptogenesis . in the canonical seesaw models this is conspired by assuming that the rh neutrinos are majorana in nature . however , this assumption does nt ensure that the left - handed physical neutrinos are majorana type . assuming that the physical neutrinos are of majorana type we investigate the connecting links between the two @xmath4-number violating phenomena occurring at two different energy scales . in the low energy effective theory with three generations of left - handed fermions , apart from the @xmath182 , one can write two more re - phasing invariants @xmath180 and @xmath181 which designates lepton number violation and cp violation @xcite . however , in the models with two rh neutrinos one of the eigen values of the physical light neutrino mass matrix is exactly zero . therefore , the corresponding phase in the majorana phase matrix can always be chosen so as to set one of the lepton number violating cp violating re - phasing invariant to zero . in the present case @xmath204 and hence the corresponding phase is arbitrary . this is ensured through @xmath74 which is redundant and pointed out in eq . ( [ emd ] ) . therefore , from eq . ( [ rdfu ] ) we can write the only @xmath4-number violating cp violating re - phasing invariant as : @xmath205\nonumber\\ & = & -\frac{1}{2}\sin 2\theta'_{12 } \cos^2\theta_{13 } \sin(\rho_2-\rho_1 + { ( \eta_2-\eta_1)\over 2}-\beta_1)\ , . \label{j_lep}\end{aligned}\ ] ] using eq . ( [ rkmult ] ) the above eq . ( [ j_lep ] ) can be rewritten as @xmath206}\nonumber\\ & \times & \left [ \sin 2\theta_{12 } \cos \theta \{-c^2 \sin 2\theta \cos \beta_1+(a^2+b^2+c^2 \cos 2\theta ) \sin \beta_1 \}\right.\nonumber\\ & & \left . \times \sqrt{(a^2+b^2)^2+c^4 + 2c^2a^2 + 2b^2c^2 \cos 2\theta}\right . \nonumber\\ & + & \left . \sin 2\theta_{12 } \sin \theta \{c^2 \sin 2\theta \sin \beta_1 + ( a^2+b^2+c^2 \cos 2\theta ) \cos \beta_1 \}\right.\nonumber\\ & & \left . \times \frac{(-a^4+b^4+c^4 + 2 b^2c^2 \cos 2\theta)}{\sqrt{(a^2+b^2)^2 + c^4 + 2 c^2a^2 + 2b^2c^2 \cos 2\theta}}\right.\nonumber\\ & + & \left . \cos 2\theta_{12}\sin \theta \frac{2ab\{(a^2+b^2)^2+c^4 + 2b^2c^2 \cos 2\theta + 2c^2a^2 \cos 2\theta \}}{\sqrt{(a^2+b^2)^2+c^4 + 2c^2a^2 + 2b^2c^2 \cos 2\theta } } \right]\ , . \label{j-2-final}\end{aligned}\ ] ] in the above eq . ( [ j-2-final ] ) the allowed values of @xmath134 is obtained from @xmath207 \right]^{1/2}\ , , \label{theta12_prime}\end{aligned}\ ] ] by fixing @xmath208 . this is shown in fig . ( [ theta12-theta ] ) . from eq . ( [ j-2-final ] ) one can see that @xmath209 as @xmath210 which is the condition for no leptogenesis " . thus we see that there is no one - to - one correspondence between the two @xmath4-number violating processes occurring at two different energy scales . however , it is always interesting to see the overlapping regions in the plane of @xmath211 versus @xmath212 as @xmath152 varies from @xmath196 to @xmath197 . this is shown in fig . ( [ j2-asy ] ) for a typical set of parameters . from fig . ( [ j2-asy ] ) one can see that for positive sign of the @xmath5-asymmetry the values of @xmath212 lie in between @xmath213 to @xmath214 for @xmath173 . this range is further reduced to @xmath215 for @xmath176 . on the other hand , for the negative sign of the @xmath5-asymmetry the values of @xmath212 lie in the range @xmath216 for @xmath173 and in the range @xmath217 for @xmath176 . thus we see that within the allowed range of parameters the contrast between the positive and negative values of @xmath211 is maximum for a given set of values of @xmath212 . therefore , we expect a knowledge of @xmath212 can precisely determine the sign of @xmath5-asymmetry since the value of @xmath193 is known . finally we note that , unlike @xmath182 , @xmath212 remains non - vanishing even if @xmath218 . now the remaining question to be addressed is how @xmath193 varies with respect to @xmath212 for different values of @xmath198 . this is shown in fig . ( [ jvalues - asy ] ) for a given set of parameters . one can see that for @xmath202 and @xmath219 both positive and negative values of @xmath193 correspond to the same set of values of @xmath212 which is unwelcome for determination of sign of the asymmetry . on the other hand for @xmath220 one can have maximal contrast between the positive and negative values of @xmath193 for the given set of values of @xmath212 and hence can be chosen for the present purpose . we have studied the connecting links between the cp violating phase(s ) giving rise to leptogenesis , occurring at a high energy scale , and the cp violating phases appearing in the low energy phenomena , i.e. , neutrino oscillation and neutrinoless double beta decay processes . this is studied in the framework of two right - handed neutrino models . the low energy leptonic cp violation is studied in a re - phasing invariant formalism . it is shown that there are only two re - phasing invariants ; ( 1 ) the lepton number conserving cp violating re - phasing invariant @xmath182 which can be determined in the future long - baseline neutrino oscillation experiments , ( 2 ) the lepton number violating cp violating re - phasing invariant @xmath212 which can be determined in the neutrinoless double beta decay experiments . it is found that there is no one - to - one correspondence between these two cp violating phenomena , occurring at two different energy scales , even though the number of parameters involving in the seesaw is exactly same as the number of low energy observable parameters . however , in a suitable parameter space we have shown that the overlapping regions in the plane of @xmath193 versus @xmath182 and @xmath193 versus @xmath212 can indeed determine the _ sign _ of the matter antimatter asymmetry of the present universe assuming that the _ size _ of the asymmetry is precisely known . to parameterize the neutrino dirac yukawa coupling in two right - handed neutrino models we follow the same procedure adopted in ref . @xcite . let @xmath221 , @xmath222 , @xmath223 be three orthonormal 3 dimensional vectors . using these basis vectors we can write the most general unitary matrix @xmath46 as : @xmath224 let us consider an arbitrary @xmath225 matrix @xmath226 which in terms of the 3-dimensional vectors @xmath227 and @xmath228 can be written as : @xmath229 without loss of generality we choose @xmath230 . as a result we get @xmath231 where @xmath232 . let @xmath122 be another unitary matrix which we choose to be of the form : @xmath233 where @xmath234 must follow @xmath235 and @xmath236 . consequently we have @xmath237 where we set @xmath238 by imposing the unitarity condition of @xmath122 . this implies that we can always write any arbitrary @xmath67 matrix @xmath239 where @xmath240 is an unitary matrix and the texture of @xmath70 , the yukawa coupling in the two right handed neutrino mass models , is given as @xmath241 where @xmath242 , @xmath154 , @xmath82 and @xmath152 are real numbers . note that by appropriately choosing the @xmath46 and @xmath122 matrices one can construct the @xmath70 matrix in twelve possible ways . in this appendix we specify the various possible textures of @xmath70 . one of the particular texture of @xmath70 has been used in section [ sec2 ] for our work . in the table - i we write all the possible textures of @xmath70 . i & @xmath243 @xmath244 & @xmath245 @xmath246 + + ii & @xmath247 @xmath248 & @xmath249 @xmath250 + + iii & @xmath251 @xmath252 & @xmath253 @xmath254 + each possible @xmath70 in table - i will lead to various forms of @xmath92 , apparent from eq . ( [ xdef ] ) . accordingly the neutrino masses and mixing angles will be modified through the @xmath7 parameters . it is our pleasure to thank prof . anjan joshipura for helpful discussions . 99 q.r . ahmed _ et al _ ( sno collaboration ) , phys . lett . * 89 * , 011301 - 011302 ( 2002 ) ; j.n . bahcall and c. pena - garay , [ arxiv : hep - ph/0404061 ] . s. fukuda _ et al _ ( super - kamiokande collaboration ) , phys . rev . lett . * 86 * , 5656 ( 2001 ) . k. eguchi _ et al _ ( kamland collaboration ) , phys . rev . lett . * 90 * , 021802 ( 2003 ) . p. minkowski , phys . lett . * b 67 * , 421 ( 1977 ) ; m. gell - mann , p. ramond and r. slansky in _ supergravity _ ( p. van niewenhuizen and d. freedman , eds ) , ( amsterdam ) , north holland , 1979 ; t. yanagida in _ workshop on unified theory and baryon number in the universe _ ( o. sawada and a. sugamoto , eds ) , ( japan ) , kek 1979 ; r.n . mohapatra and g. senjanovic , phys . . lett . * 44 * , 912 ( 1980 ) ; j. schechter and j.w.f . valle , phys . * d 22 * , 2227 ( 1980 ) . luty , phys . d*45 * , 455 ( 1992 ) ; r.n . mohapatra and x. zhang , phys . rev . d*46 * , 5331 ( 1992 ) ; a. acker , h. kikuchi , e. ma and u. sarkar , phys . * d 48 * , 5006 ( 1993 ) ; m. flanz , e.a . paschos and u. sarkar , phys . lett . * b 345 * , 248 ( 1995 ) ; m. flanz , e.a . paschos , u. sarkar and j. weiss , phys . lett . * b 389 * , 693 ( 1996 ) ; m. plumacher , z. phy . c74(1997)549 ; w. buchmuller , p. di bari and m. plumacher , annals phys . * 315 * , 305 ( 2005),[arxiv : hep - ph/0401240 ] ; j. faridani , s. lola , p.j . odonnell and u. sarkar , eur . phys . jour . * c 7 * , 543 ( 1999 ) ; r. barbieri , p. creminelli , a. strumia and n. tetradis , nucl . b * 575 * , 61 ( 2000 ) , [ arxiv : hep - ph/9911315 ] ; g. f. giudice , a. notari , m. raidal , a. riotto and a. strumia , nucl . b * 685 * , 89 ( 2004 ) , [ arxiv : hep - ph/0310123 ] ; n. sahu and u. sarkar , [ arxiv : hep - ph/0605007 ] . spergel _ et al _ astrophys.j.suppl . 148 ( 2003 ) 175 [ astro - ph/0302209 ] a. s. joshipura , e. a. paschos and w. rodejohann , jhep * 0108 * ( 2001 ) 029 [ arxiv : hep - ph/0105175 ] . s. pascoli , s. t. petcov and w. rodejohann , phys . d * 68 * , 093007 ( 2003 ) [ arxiv : hep - ph/0302054 ] . j. r. ellis and m. raidal , nucl . b * 643 * ( 2002 ) 229 [ arxiv : hep - ph/0206174 ] . g. c. branco , t. morozumi , b. m. nobre and m. n. rebelo , nucl . b * 617 * ( 2001 ) 475 [ arxiv : hep - ph/0107164 ] . n. sahu and s. uma sankar , nucl . b * 724 * ( 2005 ) 329 [ arxiv : hep - ph/0501069 ] . m. c. chen and k. t. mahanthappa , phys . d * 71 * ( 2005 ) 035001 [ arxiv : hep - ph/0411158 ] . m. n. rebelo , phys . d * 67 * , 013008 ( 2003 ) [ arxiv : hep - ph/0207236 ] . t. endoh , s. kaneko , s. k. kang , t. morozumi and m. tanimoto , phys . * 89 * ( 2002 ) 231601 [ arxiv : hep - ph/0209020 ] . p. h. frampton , s. l. glashow and t. yanagida , phys . b * 548 * ( 2002 ) 119 [ arxiv : hep - ph/0208157 ] . m. tanimoto , phys . d * 55 * , 322 ( 1997 ) [ arxiv : hep - ph/9605413 ] ; h. minakata and h. nunokawa , phys . d * 57 * , 4403 ( 1998 ) [ arxiv : hep - ph/9705208 ] ; s. m. bilenky , c. giunti and w. grimus , phys . d * 58 * , 033001 ( 1998 ) [ arxiv : hep - ph/9712537 ] ; j. arafune , m. koike and j. sato , phys . d * 56 * , 3093 ( 1997) [ erratum - ibid . d * 60 * , 119905 ( 1999 ) ] [ arxiv : hep - ph/9703351 ] ; s. pascoli , s. t. petcov and w. rodejohann , phys . b * 549 * , 177 ( 2002 ) , [ arxiv : hep - ph/0209059 ] ; s. m. bilenky , s. pascoli and s. t. petcov , phys . d * 64 * , 053010 ( 2001 ) [ arxiv : hep - ph/0102265 ] . z. maki , m. nakagawa and s. sakata , prog . * 28 * , 870 ( 1962 ) . b. pontecorvo , sov . jetp * 7 * , 172 ( 1958 ) [ zh . * 34 * , 247 ( 1957 ) ] . b. pontecorvo , sov . jetp * 6 * , 429 ( 1957 ) [ zh . * 33 * , 549 ( 1957 ) ] . particle physics data group , s. eidelman _ et al._,phys . b * 592 * , 1 ( 2004 ) . a. broncano , m. b. gavela and e. jenkins , phys . b * 552 * , 177 ( 2003 ) [ arxiv : hep - ph/0210271 ] . a. ibarra and g. g. ross , phys . b * 591 * ( 2004 ) 285 [ arxiv : hep - ph/0312138 ] . a. ibarra , jhep * 0601 * , 064 ( 2006 ) [ arxiv : hep - ph/0511136 ] . g. c. branco , m. n. rebelo and j. i. silva - marcos , phys . b * 633 * , 345 ( 2006 ) [ arxiv : hep - ph/0510412 ] . see for recent review : r.n . mohapatra and a. smirnov , [ arxiv : hep - ph/0603118 ] . c. jarlskog , phys . rev . lett . * 55 * , 1039 ( 1985 ) . j. f. nieves and p. b. pal , phys . d * 36 * , 315 ( 1987 ) ; y. liu and u. sarkar , mod . a * 16 * , 603 ( 2001 ) [ arxiv : hep - ph/9906307 ] ; p. j. odonnell and u. sarkar , phys . d * 52 * , 1720 ( 1995 ) [ arxiv : hep - ph/9305338 ] . t. morozumi , t. satou , m. n. rebelo and m. tanimoto , phys . b * 410 * ( 1997 ) 233 , [ arxiv : hep - ph/9703249 ] ; t. fijihara , s. kaneko , s. kang , d. kimura , t. morozumi and m. tanimoto , phys . d * 72 * , 016006 ( 2005 ) , [ arxiv : hep - ph/0505076 ] .
an attractive explanation for non - zero neutrino masses and small matter antimatter asymmetry of the present universe lies in leptogenesis " . at present the _ size _ of the lepton asymmetry is precisely known , while the _ sign _ is not known yet . in this work we determine the sign of this asymmetry in the framework of two right handed neutrino models by relating the leptogenesis phase(s ) with the low energy cp violating phases appearing in the leptonic mixing matrix . it is shown that the knowledge of low energy lepton number violating re - phasing invariants can indeed determine the sign of the present matter antimatter asymmetry of the universe and hence indirectly probing the light physical neutrinos to be majorana type .
the sites where the subsurface magnetic flux tubes emerge on solar surface are called emerging flux regions @xcite . the typical simple emerging flux region has one main pair of the opposite magnetic concentrations at the both ends of the emerging site . the opposite magnetic concentrations of the main pair move away from each other with the speed of 1 - 2 km@xmath1 @xcite in the developing phase of the region . inside the developing emerging site there are many magnetic flux tube emerging on the photosphere . the newly emerged magnetic flux tubes at the emerging site appear as the dark granular lanes @xcite on the photosphere . at the both ends of the dark granular lane , there are the magnetic concentrations which are called the footpoints and observed as the bright points in g band image @xcite . with h@xmath2 the emerged flux tubes are observed as dark arch filaments @xcite . the lifetime of an arch filament is 10 - 30 minutes @xcite . the rise velocity of arch filaments is 10 - 15 km s@xmath1 @xcite . @xcite found the deceleration of the apex of the small - scale arch filaments in the chromosphere . emerging flux regions show a variety of size , lifetime , total magnetic flux and field strength . especially the small emerging flux regions are called ephemeral active region ( ear ; @xcite ) . for convenience , in the following page we define larger ( i.e. non - ear ) emerging flux regions as efrs . efrs are produced by fairly large - scale flux emergence . they have a pair or more complex group of sunspots with definite penumbrae . the typical size of efrs is more than 30 mm @xcite . efrs show their emergence activities for several days and exist on the solar surface for a few month at the maximum . the total flux in an efr increases with the rate of @xmath3 mx hr@xmath1 and reaches to the order of @xmath3-@xmath4 mx @xcite . in the fairly developed efr the field strength of main spot is around 3,000 gauss @xcite . ears have simple bipolar configuration . they have no penumbra in the sunspots . the typical size varies from 5 mm to 30 mm @xcite . ears have short life time of hours or one day . the total flux in an ear is up to @xmath3 mx with the increase rate of @xmath5 mx hr@xmath1 . the magnetic field strength of main spot is from a few times 100 gauss @xcite to 2,000 gauss @xcite . various simulation studies on flux emergence have been performed by many researchers @xcite . they showed the simple bipolar emergence simulation which corresponds to the observation result . @xcite and @xcite simulated the three - dimensional magnetohydrodynamics ( mhd ) of the emerging magnetic flux . @xcite and @xcite also performed the mhd simulations of flux emergence with the sheared or twisted flux tube . they found that the flux tube with shear or twist emerges faster than that without any shear and twist . @xcite performed the mhd simulation of flux emergence in a sheet geometry . the initial stable flux sheet in the convective zone was perturbed with various wavelengths , which correspond to the convective motion . they found that irrespective of the wavelength of initial perturbation , a finite `` most unstable wavelength '' is excited . this wavelength ( 2 - 4 mm ) is inherent in the parker instability @xcite . as a result , the flux sheet is undulated and the apexes of the convex field line ( -loops ) appear consecutively on the photosphere . some of dipped field lines ( -loops ) also emerge to form the regions called `` bald patches @xcite '' . on the other hand , observational study on a large efr and bald patches was performed by @xcite . they observed a fairly large ( @xmath6 mm ) efr with the magnetogram and found that the emerged field lines undulate vertically . they revealed that there are many bald patches between the main spots . these results confirmed that the emerging flux tube does not rise altogether at a time , but each -loop component rises individually . they proposed this model as `` resistive emergence model '' . the distance between two consecutive bald patches is in the range of 2 - 6 mm , which is consistent with a theoretical argument on the flattening of emerging magnetic field just below the surface and its critical emergence length , first presented by @xcite . recently @xcite further developed the simulation performed by @xcite and obtained the result in which the undulated field line caused reconnections with neighboring -loops and created larger loops . @xcite performed the three - dimensional mhd simulation of the emergence of undulating fieldlines . these reconnection events are interpreted as a sources of ellerman bombs @xcite . the resistive emergence model is applicable to the large efr . furthermore , recently @xcite found bald patches inside a small - scale ear ( @xmath7 mm ) using solar optical telescope ( sot ; @xcite ) aboard _ hinode _ @xcite . however , @xcite and @xcite showed small ears ( 2 mm and 6 mm , respectively ) without undulated magnetic field . as stated above , the criteria of the bald patches formation are still uncertain . furthermore , in the latest simulation , the footpoints of emerged flux loops showed converging motion toward the bald patches on the photosphere @xcite . although this converging motion was observed in preceding studies @xcite , statistical analysis on that phenomenon with respect to the size and magnetic characteristics of the flux emergences has not been done yet . to clarify the criteria of forming bald patch and converging flow , we performed a statistical study about the nature of magnetic flux emergences using sot . the flux emergence phenomena from small ears to large efrs observed by sot were investigated on their morphological and magnetical characteristics . furthermore , we derived the the relations between the total magnetic flux and the maximum spatial size , the flux growth rate and the mean separation speed of the emergence event to clarify how the total flux amount controls the entire evolution of the emergence . _ hinode _ satellite has observed solar surface for over 4 years with sot . sot has broadband filter imager ( bfi ) and narrowband filter imager ( nfi ) . ca h ( 3968.5 ) filtergrams were taken by bfi with the bandpass of 3 . fe ( 6302 ) and na d ( 5896 ) polarimetric data in solar photosphere and chromosphere were observed with nfi . to search for the emerging flux phenomena , we used _ hinode _ daily quicklook movies . among the possible candidates , we selected 101 emerging flux phenomena according to the criteria as follows : ( 1 ) conspicuous presence of separating bright points in ca h image and/or opposite polarities in fe or na d image , ( 2 ) location fairly inside the solar limb ( @xmath8 , where @xmath9 is the distance from solar disk center to the location of the event and @xmath10 is the solar radius ) ( 3 ) observational time span longer than 1 hour , and ( 4 ) observational cadence higher than 10 minutes . the period in which we studied is between 2006 november 26 to 2010 august 23 . the detailed observational information of the emerging flux is available in the associated electronic tablesotsuji / electronic_table.pdf . after publication the table will be uploaded in pasj site . ] . in this section we give the description of the data reduction with the data of efr 20061201 ( # 2 in the electronic table ) as an example . first , dark - current subtraction and flat fielding were performed on the obtained sot data in the standard manner . then we processed every observed data as described below . first , we used _ soho_-michelson doppler imager ( mdi ; @xcite ) magnetogram data to calibrate sot polarimetric data . we compensated the differential rotation of two consecutive mdi data observed before and after the sot observation and interpolated them by time to estimate the distribution of magnetic field at the time of sot observation . next we deteriorated sot polarimetric data with the spatial resolution of mdi magnetogram ( @xmath11 ) . using the deteriorated sot polarimetric data and mdi magnetogram , we made a scatter plot of sot polarimetric signal ( stokes @xmath12 ) to mdi field strength ( figure [ fig:02 ] ) . the correlation coefficient of sot @xmath12 to mdi field strength was 0.95 . we performed linear fitting on the scatter plot and obtained the conversion equation from sot @xmath12 to the photospheric field strength @xmath13 , expressed as @xmath14 in the case of figure [ fig:02 ] , the offset value @xmath15 and the scaling factor @xmath16 were 4.029 gauss and 12528 gauss , respectively . note that the scatter of data points in the figure is mainly due to the doppler effect arising from the satellite s orbital motion . with this method we converted the sot polarimetric signals to the magnetic flux densities for all the samples . ( 80mm,45mm)fig01a.eps ( 80mm,45mm)fig01b.eps ( 80mm,45mm)fig01c.eps ( 80mm,80mm)fig02.eps in measuring the actual size of the solar features , we compensated the projection effect . then we applied the subsonic filter of 3 minutes to the sot ca h image sequence in order to suppress chromospheric oscillatory motions ( figure [ fig:03]b ) . as is stated in the introduction , mutually approaching anti - polarity patches were observed in undulating resistive emerging phenomena @xcite . we studied the formation process of the converging motion between the opposite polarities in efr and ear by analysing the morphological , dynamical and magnetic evolutions of our sot samples . in this section we introduce our analysis methods using the data of efr 20061201 . the morphological evolution of the magnetic flux emergence was analysed with two methods . one is the method of tracking magnetic elements using local correlation tracking ( lct ) , and the other using time - sliced diagram . the local correlation tracking ( lct ) method is commonly used to derive the horizontal velocity field @xcite . for lct , we used ` flowmap.pro ` in ssw of idl . ` flowmap.pro ` calculates the two dimensional vector flowfield by following the subtiles in the time series of two dimensional images . to examine the motion of the footpoints of flux tubes for all over the emerging site , we performed lct on sot magnetogram data and obtained velocity field of moving magnetic elements ( figure [ fig:03]c ) . the size of tracking subtile for lct was . to reduce the velocity noise due to lct error , the velocity fields were averaged both spatially and temporally over the zones of @xmath17 and 10 minutes , respectively . the standard deviation of the velocity field inside the data cube of @xmath18 minutes was @xmath19 km s@xmath1 . then we derived divergence of horizontal velocity field ( figure [ fig:03]d ) . to emphasize global and sustained flows , the divergence maps were averaged both spatially and temporally . the spatial average was performed with the width of for large efr and for small ear . the threshold between large efrs and small ears was fixed to be , which is the upper limit size of ephemeral active regions indicated by @xcite . for efr 20061201 , the spatial averaging box size was . the temporally averaging period was taken as 10 minute for all the events . ( 80mm,50mm)fig03a.eps ( 80mm,50mm)fig03b.eps ( 80mm,50mm)fig03c.eps ( 80mm,50mm)fig03d.eps ( 80mm,50mm)fig03e.eps ( 80mm,50mm)fig03f.eps to clarify the dynamics of footpoints more quantitatively , we made time - sliced diagram of sot magnetograms ( figure [ fig:03]e ) . the spatial slit was located parallel to the axis of the efr ( shown in figure [ fig:03]a ) . from the time - sliced diagram , we derived the maximum distance @xmath20 between the main spots , the mean separating speed @xmath21 of the main spots . detailed footpoints motions , such as mutually approaching anti - polarities , were examined in this diagram . to investigate the temporal evolution of emerging magnetic flux , we measured the total flux within the emerging site . positive and negative fluxes were summed up separately . the total flux of the emerging region @xmath22 was derived by subtracting the fluxes at the initial time . we plotted the variation of positive and negative fluxes , respectively ( figure [ fig:03]f ) . from this plot , we derived the maximum amount of unsigned total magnetic flux @xmath23 and the unsigned flux growth rate @xmath24 . the growth rate is defined as the total magnetic flux @xmath23 divided by the continuously emerging period @xmath25 . note that the unbalance between positive and negative fluxes in the sample plot was due to the flow of the positive flux out of the field of view . first , we introduce the sample results for large efr and small ear at section [ sec:4.1 ] and [ sec:4.2 ] , respectively . then the statistical results are shown in section [ sec:4.3 ] . in figure [ fig:03]a , there are two main spots aligned east - west direction . the size of the main spots was - . the field strength of the spots was @xmath26 gauss at the maximum . although the following spot is missing in the ca h image because of the field of view limitation ( figure [ fig:03]b ) , there are sunspots locating at the same position with the preceding negative spots in the magnetogram . the velocity field shows prominent outward motions of the main spots ( figure [ fig:03]c ) . there are positive divergence areas inside the main spots ( figure [ fig:03]d ) , which indicates the flux emergences . from the time - sliced diagram ( figure [ fig:03]e ) , we can estimate the speed of main spots to be @xmath27 km s@xmath1 for each , thus the mean separating speed @xmath21 is about 0.6 km s@xmath1 . at the end of the observation period , the distance between two main spots increased to around . we considered this value as the maximum distance @xmath20 . the emergence started at 16:00 ut on 1 december and lasted until the observation end at 24:00 ut on 2 december . thus the active emergence period was taken as 32 hours . the maximum amount of total magnetic flux were @xmath28 mx for positive polarity and @xmath29 mx for negative polarity ( figure [ fig:03]f ) . as the following positive spot flowed out of the field of view , measured positive flux was less than that of negative one . thus we took the maximum negative flux as @xmath23 . mean flux growth rate @xmath30 for this event was @xmath31 mx hr@xmath1 . let us look at the central part of the region where there are many small positive or negative magnetic patches ( figure [ fig:03]a ) . these patches correspond to the ca h bright points ( figure [ fig:03]b ) . although the magnetic patches seem to move with apparently random velocities in figure [ fig:03]c , these patches are located in the converging region of divergence map ( figure [ fig:04]d ) . thus these magnetic patches accumulated and stagnated to the localized area . in fact , the time - sliced diagram shows that these magnetic patches actually approach to each other with speed of @xmath32 km s@xmath1 . so we identify these mutually approaching area as a -loop formation in the efr . in the following , we denote the area as the `` stagnation zone ( sz ) '' , where the anti - polarities mutually approach and accumulate . we selected a small - scale magnetic emergence event which emerged on 13 february 2007 ( # 9 in the electronic table ) as the sample case of ear . figure [ fig:04]a shows the magnetogram of the region . two magnetic concentrations in the magnetogram correspond to the ca h bright points in figure [ fig:04]b . the field strength of the two magnetic concentrations was @xmath33 gauss at the maximum . the velocity field of the small ear shows separative and anti - clockwise rotational motion of two magnetic concentrations ( figure [ fig:04]c ) . figure [ fig:04]d is the divergence map derived from the velocity field and averaged spatially with . the divergence map shows the positive area at the central region of the ear , which indicates that there is no converging flow inside the emerging site . in the time - sliced diagram we can see the simple separating motion of two main magnetic concentrations and no stagnation zone ( figure [ fig:04]d ) . the maximum distance between the main concentration @xmath20 was . the mean speed of separating motion for main concentrations @xmath34 is about 1.5 km s@xmath1 . the temporal evolution of total flux shows that the maximum amounts of total fluxes are @xmath35 mx for positive polarity and @xmath36 mx for negative polarity ( figure [ fig:04]f ) . we took the maximum positive flux as @xmath23 for this event . the time - sliced diagram and the total fluxes evolution plot indicate that the flux emergence began at 03:15 ut and last until the observation end at 04:30 ut . thus the active emergence period was 75 minutes and the mean flux growth rate @xmath30 was @xmath37 mx hr@xmath1 . ( 80mm,45mm)fig04a.eps ( 80mm,45mm)fig04b.eps ( 80mm,45mm)fig04c.eps ( 80mm,45mm)fig04d.eps ( 80mm,45mm)fig04e.eps ( 80mm,45mm)fig04f.eps table [ tab:01 ] in appendix [ seq : a1 ] shows the measured quantities of all the samples . if there was no magnetic observation , ca h data were used to derive @xmath20 and @xmath21 , and to judge the existence of szs . there is no data of the total flux @xmath23 and the flux growth rate @xmath30 for the observations without the magnetogram . figure [ fig:05 ] shows the statistical characteristics of measured quantities . ( 80mm,50mm)fig05a.eps ( 80mm,50mm)fig05b.eps ( 80mm,50mm)fig05c.eps ( 80mm,50mm)fig05d.eps ( 80mm,50mm)fig05e.eps ( 80mm,50mm)fig05f.eps in figure [ fig:05]a , the histogram of the maximum separation distance @xmath20 indicates that szs are rarely found in small @xmath20 regions . the flux emergence phenomena with the separation size of more than 25 mm always have szs . below the threshold of 25 mm , there are both efrs and ears with / without szs . to clarify the conditions to have szs , we categorized all the events to three groups as follow . group i : maximum separation distance @xmath38 mm and with szs . group ii : maximum separation distance @xmath39 mm and with szs . group iii : maximum separation distance @xmath39 mm and without sz . group i , ii and iii are indicated in the figure [ fig:05 ] by white , gray and black bars / squares , respectively . figure [ fig:05]b shows the histogram of the maximum fluxes @xmath23 , where we present the maximum flux separately for these three groups . group i has the maximum flux of @xmath40 mx , while group ii and iii have @xmath41 mx and @xmath42 mx , respectively . figure [ fig:05]c shows the relation between the maximum flux @xmath23 and the maximum separation distance @xmath20 . in the plot the three group i , ii and iii are clearly separated . the maximum separation distance @xmath20 depends on the maximum flux amount @xmath23 . the scatter plot implies a power - law relation of @xmath43 where @xmath20 is in mm and @xmath23 is in mx . we can see the trend in which the small ears have low values of maximum flux while large efrs have high values of maximum flux . the power - law relation is consistent with the result of @xcite , although the index of power was 0.18 instead of 0.27 . from the result of categorization , we found that the sz features are associated with the magnetic flux emergence of more than around @xmath0 mx . we also measured the typical size of elementary structures of emergence @xmath44 defined as the distance between two footpoints of individual -loops at their emergence period . some small footpoints of emerged loops might be finally transported to the border of supergranules by the local convection . others cause cancellation between the opposite polarities and disappear , which enlarges the distance between two footpoints of individual -loops , ( i.e. @xmath44 ) . thus the @xmath44 varies with time . for the accurate descriptions , we adopted the @xmath44 at the epoch when the -loops were observed as ca h filaments ( @xmath4510 minutes after the start of the emergence ; @xcite ) . figure [ fig:05]d shows the scatter plot of @xmath20 and @xmath44 , which suggest that @xmath44 takes the values in the range of 2 - 6 mm irrespective of @xmath20 . thus elementary and unitary -loops in any emerging flux region has a common size of around 4 mm , which is consistent with the most unstable wavelength ( 2 - 4 mm ) of parker instability and preceding observation / simulation studies @xcite . the relation between the maximum flux @xmath23 and the flux growth rate @xmath30 is shown in figure [ fig:05]e . in the scatter plot , the data points distribute along the relation of @xmath46 where @xmath25 is the emergence duration , @xmath30 is in mx hr@xmath1 and @xmath23 in mx . from equation ( [ eq:03 ] ) , we can derive the emergence duration @xmath25 in the unit of hour as a function of @xmath23 , which is @xmath47 equation ( [ eq:04 ] ) indicates that an emergence event with large maximum flux shows relatively rapid magnetic flux growth . according to equation ( [ eq:04 ] ) , when a flux tube of @xmath48 emerges with @xmath49 , the flux tube with @xmath50 emerges with @xmath51 . @xmath25 does not depend linearly but non - linearly on @xmath23 . a tube with more magnetic flux emerges with less time than in the case of linear dependency . while the equation ( [ eq:04 ] ) is consistent with previous observations such as @xcite and @xcite , the empirical relation ( [ eq:04 ] ) was first derived with wide range of magnetic parameters by _ hinode _ high - resolution samples . figure [ fig:05]f presents the relation between @xmath23 and @xmath21 , which indicates that the larger size efrs show the separating speed less than 1 km s@xmath1 , while the small scale ears footpoints separate with various speed up to 4 km s@xmath1 . the mean separating speed @xmath21 can be written as @xmath52 from equations ( [ eq:02 ] ) and ( [ eq:04 ] ) , equation ( [ eq:05 ] ) reduces to @xmath53 where @xmath21 is in km s@xmath1 and @xmath23 in mx . equation ( [ eq:06 ] ) indicates that the footpoints of emerged flux tube with less magnetic flux separate each other with larger speed . this relation is plotted with solid line in figure [ fig:05]e , which is consistent with the observed values . the sz features are associated with the magnetic flux emergence of more than around @xmath0 mx . @xcite performed three - dimensional mhd simulation of emerging magnetic flux and suggested that emerging field lines take the evolutionary path of a simple expansion if they emerge with a large aspect ratio ( the ratio of their height to their footpoint distance ) ; otherwise , field lines are inhibited from expanding and they show an undulating behavior ( i.e. sz formation ) . our results provide a new criterion of the total magnetic flux regarding the formation of szs . another notable result is that the flux emergence phenomena with / without sz follow the relation between the total flux and the maximum spatial size as described in equation ( [ eq:02 ] ) . this relation is derived from wider range of magnetic parameters than the previous studies @xcite . there authors gave the result for only narrow range of magnetic parameters . we present a comprehensive result on this relation compared to the previous works . ( 80mm,80mm)fig06.eps let us try to derive the power - law relation between the maximum flux @xmath23 and the maximum separation distance @xmath20 , @xmath54 from the viewpoint of simple dimensional analysis . figure [ fig:06 ] shows the schematic image of flux emergence from the convection layer . first , @xmath20 is estimated as follows . initially horizontal flux tube in the convection layer will rise with the typical length @xmath55 where @xmath56 is the local pressure scale height . the maximum separation distance @xmath20 between two main spots depends on the most unstable wavelength of parker instability @xmath57 at the initial depth of the flux tube , @xmath58 thus @xmath20 is proportional to @xmath56 at the depth where initially the flux tube locates . next the total flux @xmath23 is estimated as follows . from equipartition arguments , the magnetic and kinetic energy in the solar convection layer will balance to each other , @xmath59 where @xmath60 , @xmath61 and @xmath62 are the field strength inside the flux tube , the mass density around the flux tube and the mean convection velocity . from mixing length theory @xcite , @xmath63 is given as @xmath64 from equations ( [ eq:09 ] ) and ( [ eq:10 ] ) , @xmath65 now we assume that the solar convection layer can be approximated by an adiabatically stratified atmosphere @xcite , @xmath66 @xmath25 and @xmath67 are temperature and adiabatic index @xmath68 , where @xmath69 and @xmath70 are the specific heats at constant pressure and volume , respectively . the local scale height @xmath56 is proportional to the temperature @xmath25 , thus @xmath71 from equation ( [ eq:11 ] ) and ( [ eq:13 ] ) , @xmath72 let us think about the flux tube width @xmath73 . if the @xmath73 is much larger or smaller than the local mixing length @xmath74 , the flux tube will be disintegrated by the convection flows or accumulated at the convection boundary . thus the flux tube width is expected to be comparable to the mixing length , @xmath75 thus the total flux @xmath23 can be estimated as @xmath76 from equation ( [ eq:08 ] ) and ( [ eq:16 ] ) , @xmath77 thus the power - law index of the relation between the maximum flux @xmath23 and the maximum separation distance @xmath20 is derived to be @xmath78 . for example , @xmath79 is 0.30 with the adiabatic index @xmath80 ( ideal gas case realised in deep convective layers ) . with @xmath81 at near the solar surface where the ionization status is changing rapidly @xcite , @xmath79 is 0.25 . these calculated values are comparable to the observed value @xmath82 . from the argument above , we get the view of the emergence depicted as in figure [ fig:06 ] . magnetic tubes of large flux are anchored in deep layers and appear with large separation between two main spots on the solar photosphere . we derived the relation between the maximum flux @xmath23 and the flux growth rate @xmath30 to be equation ( [ eq:03 ] ) . ( 80mm,60mm)fig07.eps by a simple model of an emerging tube with uniform magnetic flux density , let us try to derive the power - law relation between the maximum flux and the flux growth rate , @xmath83 figure [ fig:07 ] shows the schematic image of the flux tubes just beneath the photosphere . around the photosphere , the plasma @xmath84 is almost 1 . thus the magnetic pressure @xmath85 where @xmath86 is the gas pressure at the photosphere . so we can think that flux densities are nearly constant irrespective of the spatial size or total flux of the magnetic tube , @xmath87 the rise velocity @xmath63 of the flux tube is estimated as follows . when the apex of flux tube reaches to the underneath the solar surface , the rise motion is suppressed and the tube top becomes flattened @xcite . our observation ( section [ sec:4.3.2 ] ) showed that the emergence occurs in unitary form irrespective of total magnetic flux . in these situations the rise velocity @xmath63 from the photosphere will not depend upon the total magnetic flux of the tube , @xmath88 assuming that the aspect ratio of flux tube width @xmath89 is constant just beneath the photosphere , @xmath90 where @xmath73 and @xmath91 are horizontal and vertical width of the flux tube , respectively . the flux growth rate and the total flux are described as @xmath92 @xmath93 respectively . from equation ( [ eq:22 ] ) and ( [ eq:23 ] ) , the relation between @xmath23 and @xmath30 are written as @xmath94 thus the power - law index is @xmath95 which is consistent with the observed value 0.57 . in this discussion we have not considered the factors such as magnetic field stratification inside the flux tube , realistic aspect ratio of tube and so on . if we include these factors then we can get more realistic interpretation of equation ( [ eq:03 ] ) . we investigated the morphological , dynamical and magnetical characteristics of various flux emergence phenomena using high - resolution _ hinode _ sot data . to estimate the magnetic field density of sot data we used _ mdi magnetogram data for the calibration . from 101 samples of flux emergence events , we derived the total flux , flux growth rate , maximum separation and mean separation speed . the sz features are associated with the magnetic flux emergence of more than around @xmath0 mx . the magnetic flux growth rate , emergence duration and mean separation speed were found to follow the power - law functions of the total magnetic flux with the indices of 0.57 , 0.43 and -0.16 , respectively . the typical size of elementary emergence structures is around 4 mm , which is consistent with the most unstable wavelength ( 2 - 4 mm ) of parker instability . the mean separating speed @xmath21 decreases with larger magnetic flux . we got a physical view of solar flux emergence that emerging magnetic fields float and evolve balancing to the surrounding turbulent atmosphere from the discussion on the observed power - law relations . these observational results should be verified by future numerical studies . possible influence of twisting or pre - existing magnetic field could be studied with the data of horizontal magnetic field in the emerging site and will be reported in the near future . we are grateful to the operation team of _ hinode _ and _ soho_. this work was supported by the grant - in - aid for the global coe program `` the next generation of physics , spun from universality and emergence '' from the ministry of education , culture , sports , science and technology ( mext ) of japan , _ hinode _ is a japanese mission developed and launched by isas / jaxa , collaborating with naoj as a domestic partner , nasa and stfc ( uk ) as international partners . scientific operation of the _ hinode _ mission is conducted by the _ hinode _ science team organized at isas / jaxa . this team mainly consists of scientists from institutes in the partner countries . support for the post - launch operation is provided by jaxa and naoj ( japan ) , stfc ( u.k . ) , nasa , esa , and nsc ( norway ) . rcrrrrc|rcrrrrc # & date & & & & & & # & date & & & & & + 1&20061126&15.2&0.97&1.2e19&2.6e18&n&51&20080807&7.4&1.87&&&n + 2&20061201&71.0&0.62&3.0e21&9.5e19&y&52&20080809&3.9&3.09&&&n + 3&20061209&16.6&0.36&9.2e19&7.1e18&y&53&20081014&12.2&1.07&&&y + 4&20061226&24.9&0.74&&&y&54&20081018&7.5&1.20&3.9e18&2.3e18&n + 5&20070104&17.4&0.58&2.0e20&2.4e19&y&55&20081019&8.4&1.09&6.5e18&3.0e18&n + 6&20070117&17.3&0.26&&&y&56&20090118&14.3&0.73&1.7e20&3.2e19&y + 7&20070204&10.7&0.33&1.2e20&1.3e19&y&57&20090226&17.6&0.56&2.6e19&3.0e18&n + 8&20070205&19.2&1.21&1.5e20&3.3e19&y&58&20090314&10.9&1.45&&&n + 9&20070213&7.2&1.47&5.0e18&3.7e18&n&59&20090402&8.8&0.86&3.6e19&1.3e19&n + 10&20070219&8.5&2.72&3.6e18&4.1e18&n&60&20090426&18.5&1.44&1.2e19&3.4e18&y + 11&20070308&16.8&1.38&5.7e19&1.7e19&y&61&20090601&32.6&2.48&2.7e20&7.3e19&y + 12&20070328&7.8&0.29&&&n&62&20090603&30.7&0.54&3.1e20&2.0e19&y + 13&20070331&11.6&1.76&6.6e18&3.6e18&n&63&20090623&7.4&2.52&1.1e19&1.3e19&n + 14&20070414&10.7&1.70&2.1e19&1.2e19&y&64&20090704&63.4&0.35&&&y + 15&20070419&16.7&1.09&4.5e19&1.0e19&y&65&20090707&51.6&0.54&2.3e21&8.8e19&y + 16&20070601&15.3&0.50&&&y&66&20090821&5.9&1.01&4.6e18&2.8e18&n + 17&20070603&14.2&0.73&&&y&67&20090928&10.2&1.18&2.8e19&1.2e19&n + 18&20070702&8.2&0.93&1.8e19&7.2e18&n&68&20091009&9.8&1.47&2.3e18&1.2e18&n + 19&20070807&32.0&0.46&4.4e20&2.3e19&y&69&20091015&9.4&2.09&1.2e19&9.6e18&n + 20&20070825&12.9&1.90&4.2e19&2.2e19&y&70&20091019&10.0&1.29&4.5e18&7.8e18&n + 21&20070918&9.6&2.49&9.7e18&1.8e19&n&71&20091023&34.9&0.44&&&y + 22&20070929&20.6&3.99&1.3e19&9.4e18&y&72&20091025&9.7&1.00&&&n + 23&20071001&22.5&1.48&1.1e20&2.7e19&y&73&20091026&49.8&1.91&1.7e21&2.4e20&y + 24&20071008&11.2&1.70&2.5e19&1.4e19&y&74&20091207&11.2&1.06&6.0e18&2.1e18&n + 25&20071009&16.3&1.42&2.0e19&6.4e18&y&75&20091215&80.5&2.00&9.0e20&8.0e19&y + 26&20071015&6.4&0.59&&&n&76&20091226&19.7&1.03&1.4e20&2.7e19&y + 27&20071020&8.8&0.63&&&n&77&20091229&32.6&1.03&5.5e20&6.3e19&y + 28&20071026&8.7&1.36&1.4e19&7.7e18&n&78&20091230&67.3&0.57&3.0e21&9.1e19&y + 29&20071108&4.4&3.69&2.1e18&6.4e18&n&79&20100108&28.4&0.67&6.0e20&5.1e19&y + 30&20071111&19.1&3.35&1.6e19&1.0e19&y&80&20100109&45.0&0.69&2.6e21&1.4e20&y + 31&20071117&7.4&1.41&1.2e19&8.1e18&n&81&20100110&95.6&0.57&6.7e21&1.4e20&y + 32&20071123&24.8&0.63&&&y&82&20100124&15.7&0.94&1.3e20&2.8e19&y + 33&20071127&12.5&2.51&4.3e18&3.1e18&n&83&20100217&25.3&0.71&1.3e20&1.3e19&y + 34&20071129&13.2&1.11&&&y&84&20100221&10.7&1.75&7.9e19&4.7e19&y + 35&20071207&13.2&0.33&&&n&85&20100222&32.5&2.20&3.3e19&8.1e18&y + 36&20071208&35.6&0.65&1.2e21&8.1e19&y&86&20100313&11.6&1.22&2.4e18&9.0e17&n + 37&20071209&17.2&0.96&9.6e19&1.9e19&y&87&20100322&4.5&1.17&1.8e18&1.6e18&n + 38&20071210&24.9&1.56&1.6e20&3.6e19&y&88&20100325&11.3&1.46&3.8e19&1.8e19&y + 39&20071211&27.1&0.59&9.5e20&7.5e19&y&89&20100414&10.3&1.28&3.1e19&1.4e19&n + 40&20071212&15.4&0.75&&&y&90&20100612&46.0&0.19&1.1e20&7.2e19&y + 41&20071213&24.7&0.66&6.3e20&6.0e19&y&91&20100613&28.1&0.65&&&y + 42&20080105&11.9&0.45&&&n&92&20100618&13.8&0.51&5.8e19&7.7e18&y + 43&20080106&13.9&1.00&&&n&93&20100619&38.0&0.55&1.1e21&5.9e19&y + 44&20080113&18.1&1.09&&&y&94&20100723&18.5&1.03&2.4e20&4.8e19&y + 45&20080124&18.2&1.30&3.0e19&7.8e18&y&95&20100727&13.6&0.80&6.0e19&1.3e19&y + 46&20080419&17.1&1.41&7.7e19&2.3e19&y&96&20100728&48.4&0.56&1.8e21&7.6e19&y + 47&20080610&21.8&0.84&8.3e18&1.1e18&n&97&20100805&28.4&0.42&5.2e20&2.8e19&y + 48&20080615&16.2&0.33&&&y&98&20100809&10.6&0.27&1.7e20&1.5e19&y + 49&20080620&4.6&2.08&&&n&99&20100812&11.6&0.29&7.1e17&1.0e18&n + 50&20080630&3.6&1.56&2.6e17&4.1e17&n&100&20100814&43.9&1.03&4.2e20&3.5e19&y + & & & & & & & 101&20100822&12.4&2.75&7.9e18&6.3e18&n + 99 archontis , v. , & hood , a. w. 2009 , , 508 , 1469 berger , t. e. , lfdahl , m. g. , shine , r. s. , & title , a. m. 1998 , , 495 , 973 bernasconi , p. n. , rust , d. m. , georgoulis , m. k. , & labonte , b. j. 2002 , , 209 , 119 bhatnagar , a. , & livingston , w. 2005 , fundamentals of solar astronomy , world scientific series , 374 brants , j. j. & zwaan , c. 1982 , , 80 , 251 brants , j. j. 1985 , , 95 , 15 brants , j. j. , & steenbeek , j. c. m. 1985 , , 96 , 229 bruzek , a. 1967 , , 2 , 451 bruzek , a. 1969 , , 8 , 29 centeno , et al . 1988 , , 666 , 137 cheung , m. c. m. , schssler , m. tarbell , t. d. , & title , a. m. 2008 , , 687 , 1373 cheung , m. c. m. , rempel m. , title , a. m. , & schssler , m. 2010 , , 720 , 233 chou , d. , & zirin , h. 1988 , , 333 , 420 ellerman , f. 1917 , , 46 , 298 fan , y. 2001 , , 554 , 111 foukal , p. 2004 , solar astrophysics , wiley - vch , 173 guglielmino , s. , l. , et al . 2008 , , 688 , 111 hagenaar , h. j. 2001 , , 555 , 448 harvey , k. l. , & martin , s. f. 1973 , , 32 , 389 harvey , k. l. , harvey , j. w. , & martin , s. f. 1975 , , 40 , 87 ichimoto , k. , et al . 2004 , spie , 5487 , 1142 isobe , h. , tripathi , d. , & archontis , v. 2007 , , 657l , 53 kitai , r. 1983 , 87 , 135 kosugi , t. , et al . 2007 , , 243 , 3 kurokawa , h. , kawaguchi , i. , funakoshi , y. , & nakai , y. 1982 , , 79 , 77 lamb , d. a. , deforest , c. e. , hagenaar , h. j. , parnell , c. e. , & welsch , b. t. 2010 , , 720 , 1405 loughhead , r. e. , & bray , r. 1961 , australian journal of physics , 14 , 347 magara , t. 2001 , , 549 , 608 magara , t. & longcope , d. w. 2001 , , 559 , l55 magara , t. & longcope , d. w. 2003 , , 586 , 630 martin , sara f. 1988 , , 117 , 243 matsumoto , r. , & shibata , k. 1992 , , 44 , 167 matsumoto , r. , tajima , t. , shibata , k. , & kaisig , m. 1993 , , 414 , 357 matsumoto , t. , kitai , r. , shibata , k. , otsuji , k. , naruse , t. , shiota , d. , & takasaki , h. 2008a , , 60 , 95 matsumoto , t. , et al . 2008b , , 60 , 577 matsumoto , t. , & kitai , r. 2010 , , 716 , l19 murray , m. j. , et al . 2006 , , 460 , 909 november , l. j. , & simon , g. w. 1988 , , 333 , 427 nozawa , s. , shibata , k. , matsumoto , r. , sterling , a. c. , tajima , t. , uchida , y. , ferrari , a. , & rosner , r. 1992 , , 78 , 267 nozawa , s. 2005 , , 57 , 995 otsuji , k. , et al . 2007 , , 59 , 649 otsuji , k. , kitai , r. , matsumoto , t. , ichimoto , k. , ueno , s. , nagata , s. , isobe , h. , & shibata , k. 2010 , , 62 , 893 pariat , e. , aulanier , g. , schmieder , b. , georgoulis , m. k. , rust , d. m. , & bernasconi , p. n. 2004 , , 614 , 1099 parker , e. n. 1966 , , 145 , 811 scherrer , p. h. , et al . 1995 , , 162 , 129 shibata , k. , tajima , t. , steinolfson , r.s . , & matsumoto , r. 1989 , , 345 , 584 shimizu , t. , et al . 2007 , , 249 , 221 stix , m. 1989 , springer - verlag , the sun , 198 strous , l. h. , scharmer , g. , tarbell , t. d. , title , a. m. , & zwaan , c. 1996 , , 306 , 947 strous , l. h. , & zwaan , c. 1999 , , 527 , 435 suematsu , y. , et al . 2007 , , 249 , 197 titov , v. s. , priest , e. r. , & demoulin , p. 1993 , , 276 , 564 tsuneta , s. , et al . 2007 , , 249 , 167 watanabe , h. , kitai , r. , okamoto , k. , nishida , k. , kiyohara , j. , ueno , s. , hagino , m. , ishii , t. t. , & shibata , k. 2008 , , 684 , 736 zirin , h. 1972 , , 22 , 34 zwaan , c. , brants , j. j. , & cram , l. e. 1985 , , 95 , 3 zwaan , c. 1987 , annual review of astronomy and astrophysics , 25 , 83
we studied 101 flux emergence events ranging from small ephemeral regions to large emerging flux regions which were observed with _ hinode _ solar optical telescope filtergram . we investigated how the total magnetic flux of the emergence event controls the nature of emergence . to determine the modes of emergences , horizontal velocity fields of global motion of the magnetic patches in the flux emerging sites were measured by the local correlation tracking . between two main polarities of the large emerging flux regions with more than around @xmath0 mx , there were the converging flows of anti - polarity magnetic patches . on the other hand , small ephemeral regions showed no converging flow but simple diverging pattern . when we looked into the detailed features in the emerging sites , irrespective of the total flux and the spatial size , all the emergence events were observed to consist of single or multiple elementary emergence unit(s ) . the typical size of unitary emergence is 4 mm and consistent with the simulation results . from the statistical study of the flux emergence events , the maximum spatial distance between two main polarities , the magnetic flux growth rate and the mean separation speed were found to follow the power - law functions of the total magnetic flux with the indices of 0.27 , 0.57 , and -0.16 , respectively . from the discussion on the observed power - law relations , we got a physical view of solar flux emergence that emerging magnetic fields float and evolve balancing to the surrounding turbulent atmosphere .
in the standard model ( sm ) , flavor changing neutral currents ( fcncs ) are forbidden at tree level . they can occur starting at one loop . for the @xmath2 transitions , the diagrams involving the top quark dominate the short distance rate . this is essentially a consequence of non - decoupling in spontaneously broken gauge theories : heavy fermions in the loops give contributions that do not vanish as the mass increases but rather grow . thus , fcnc processes at relatively low energies have the potential to explore high energy scales such as the weak scale or even beyond . extensions of the sm , such as supersymmetry , technicolor , etc . would result in new contributions to fcnc loops , leading to deviations from the sm predictions . this sensitivity of fcnc processes to high energy scales , and therefore to new physics makes them of great interest . this is particularly true in @xmath3 decays since it is possible that the third generation is involved in electroweak symmetry breaking . however , @xmath3 decays are affected by hadronic uncertainties that may obscure the interesting short distance processes . such uncertainties result from the fact that hadronization is a non - perturbative problem only tractable from first principles by lattice gauge theory . the use of inclusive decay modes greatly circumvent this problems . there , the uncertainties are mostly not from hadronization ( only relevant in the initial state ) but from perturbative qcd . exclusive modes , on the other hand , are largely affected by theoretical uncertainties from hadronic matrix elements of the short distance operators . while lattice calculations progress toward greater precision and accuracy , we can ask the question : can rare @xmath3 decays be used as tests of the one - loop structure of the sm with our current knowledge of hadronic matrix elements ? the comparison with the program of electroweak precision measurements , mostly performed at the @xmath4 pole , is interesting . that program relied on processes with large tree - level sm amplitudes ( large sm background ) , but exquisite experimental precision made it possible to achieve sensitivity to one loop contributions since the accuracy of our theoretical knowledge matches this precision . on the other hand , rare @xmath3 decays and fcnc processes in general , start at one loop but are affected by large theoretical uncertainties . these considerations are still valid even if we are talking about new physics entering at tree level . very large data samples of @xmath5 and @xmath6 decays will be available soon . how good a test of the sm will these be ? the inclusive @xmath7 decay already constrains physics beyond the sm . even more constraints will result once @xmath6 modes begin to be measured . inclusive modes are theoretically cleaner . the uncertainties are , in principle , from perturbative qcd and the ope for heavy quarks . in practice , there is the added problem that cuts are needed to make contact with experiment @xcite . this introduces additional uncertainties . in addition , the experimental signals require a very clean environment . exclusive decays are easier experimentally . but , as e mentioned earlier , they are affected by large hadronic uncertainties coming as unknown form - factors . these are the kinds of things the lattice will one day compute precisely . but for now ( short of using models ) we have to rely on symmetries and other related tricks in order to extract the short distance physics from these modes . the factorization of short and long distance physics takes place in the effective hamiltonian at low energies . this is obtained by integrating out the heavy fields ( e.g. the @xmath8 , the top quark , the gluino , etc . ) and has the form @xmath9 here , the basis @xmath10 constitute a complete set of operators leading to @xmath2 transitions , whereas the wilson coefficient functions @xmath11 encode the short distance information coming from integrating out the heavy degrees of freedom . for instance in the sm the @xmath11 come from loop integrals involving the @xmath8 and the top quark and they will depend on their masses . if physics beyond the sm contributes to these decays it will shift the values of the wilson coefficients with terms now depending on gluino , squark or technipion masses . the @xmath12 dependence of the wilson coefficients is in principle canceled by that of the matrix elements of the operators . at leading order in @xmath13 the operators mediating @xmath2 transitions are @xmath14 among these , only @xmath15 contributes to processes with the photon on - shell , such as @xmath5 . mixing with other operators occurs due to the strong interactions . most notably with the gluonic dipole operator @xmath16 , and especially with @xmath17 , which is generated by tree - level @xmath8 exchange . the hadronic matrix elements can be parametrized in terms of form - factors . for the @xmath18 decay the hadronic matrix elements of the operators @xmath19 , @xmath20 and @xmath21 can be written as @xmath22 with @xmath23 , and @xmath24 unknown functions of @xmath25 . in the su(3 ) limit @xmath26 in ( [ bp_sem ] ) are the same as the form - factors entering in the semileptonic decay @xmath27 . for @xmath28 decays of the `` semileptonic '' matrix elements over vector and axial vector currents are @xmath29 and for the fcnc magnetic dipole operator @xmath30 @xmath31 where @xmath32 denotes the polarization four - vector of the @xmath33 . in the @xmath34 limit these form - factors also describe @xmath35 , as well as @xmath36 decays . the form - factors defined in eqns.([bp_sem])-([sig2kst ] ) are not calculable in a controlled approximation with the notable exception of lattice gauge theory . we will see however , that symmetries and generally the behavior of hadrons in certain limits will allow us to gain important insight on these quantities . notice that @xmath37 and @xmath38 does not contribute to the amplitude to the radiative decay into an on - shell photon , i.e. in @xmath1 . so now @xmath39 , where the short distance is basically the same as in inclusive decays . in the hql @xcite @xmath41 the form - factors @xmath42 corresponding to the dipole operator are not independent of the semileptonic form - factors @xmath43 . instead , they obey the following relations @xcite @xmath44 & = & \frac{3m_b^2-q^2+m_v^2}{2m_b}\frac{v(q^2 ) } { m_b+m_v}-\frac{m_b+m_v}{2m_b } a_1(q^2 ) , \label{rel_3}\\ t_3(q^2 ) & = & \frac{m_b^2-q^2 + 3m_v^2}{2m_b } \frac{v(q^2)}{m_b+m_v}+ \frac{m_b^2-m_v^2}{m_b q^2 } m_v a_0(q^2 ) \nonumber\\ & \hspace*{-1cm}-&\hspace*{-0.5cm}\frac{m_b^2+q^2-m_v^2}{2m_b q^2 } \big[(m_b+m_v)a_1(q^2)-(m_b - m_v)a_2(q^2)\big ] . \nonumber \end{aligned}\ ] ] in terms of the symmetries of the hqet , eqns . ( [ rel_3 ] ) result from the heavy quark _ spin _ symmetry ( hqss ) that arises in the heavy quark limit due to the decoupling of the spin of the heavy quark @xcite . these relations imply , for instance , that the @xmath35 and @xmath45 are described by the same form - factors . additionally , if @xmath34 symmetry is assumed , these also are the @xmath40 form - factors . we now consider the large energy limit ( lel ) @xcite for heavy - to - light transitions into a vector meson as the ones we are studying . as a result , one recovers the hqss form - factor relations ( [ rel_3 ] ) , but now there are additional new relations among the form - factors defined in ( [ vcurr]-[sig2kst ] ) . these read as @xcite @xmath46 and @xmath47 and will receive corrections that roughly go as @xmath48 . it is apparent from eqns . ( [ le4])-([t3 ] ) that , in the lel regime , the @xmath49 decays are described by only two form - factors : @xmath50 and @xmath51 , instead of the seven apriori independent functions in the general lorentz invariant ansatz of the matrix elements . here , @xmath50 and @xmath51 are functions of the heavy mass @xmath52 and the hadronic energy @xmath53 , and refer to the transverse and longitudinal polarizations , respectively . this simplification leads to new relations among the form - factors . for instance , the ratio of the vector form - factor @xmath54 to the axial - vector form - factor @xmath55 , @xmath56 is independent of any of these unknown , non - perturbative functions @xmath57 and is determined by purely kinematical factors . here , @xmath58 denotes the energy of the final light vector meson . a similar relation holds for @xmath59 and @xmath60 , since they both are also proportional to the `` transverse '' form - factor @xmath50 . as we will see below , these predictions have important consequences for observables at large recoil energies ( low @xmath61 ) . these corrections result from the exchange of hard gluons . there are two kinds of them : factorizable and non - factorizable gluon exchange . the first kind corresponds to either the renormalization of the heavy - light currents which already appears in ref . @xcite , or hard gluon exchange with the spectator quark . the latter can be computed in the brodsky - lepage formalism for exclusive processes at large momentum transfers @xcite . this was done in ref . @xcite , where it was found that the form - factor relations in ( [ le4])-([t3 ] ) receive @xmath13 corrections that are typically @xmath62 or smaller . an interesting case is the ratio of vector to axial - vector form - factors in eqn . ( [ rv ] ) , which receives _ no @xmath13 corrections _ to leading order in the @xmath63 expansion . this has an interesting explanation and we will come back to this point below . the second kind of hard gluon corrections are the so called non - factorizable ones @xcite . these are mediated by diagrams where the gluon exchanged with the spectator comes from either the insertion of the operator @xmath64 , or from the insertion of four - fermion operators @xmath65 with the gluon attached to the quark loop . these corrections can not be absorbed by form - factors or renormalization of the currents . thus they are genuinely distinct hard corrections that do not occur in inclusive decays . in refs . @xcite the effects of all these contributions are computed and are rather large in @xmath1 . the effective `` exclusive '' wilson coefficient @xmath66 is shifted by roughly @xmath67 , whereas the effect in @xmath68 is smaller . the authors go on to compute the exclusive rate by making use of light - cone qcd sum rules for the form - factors , resulting in a rate which is approximately a factor of two larger than the experimental rate . this may be signaling a problem with the form - factor calculation rather than one with the non - factorizable hard gluons . in ref . @xcite an effective field theory including collinear quarks and gluons is developed . this theory includes leet but , unlike leet , it has the correct infrared behavior . sudakov logarithms are accounted for in this treatment . concerning its application to exclusive decays , this `` complete '' leet does not modify the form - factor relations implied by eqns.([le4]-[t3 ] ) . a main ingredient of leet is preserved when the interaction with collinear gluons is taken into account : energetic quarks still are two component spinors in the lel . thus the complete leet including collinear gluons confirms that the leet relations are valid in the large energy limit of qcd . just as in hqet the @xmath69 corrections are potentially a very important source of uncertainty , the lel predictions are affected by @xmath63 corrections . unlike in hqet , it is not clear that leet is a framework where the @xmath63 corrections can be estimated . even when collinear gluons are incorporated , it is difficult to separate degrees of freedom to be integrated out so that an effective field theory can emerge . perhaps , we can still estimate the size of these corrections with this caveat in mind . more work is needed , and certainly once experimental tests of leet predictions become available we hope to understand more about them . the forward - backward asymmetry for leptons as a function of the dilepton mass squared @xmath70 is defined as @xmath71 where @xmath72 , and @xmath73 is the angle between the @xmath74 and the @xmath75 in the dilepton center - of - mass frame . is defined with respect to @xmath76 . but the expressions there correspond to the current definition . this explains the sign difference with respect to ref . @xcite . ] the asymmetry is proportional to the wilson coefficient @xmath77 and vanishes with it . furthermore , it is proportional to a combination of @xmath78 and @xmath79 such that it has a zero in the physical region if the following condition is satisfied @xcite @xmath80=\frac{m_b}{q^2_0}\,c_7^{\rm eff.}\,\left\ { \frac{t_1}{v}(m_b+m_{k^ * } ) - ( m_b - m_{k^*})\,\frac{t_2}{a_1}\right\}~,\label{afbz}\ ] ] where @xmath81 is the position of the @xmath82 zero and all @xmath61-dependent quantities ) with respect to a similar expression in ref.@xcite is due to a sign in the definition of @xmath54 . ] are evaluated at @xmath81 . in inclusive @xmath83 , the zero of the asymmetry implies the relation : @xmath84=-2 ( m_b^2/q_0 ^ 2)\,c_7^{\rm eff.}$ ] . thus , in principle the condition in ( [ afbz ] ) is affected by the presence of the hadronic form - factors making it a priori a more uncertain relation . however , by making use of the hqss relations ( [ rel_3 ] ) for @xmath85 and @xmath86 , the form - factor eqn . ( [ afbz ] ) simplifies to @xmath87 } { c_7^{\rm eff.}}=-\frac{m_b}{q^2_0}\left\ { \frac{2m_b\,k^2}{(m_b+m_{k^*})^2}r_v + \frac{(m_b+m_{k^*})^2}{2m_br_v } + 2(m_b - e_{k^ * } ) \right\ } \label{con2}\ ] ] where @xmath88 is the ratio of vector to axial - vector form - factors defined in ( [ rv ] ) and evaluated at @xmath89 . thus , the determination of the zero of @xmath90 in @xmath40 gives a relation between the short distance wilson coefficients @xmath78 and @xmath79 where the only uncertainty from form - factors is in the ratio @xmath88 . as mentioned earlier , since ( [ con2 ] ) was derived using the heavy quark _ spin _ symmetry , it is expected to receive small corrections . in principle , information on @xmath91 could be extracted from the @xmath35 decay . but it turns out that this may not be necessary . if we plot the asymmetry vs. the dilepton mass in a variety of models we note that the asymmetry is agreed upon with exceptional accuracy ! in ref.@xcite it was noted that this feature must emerge from a `` factorization '' of the soft physics in such a way that @xmath92 so that the soft physics cancels in the ratio @xmath88 . it was first recognized in ref . @xcite that this is precisely what the leet predicts : @xmath93 as extracted from eqns . ( [ le4 ] ) and ( [ le5 ] ) . then , the condition for the vanishing of @xmath90 reads now @xmath80=-2\frac{m_bm_b}{q_0 ^ 2}\;c_7^{\rm eff.}\ , \left(1-\frac{m_{k^*}^2}{2m_be}\right)~. \label{condlet}\ ] ] thus , in the lel the position of the zero of @xmath90 in @xmath40 is predicted in terms of the short distance wilson coefficients @xmath78 and @xmath79 . the hard gluon corrections discussed above would shift the position of the zero by a calculable amount given in ref.@xcite . there it is pointed out that non - factorizable contributions change the value of @xmath81 from what it would be obtained if the `` inclusive '' wilson coefficients are used by an amount around @xmath94 . this is roughly @xmath95 . the plot in fig.[afb ] takes all these corrections into account . we finally discuss another application of leet . in ref . @xcite the leet prediction for @xmath88 in eqn.([rv ] ) was used in combination with the @xmath1 and @xmath7 data , and heavy quark spin symmetry in order to constrain the semileptonic form - factors @xmath54 and @xmath55 at @xmath96 . here we update this analysis by incorporating the factorizable and non - factorizable hard corrections from refs.@xcite . since the nlo corrections in the exclusive mode are not canceled by the ones in the inclusive decay , we will not normalize the exclusive branching ratio to the inclusive one . instead we make use of the exclusive data only in order to extract the @xmath1 form - factor . the branching ratio reads @xmath97 where the form - factor @xmath85 was defined in ( [ sig2kst ] ) , and @xmath98 is the nlo effective wilson coefficient for the exclusive mode , which differs from the one entering in inclusive decays . we use the running mass @xmath99gev , and @xmath100 . the value of @xmath98 was computed in ref . @xcite to be @xmath101 . we make use of the world average for the neutral mode @xcite @xmath102 . we obtain @xmath103 , where the error mainly reflects the experimental uncertainty , the uncertainty in the @xmath104 quark mass and the scale dependence in @xmath98 . this result assumes the sm for the calculation of the short distance coefficient @xmath98 . however , this is not a very strong assumption given that the agreement of the inclusive calculation with the @xmath7 , and the fact that the new physics is likely to affect this rate and the exclusive one similarly . armed with this extracted value of @xmath105 , now we can go to the hqss relations and turn them into a relation between @xmath106 and @xmath107 . this is @xmath108 eqn . ( [ expconst ] ) results in a constraint in the @xmath109 plane , which in fig . [ fitcon ] corresponds to the band descending from left to right . in addition , the expression ( [ rv ] ) for @xmath88 gives another constraint corresponding to a straight line going through the origin . this leet prediction will surely receive corrections at next to leading order in the @xmath63 and @xmath110 expansions . these corrections have not been computed and we simply guess they are of order @xmath111 , giving the cone in fig . [ fitcon ] . we should notice , however , that the leet prediction for @xmath88 does not receive hard gluon or collinear gluon corrections . we will come back to this point . the ellipses in fig . [ fitcon ] correspond to @xmath112 and @xmath113 c.l . the fit gives @xmath114 the results in eqn . ( [ result ] ) differ from ref . @xcite where the nlo corrections had not been taken into account . these lift the value of the wilson coefficient by roughly @xmath67 resulting in a lower value for @xmath105 , and through the fit in fig . [ fitcon ] , in lower values for both @xmath106 and @xmath107 . also , here we use the most recent average of the neutral mode resulting in a thinner band than the one in ref . @xcite . we compare our findings for @xmath106 and @xmath107 with several model predictions . for illustration , we take the bauer - stech - wirbel ( bsw ) model from ref . @xcite ( cross ) , the modified version of the isgur - scora - grinstein - wise ( isgw2 ) model from ref . @xcite ( diamond ) , a recent relativistic constituent quark model prediction by melikhov and stech ( ms ) @xcite ( star ) , a recent calculation in the light cone qcd sum rule ( lcsr ) formalism of ref . @xcite ( diagonal cross ) and the prediction by ligeti and wise ( lw ) from ref.@xcite ( square ) . as it can be seen from fig . [ fitcon ] , the latter is even more excluded now . as discussed in more detail in ref.@xcite , this is likely to be due largely to the use of a monopole form - factor to extrapolate from the charm data to the maximum recoil in a @xmath3 decay . corrections to the heavy quark flavor symmetry , although large , are unlikely to be the sole explanation for such a discrepancy . a new finding in this update is the apparent exclusion of the ms and lcsr predictions for @xmath106 and @xmath107 . these models , were in agreement with the fit in ref . of course , this exclusion refers to the predicted central values for the models . in some cases the predictions have large errors and modifications in the calculations may bring them into agreement with the fit . a potentially large isospin violation splitting the neutral and charged modes was found in ref.@xcite in the context of qcd factorization . this would shift the value of @xmath98 in the neutral mode by a few percent , resulting in a similarly small shift in the value of @xmath105 extracted from eqn.([rgamma ] ) . finally , we comment on the leading order leet prediction for @xmath88 , eqn . ( [ rv ] ) . making use of the hqss relations ( [ rel_3 ] ) , the transverse helicity amplitudes for a generic @xmath115 transition can be written as @xmath116 where @xmath117 is a factor depending on the mode under consideration ( e.g. wilson coefficients , coupling constants , etc ... ) and @xmath118 is the momentum of the vector meson . thus , we see from the form of @xmath88 in the large energy limit , that the `` @xmath119 '' helicity vanishes @xmath120 in the lel regime , up to residual terms of order @xmath121 . this is not a surprise : in the limit of an infinitely heavy quark decaying into a light quark , the helicity of the latter is `` inherited '' by the final vector meson . in the sm , the @xmath122 structure in semileptonic decays is reflected in the dominance of the @xmath123 transverse helicity . on the other hand , the amplitude to flip the helicity of the fast outgoing light quark is suppressed by @xmath124 . quark models tend to have this concept built in , which explains the agreement on the position of the zero in fig . this is also the reason why @xmath13 corrections from hard gluon exchange between the spectator quark and the fast light quark do not affect eqn . ( [ rv ] ) : they are not helicity - changing at leading order in @xmath124 . the same is true for the ratio of @xmath59 and @xmath60 . and @xmath107 from @xmath125 data plus hqss ( thicker band ) together with the relation from the lel ( cone).the ellipses correspond to @xmath112 and @xmath113 confidence level intervals . central values of model predictions are also shown and correspond to bsw @xcite ( vertical cross ) , isgw2 @xcite ( diamond ) , ms @xcite ( star ) , lcsr @xcite ( diagonal cross ) and lw @xcite ( square ) , respectively . this updates ref . @xcite to include hard corrections as computed in ref . @xcite , as well as the most recent data . ] i thank gudrun hiller , with whom part of the work reported here was done , for discussions and a careful reading of the manuscript . i also thank the organizers of hf9 at caltech for such an excellent meeting , in spite of the difficult circumstances they had to face during its course . this work was supported by the director , office of science , office of high energy and nuclear physics of the u.s . department of energy under contract de - ac0376sf00098 . m. dugan and b. grinstein , phys . lett . * b255 * , 583 ( 1991 ) . j. charles _ et al . _ , phys . rev . * d60 * , 014001 ( 1999 ) . g. p. lepage and s. j. brodsky , phys . * d22 * , 2157 ( 1980 ) . the first application to @xmath3 decays is in a. szczepaniak , e. m. henley and s. j. brodsky , phys . b * 243 * , 287 ( 1990 ) . m. beneke and t. feldmann , nucl . * b592 * , 3 ( 2001 ) . s. w. bosch and g. buchalla , `` the radiative decays @xmath126 at next - to - leading order in qcd '' , hep - ph/0106081 . stefan bosch , these proceedings . m. beneke , t. feldmann and d. seidel , nucl . phys.*b612 * , 25 ( 2001 ) . c. w. bauer , s. fleming and m. e. luke , phys . * d63 * , 014006 ( 2001 ) ; c. w. bauer , s. fleming , d. pirjol and i. w. stewart , phys . rev . * d63 * , 114020 ( 2001 ) . g. burdman , phys . rev . * d57 * , 4254 ( 1998 ) . a. ali , p. ball , l. t. handoko and g. hiller , phys . rev.*d61 * , 074024 ( 2000 ) . g. burdman and g. hiller , phys . rev . * d63 * , 113008 ( 2001 ) . a. ryd , for the babar collaboration , these proceedings ; m. nakao , for the belle collaboration , these proceedings ; k. honscheid , for the cleo collaboration , these proceedings . m. wirbel , b. stech and m. bauer , z. phys . * c29 * , 637 ( 1985 ) . n. isgur and n. scora , phys . rev . * d52 * , 2783 ( 1995 ) . d. melikhov and b. stech , phys . rev . * d62 * , 014006 ( 2000 ) . z. ligeti and m. wise , phys . rev . * d60 * , 117506 ( 1999 ) . a. l. kagan and m. neubert , `` isospin breaking in @xmath127 decays '' , hep - ph/0110078 .
i discuss recent progress in our understanding of exclusive rare and semileptonic decays . i show the impact of hqet when combined with the predictions in the large energy limit of qcd , focusing first on applications to @xmath0 . i also discuss the constraints on semileptonic form - factors that appear in hqet / leet with the use of radiative decay data , and update these to include the effects of next - to - leading order contributions in @xmath1 , as well as the latest data . 22.0 cm 16.0 cm -1.0 in @=11 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbciteacitea , tempcnta = tempcntbtempcnta @=12 lbnl-49221 gustavo burdman + + _ berkeley , ca 94720 _
despite being an extremely succesful theory to predict the outcome of experiments with particles and other microscopic objects , quantum mechanics ( qm ) is believed by many to be incomplete or , at least , not fully understood . in particular , the strange " or non - classical phenomena of qm , like self - interference and born s rule , are described in terms of abstract mathematical objects . although some have been ready to interpret the complex - valued wavefunction as a real object , wavefunctions are generally seen as mathematical tools serving to calculate probabilities from their square moduli . contrasting to real - valued mathematics and one - to - one mapping between real variables and observables of classical theories , the standard description is thus sometimes considered as purely operational . in this paper a model is presented that , although very simple in terms of mathematical development and formalism , seems to be able to predict probability fields of at least free particles , and particularly self - interference , without using schrdinger equation or wavefunctions . unlike the standard qm picture , the proposed model only uses integer - valued quantities and arithmetic operations . in particular , it assumes a discrete spacetime under the form of an euclidean lattice . the proposed approach describes individual particle trajectories as random walks . transition probabilities are simple functions of a few quantities that are either randomly associated to the particles during their preparation , or stored in the lattice sites they visit during the walk . non - relativistic qm predictions are retrieved as probability distributions of similarly - prepared ensembles of particles . the proposed model goes beyond ensemble interpretations @xcite in the sense that it describes the behavior of individual particles . with respect to de broglie bohm mechanics @xcite , or deterministic trajectory representation @xcite , the proposed model introduces a non - deterministic behavior but does not appeal to non - localities . intrinsec ideterminism is already contained in stochastic interpretations of qm that are , however , mostly aimed at retrieving the schrdinger equation from a classical equation of motion plus a stochastic force . the born probability rule remains unexplained in this context @xcite , or is founded on the definition of probability density of particles as the squared intensity of an associated wave @xcite . this latter assumption is not used in the proposed approach , which in contrast predicts nonclassical consequences of born rule ( as double - slit interference ) only from the random walk features . the idea of lattice or discrete - time algorithms that reproduce particle propagation in the continuum limit is also not new . however , the proposed model uses the lattice only as the support for particle motion , not for wavefunctions or other mathematical operators as , e.g. , in @xcite . while other random walks or spacetime quantizations @xcite are able to reproduce the emergence of schrdinger equation from pure combinatorics , again the born probability rule and thus intereference are not explained in such models , while naturally emerges in the proposed one . in order to reproduce interference , antiparticles are not appealed to , as in some abstract lattice gas models for waves @xcite , nor negative probabilities . the paper is organized as follows . section 2 presents the assumptions concerning spacetime , which naturally lead to heisemberg s uncertainty principle . section 3 describes the model for free motion without quantum forces and retrieves de broglie relation and schrdinger equation . section 4 describes the model with quantum forces and shows numerical results for several self - interference scenarios . section 5 is just a short introduction to possible extensions to treat interacting particles and the transition to relativistic qm . the proposed model assumes that the spacetime is inherently discrete . limiting for simplicity the analysis to one dimension @xmath0 , that means that only values @xmath1 , @xmath2 and @xmath3 , @xmath4 are meaningful . noninteger values of space and time are simply impossible in this picture . the two fundamental quantities @xmath5 and @xmath6 are the size of the lattice that constitutes the space and the fundamental temporal resolution , respectively . under this assumption , a particle s history consists of a succession of points @xmath7 in the spacetime , where @xmath8 is the discrete index that describes advance in history , here denoted as iteration " . advance in time is unidirectional and unitary , that is , @xmath9 follows necessarily @xmath10 . advance in space is still unitary but bidirectional . if in a certain iteration a particle resides at the location @xmath11 of the spatial lattice , in the next iteration the particle can only reside at locations @xmath12 , @xmath11 , or @xmath13 . the local velocity of this mechanism , @xmath14 , is a random variable that can take only the three discrete values @xmath15 , as described in sect . [ sec : single ] and sect . [ sec : interf ] . consider for the moment only free motion , without interference , such that @xmath16 . the observable velocity of the particle as the result of a observation process lasting @xmath17 iterations or time steps ( @xmath17 is arbitrary ) would be @xmath18 the maximum velocity that a particle can reach is the speed of light @xmath19 . light trajectory in the positive direction corresponds to @xmath20 , @xmath21 . consequently to ( [ eqn : v ] ) , one constraint to the fundamental lattice quantities is necessarily @xmath22 another consequence of ( [ eqn : v ] ) is that to determine the average velocity of a particle , an observer should wait in principle a time @xmath17 tending to infinity . every observation lasting a finite number of iterations @xmath17 will give an approximation of @xmath23 . consider , e.g. , @xmath24 . the observed velocity can be @xmath25 , @xmath26 or @xmath27 . thus the uncertainty on @xmath23 is @xmath19 in absolute value . for @xmath28 , the possible outcomes for the sample mean are @xmath19 , @xmath29 , @xmath26 , @xmath30 , and @xmath27 . thus the uncertainty on @xmath23 is @xmath29 in absolute value . extending these considerations , the uncertainty on @xmath23 after an observation lasting @xmath17 iterations is @xmath31 in lattice units . moreover , an observation lasting @xmath17 iterations necessarily implies a change in the position of the particle . the span of the particle during the observation ranges from @xmath32 to @xmath33 . thus the uncertainty on the position of the particle at the end of the observation is @xmath34 in lattice units . using the two results above , and denoting @xmath35 and @xmath36 the uncertainties of velocity and position as a function of observation horizon @xmath17 , the relationship @xmath37 holds . the latter equation resembles the heisenberg uncertainty principle since it fixes an inverse proportionality between the uncertainty with which the velocity of a particle can be known and the uncertainty with which its position can be known . multiplying by the particle mass @xmath38 , and comparing ( [ eqn : heis ] ) to heisenberg uncertainty principle , one obtains that the two fundamental lattice quantities are related to the planck constant , @xmath39 the term @xmath40 ( the solid angle of a sphere ) holds for three - dimensional spaces . in our example case of a one - dimensional space , this term reduces to 2 , the measure of the unit 1-sphere . thus , combining ( [ eqn : cc ] ) with the accordingly modified ( [ eqn : hbar ] ) , the values for the fundamental lattice quantities are obtained as @xmath41 and @xmath42 note that the compton wavelength is retrieved as twice the fundamental lattice size @xmath5 . the role of mass is not completely clear at this point . likely , general relativity will serve to integrate it into the picture . this section will first describe the equations of motion of a free particle in the proposed model . then , the stochastic variables associated with the motion will be analyzed . finally , the equivalence with the wavefunction picture and schrdinger equation will be retrieved . this section describes the propagation rules of a particle on the lattice , or its _ dynamics_. as stated in the previous section , time can only increase by one unit at each iteration . time is re - initialized to zero whenever the particle interacts with the environment ( external forces ) . this event is called _ preparation _ in the following . we might introduce a stochastic variable representing time at an iteration @xmath43 , @xmath44 where @xmath45 is the most recent iteration when the particle has undergone preparation , that is , has interacted with the environment and has been actualized . however , @xmath46 would have a deterministic distribution , @xmath47 and thus it will be often replaced by its support @xmath48 in the following . consider now spatial dynamics . at each iteration , the particle might jump to one of the nearest neighboring sites of the lattice , or stay at rest . the actual local trajectory is not deterministic , i.e. , it is not a prescribed function of previous parts of trajectory . rather , the local trajectory has the characteristics of a random walk . this point is very important and it implies that an intrinsic randomness affects the particle motion . generally , there is a different transition probability for each of the three possible transitions . in free motion without interference or external forces , these probabilities do not change with time . let me denote the transition probabilities @xmath49 , @xmath50 , and @xmath51 , respectively . of course , @xmath52 moreover , the proposed model assumes that the expected value of @xmath53 is imprinted to the particle . this imprint is to be attributed to the preparation and is actualized every time the particle interacts with the environment ( in a way to be considered later ) . let me denote this expected value as _ momentum propensity _ @xmath54 , @xmath55=a - c . \label{eqn : p}\ ] ] another characteristic of the random motion is the expected value of the squared velocity , that is , @xmath56 = a+c \label{eqn : e}\ ] ] that can be reinterpreted as an _ energy propensity _ ( define the stochastic variable energy " as @xmath57 ) . combining ( [ eqn : abc])([eqn : e ] ) , obtain @xmath58 the energy @xmath59 must be a function of @xmath54 . a well - known result of special relativity states that energy of a particle is the sum of the rest energy and the kinetic energy . following this suggestion , the proposed model assumes that @xmath60 equation ( [ eqn : energy ] ) might be also interpreted in the following way : energy is the average of the time energy " and the space energy " , where the former contribution is always one , since time can only advance by one unit . consequently , ( [ eqn : a ] ) can be rewritten as @xmath61 appendix [ app : matter ] shows that it is possible to retrieve the de broglie relation @xmath62 for matter waves " with the proposed model , involving the energy propensity @xmath59 . the equations in the previous section , in fixing the probability of each jump at each iteration @xmath43 , define the trajectory of the particle as a random walk . we can introduce now the stochastic variables that describe the position of the particle , @xmath63 . that is defined as the cumulated sum of the velocity that the particle has experienced since its preparation , @xmath64 let me calculate the _ probablity mass function _ of this stochastic variable , @xmath65 i.e. , the probability that the particle has crossed @xmath11 sites after @xmath43 iterations . consider the scenario where particles are emitted from a source located at the site @xmath66 of the lattice with an intrinsic value of @xmath54 , and thus of @xmath59 , determined by the preparation . time interval between two emissions is very large , so to exclude any interactions between successive particles . moreover , the single source excludes quantum interference . after one iteration or , equivalently , time step , the particle has a probability @xmath67 to be at the site @xmath68 , a probability @xmath69 to be at the site @xmath70 , and a probability @xmath19 to be at the site @xmath71 . after two iterations , the probabilities are : @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 . note that , since the functions @xmath77 and @xmath78 are symmetric , the probability function is symmetric with respect to @xmath70 . in general , the position probability function is described by the recursive equation @xmath79 deriving a closed formula for the probability mass function @xmath80 is tedious but straightforward at this point . with the initial condition @xmath81 , the result is @xmath82 with support @xmath83 , and can be verified by inspection . from this formula , the probability that a particle is at the event horizon , i.e. , @xmath84 , is easily retrieved as @xmath85 . similarly , @xmath86 . the function ( [ eqn : rhop ] ) has a limit for large @xmath10 s that can be derived in two equivalent ways . on the one hand , the equation of motion ( [ eqn : xstoc ] ) can be reviewed in the continuum limit as a stochastic differential equation reading @xmath87dn + \sqrt{\mathrm{var}}[\mathcal{v}_{n}]db , \label{eqn : sde}\ ] ] where @xmath88 is a brownian motion with zero mean and unit variance . now , from ( [ eqn : p])([eqn : a1 ] ) , the identity @xmath89=e - p^2=b$ ] follows . consequently , the continuum limit of @xmath90 is a gaussian function with mean @xmath91 and variance @xmath92 , that is , @xmath93 other derivations of ( [ eqn : normal ] ) are illustrated in appendix [ app : normal ] . appendix [ app : specrel ] shows that ( [ eqn : normal ] ) is invariant to lorentz transformations . a second stochastic variable of interest is the cumulated energy that the particle has experienced , defined by @xmath94 its pmf follows the binomial distribution with @xmath10 trials and @xmath95 probability of success , i.e. @xmath96 with support @xmath97 ( @xmath59 is the energy propensity , not the neper number ) . to describe particle dynamics as seen " by the lattice sites , we can define new stochastic variables . we will denote these variables with a time subscript _ and _ a position superscript , instead of the only time subscript as for the particle variables . in principle , any particle stochastic variable can be transformed into a lattice stochastic variable by imposing that @xmath98 . clearly , @xmath99 can only take the value @xmath11 in the single - source scenario we are considering . similarly , a variable @xmath100 would be deterministic , with its support including only the value @xmath10 . let me introduce a true stochastic variable , the _ site occupancy _ , defined as @xmath101 define @xmath102 $ ] . clearly , @xmath103 and its pmf is given by ( [ eqn : rhop ] ) . consider now @xmath104 , a fully stochastic variable representing the cumulated energy of the particle seen by a site . its pmf is calculated ( see appendix [ app : phis ] ) as @xmath105 with support @xmath106 . it can be also proved that @xmath107 = |\xi|+\frac{(|\xi|-\tau)(|\xi|-\tau+1)}{(2\tau-1 ) } = \frac{\xi^2+\tau^2-\tau}{(2\tau-1 ) } \label{eqn : es}\ ] ] and @xmath108 = 2\frac{(\xi^2-\tau^2)(\xi^2-(\tau-1)^2)}{(2\tau-1)^2(2\tau-3)}. \label{eqn : varl}\ ] ] equation ( [ eqn : es ] ) can be easily verified by inspection of a few sites , provided that @xmath109 . for example , for @xmath66 , one easily obtains @xmath110=1 $ ] . the cumulated energy of every particle reaching the site @xmath111 is obviously 1 . generalizing this result , clearly @xmath112=\tau$ ] . for a site like @xmath113 the prediction is less trivial . the possible values of the cumulated energy can be 2 , if the particle follows a back - and - forth path , or 0 , if it stays at rest for two time steps . using ( [ eqn : es ] ) , one obtains for this case @xmath114=2/3 $ ] , which is a weighted mean between the two possible values of cumulated energy . equation ( [ eqn : varl ] ) also can be verified by inspection . for instance , when @xmath115 or @xmath116 , it is apparent that @xmath104 can only take the value @xmath117 . in fact , ( [ eqn : varl ] ) yields @xmath118 = 0 $ ] . the continuum limit counterpart of ( [ eqn : prs ] ) is calculated as @xmath119 } } \exp\left(- \frac{(\sigma - e[\mathcal{s_{\tau}^{\xi}}])^2}{2\mathrm{var}[\mathcal{s}_{\tau}^{\xi}]}\right ) \label{eqn : phis}\ ] ] to retrieve the predictions of schrdinger s equation , a key element of the model is introduced . the proposed model assumes that whenever the particle interacts with the environment , its momentum propensity @xmath54 is properly reset . now , in the free motion scenario , consider for the moment that @xmath54 is a continous variable determined _ randomly _ during the preparation at the particle source . consequently , the probability of releasing a particle with a momentum propensity @xmath54 is uniform over the interval between -1 and + 1 , spannnig 2 , and thus the probability density of the momentum propensity is @xmath120 . when the source releases a large number of particles in succession , each one with a randomly determined value of @xmath54 , the probability of finding a particle at the location @xmath121 is given by the _ ensemble average _ and is calculated as @xmath122 introducing ( [ eqn : rhop ] ) into ( [ eqn : intp ] ) , and after some manipulations ( see appendix [ app : p ] ) , obtain @xmath123 . \label{eqn : p}\ ] ] moreover , it is easily verified that @xmath124 as obviously required . now , compare this result with the predictions of qm , i.e. , the particular solution of the schrdinger equation . for a single perfectly localized source at @xmath125 , the probability density is calculated ( see [ app : single ] ) as @xmath126 thus it is inversely proportional to time and it does not depend on @xmath0 . normalizing to lattice units and using ( [ eqn : hbar ] ) allows reducing ( [ eqn : f0 ] ) to @xmath127 this result compares with ( [ eqn : p ] ) , with @xmath128 replacing @xmath129 . the two functions of @xmath10 are very similar and , indeed , practically coincident for @xmath10 sufficiently large . in other terms , _ the square modulus of the wavefunction predicted by the schrdinger equation is the continuum limit of the probability in the proposed model_. note that only the particular formulation of energy propensity ( [ eqn : energy ] ) yields this result . other values for @xmath59 and consequently for @xmath69 ( for instance , @xmath130 ) , would yield position - dependent mass probability functions . define now the _ action _ at any site as @xmath131 dp . \label{eqn : action}\ ] ] note that the expected value given by equation ( [ eqn : es ] ) does not depend on @xmath54 . therefore , by putting it into ( [ eqn : action ] ) , obtain @xmath132 now , compare this result with the phase of the wavefunction ( [ eqn : psi2 ] ) . the latter , usually interpreted as the action of the particle , is @xmath133 which , in lattice units , becomes @xmath134 the latter equation corresponds to the second term in the right - hand side of ( [ eqn : s ] ) , that is , @xmath135 , multiplied by @xmath136 to obtain a phase angle . the correspondence is almost perfect , except for the term @xmath137 that in the proposed model replaces the term @xmath128 predicted by qm . for large values of @xmath10 , however , the two results are practically coincident . in other terms , _ the phase of the wavefunction predicted by the complex schrdinger equation is the continuum limit of the action in the proposed model_. the results in the previous sections have been derived for a probability density of the momentum propensity @xmath120 . consider now a generic function @xmath138 . apply ( [ eqn : intp ] ) to the continuum limit ( [ eqn : normal ] ) of the probability function @xmath139 . it can be shown that ( [ eqn : normal ] ) approximates a dirac delta function , whence @xmath140 the propability density function @xmath141 obeys the following partial differential equation @xmath142 introducing now the continuum - limit approximation of @xmath143 , @xmath144 one recognizes in ( [ eqn : schro1 ] ) the continuity equation @xmath145 on the other hand , the relationship @xmath146 also holds . sect . [ sec : wave ] has shown the equivalence of @xmath141 to @xmath147 , while sect . [ sec : phase ] that of @xmath148 with @xmath149 for large @xmath10 s . with these two substitutions , and reintroducing physical units instead of lattice units , equations ( [ eqn : continuity])([eqn : hjb ] ) become the continuity equation @xmath150 and the hamilton jacobi equation @xmath151 of the de broglie bohm formulation of qm , which in turn are equivalent to schrdinger s equation for a free particle . in the last sections , the predictions of the proposed model were shown in closed form , using mathematical equations in terms of _ a priori _ probabilities and probability fluxes . the probability of a number of observable were calculated and found to be in accord to the predictions of qm . now , i will present numerical simulations of the random walk of _ single _ particles and i will calculate the _ a posteriori _ probabilities as frequencies over a large number of emissions . thus , this section is aimed at reproducing numerically a true experiment . table [ tab : pseudo ] shows the algorithm used for such simulations . the two for - cycles are for the successively released @xmath152 particles , and for time up to @xmath153 . each particle experiences the choice of two randomly - selected values : ( i ) the momentum propensity and ( ii ) at each time step , its local velocity @xmath154 as a function of @xmath54 . the final code line represents the counting of the particle that arrive at a certain location at time @xmath153 . from this number of arrivals , an _ a posteriori _ frequency @xmath155 is calculated as the ratio to the total number of particles emitted . @xmath156 @xmath157 @xmath158 random value beween -1 and + 1 @xmath159 @xmath160 random value + 1 , 0 or -1 with prob . given by ( [ eqn : a ] ) @xmath161 @xmath162 @xmath163 figure [ fig : buildprob ] shows the frequency @xmath155 after a time @xmath164 for different values of @xmath152 . as the the number of particles emitted in the ensemble increases , a frequency distribution builds up . for large @xmath152 , the frequency clearly tends to the a priori probability @xmath165 , that is , a constant value given by ( [ eqn : p ] ) . as a function of @xmath11 after @xmath164 . from top - left to bottom - right , @xmath166 500 , 5000 , 10000 , and 50000 , respectively . ] after having reproduced the predictions of the schrdinger equation for a free particle , let me proceed now to a second puzzling aspect of qm : particle self - interference . double - slit experiment usually serves to visualize this phenomenon . however , the core of self - interference is isolated and better illustrated by a double - source preparation , where particles can be emitted by two alternative sources and the two possible paths interefere with each other . additional free - particle scenarios leading to self - intereference are multiple - source preparations and the particle in a ring " situation . the representation of these scenarios using only the process illustrated in sect . [ sec : single ] would give just the superimposition of probability densities of the type ( [ eqn : p ] ) and no interference would arise . instead , interference is originated in the proposed model from the interaction of successive particles emitted with the lattice , a process denoted here as quantum force " , which is illustrated in the next sections . in short , the emergence of quantum forces is the result of an exchange of information between the lattice and the particles . let me call this piece of information _ boson _ , in analogy with force - mediating particles . in the proposed model , each particle carries two counters . the spatial counter records the number of space steps crossed and thus it is represented by the stochastic variable @xmath167 . the temporal counter records the number of time steps crossed and thus is always certainly equal to @xmath10 . on the other hand , each lattice site maintains a sort of register " storing the value of the spatial counter of the last particle that has visited the site . therefore , this register can be described by the stochastic variable @xmath99 . when a particle visits a site , the site senses the value of the particle s spatial counter , say , @xmath168 . this value is compared with the value @xmath169 sensed at the last particle visit and stored in the site register . when the difference @xmath170 is nonzero , an exchange of information between the particle and the lattice site takes place . first , a couple of bosons " is created . one boson is taken by the particle ( later it will be explained how this boson acts on the particle ) , while the second stays at the site . the pair @xmath171 for which the bosons have been created distinguishes this couple from other possible bosons . for this reason let me call them @xmath172-bosons . the second event is that the value @xmath169 in the site register is replaced with the value @xmath168 , while the particle counter is replaced with the value @xmath169 . in other words , the site register and the particle counter are exchanged , @xmath173 thirdly , both new bosons are given a particular momentum . the particle boson takes the momentum @xmath174 that equals the momentum of the boson of the same type that was previously resident at the site , if there was one : @xmath175 where @xmath176 is the lifetime of the resident boson , that is , the number of iterations spanned from when it was created ( @xmath177 for a new boson ) . the boson that stays at the site suppresses the previously resident boson of the same type ( the site can carry multiple bosons only if they are characterized by different values @xmath171 ) and takes a momentum @xmath178 where @xmath179 is the particle sample momentum sensed at the site , a rational number given by the ratio of the particle space counter and its time counter ( @xmath180 ) . again , there is an exchange of momentum between the site and the particle , through their respective bosons . the momentum of the resident boson decays with its lifetime , according to the rule @xmath181 note that rule ( [ eqn : wdecay ] ) may be interpreted as a discrete analogous of an exponential decay . note also that , since @xmath182 and @xmath183 , also @xmath184 . each time a particle visits a site , it takes a new @xmath172-boson if the value of its spatial counter , @xmath168 , is different from the value @xmath169 of the site register . if a new boson is not taken , the particle keeps the boson it has , if it has one . a new boson suppresses a possible boson of the same type , as the particle can carry multiple bosons only if they are characterized by different values @xmath171 . each @xmath172-boson is created with a momentum @xmath174 given by ( [ eqn : pboson0 ] ) , but this value decreases over its lifetime according to the rule @xmath185 where @xmath186 is the lifetime of the boson , that is , the number of iterations spanned from when it was created ( @xmath187 for a new boson ) . rule ( [ eqn : pdecay ] ) may be interpreted as a discrete analogous of a decay that is @xmath188 . finally , the effective momentum propensity that regulates the particle motion at a given time is not the particle momentum @xmath54 but the total momentum @xmath189 where the summation is taken over all bosons carried by the particle and the @xmath186 index has been removed for simplicity since , at a certain time step , each boson has its own lifetime . simple as it is , the mechanism given by ( [ eqn : pboson0])([eqn : peff ] ) is capable of accounting for self - interference , as it will be proven in next section . with the mechanism illustrated in sect . [ sec : qforce ] , derive now the probability function @xmath190 for the two - source scenario , that is , two lattice sources separated by @xmath191 sites , labelled 1 ( source at @xmath192 ) and 2 ( source at @xmath193 ) , respectively . the probability that a particle is emitted from source 1 is @xmath194 , while @xmath195 is the probability of emission from source 2 . in this scenario , any site register @xmath99 can take only the values @xmath196 ( particle coming from source 1 ) and @xmath197 ( particle coming from source 2 ) . let me label these two alternative events 1 " and 2 " in the following . four types of @xmath172 events are possible , namely , 11 " , 12 " , 21 " , and 22 " . the probabilities of these events are easily calculated as @xmath198 , @xmath199 , @xmath200 , @xmath201 . the events 11 " and 22 " do not create any boson ( the particle counter is equal to the site register ) . consider now the bosons 12 " . the probability that the particle carries such boson equals the probability that it carries a new boson , _ or _ a boson that is one iteration - old , _ or _ a boson that is two iterations - old , etc . , all these events being mutually exclusive . such probability is therefore calculated as @xmath202=p_{12}+p_{12}(1-p_{12})+p_{12}(1-p_{12})^2+\ldots = p_{12 } \sum_{k=0}^{\tau-1 } ( 1-p_{12})^k \label{eqn : prboson}\ ] ] however , we consider large @xmath10 s probabilities to retrieve @xmath190 , so let @xmath10 in ( [ eqn : prboson ] ) tend to infinity . boson lifetime affects the momentum exchange , as per ( [ eqn : pdecay ] ) . thus the mean value of the boson momentum is calculated as @xmath203 = p_{12 } \sum_{k=0}^{\infty } ( 1-p_{12})^k p_{12}^{(k ) } \label{eqn : p12}\ ] ] from the definition ( [ eqn : pdecay ] ) , @xmath204 after some manipulations , the latter formula turns out to be formally equivalent to @xmath205 as it can be verified by inspection . therefore , ( [ eqn : p12 ] ) is rewritten as @xmath206}{p_{12}^{(0 ) } } = p_{12 } \sum_{k=0}^{\infty } ( 1-p_{12})^k ( -1)^k { -1/2 \choose k } = p_{12 } ( 1-(1-p_{12}))^{-1/2 } = \sqrt{p_{12}}\ ] ] after having recognized in the right - hand side the binomial series @xmath207 multiplied by @xmath208 . finally , the mean value of the boson momentum ( denoted for simplicity with the symbol @xmath209 without a superscript ) is @xmath210 and , with the same notation , @xmath211 since bosons 12 " and 21 " give the same contribution to the combined momentum ( while the events 11 " and 22 " do not generate bosons ) . consider now the quantity @xmath212 that is given by ( [ eqn : pboson0 ] ) and apply rule ( [ eqn : wdecay ] ) to compute @xmath213 where @xmath214 . as the site boson lifetime @xmath176 increases , the product in ( [ eqn : product ] ) tends to the sinc of @xmath215 , that is @xmath216 for large @xmath10 s the sample momentum @xmath217 tends to @xmath218 . consider now @xmath219 as a realization of the stochastic variable @xmath220 . using ( [ eqn : p ] ) , ( [ eqn : sde ] ) , and replacing @xmath54 with @xmath221 , it turns out that @xmath222 in particular , @xmath223 for large @xmath10 s . replacing this value into ( [ eqn : ep1 ] ) , obtain @xmath224 now the steady - state probability density of @xmath219 , denoted as @xmath225 , can be calculated given the probability density of @xmath54 , @xmath120 . using the rule @xmath226 , obtain @xmath227 finally , use substitution ( [ eqn : fsut ] ) to calculate the ensemble average @xmath228 this result is in perfect agreement with qm predictions , derived in appendix [ app:2slit ] from schrdinger s equation , under the equivalence rule @xmath229 derived in sect . [ sec : single ] and after transformation to lattice units . similar considerations apply for scenarios with @xmath230 sources , where @xmath231 is the probability of the i - th source ( @xmath232 ) and @xmath233 is the distance in lattice units between sources @xmath234 and @xmath235 , for @xmath236 . in these cases , there are @xmath237 possible types of bosons , each of which ultimately contributes to a term @xmath238 to the momentum pdf . since contribution of boson @xmath239 is the same as that of boson @xmath240 , they can be lumped to yield @xmath241 @xmath242 again , this result is in perfect agreement with qm predictions , derived in appendix [ app : multi ] from schrdinger s equation , after transformation to lattice units . also the well - known lattice in a ring " scenario can be simulated with the presented model , since the particle has no interactions with the environment . now the useful lattice does not extend infinitely in both directions but only to values @xmath243 $ ] , where @xmath176 is the ring circumference in lattice units . due to ring periodicity , it is now necessary to distinguish the stochastic variable @xmath244 , the site reached after @xmath10 time steps , from the particle spatial counter @xmath63 defined by ( [ eqn : xstoc ] ) . when @xmath245 , @xmath244 is reset to zero , and when @xmath246 , @xmath244 is reset to @xmath247 ( in other words , @xmath248 ) . the register @xmath99 at a site @xmath11 has not the same distribution as @xmath244 but as @xmath63 . therefore it can take the obvious value @xmath11 ( suppose for simplicity @xmath249 ) , but also the value @xmath250 , if the particle has reached the site after having crossed the entire ring once , the value @xmath251 , if the ring has been crossed twice , etc . also values @xmath252 , @xmath253 , @xmath254 are possible , if the ring is crossed in the opposite direction . summarizing , @xmath99 can take infinitely many distinct values @xmath255 . consequently , all possible particle paths interefere , with a path difference that is always a multiple of @xmath176 . in the proposed model this situation is equivalent to a multiple - source preparation , with infinitely many equally - probable and equally - spaced sources separated by a distance @xmath176 in lattice units . in this scenario , the probability density function of the particle momentum is @xmath256 , that is , we consider an ensemble of particles having the same momentum @xmath54 . we are interested in calculating the steady - state probability density of the total momentum @xmath257 . equation ( [ eqn : ep ] ) is replaced by @xmath258 in the limit for @xmath259 . it can be proven that this limits reads @xmath260\ ] ] where the operator @xmath261 $ ] denotes the rounding function . in other words , for a well - defined particle momentum @xmath54 , the steady - state momentum @xmath219 can only take one of the discrete values @xmath262 , that for @xmath263 $ ] . this result coincides precisely with qm predictions derived in appendix [ app : ring ] , after transformation to lattice units . this section is aimed at reproducing numerically the self - interference scenarios analysed in the previous sections . the model described in the previous sections is in principle implemented by the algorithm shown in table [ tab : codeint ] . with respect to the algorithm of table [ tab : pseudo ] , more instructions are now needed to represent the creation of pairs of bosons , their initialization , and the decay of their momentum . initialization : @xmath264 random value bewteen @xmath265 and @xmath266 @xmath267 @xmath268 random value beween -1 and + 1 @xmath269 @xmath270 @xmath271 particle dynamics : @xmath159 @xmath272 @xmath160 random value + 1 , 0 or -1 with prob . given by ( [ eqn : a ] ) @xmath161 @xmath273 @xmath274 momentum decay for particle bosons , see ( [ eqn : pdecay ] ) : @xmath275 @xmath276 momentum decay for lattice bosons , see ( [ eqn : wdecay ] ) : @xmath277 @xmath278 boson creation : site @xmath279 boson @xmath280 12 " boson @xmath280 21 " no boson @xmath69 boson momentum reinitialization : @xmath281 @xmath282 @xmath283 @xmath284 @xmath285 @xmath286 @xmath162 @xmath287 numerical results of fig . [ fig : interftot ] have been obtained with a modified version of the algorithm above , aimed at fastening calculations . in fact , the interference pattern starts to build when the frequency of @xmath172 events approaches the theoretical probabilities @xmath288 introduced in sect . [ sec : apriori ] . that requires that a sufficiently large number of lattice sites are visited by a sufficiently large number of particles . lattice training " process requires in turn that a large number , say @xmath289 , of particles are emitted . when we add the number of particle emissions required for the statistical build up of the intereference pattern , then it becomes clear that complete simulations would require extensive computing times . to fasten computations , i have considered the lattice training " already completed . in practice , the values @xmath169 found by a particle are not acessed in the site registers but randomly attributed with probabilities that equal the steady - state ones . in the two - slit scenario , @xmath169 equals @xmath217 with probability @xmath194 , and @xmath290 with probability @xmath291 . similarly , particle bosons are directly initialized with their expected steady - state momenta , see ( [ eqn : sinc ] ) . in fact , the convergence of the process ( [ eqn : product ] ) requires much less iterations than the training of the lattice , and it is completed well before @xmath289 emissions . the latter statement can be observed in fig . [ fig : w12 ] that shows pairs of values @xmath292 and @xmath293 in a lattice after @xmath294 and for @xmath164 ( no gap between two successive emissions ) . in lattice of size @xmath153 after @xmath294 particle emissions , vs. respective steady - state values.,scaledwidth=50.0% ] figure [ fig : interftot ] shows the frequency @xmath155 calculated for the two - slit scenario ( @xmath295 ) after a time @xmath164 for different values of @xmath296 , with the lattice already trained as specified above . as the the number of particles emitted in the ensemble increases , a frequency distribution builds up . for large @xmath152 , the frequency clearly tends to the a priori probability @xmath297 given by ( [ eqn : p2slit ] ) . as a function of @xmath11 after @xmath164 . from top - left to bottom - right , @xmath166 500 , 5000 , 10000 , and 50000 , respectively . ] two more scenarios are presented in fig . [ fig : interslit ] . the former is a case of two sources at distance @xmath295 with non - equal probabilities @xmath298 , @xmath299 . the second scenario is that of three equally - probable sources located at @xmath300 . the frequency values have been obtained with the accelerated algorithm discussed above . in the first scenario , two types of bosons are created , both having creation probability @xmath301 and distance @xmath191 . in the second scenario , there are two types of bosons with @xmath295 and four types of bosons with @xmath302 , all of them with probability equal to 1/9 . in both cases , the figure shows a precise agreement between large @xmath10 s prediction of the model with the theoretical probability densities . with respective probability 0.1 and 0.9 ( left ) , three equally - probable sources at @xmath300 ; both scenarios with @xmath303 , @xmath304 ; red curves : theoretical probability densities . the proposed approach has been proven capable of describing trajectories of individual particles in an ensemble of similarly - prepared particles in a simple and realistic way . simple and realistic means that the ontology of the proposed model includes real particles , a real discrete spacetime lattice , both capable of storing and exchanging a few pieces of information , and arithmetic operations . in fact , all quantities modeled are represented either by integers or rational numbers , and operations on them are products and sums . the predictions of the model have been shown to tend to the predictions of qm in the continuum limit for free particles and , more remarkably , also in the case of quantum interference . in my opinion , the latter evidence makes the model a successful candidate to provide both qualitative and _ quantitative _ explanation for several quantum phenomena . however , other qm aspects have still to be added to the model . the results of this paper concern free particles only . however , the proposed model seems naturally capable to integrate also external forces into the picture . each interaction of the particle with its sourrounding is indeed expected to modify its intrinsic properties . external forces shall be then treated as permanent variations of the particle intrinsic momentum propensity @xmath54 . relativistic newton s second law shall be considered , to prevent that @xmath54 becomes larger than unity under the action of persistent forces . a special case of external force is when the particle hits an infinite potential barrier ( that is , a site that carries such an information ) . when that happens , @xmath54 shall be assumed to instantaneously change its sign , similarly to what happens to the velocity of a classical particle . moreover , the multiple counters carried by the particle shall be properly reset . extension to two- and three - dimensional spaces seems also natural . a set of three momentum propensities , @xmath305 , @xmath306 , and @xmath307 shall be introduced , fulfilling the condition that the total energy @xmath308 . this condition implies that @xmath309 , thus fixing a constraint to the probability densities of the three propensities . bohm d. , hiley b.j . , the undivided universe : an ontological interpretation of quantum theory , routledge , london , 1993 . floyd e. , welcher weg ? a trajectory representation of a quantum diffraction experiment , found . phys . , 37(9):14031420 , 2007 . ballentine l.e . , quantum mechanics : a modern development , world scientific , singapore , 2006 . neumaier a. , ensembles and experiments in classical and quantum physics , int . b 17:29372980 , 2003 . nelson e. , derivation of the schrdinger equation from newtonian mechanics , phys . rev . 150:10791085 , 1966 . fritsche l. , haugk m. , a new look at the derivation of the schrodinger equation from newtonian mechanics , ann . ( leipzig ) 12(6):371-403 , 2003 . carroll r. , remarks on the schrdinger equation , inter . evolution equations 1:2356 , 2005 . grssing g. , sub - quantum thermodynamics as a basis of emergent quantum mechanics , entropy 12:19752044 , 2010 . dowker f. , henson j. , j. stat . phys . 115(516):13271339 , 2004 . bialynicki - birula i. , weyl , dirac , and maxwell equations on a lattice as unitary cellular automata , phys . d 49(12):69206927 . de raedt h. , de raedt k. , michielsen k. , event - based simulation of single - photon beam splitters and mach - zender interferometers , europhys . 69(6 ) , 861867 , 2005 . , quantum mechanics in a two - dimensional spacetime : what is a wavefunction ? phys . 324:12111218 , 2009 . janaswamy r. , transitional probabilities for the 4-state random walk on a lattice , j. phys . 41 , 2008 . badiali j.p . , entropy , time - irreversibility and the schrdinger equation in a primarily discrete spacetime , j. phys . a : math . gen . 38(13):28352848 , 2005 . chen h. , chen s. , doolen g. , lee y.c . , simple lattice gas models for waves , complex systems 2:259267 , 1988 . in some interpretations of quantum phenomena , a particle is associated with a matter wave , whose frequency is proportional to its energy via the planck constant . in the proposed model , the frequency is retrieved as the reciprocal of the _ average return time _ to any position of the lattice . to see that , define the probability mass function @xmath310 as the probability that a particle returns at an arbitrary position for the first time after a time @xmath311 . for example , @xmath312 , @xmath313 , etc . the general expression for @xmath310 is @xmath314 now , define the average return time as @xmath315 = \frac{\sum_{n=1}^{\infty}np(n)}{\sum_{n=1}^{\infty}p(n)}\ ] ] using the results @xmath316 one can find that @xmath317 @xmath318 and consequently @xmath319 as a function of @xmath69 . let me now introduce the energy with the substitution @xmath320 . after some tedious but straightforward manipulations of ( [ eqn : pn1])([eqn : pn2 ] ) , find the frequency @xmath321 as @xmath322 that is the relationship sought . it is easy to verify ( fig . [ fig : matter ] ) that for @xmath323 , @xmath324 , while for @xmath325 , also @xmath326 . moreover , for small values of @xmath59 , the relationship ( [ eqn : frequency ] ) is approximated by @xmath327 which is precisely the de broglie relation in lattice units . consider ( [ eqn : rhop ] ) as a binomial distribution @xmath329 with @xmath330 , @xmath331 , @xmath332 . for @xmath43 large enough an approximation of @xmath90 is a normal distribution with mean @xmath333 and variance @xmath334 . recalling that @xmath335 , obtain ( [ eqn : normal ] ) . yet a third possible method would start from expressing the recursive equation ( [ eqn : process ] ) as @xmath336 the latter difference equation has a continuum limit described by the differential equation @xmath337 which is a convective diffusion equation with @xmath59 playing the role of the diffusivity and @xmath54 of the convection velocity . note , however , that in ( [ eqn : normal ] ) the correct result for the diffusivity is @xmath320 and not @xmath59 as it would be predicted by ( [ eqn : process_pde ] ) . equations ( [ eqn : rhop ] ) or ( [ eqn : normal ] ) are invariant with respect to lorentz transformations . to see that , let me take a particle having momentum propensity @xmath54 and a reference frame that is moving at velocity @xmath338 with respect to the fundamental lattice . in this refercne frame , the fundamental lattice dimensions ( primed quantities ) are deformed as follows : @xmath339 where @xmath340 and @xmath341 , so that the relationship @xmath342 still holds . in both reference frames , the probability of having the particle at a site of the fundamental lattice @xmath346 must be the same . in the primed reference frame , ( [ eqn : normal ] ) is calculated as @xmath347 now , let me assume that the momentum propensity in the moving reference frame is @xmath348 in agreement with relativistic velocity addition formula . consequently , since @xmath349 , one easily verifies that @xmath350 with these relationships , it is easy verified that @xmath351 and @xmath352 using ( [ eqn : xpt ] ) and ( [ eqn : bt ] ) , it is easily verified that the quantity @xmath353 given by ( [ eqn : normalprime ] ) approximates @xmath354 given by ( [ eqn : normal ] ) when @xmath355 , that is , for large @xmath10 s . in reality , all the formulae above are valid for large @xmath10 s , since , for instance , i do not impose that @xmath356 or @xmath357 are integers . in our laboratory system ( primed reference frame moving with an unknown velocity @xmath338 with respect to the fundamental lattice ) , we prepare an experiment with a momentum propensity that we label as @xmath358 and we observe a probabilty mass function at a point in the spacetime that we label as @xmath359 , @xmath360 . although our assumptions on these values is incorrect , we measure the correct probability mass function using the @xmath90 formula . if @xmath361 , that is , in the reference frame of the particle , we obtain the following results : @xmath362 , @xmath363 , @xmath364 , @xmath365 , @xmath366 , @xmath367 . thus the particle itself sees " a broader lattice ( @xmath368 , @xmath369 ) . for @xmath370 , the lattice becomes infinitely large . the number of paths leading to a certain site is given by @xmath371 where @xmath372 , @xmath373 , and @xmath374 are the number of moves with @xmath375 , respectively . clearly , @xmath376 and , in order to reach exactly the site in question , @xmath377 . with these constraints , ( [ eqn : npath ] ) becomes @xmath378 the range of @xmath374 derives from the fact that all the binomial arguments are positive integers . therefore , @xmath379 , @xmath380 , @xmath381 . consequently , @xmath382 . for each value of @xmath374 there will be a number @xmath388 of paths with the same cumulated energy . the value taken by @xmath104 equals the sum @xmath389 since only these moves contribute to the cumulated energy . the relative probability of these @xmath388 paths over the totality of @xmath386 paths is thus ( [ eqn : prs ] ) . the wavefunction for a free particle is @xmath394 where the wavenumber @xmath186 is related to the momentum of the particle and @xmath395 for a single perfectly localized source at @xmath125 , ( [ eqn : phi ] ) reads @xmath396 and consequently ( [ eqn : psi ] ) is rewritten as @xmath397dk \label{eqn : psi1}\ ] ] that , integrated , yields @xmath398 . \label{eqn : psi2}\ ] ] the probability density is easily calculated as @xmath399 the solution of the schrdinger equation for the case where the two sources are equally probable is based on the linear superimposition of the two waveforms , @xmath400 where both @xmath401 and @xmath402 are obtained from ( [ eqn : psi2 ] ) by replacing the term @xmath403 ( that was valid for a source at @xmath125 ) with a term @xmath404 and @xmath405 , respectively . thus , @xmath406+\exp\left[i\frac{m(x+\frac{\delta}{2 } x)^2}{2\hbar t}\right]\right\}. \label{eqn:2psi}\ ] ] the probability density is given by @xmath407 where @xmath408 and @xmath409 are the two independent action values , that is , the phases of the two exponentials in ( [ eqn:2psi ] ) . as @xmath410 , obtain @xmath411 and thus an interference term arises with respect to ( [ eqn : f0 ] ) , due to the presence of two possible sources . the interference is related to the phase difference between the two waveforms . in the case the probability of the two sources is different , say , @xmath412 , equation ( [ eqn : psi2slit ] ) is replaced by @xmath413 . consequently , ( [ eqn : p2slit ] ) is replaced by @xmath414 if @xmath415 , ( [ eqn : p2slit ] ) is retrieved . in the case of @xmath416 sources , ( [ eqn : p2slit ] ) is generalized as follows . @xmath417 where @xmath418 is the locations of the @xmath234-th source and @xmath232 . if the sources are equally probable and equally spaced by a distance @xmath419 , ( [ eqn : p2slit ] ) is particularized as @xmath420 the particle in a ring " scenario can be described by assuming interference between two plane waves , with opposite wavenumbers , representing the two possible directions of motion along the ring . equation ( [ eqn : psi ] ) is still valid , but with @xmath421 , where @xmath191 is here the dirac delta function . as the term @xmath422 does not affect the amplitude of the wavefunction , it can be neglected to write @xmath423 however , periodicity imposes that @xmath424 ) , where @xmath425 is the ring radius , or @xmath426 , which implies that @xmath427 . this condition is satisfied for @xmath428 , where @xmath8 , or for @xmath429 in agreement with the somehow standard derivation that considers angular momentum and polar coordinates . in terms of momentum @xmath430 ( here the dimensional momentum , not the momentum propensity as in the paper body ) , ( [ eqn : kn ] ) reads @xmath431 where @xmath432 is the ring circumference .
this paper presents a simple model that mimics quantum mechanics ( qm ) results in terms of probability fields of free particles subject to self - interference , without using schrdinger equation or wavefunctions . unlike the standard qm picture , the proposed model only uses integer - valued quantities and arithmetic operations . in particular , it assumes a discrete spacetime under the form of an euclidean lattice . the proposed approach describes individual particle trajectories as random walks . transition probabilities are simple functions of a few quantities that are either randomly associated to the particles during their preparation , or stored in the lattice sites they visit during the walk . non - relativistic qm predictions , particularly self - interference , are retrieved as probability distributions of similarly - prepared ensembles of particles . extension to interacting particles is discussed but not detailed in this paper .
the primary motivation for a next generation @xmath3 linear collider is the study of collisions at centre of mass energies beyond 200gev . however , by adding an auxiliary positron source , it may be possible to run with high luminosity ( @xmath4 ) and polarised beams at the peak of the resonance @xcite . this luminosity would allow the accumulation of @xmath5 hadronic decays in about 70 days running , opening up the possibility to repeat the electroweak measurements currently performed at lep and sld with much higher precision . a sample of @xmath5 decays represents a gain of about 100 over the lep1 data sample . however , with the beam polarisation and good b - tagging capability at a linear collider , the more relevant comparison is with sld , where the gain will be more like a factor of 2000 . with beam polarisation , the variable that is most sensitive to the effective weak mixing angle @xmath6 is @xmath0 , the left - right asymmetry in decays . using @xmath0 , sld at slac was able to measure the weak mixing angle to a better accuracy than any single observable at lep with less than a tenth of the luminosity @xcite . although the present sld analysis technique is limited by the precision of the electron polarisation measurement , the availability of positron polarisation at a linear collider allows the polarisation to be measured from cross section asymmetries . using a scheme first proposed by a. blondel for lep @xcite , external measurements are then only needed for polarisation differences . this large data sample would also contain about @xmath7 b hadrons , a large enough sample to study rare cp - violating b decays . the linear collider offers a number of advantages over other facilities : the beam polarisation allows a very good initial state tag from the b - quark forward - backward asymmetry , the data sample is larger than that from asymmetric @xmath8 b factories , whilst the environment is much cleaner than that at hadron colliders . competitive measurements of the angles of the ckm unitarity triangle may thus be possible . the precision attainable on such electroweak and cp - violation measurements is studied in this paper . the basic assumptions used are described in section [ s : setup ] , followed by discussions of the lineshape parameters in section [ s : lines ] and in section [ s : rb ] . the measurement of @xmath0 , including the blondel scheme for the polarisation determination , is described in detail in section [ s : alr ] , followed by @xmath1 in section [ s : ab ] . the potential for measurement of the ckm unitarity triangle angles @xmath9 , @xmath10 , @xmath11 and @xmath12 is investigated in detail in section [ s : cp ] . finally , the conclusions are given in section [ s : conc ] . the tesla linear collider should be able to deliver a luminosity of @xmath13 at the -pole with the aid of an auxiliary positron source @xcite . with this luminosity , the sample of @xmath5 hadronic ( corresponding to @xmath14 ) assumed in this paper can be produced in about 70 days running a small cost to the overall physics program of the collider . the detector , trigger and data acquisition must of course be able to cope with the trigger rate of 200hz from hadronic decays . it will be assumed that the electron and positron beams can be polarised with polarisations of @xmath15% and @xmath16% . this corresponds to an equivalent polarisation of @xmath17 . additionally , the polarisation should be switchable randomly from train to train . the energies of both the electron and positron beams can be measured with a spectrometer similar to the one currently being installed in lep @xcite . fast energy scans should be possible , so that the spectrometer can be calibrated relative to @xmath18 from lep . with this scheme a precision of 1mev might be reached . it should be noted that the relevant parameter to correct the electroweak measurements is the difference @xmath19 , so that the error on @xmath18 does not enter when the spectrometer is calibrated relative to the mass . the parameters that are measured at lep in the scan around the resonance are : * the mass of the @xmath20 * the width of the @xmath21 * the hadronic peak cross section @xmath22 * the ratio of the hadronic and leptonic partial widths @xmath23 since @xmath18 is used to calibrate the spectrometer , its precision clearly can not be improved at tesla . with the large data sample , the error on @xmath24 is equal to the error on @xmath25 , so 1mev is within reach . the error on the hadronic peak cross section is dominated by uncertainties coming from the luminosity measurement , mainly from theory . it is possible that with the high luminosity available at tesla , processes with low cross section but small uncertainties like @xmath26 can be used , but here it will be assumed that no improvement on @xmath27 will be possible . the partial width ratio @xmath28 receives substantial contributions from systematic uncertainties due to the leptonic and hadronic event selection . the uncertainties from the leptonic event selection , especially for muons , are mainly of a statistical nature , so can be assumed to scale with the luminosity . the uncertainty on the hadronic event selection will be assumed to stay at the best value reached at lep ( 0.05% ) . a more careful study on this subject is in progress @xcite . the interesting physics parameters that can be derived from the lineshape parameters are * the mass of the @xmath20 * the strong coupling constant at the -mass @xmath29 * the radiative correction parameter normalising the strength of the -couplings to the fermions @xmath30 * the number of light neutrino species @xmath31 the possible improvements in these parameters are summarised in table [ tab : line ] . it can be seen that an interesting gain can be obtained for @xmath32 and @xmath33 . @xmath34 is completely determined by @xmath27 , so no change is expected if the luminosity determination can not be improved . .possible improvements in the physics quantities derived from the -lineshape after high luminosity -running at tesla . for @xmath35 and @xmath36 is assumed . [ cols="^,^,^",options="header " , ] the expected invariant mass spectra for @xmath37 and @xmath38 are shown in figure [ f : b0pi0bg](c ) and ( d ) . the background histograms are extrapolations from the small number of surviving events , so their shapes are somewhat uncertain . nevertheless , it is clear that a good understanding of the background shape will be necessary to measure the signal branching ratios , and this may be rather difficult to achieve . in summary , the measurement of the branching ratios @xmath39 and @xmath40 would be an important contribution to the understanding of cp violation in b decays , since it can not be performed at hadron colliders . however , it looks rather difficult with the present detector and uncertainties in the level of background . to proceed further probably requires a more realistic calorimeter simulation and a more detailed study of the @xmath41 reconstruction and background rejection . the study of time dependent asymmetries in decays provides a method of determining the ckm angle @xmath11 and the small angle @xmath12 ( which is the ` tilt ' of the second unitarity triangle formed from the relation @xmath45 ) . mesons are not produced at all in @xmath46 decays ( so are inaccessible to @xmath8 b factories unless they run at a higher resonance ) and are only produced in 10% of b decays at higher energies . the mixing frequency is also at least 25 times higher than that for mesons , so requiring very good time resolution to see the oscillations and asymmetries . both of these factors make studies of decays much more challenging experimentally than studies of decays . searches for oscillations at lep , sld and the tevatron have so far been carried out mainly using semi - inclusive modes such as @xmath49 and @xmath50 . in these analyses , the energy of the ( which is necessary to calculate the decay proper time ) is estimated using reconstructed and missing energy . however , the limit on is now @xmath51 @xcite , and measuring significantly higher frequencies can only be done with rare fully reconstructed decays . the decay chain @xmath52 , with @xmath53 , @xmath54 provides a suitable decay mode to measure or set a limit on , whilst the similar @xmath55 provides a way of measuring @xmath56 . these channels have been studied using the full geant simulation brahms . the decay chain was fully reconstructed , requiring two oppositely charged tracks to make a @xmath57 candidate with @xmath58gev , combined with another track to make a candidate with @xmath59gev and a further opposite signed track to make a candidate . the 3 tracks from the were required to make a good vertex , as were the extrapolated track from the and the @xmath60 from the . the latter vertex was used to define the decay length . the energy was additionally required to exceed 20gev , but no particle i d ( ) cuts were applied . the resulting mass distribution is shown in figure [ f : bsrec](a ) , for @xmath52 signal events , and the same events reconstructed assuming the @xmath60 from the decay to have the kaon mass . the latter distribution , which forms a ` reflection ' background for the reconstruction of @xmath61 is reasonably separated from the signal , without the need for particle i d . the signal mass resolution is about 8mev , without the application of any kinematic fit . the reconstruction efficiency for this selection is about 20% , but many of the losses occur in requiring good vertices , which may be improved with better tracking code . assuming @xmath62 ( by analogy with @xmath63 @xcite ) and @xmath64 @xcite gives a total of about 500 reconstructed events in @xmath5 decays . the proper time of each decay was reconstructed from the measured decay length ( between the reconstructed primary vertex and the intersection of the @xmath65 and @xmath60 tracks ) and reconstructed energy of each candidate . the energy resolution is extremely good ( 0.5% ) and the proper time resolution is dominated by the decay length reconstruction . this was simulated separately for each of the vertex detector options available in brahms ccd vertex detectors surrounding 1 cm and 2 cm radius beampipes , and an aps vertex detector with a 2 cm beampipe . the resulting proper time resolutions are shown in figure [ f : bsrec](b)(d ) , together with the resolutions obtained from a double gaussian fit to the distribution of reconstructed minus true proper time . the resolution clearly depends strongly on which vertex detector option is chosen . the sensitivity to oscillations as a function of can be quantified by means of the amplitude method @xcite . the amplitude @xmath66 of oscillations at each ` test ' value of is measured , in a method analogous to fourier analysis . oscillations are excluded if the value @xmath67 is excluded by the measurement , whilst exclusion of @xmath68 is required to establish a signal . the significance of the analysis is then quantified by the inverse of the error on @xmath66 , @xmath69 , which must be at least 5 to claim a signal ( ` @xmath70 discovery ' ) , and at least @xmath71 to set a 95% confidence level exclusion . the expected error can be calculated approximately by : @xmath72 where @xmath73 is the number of events in the sample , the signal purity , @xmath74 the mistag and @xmath75 the proper time resolution . the results of this calculation for @xmath76 and @xmath77 are shown in figure [ f : dmslim ] for each of the vertex detector options . the background to the @xmath52 signal is expected to be combinatorial and very small , due to the very good mass resolution and two intermediate mass constraints ( @xmath57 and ) . hence values of close to one are expected to be realistic , although this has not been studied with simulation . with @xmath77 , it can be seen that oscillations can be discovered between 20 and @xmath78 , depending on the vertex detector options . this is comparable with the reach of cdf and atlas , though slightly lower than the @xmath79 expected by lhc - b @xcite . the larger reach of lhc - b is due to the much higher numbers of reconstructed events30000 as opposed to 500 for the linear collider , though the proper time resolutions are similar and the tagging better at the linear collider . however , such high values of should also be visible through their effects on the width difference ( @xmath80 ) , and are currently disfavoured within the framework of the standard model . the decay @xmath55 can be used to measure the ckm angle @xmath56 . however , its branching ratio is expected to be at least an order of magnitude smaller than that for @xmath52 , so only around 50 reconstructed events are expected in @xmath5 decays not enough to measure the 4 time dependent decay rates necessary for this analysis . lhc - b expects around 2500 events per year in this channel @xcite so the linear collider is clearly not competitive , even with the large statistical advantage in tagging power . the decay @xmath82 is the analogue of the @xmath83 decay , and can be used to measure the ckm angle @xmath12 . this asymmetry is expected to be very small in the standard model , so a significant asymmetry would be a sign of new physics . however , as with the other decay modes discussed above , the relatively small number of mesons produced at the linear collider make it uncompetitive with lhc - b , and also with atlas and cms in this decay channel which is easy to trigger on . the reconstruction of @xmath82 is similar to @xmath83 , and with a 30% reconstruction efficiency 700 events are expected in @xmath5 . atlas and lhc - b expect around 40000 events @xcite , so again the linear collider is at a severe statistical disadvantage . various other strategies , involving the decays of and mesons , have been proposed to extract or constrain @xmath11 . some of these have been studied for babar @xcite and lhc - b @xcite . one method is to use rare @xmath86 and @xmath87 decays , however the combined branching ratios are so small that only a few events per year are expected at babar , and a few tens of events at lhc - b . in @xmath5 the linear collider should collect similar numbers to lhc - b , so if these analyses are possible at all , the linear collider should be able to make a useful contribution . however , these channels have not yet been studied in detail . another proposed method is to make use of flavour symmetries in @xmath88 and @xmath89 decays . the feasibility of reconstructing two body b decays has been discussed in section [ s : alpha ] . the reconstruction of @xmath90 has not yet been studied but should present no major problems . again , if this analysis is possible , the linear collider should be able to make a useful contribution . however , the measurement of @xmath91 is also required , and if the penguin decay @xmath92 is also important , the linear collider will be at a disadvantage in not having particle id the mass resolution can not be used effectively in final states containing reconstructed @xmath41 . the possibilities for doing electroweak and cp violation physics using a sample of @xmath5 hadronic decays collected at a linear collider factory with polarised beams have been investigated . huge improvements in the measurements of the coupling parameters @xmath93 and @xmath1 will be possible . @xmath1 can be measured to a precision on @xmath94 and @xmath93 to a precision of @xmath95 , corresponding to an error in @xmath6 of @xmath96 . however , to make this measurement useful the hadronic cross section below the @xmath97 needs to be measured to about 1% precision , in order to calculate the running of the fine structure constant up to the mass . if the higgs has already been found , a stringent test of the standard model will be possible , and if not the higgs mass can be predicted with 5% precision . some interesting predictions in other models , such as the mssm , can also be made @xcite . however , the partial width measurements can not be improved very much due to systematic error limitations . for @xmath98 a factor five may be possible , while for @xmath32 and @xmath33 only a factor two is possible . for the cp violation studies , whilst no ` golden ' measurements have been found that can be performed better than at any other facility , many useful measurements can be made that are complementary , and of a competitive precision , to those at @xmath8 b factories , the tevatron and lhc . these measurements can be made with the standard linear collider detector optimised for high energy running , and do not require dedicated particle identification detectors . useful contributions can be made to the study of the ckm angles @xmath9 and @xmath10 in a variety of decay modes . however , the study of @xmath11 , which mainly requires meson decays , does not appear to be competitive with the lhc , due to the much larger numbers of b decays produced at the latter machine . in summary , with relatively little investment in time and money a large improvement in the electroweak precision observables can be achieved , and some competitive and complimentary cp violation measurements can be performed . the factory option is thus an attractive supplement to the already rich linear collider physics program . the authors would like to thank r. heuer , f. jegerlehner , t. mannel , t. ohl , p. rowson , m. winter and m. woods for useful discussions . e. torrence , _ determination of the lep centre - of - mass energy _ , talk presented at the eps99 conference , tampere , finland , 1521 july 1999 , to appear in the conference proceedings . m. winter , private communication delphi collaboration , p.abreu et al . , _ a precise measurement of the partial decay width ratio @xmath99 _ cern - ep/98 - 180 , submited to eur . j. f. jegerlehner , _ sigma hadronic and precision tests of the sm , _ talk presented at the lnf spring school and euroda@xmath100ne collaboration meeting , frascati 12.-17 . april 99 + ` http://www.ifh.de/~fjeger/frascati99.ps.gz ` brahms version 1.02 , a monte carlo for a detector at a 500/800gev linear collider , t. behnke , g. blair , m. elsing , k. mnig , m. pohl , for a detailed description see http:///www.hep.ph.rhbnc.ac.uk/ blair / detsim / brahms / summary.ps . combined results on oscillations : results for summer 1998 conferences , lep b oscillations working group , lepbosc 98/3 , september 4th , 1998 , for further details see http://www.cern.ch / lepbosc / combined_results / vancouver_1998/.
a future linear collider such as tesla may be able to run on the resonance with very high luminosity and polarised electron and positron beams . the possibilities of measuring electroweak quantities with high precision are investigated . huge improvements with respect to the present precision can be expected , especially for the asymmetries @xmath0 and @xmath1 where beam polarisation can be exploited . the very large sample of @xmath2 events also allows studies of various cp - violating b decays . the precision achievable on the ckm unitarity triangle angles is comparable to experiments at b factories and future hadron colliders . desy 99 - 157 + october 12 , 1999 + * electroweak and cp violation physics + at a linear collider z - factory + * r. hawkings + desy - flc / universitt hamburg + k. mnig + desy zeuthen
one of the interesting discoveries from the studies of asteroid rotation is the `` slivan state '' @xcite , which is the spin vector alignment and rotation period clustering of relatively large asteroids ( @xmath2 mag and @xmath3 km ) in the koronis family , which is a relatively old family ( 1.83 gyr * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) at outer main - belt ( semimajor axis of @xmath4 2.87 au ) . later , @xcite used the yarkovsky - okeefe - radzievskii - paddack effect ( yorp ; * ? ? ? * ) to explain how the spin states of the koronis family members could be driven within 23 gyr into the @xmath5 spin - orbit secular resonance between the precession rate of the asteroid s spin axis and saturn s longitude of node . this discovery reveals the subtle influence of thermal effect on asteroid rotation . such spin vector alignment has also been reported in the flora family @xcite . however , this was not seen in the maria family @xcite . determination of asteroid s spin vectors is extremely time - consuming as it requires intensive lightcurve observations at various phase angles . therefore , it is difficult to verify the `` slivan state '' in asteroid families . however , as @xcite pointed out , rotation period clustering should be observed in the koronis family once the `` slivan state '' is established , in which the prograde rotators are confined in rotation periods between 710 hr and the retrograde rotators are driven to rotation periods of @xmath6 or @xmath7 hr . from this point of view , rotation period clustering can be used as proxy of the `` slivan state '' for the koronis family even if the rotational directions of family members are determined . it is believed that an asteroid family was formed in a catastrophic break - up of its parent body which resulted in an initial dispersion of the fragments semimajor axes . further dispersion occurred as the family members drifted away from the family s center due to the yarkovsky effect . according to @xcite , the prograde / retrograde rotators would increase / decrease their semimajor axes ( hereafter , outer / inner sides ) . therefore , a simple assumption can be made on the family members outside the initial dispersion in semimajor axis . those at the inner side ( hereafter , inner - side members ) are expected to be mostly retrograde rotators and those at the outer side ( hereafter , outer - side members ) should be mostly prograde rotators . in this way , we should be able to see more outer - side members of the koronis family in rotation periods of 710 hr and correspondingly less inner - side members in rotation periods of 513 hr . the ptf / iptf is an all - sky time - series survey , which has six - year archived data and has been conducting several dedicated observations for asteroid spin - rate study . the large number of asteroid rotation periods obtained from archived data and dedicated observations of ptf / iptf is very useful to test the aforementioned idea on the koronis family @xcite . in section 2 , we describe our algorithm in detail . the data used in the analysis are given in section 3 . results and discussion are presented in section 4 . a summary and conclusion can be found in section 5 . an asteroid family is a group of asteroids , which share similar proper elements of semimajor axis ( @xmath8 ) , eccentricity ( @xmath9 ) and inclination ( @xmath10 ) , resulting from a catastrophic break - up event . when an asteroid family was formed , the fragments followed the size - velocity relation , which describes the relationship between the sizes of asteroids and their velocity dispersions . different velocity dispersions in the break - up events would produce different initial dispersions in semimajor axis ( @xmath11 ; see * ? ? ? * and references therein ) . this can be used in age determination of asteroid families ( see examples in * ? ? ? * ) . to calculate @xmath11 , we follow the approach of @xcite , @xmath12 where @xmath13 is the mean motion of the break - up fragment , @xmath14 is its transverse velocity , and @xmath15 refers to other terms related to eccentricity , which is neglected when circular orbit is considered . moreover , we assumed size - velocity relationship @xmath16 for the velocity dispersion of a fragment with size @xmath17 , where @xmath18 is the velocity dispersion of our reference diameter @xmath19 km . therefore , @xmath20 in addition , the diameter @xmath17 is related to @xmath21 by @xmath22 where @xmath23 is geometric albedo and we adopt the average value @xmath24 for the koronis family @xcite . we can quantify @xmath11 on the plot of @xmath8 vs. @xmath21 . in addition , the orbital motion of the family members are modified by the yarkovsky effect , where prograde and retrograde rotators would move outward and inward along semimajor axis , respectively . after certain amount of time , some members would leave @xmath11 . therefore , we can expect that family members located beyond the @xmath11 interval should be subdivided in two categories : those moving to greater ( i.e. , outer - side members ) and smaller ( i.e. , inner - side members ) semi - major axes are expected to be mostly prograde and retrograde rotators , respectively . therefore , a clustering of the rotation period around 7 - 10 hr is expected among the former objects in case they are in a slivan state , whereas the objects moving to smaller semi - major axis should exhibit a lack of rotation periods between 5 and 13 hr . @xcite used analytic / synthetic proper elements to calculate asteroid dynamical family memberships with the hierarchical clustering method ( hcm ; * ? ? ? * ; * ? ? ? * ) and generated the data set , nesvorny hcm asteroid families v2.0 . to select the koronis family members , we adopt both catalogs calculated with analytic and synthetic proper elements . however , we exclude the members of the karin family , a sub - family in the koronis family , from the selected sample . to ensure the membership of the aforementioned selected objects , we also use the asteroid family catalog of @xcite as a reference . in addition , the sdss colors @xcite of these dynamically selected members , if available , are used to identify the spectral types of the family members . the rotation periods of the koronis family members used in this study are collected from two data sets and each is briefly described below . _ the archived ptf asteroid rotation periods _ * ; hereafter ptf - waszczak ) . the data set used 5-yr archived data of ptf / iptf to obtain 54,296 sparsely - sampled asteroid lightcurves , from which @xmath4 8,300 reliable rotation periods were determined by fitting rotation and phase - function model simultaneously . the data set is sensitive to detect spin rate of @xmath25 rev / day ( see figure 7 in @xcite ) . most objects in the data set are main - belt asteroids of @xmath26 km . therefore , this data set includes plenty of members of asteroid families and is the main source used in this study to verify rotation period clustering of asteroid families . _ the dedicated ptf spin - rate study _ * ; * ? ? ? * ; hereafter ptf - chang ) . there are 1,751 reliable rotation periods obtained from three dedicated surveys . each survey used the palomar 48-inch oschin schmidt telescope to observe 87 deg@xmath27 in mould-@xmath28 band with 20 min candence in the ecliptic plane . they were carried out in february 1518 of 2013 and in january 69 and february 2023 of 2014 , respectively . the collected asteroid lightcurves were analyzed by the traditional second - order fourier series method to find rotation period @xcite . the surveys have relatively lower ability to detect rotation period @xmath29 hr ( see figure 5 in @xcite ) . the data set mainly contains main - belt asteroids of hundred meter to 20 km in size . we discard the rotation periods having difference @xmath30 % between both data sets and obtain 173 unique asteroids in total for the koronis family ( see table [ table1 ] ) . among 5146 hcm selected koronis family members , 1131 objects have sdss colors . their colors show a preferential concentration of s - type in figure [ fig_sdss ] . this is expected for the koronis family as pointed out by @xcite . therefore , those non - s - type objects in the hcm selected koronis family members are considered as interlopers , which only make up a very minor population ( i.e. , @xmath31 15% ) and would have limited affect in the following analysis . the typical `` v '' shape of asteroid families is apparent in the plot of @xmath8 vs. @xmath21 of the koronis family ( see figure [ fig_ah ] ) . the boundaries at @xmath32 and @xmath33 au are due to the powerful 5/2 and 7/3 mean - motion resonances with jupiter . we also notice that a large part of the sample of @xcite are at least @xmath42 mag brighter than ours . in figure [ fig_ah ] , @xmath11 of @xmath34 50 and 100 m / s are drawn in solid and dashed lines , respectively , where red represents the inner side of the koronis family and green is the outer side . the family members locating at the outside the red and green lines are inner- and outer - side members , respectively . to inspect rotation period clustering of the slivan state , we use a bimodal distribution to divide the rotation periods into `` clustered '' ( i.e. , presenting the slivan state ) and `` non - clustered '' ( i.e. , not presenting the slivan state ) . therefore , the clustered rotation period for the inner - side members is @xmath6 or @xmath7 hr , and that for the outer - side members is between 710 hr . if the rotation period does not fulfill the aforementioned criteria , it is classified as non - clustered rotation period . the results are given in figure [ fig_cluster ] . the ratios of clustered to non - clustered outer - side members are 0.14 for @xmath35 m / s and 0.09 for @xmath36 m / s . this indicates that most of outer - side members are probably not in the slivan state . the ratios of clustered to non - clustered inner - side members are 0.84 for @xmath35 m / s and 0.38 for @xmath36 m / s . although these ratios of inner - side members are higher than the outer - side members , its distributions are still dominated by non - clustered rotation periods . therefore , we conclude that rotation period clustering is not observed in our results . if the inner-/outer - side members are dominated by retrograde / prograde rotators as the general scenario of yarkovsky drifting described in section [ method ] , our method should work properly and consequently rotation period clustering should be seen in our analysis on the koronis family . this means that determination of the sense of rotation in our sample would be helpful to verify whether retrograde / prograde rotators are dominant in the inner-/outer - side members of the koronis family . we note that , the old koronis family could have undergone a certain degree of collision after the initial catastrophic break - up event , such that the spin states of the members could have altered and consequently the dispersion in semimajor axis would not evolve in the way as described in section [ method ] . such intermittent collisional process capable of changing the rotational configuration is expected to be more frequent for the small members of the koronis family . figure [ fig_dp ] is the plot of @xmath17 vs rotation period and it shows that we have much smaller diameter in our sample than @xcite . this might be the reason why the rotation period clustering is not observed in our result . therefore , numerical simulation of yarkovsky drifting for the koronis family members including random collisional effect would be helpful to understand whether the scenario of the slivan state should be applicable to the general population of for the koronis family . since the koronis family is relatively old , a population of interlopers in our sample could jeopardize our method to separate prograde and retrograde rotators . however , the koronis family members with available sdss colors display a concentration of s - type in a c - type dominant environment ( i.e. , the outer main belt ) . therefore , we believe that the population of interlopers should be minor in the hcm selected members , unless these interlopers happen to be s - type asteroids . @xcite reported a `` stray '' koronis family member , ( 263 ) dresda . this object is a prograde rotator with @xmath37 mag locating at the outer side . however , its spin status is not in the slivan state ( i.e. , rotation period of 16.81 hr and obliquity of @xmath38 ) . because of the initial spin state and location of dresda , the yorp effect could not drive it into the slivan state @xcite . such non - clustered spin state evolution was also studied by @xcite . therefore , the `` stray '' dresda may provide another scenario for spin state evolution for the koronis family . moreover , the yorp effect could also destroy the slivan state , especially it works more efficiently on small asteroid . we suspect that the non - clustering rotation period distribution is probably caused by this mechanism . asteroid families are believed to have been produced by catastrophic break - up events , forming swarms of fragments distributed in intervals of orbital proper elements @xmath11 , @xmath39 , @xmath40 , reminiscent of the orbital elements of their parent bodies . in the subsequent orbital evolution , prograde and retrograde rotators would have drifted out of the @xmath11 to the far and near sides of the family by yarkovsky effect , respectively . if the slivan state exists in the bulk of koronis family , the outer - side members would show rotation period clustering within 7 - 10 hr and the inner - side members would avoid rotation period interval of 513 hr @xcite . with the large data set of rotation periods reported in ptf - waszczak and ptf - chang , we do not see rotation period clustering in the inner-/outer - side members of the koronis family with @xmath41 50 and 100 m / s . available sdss colors for the koronis family members shows 85% of them are of s - type , we therefore believe that our method can not be seriously affected by the limited contribution of interlopers . it would be very important to measure the pole orientations of our sample , and thereby to verify our conclusion . moreover , the `` stray '' spin state of the family member , ( 263 ) dresda , perhaps represents another pathway for the spin state evolution of the koronis family . finally it is noted that intermittent collisional process in the asteroid belt could lead to further disruption and change in the spin states of small objects not accounted for by the secular yorp effect . we thank the referee , alberto cellino , for his useful suggestions and comments to improve the content of the paper . we also thank chung - ming ko for his english correction . this work is supported in part by the national science council of taiwan under the grants most 104 - 2112-m-008 - 014-my3 and most 101 - 2119-m-008 - 007-my3 , and also by macau science and technology fund no . 017/2014/a1 of msar . binzel , r. p. 1988 , , 73 , 303 cellino , a. , michel , p. , tanga , p. , et al . 1999 , , 141 , 79 cellino , a. , bus , s. j. , doressoundiram , a. , & lazzaro , d. 2002 , asteroids iii , 633 chang , c .- k . , ip , w .- h . , lin , h .- w . , et al . 2014 , , 788 , 17 chang , c .- k . , ip , w .- h . , lin , h .- w . , et al . 2015 , arxiv:1506.08493 farinella , p. , & vokrouhlicky , d. 1999 , science , 283 , 1507 greenberg , r. , bottke , w. f. , nolan , m. , et al . 1996 , , 120 , 106 harris , a. w. , young , j. w. , bowell , e. , et al . 1989 , , 77 , 171 harris , a. w. , & lagerros , j. s. v. 2002 , asteroids iii , 205 kim , m .- j . , choi , y .- j . , moon , h .- k . , et al . 2014 , , 147 , 56 kryszczyska , a. , colas , f. , poliska , m. , et al . 2012 , , 546 , a72 kryszczyska , a. 2013 , , 551 , a102 masiero , j. r. , mainzer , a. k. , grav , t. , et al . 2011 , , 741 , 68 milani , a. , cellino , a. , kneevi , z. , et al . 2014 , , 239 , 46 nesvorny , d. , nesvorny hcm asteroid families v2.0 . ear - a - vargbdet-5-nesvornyfam - v2.0 . nasa planetary data system , 2012 parker , a. , ivezi , . , juri , m. , et al . 2008 , , 198 , 138 rubincam , d. p. 2000 , , 148 , 2 slivan , s. m. 2002 , , 419 , 49 slivan , s. m. , binzel , r. p. , kaasalainen , m. , et al . 2009 , , 200 , 514 spoto , f. , milani , a. , & kneevi , z. 2015 , , 257 , 275 vokrouhlick , d. 1998 , , 335 , 1093 vokrouhlick , d. , nesvorn , d. , & bottke , w. f. 2003 , , 425 , 147 vokrouhlick , d. , & farinella , p. 1998 , , 335 , 351 vokrouhlick , d. , & farinella , p. 1999 , , 118 , 3049 vokrouhlick , d. , bro , m. , bottke , w. f. , nesvorn , d. , & morbidelli , a. 2006 , , 182 , 118 waszczak , a. , chang , c .- k . , ofek , e. o. , et al . 2015 , arxiv:1504.04041 zappala , v. , cellino , a. , farinella , p. , & knezevic , z. 1990 , , 100 , 2030 zappala , v. , cellino , a. , farinella , p. , & milani , a. 1994 , , 107 , 772 zappal , v. , cellino , a. , delloro , a. , & paolicchi , p. 2002 , asteroids iii , 619 rrrrrrrrrrr 975 & 2.872 & 0.045 & 0.036 & 10.20 & 22.2@xmath42 & - & - & - & 7.22 & 0.20 + 1100 & 2.868 & 0.047 & 0.035 & 10.47 & 16.9@xmath42 & - & - & - & 14.53 & 0.10 + 1223 & 2.875 & 0.048 & 0.037 & 10.30 & 22.8@xmath42 & - & - & - & 7.83 & 0.28 + 1423 & 2.870 & 0.044 & 0.037 & 10.82 & 19.6@xmath42 & 0.10 & -0.05 & s & 12.29 & 0.96 + 1497 & 2.918 & 0.080 & 0.035 & 11.30 & 16.3@xmath42 & - & - & - & 3.30 & 0.20 + 1618 & 2.870 & 0.043 & 0.036 & 11.68 & 17.5@xmath42 & 0.09 & 0.02 & s & 43.32 & 0.65 + 1802 & 2.872 & 0.044 & 0.039 & 11.46 & 12.4@xmath42 & - & - & - & 3.16 & 0.37 + 1912 & 2.860 & 0.045 & 0.037 & 11.66 & 26.3 & 0.14 & -0.01 & s & 4.63 & 0.47 + 1913 & 2.928 & 0.071 & 0.039 & 11.64 & 13.4@xmath42 & - & - & - & 13.97 & 0.27 + 1955 & 2.920 & 0.073 & 0.037 & 11.97 & 9.8@xmath42 & 0.14 & -0.02 & s & 5.57 & 0.39 + 2123 & 2.868 & 0.048 & 0.036 & 11.33 & 14.5@xmath42 & 0.09 & 0.00 & s & 16.30 & 0.19 + 2498 & 2.865 & 0.046 & 0.034 & 12.35 & 11.3@xmath42 & - & - & - & 3.06 & 0.15 + 2620 & 2.880 & 0.047 & 0.036 & 12.62 & 16.9 & - & - & - & 3.39 & 0.56 + 2713 & 2.883 & 0.046 & 0.038 & 11.42 & 15.1@xmath42 & 0.09 & -0.12 & s & 3.56 & 0.61 + 2726 & 2.876 & 0.049 & 0.037 & 11.99 & 10.9@xmath42 & 0.13 & 0.00 & s & 4.75 & 0.26 + 2811 & 2.936 & 0.070 & 0.035 & 11.59 & 12.3@xmath42 & - & - & - & 3.25 & 0.32 + 2814 & 2.876 & 0.044 & 0.034 & 12.03 & 9.9@xmath42 & - & - & - & 3.37 & 0.18 + 2833 & 2.856 & 0.046 & 0.037 & 11.70 & 25.8 & 0.12 & 0.00 & s & 3.49 & 0.11 + 2931 & 2.855 & 0.045 & 0.037 & 11.63 & 11.7@xmath42 & 0.12 & -0.06 & s & 36.82 & 0.23 + 2953 & 2.857 & 0.043 & 0.038 & 11.64 & 12.6@xmath42 & - & - & - & 6.23 & 0.15 + 2985 & 2.866 & 0.046 & 0.033 & 11.92 & 10.5@xmath42 & - & - & - & 6.06 & 0.48 + 3131 & 2.859 & 0.049 & 0.037 & 11.94 & 23.2 & 0.15 & 0.00 & s & 19.75 & 0.70 + 3377 & 2.882 & 0.043 & 0.037 & 12.47 & 7.7@xmath42 & 0.12 & -0.04 & s & 6.76 & 0.33 + 3726 & 2.853 & 0.044 & 0.037 & 12.02 & 10.4@xmath42 & 0.07 & -0.01 & s & 3.51 & 0.11 + 3791 & 2.884 & 0.047 & 0.037 & 11.82 & 11.7@xmath42 & - & - & - & 10.58 & 0.18 + 3975 & 2.883 & 0.049 & 0.037 & 12.23 & 9.1@xmath42 & 0.09 & -0.01 & s & 8.16 & 0.71 + 4351 & 2.888 & 0.047 & 0.036 & 12.43 & 8.2@xmath42 & - & - & - & 2.92 & 0.27 + 4365 & 2.879 & 0.071 & 0.035 & 12.60 & 7.4@xmath42 & - & - & - & 9.84 & 0.69 + 4506 & 2.925 & 0.073 & 0.037 & 12.33 & 8.8@xmath42 & 0.05 & 0.07 & s & 5.52 & 0.76 + 4664 & 2.866 & 0.050 & 0.036 & 12.36 & 19.0 & - & - & - & 3.11 & 0.31 + 4965 & 2.855 & 0.047 & 0.037 & 13.86 & 9.6 & - & - & - & 12.55 & 0.35 + 5187 & 2.859 & 0.041 & 0.036 & 12.34 & 8.9@xmath42 & - & - & - & 7.52 & 0.30 + 5494 & 2.851 & 0.046 & 0.037 & 12.62 & 8.9@xmath42 & - & - & - & 5.73 & 0.98 + 5523 & 2.877 & 0.042 & 0.038 & 13.42 & 8.1@xmath42 & - & - & - & 53.33 & 0.31 + 5582 & 2.850 & 0.045 & 0.037 & 13.37 & 7.6@xmath42 & 0.32 & -0.35 & v & 4.50 & 0.23 + 5677 & 2.869 & 0.042 & 0.040 & 13.22 & 8.8@xmath42 & 0.08 & 0.02 & s & 5.08 & 0.20 + 5759 & 2.874 & 0.050 & 0.036 & 13.68 & 7.0@xmath42 & - & - & - & 4.20 & 0.61 + 5941 & 2.891 & 0.044 & 0.036 & 13.25 & 12.6 & - & - & - & 10.19 & 0.94 + 6006 & 2.891 & 0.044 & 0.037 & 12.78 & 7.5@xmath42 & 0.11 & 0.00 & s & 2.77 & 0.28 + 6210 & 2.887 & 0.043 & 0.035 & 12.58 & 8.7@xmath42 & - & - & - & 2.92 & 0.25 + 6262 & 2.888 & 0.043 & 0.037 & 12.60 & 7.8@xmath42 & - & - & - & 8.02 & 0.49 + 6686 & 2.848 & 0.045 & 0.036 & 12.68 & 8.6@xmath42 & - & - & - & 4.71 & 0.42 + 6829 & 2.860 & 0.040 & 0.036 & 12.98 & 7.3@xmath42 & - & - & - & 27.82 & 0.28 + 6980 & 2.846 & 0.043 & 0.037 & 12.97 & 8.8@xmath42 & - & - & - & 3.25 & 0.40 + 7028 & 2.932 & 0.070 & 0.035 & 12.77 & 7.3@xmath42 & - & - & - & 10.43 & 0.24 + 7120 & 2.879 & 0.041 & 0.040 & 13.09 & 13.6 & - & - & - & 2.87 & 0.18 + 7210 & 2.864 & 0.052 & 0.036 & 12.55 & 8.2@xmath42 & 0.09 & -0.04 & s & 6.24 & 0.29 + 7244 & 2.857 & 0.052 & 0.036 & 13.08 & 7.5@xmath42 & - & - & - & 35.03 & 0.83 + 7340 & 2.851 & 0.051 & 0.036 & 13.76 & 5.6@xmath42 & - & - & - & 10.70 & 0.76 + 7372 & 2.888 & 0.049 & 0.038 & 12.66 & 16.6 & - & - & - & 2.56 & 0.19 + 8102 & 2.890 & 0.051 & 0.037 & 13.55 & 5.4@xmath42 & - & - & - & 3.33 & 0.32 + 8234 & 2.898 & 0.045 & 0.035 & 13.74 & 10.1 & - & - & - & 33.09 & 0.49 + 8361 & 2.884 & 0.040 & 0.040 & 13.11 & 6.3@xmath42 & 0.10 & 0.06 & s & 3.60 & 0.75 + 8571 & 2.843 & 0.044 & 0.036 & 13.18 & 6.2@xmath42 & - & - & - & 5.30 & 0.54 + 8657 & 2.845 & 0.048 & 0.037 & 13.32 & 5.9@xmath42 & - & - & - & 13.47 & 0.67 + 9078 & 2.843 & 0.042 & 0.039 & 12.99 & 6.8@xmath42 & - & - & - & 45.65 & 0.44 + 9122 & 2.879 & 0.038 & 0.037 & 14.12 & 8.5 & 0.09 & -0.01 & s & 7.92 & 0.27 + 9306 & 2.861 & 0.053 & 0.036 & 12.81 & 7.2@xmath42 & - & - & - & 3.79 & 0.18 + 9337 & 2.846 & 0.050 & 0.037 & 13.32 & 5.9@xmath42 & - & - & - & 3.14 & 0.13 + 9726 & 2.848 & 0.039 & 0.037 & 14.82 & 6.1 & 0.10 & 0.01 & s & 5.55 & 0.28 + 9916 & 2.947 & 0.069 & 0.038 & 13.16 & 6.2@xmath42 & 0.10 & 0.01 & s & 15.48 & 0.45 + 9999 & 2.874 & 0.053 & 0.037 & 12.91 & 7.1@xmath42 & - & - & - & 3.48 & 0.15 + 10025 & 2.951 & 0.077 & 0.036 & 12.69 & 8.2@xmath42 & - & - & - & 4.44 & 0.48 + 10052 & 2.870 & 0.037 & 0.038 & 13.09 & 6.1@xmath42 & - & - & - & 4.07 & 0.24 + 11762 & 2.840 & 0.042 & 0.036 & 13.13 & 13.4 & - & - & - & 6.22 & 0.53 + 12186 & 2.898 & 0.047 & 0.037 & 13.77 & 9.9 & 0.12 & 0.00 & s & 11.20 & 0.29 + 12356 & 2.897 & 0.050 & 0.035 & 13.14 & 6.0@xmath42 & 0.10 & -0.02 & s & 6.45 & 0.38 + 12540 & 2.899 & 0.050 & 0.036 & 13.06 & 7.2@xmath42 & - & - & - & 3.55 & 0.67 + 12564 & 2.894 & 0.041 & 0.037 & 14.28 & 5.4@xmath42 & - & - & - & 7.04 & 0.44 + 12838 & 2.887 & 0.039 & 0.039 & 12.77 & 15.8 & - & - & - & 10.91 & 0.48 + 13266 & 2.837 & 0.045 & 0.038 & 13.84 & 9.6 & - & - & - & 4.85 & 0.64 + 13306 & 2.839 & 0.048 & 0.037 & 13.16 & 7.1@xmath42 & - & - & - & 5.43 & 0.41 + 13806 & 2.861 & 0.054 & 0.036 & 14.38 & 7.5 & - & - & - & 20.70 & 0.26 + 14164 & 2.835 & 0.042 & 0.036 & 14.31 & 7.8 & 0.09 & 0.01 & s & 11.86 & 0.71 + 14243 & 2.835 & 0.043 & 0.038 & 14.19 & 8.2 & - & - & - & 10.20 & 0.73 + 14246 & 2.905 & 0.045 & 0.036 & 13.14 & 6.8@xmath42 & 0.10 & -0.03 & s & 5.77 & 0.70 + 14382 & 2.906 & 0.046 & 0.036 & 13.96 & 5.8@xmath42 & - & - & - & 6.58 & 0.29 + 14779 & 2.903 & 0.043 & 0.035 & 13.32 & 6.8@xmath42 & - & - & - & 35.07 & 0.64 + 14977 & 2.895 & 0.039 & 0.038 & 14.71 & 4.9@xmath42 & - & - & - & 44.03 & 0.71 + 14992 & 2.887 & 0.038 & 0.040 & 13.85 & 5.6@xmath42 & 0.06 & -0.02 & s & 12.96 & 0.47 + 15195 & 2.945 & 0.078 & 0.038 & 14.76 & 6.3@xmath42 & - & - & - & 13.33 & 0.41 + 15339 & 2.927 & 0.082 & 0.036 & 13.51 & 5.0@xmath42 & - & - & - & 5.61 & 0.25 + 15409 & 2.896 & 0.051 & 0.033 & 13.76 & 6.7@xmath42 & - & - & - & 2.72 & 0.13 + 15414 & 2.950 & 0.086 & 0.036 & 13.79 & 4.2@xmath42 & - & - & - & 2.71 & 0.21 + 15421 & 2.834 & 0.044 & 0.038 & 14.37 & 7.6 & - & - & - & 7.54 & 0.35 + 15460 & 2.859 & 0.036 & 0.036 & 14.11 & 5.4@xmath42 & - & - & - & 7.27 & 0.22 + 15564 & 2.951 & 0.081 & 0.036 & 13.71 & 5.3@xmath42 & - & - & - & 6.08 & 0.38 + 16131 & 2.835 & 0.048 & 0.036 & 14.24 & 8.0 & - & - & - & 2.73 & 0.49 + 16199 & 2.834 & 0.048 & 0.038 & 14.30 & 7.8 & - & - & - & 3.96 & 0.73 + 16326 & 2.900 & 0.073 & 0.035 & 13.50 & 5.6@xmath42 & - & - & - & 3.07 & 0.28 + 16483 & 2.834 & 0.041 & 0.039 & 13.48 & 5.1@xmath42 & - & - & - & 12.03 & 0.17 + 16541 & 2.848 & 0.040 & 0.041 & 13.04 & 14.0 & 0.09 & -0.04 & s & 6.15 & 0.33 + 16626 & 2.892 & 0.037 & 0.037 & 13.69 & 10.3 & - & - & - & 6.81 & 0.37 + 16909 & 2.844 & 0.054 & 0.037 & 13.05 & 6.9@xmath42 & - & - & - & 6.79 & 0.38 + 16917 & 2.902 & 0.040 & 0.034 & 13.39 & 11.9 & - & - & - & 321.55 & 0.69 + 17163 & 2.885 & 0.055 & 0.038 & 13.89 & 4.9@xmath42 & 0.06 & -0.04 & s & 4.11 & 0.23 + 17605 & 2.894 & 0.037 & 0.038 & 14.12 & 8.5 & 0.09 & -0.02 & s & 6.32 & 0.58 + 17694 & 2.840 & 0.038 & 0.037 & 13.67 & 10.4 & - & - & - & 4.65 & 0.33 + 17779 & 2.830 & 0.042 & 0.039 & 13.39 & 11.9 & 0.09 & -0.06 & s & 6.51 & 0.40 + 18930 & 2.899 & 0.040 & 0.040 & 14.37 & 7.5 & 0.14 & 0.00 & s & 13.23 & 0.79 + 19083 & 2.862 & 0.054 & 0.041 & 13.35 & 12.1 & 0.08 & -0.05 & s & 3.05 & 0.25 + 20774 & 2.900 & 0.043 & 0.041 & 13.48 & 6.5@xmath42 & 0.16 & -0.02 & s & 3.65 & 0.17 + 21027 & 2.907 & 0.045 & 0.040 & 14.26 & 7.9 & - & - & - & 3.53 & 0.27 + 21566 & 2.901 & 0.040 & 0.034 & 13.84 & 5.2@xmath42 & - & - & - & 3.08 & 0.25 + 21580 & 2.906 & 0.041 & 0.035 & 13.10 & 6.2@xmath42 & 0.05 & 0.00 & s & 3.63 & 0.28 + 21993 & 2.914 & 0.048 & 0.037 & 14.37 & 7.6 & - & - & - & 5.83 & 0.68 + 22497 & 2.917 & 0.045 & 0.036 & 13.73 & 10.1 & - & - & - & 5.95 & 0.26 + 22899 & 2.910 & 0.047 & 0.039 & 13.68 & 5.7@xmath42 & - & - & - & 5.02 & 0.14 + 23056 & 2.896 & 0.056 & 0.035 & 13.85 & 4.3@xmath42 & - & - & - & 6.65 & 0.42 + 23543 & 2.834 & 0.047 & 0.033 & 12.84 & 7.2@xmath42 & - & - & - & 6.03 & 0.22 + 23873 & 2.863 & 0.035 & 0.038 & 13.71 & 5.7@xmath42 & - & - & - & 13.86 & 1.23 + 24874 & 2.897 & 0.035 & 0.040 & 14.70 & 4.6@xmath42 & - & - & - & 3.24 & 0.23 + 25112 & 2.862 & 0.053 & 0.042 & 14.74 & 6.4 & 0.11 & -0.07 & s & 10.13 & 0.35 + 25171 & 2.918 & 0.042 & 0.035 & 14.47 & 7.2 & 0.06 & -0.05 & s & 11.44 & 0.33 + 25241 & 2.916 & 0.043 & 0.034 & 13.32 & 6.0@xmath42 & - & - & - & 5.15 & 0.58 + 25285 & 2.916 & 0.041 & 0.034 & 13.82 & 7.6@xmath42 & - & - & - & 35.34 & 0.76 + 27502 & 2.948 & 0.081 & 0.037 & 14.55 & 6.9@xmath42 & - & - & - & 6.96 & 0.34 + 30305 & 2.901 & 0.055 & 0.036 & 14.44 & 4.4@xmath42 & - & - & - & 32.64 & 0.57 + 31050 & 2.904 & 0.055 & 0.036 & 14.36 & 5.1@xmath42 & 0.11 & -0.06 & s & 41.97 & 0.42 + 32873 & 2.855 & 0.038 & 0.039 & 14.90 & 5.9 & - & - & - & 4.64 & 0.54 + 34087 & 2.828 & 0.053 & 0.038 & 14.41 & 7.4 & 0.12 & 0.07 & s & 4.17 & 0.53 + 34469 & 2.905 & 0.049 & 0.040 & 14.70 & 6.5 & 0.07 & -0.01 & s & 6.04 & 0.63 + 35513 & 2.917 & 0.054 & 0.034 & 14.62 & 6.7 & - & - & - & 13.58 & 0.53 + 36036 & 2.895 & 0.057 & 0.037 & 14.94 & 5.8 & - & - & - & 9.46 & 0.25 + 36056 & 2.868 & 0.060 & 0.035 & 13.34 & 12.2 & - & - & - & 14.48 & 0.30 + 36281 & 2.839 & 0.057 & 0.038 & 14.56 & 6.9 & 0.09 & 0.03 & s & 16.14 & 0.24 + 37346 & 2.924 & 0.076 & 0.033 & 14.38 & 3.6@xmath42 & - & - & - & 3.26 & 0.44 + 38272 & 2.840 & 0.060 & 0.037 & 13.43 & 5.6@xmath42 & - & - & - & 3.90 & 0.15 + 38963 & 2.929 & 0.047 & 0.038 & 14.96 & 5.8 & 0.12 & -0.02 & s & 2.73 & 0.43 + 38992 & 2.918 & 0.054 & 0.037 & 14.98 & 5.7 & - & - & - & 6.92 & 0.61 + 41797 & 2.871 & 0.034 & 0.038 & 14.64 & 6.7 & 0.08 & 0.23 & s & 10.14 & 0.39 + 43883 & 2.836 & 0.059 & 0.034 & 15.66 & 4.2 & - & - & - & 28.59 & 0.32 + 45259 & 2.900 & 0.059 & 0.037 & 14.75 & 6.3 & - & - & - & 11.61 & 0.79 + 45303 & 2.926 & 0.051 & 0.037 & 14.13 & 8.4 & - & - & - & 8.39 & 0.25 + 45586 & 2.908 & 0.059 & 0.033 & 13.83 & 9.7 & - & - & - & 9.19 & 0.31 + 45628 & 2.873 & 0.034 & 0.042 & 15.22 & 5.1 & 0.03 & -0.03 & s & 3.48 & 0.39 + 46583 & 2.927 & 0.056 & 0.036 & 15.53 & 4.4 & 0.09 & -0.03 & s & 8.52 & 0.27 + 47691 & 2.944 & 0.066 & 0.032 & 13.28 & 5.3@xmath42 & - & - & - & 3.87 & 0.54 + 50493 & 2.841 & 0.063 & 0.037 & 15.40 & 4.7 & - & - & - & 4.64 & 0.40 + 50811 & 2.937 & 0.076 & 0.032 & 14.65 & 3.6@xmath42 & - & - & - & 5.16 & 0.42 + 51987 & 2.935 & 0.090 & 0.036 & 14.61 & 6.8@xmath42 & 0.08 & -0.01 & s & 2.82 & 0.22 + 52179 & 2.942 & 0.073 & 0.036 & 15.39 & 4.7@xmath42 & 0.08 & 0.01 & s & 4.97 & 0.81 + 53298 & 2.920 & 0.076 & 0.038 & 13.45 & 6.5@xmath42 & 0.16 & -0.01 & s & 5.51 & 0.43 + 54272 & 2.832 & 0.063 & 0.038 & 15.20 & 5.1 & 0.11 & -0.05 & s & 31.86 & 0.58 + 55004 & 2.946 & 0.084 & 0.036 & 13.99 & 9.0@xmath42 & - & - & - & 3.81 & 0.39 + 56558 & 2.935 & 0.081 & 0.034 & 14.92 & 5.9@xmath42 & 0.07 & 0.12 & s & 3.56 & 0.24 + 60552 & 2.902 & 0.034 & 0.036 & 15.62 & 4.2 & - & - & - & 29.24 & 0.56 + 60679 & 2.950 & 0.087 & 0.033 & 14.72 & 6.4@xmath42 & 0.06 & 0.02 & s & 5.55 & 0.87 + 62298 & 2.925 & 0.057 & 0.040 & 14.20 & 8.2 & - & - & - & 84.86 & 0.51 + 63660 & 2.871 & 0.064 & 0.038 & 13.65 & 4.7@xmath42 & - & - & - & 14.85 & 0.21 + 63795 & 2.934 & 0.056 & 0.039 & 15.92 & 3.7 & - & - & - & 3.87 & 0.62 + 66742 & 2.938 & 0.059 & 0.037 & 14.31 & 7.8 & - & - & - & 12.70 & 0.23 + 69934 & 2.938 & 0.058 & 0.037 & 15.10 & 5.4 & 0.11 & 0.02 & s & 25.11 & 0.46 + 71049 & 2.941 & 0.053 & 0.036 & 14.35 & 5.4@xmath42 & 0.08 & -0.00 & s & 56.30 & 0.46 + 72653 & 2.920 & 0.064 & 0.037 & 14.37 & 7.6 & 0.12 & -0.01 & s & 10.79 & 0.46 + 76069 & 2.928 & 0.063 & 0.039 & 15.73 & 4.0 & - & - & - & 9.08 & 0.68 + 76142 & 2.903 & 0.064 & 0.037 & 15.35 & 4.8 & - & - & - & 5.50 & 0.70 + 78596 & 2.853 & 0.066 & 0.037 & 15.40 & 4.7 & - & - & - & 28.51 & 0.67 + 79988 & 2.911 & 0.064 & 0.037 & 15.41 & 4.7 & - & - & - & 7.33 & 0.41 + 81205 & 2.907 & 0.076 & 0.038 & 15.05 & 5.5@xmath42 & - & - & - & 4.95 & 0.78 + 82730 & 2.899 & 0.062 & 0.035 & 15.78 & 3.9 & 0.11 & -0.00 & s & 17.33 & 0.56 + 83422 & 2.952 & 0.081 & 0.034 & 14.67 & 6.6@xmath42 & - & - & - & 9.64 & 0.55 + 97059 & 2.890 & 0.066 & 0.037 & 14.97 & 5.7 & 0.11 & -0.03 & s & 3.94 & 0.41 + 104188 & 2.948 & 0.053 & 0.037 & 15.89 & 3.8 & - & - & - & 50.10 & 0.83 + 112888 & 2.895 & 0.030 & 0.037 & 15.46 & 4.6 & - & - & - & 35.36 & 0.40 + 113049 & 2.886 & 0.066 & 0.036 & 15.09 & 5.4 & - & - & - & 6.44 & 0.51 + 113515 & 2.934 & 0.090 & 0.037 & 15.03 & 5.6@xmath42 & - & - & - & 4.57 & 0.69 + 116503 & 2.855 & 0.065 & 0.035 & 15.86 & 3.8 & 0.18 & -0.14 & s & 11.82 & 0.68 + 132830 & 2.946 & 0.061 & 0.037 & 15.99 & 3.6 & 0.12 & -0.02 & s & 5.91 & 0.40 + 133528 & 2.918 & 0.067 & 0.037 & 15.50 & 4.5 & 0.10 & -0.04 & s & 3.05 & 0.35 + 135223 & 2.942 & 0.085 & 0.038 & 14.91 & 5.9@xmath42 & 0.09 & 0.08 & s & 7.93 & 0.78 + 198605 & 2.907 & 0.067 & 0.037 & 15.79 & 3.9 & 0.06 & 0.00 & s & 160.77 & 0.38 + 268623 & 2.903 & 0.066 & 0.039 & 15.76 & 4.0 & - & - & - & 7.01 & 0.68 +
rotation period clustering in prograde / retrograde rotators might indicate the preliminary indication of the slivan state in the koronis family as a result of the yorp effect . we follow the general scenario of dispersion in semimajor axis of the asteroid family members to separate prograde and retrograde rotators in the koronis family . from the available rotation periods obtained from ptf / iptf , we can not found the rotation period clustering of objects with @xmath0 mag in the koronis family . this could be the result of the intermittent collisional process of small asteroids ( @xmath1 km ) which leads to astray yarkovsky drifting . measurement of the pole orientations of our sample will verify our preliminary result and validate our method .
over the past few decades photoemission based tools like photoelectron spectroscopy ( pes ) and angle - resolved photoelectron spectroscopy ( arpes ) have proven extremely successful in studying the chemical and electronic structure of solid state materials and surfaces@xcite . as a result , physics of the photoemission phenomena has been well investigated with regards to explaining the angle resolved electron energy spectra obtained using uv and x - ray light sources . more recently , photoemission has gained popularity as a source of electrons for several applications like free electron lasers ( fels)@xcite and ultrafast electron diffraction ( ued)@xcite experiments . the quantum efficiency ( qe ) and the transverse ( to the normal on the photoemission surface ) momentum spread or the rms transverse momentum are the most critical figures of merit of the photoemission based electron sources ( or photocathodes ) that limit the performance of such applications@xcite . for example , the transverse coherence length of the electron beam in ued which limits the largest lattice size that can be studied is inversely proportional to the rms transverse momentum of electrons emitted from the cathode@xcite . the transverse momentum spread also limits the smallest possible electron beam emittance which defines the shortest possible lasing wavelength of an fel@xcite . the qe determines the drive laser power needed to obtain the electron bunch charge required for the particular application ; a low qe can implies high drive laser power often making the drive laser system prohibitively complex and expensive@xcite . high drive laser power can also limit the smallest possible rms transverse momentum through ultrafast laser heating of the electron gas@xcite . hence a high qe is required . despite the technological importance of solid state photoemission as an electron source , the physics that governs the relevant photoemission properties of qe and rms transverse momentum is not well understood . the first theory to model these photoemission parameters from metal cathodes was proposed by dowell and schmerge and followed a three step photoemission model@xcite . this theory successfully explained the qe and rms transverse momentum obtained from polycrystalline or disordered cathodes but did not model photoemission very close to the threshold accurately . an extension to this theory was developed recently , to model photoemission near the threshold@xcite . it showed that the smallest possible rms transverse momentum from polycrystalline surfaces is thermally limited by the temperature of the lattice . however these models did not include the effects of band structure , polarization and angle of incident light ( the vectorial photoelectric effect@xcite and did not model emission from single crystal surfaces of metals . a technique to estimate the rms transverse momentum spread from single crystal faces of metal cathodes using the band structure calculated from density functional theory was developed by schroeder _ _ et al__@xcite . however , this technique does not estimate the qe and assumes uniform probability of photoemission from any given electron state which is generally not true . in this paper , we present a scheme to calculate the qe and transverse momentum spread accurately using an one - step photoemission model . our model is a 3-d expansion of the 1-d photoemission model developed by miller _ _ et al.__@xcite to explain ultraviolet photoemission spectra ( ups ) obtained from single crystals surfaces of noble metals . photoemission is modelled as a one step process of the transition of electrons from the initial bulk or surface state ( ss ) inside the metal to a time - reversed leed like free electron state under the influence of the electromagnetic field of the incident light . we obtain the rate of such a transition using the fermi golden rule to calculate the qe and the rms transverse momentum of emitted electrons . this photoemission model takes into account the effects of band structure , polarization and angle of incident light . using the ( 111 ) surface of silver as an example , we show that our model predicts the qe near threshold accurately and explains the effects of polarization of incident light and angle of incidence quantitatively . we show that the emission from an ag(111 ) surface at threshold , at an angle of incidence near 60@xmath0 is dominated by the electrons emitted from the shockley surface state@xcite resulting in a qe of greater than @xmath1 very close to the photoemission threshold . we also calculate the rms transverse momentum of electrons emitted from the ag(111 ) surface . we show that the ag(111 ) surface can simultaneously provide a high qe and a low rms transverse momentum very close to the thermal limit@xcite and hence can be used as an excellent electron source . the dependence of qe on the polarization of incident light and angle of incidence is called the vectorial photoelectric effect and has been investigated experimentally , but has been modeled only empirically@xcite . using our scheme to calculate the qe we show that the vectorial photoelectric effect results from the variation of the overlap integral with the angle of incidence and polarization of incident light and can be modeled without use of any empirical data . we assume that the normal to the solid - vacuum interface is along the @xmath2 direction and the classical interface is located at @xmath3 , with @xmath4 being the vacuum side . the hamiltonian of the photon - electron interaction is given by @xmath5 where @xmath6 is the momentum operator , @xmath7 is the vector potential of the incident light , @xmath8 is the unit charge , @xmath9 is the speed of light and @xmath10 is the mass of a free electron in vacuum . the vector potential of incident light inside the metal surface can be given by @xmath11 , where @xmath12 is the magnitude of the incident vector potential just outside the surface , @xmath13 is the polarization vector inside the surface , @xmath14 is the decay length of the incident light in the metal and @xmath15 is the location of the interface adjusted to account for the spilling over of the electron cloud into vacuum@xcite due to the surface state . for the ag(111 ) surface @xmath15 is determined by wave function matching of the shockley surface state at the solid - vacuum interface as shown in section iii b. note that the magnitude of polarization vector @xmath13 is not unity and takes into account the reflection at the surface as given in section ii b. the incident photon flux per unit area is given by @xmath16where @xmath17 is the frequency of incident light , @xmath18 is the dielectric constant of vacuum and @xmath19 is the angle of incidence@xcite . for ultra - violet light , the wavelength is long enough that the @xmath20 term in equation [ eq : hamiltonian ] can be ignored everywhere except at the metal - vacuum interface . at the metal - vacuum interface , there is a sharp discontinuity in @xmath7 in the @xmath2 direction and @xmath20 results in a delta function at @xmath21 . the hamiltonian is then given by @xmath22 where @xmath23 is the @xmath2 component of @xmath13 , @xmath24 is the heaviside function , @xmath25 and c is a constant that depends only on the photon energy and the properties of the solid . the constant c can be obtained by fitting the calculations of the 1d model to the photoemission electron spectroscopy data@xcite . photoemission from single crystal surfaces can be modeled as a transition process of an electron between an initial bulk or surface state ( ss ) inside the lattice with wave function @xmath26 to a time reversed leed like free electron state in vacuum with wave function @xmath27 under the influence of incident light@xcite . the total transition rate of this process is given by fermi s golden rule as @xmath28 where the summations are over all possible initial and final states , @xmath29 and @xmath30 are the energies of the initial and final states respectively , the @xmath31 function enforces the conservation of energy and @xmath32 is the fermi - dirac distribution . @xmath33 is the boltzmann constant and @xmath34 is the temperature of the lattice . note that we have assumed the fermi level to be 0 . the expression for the transition rate includes a factor of 2 to account for the two possible electron spins . we work within the box approximation to assume that the volume under consideration extends from @xmath35 to @xmath36 in all directions and l@xmath37 . within this assumption we can convert the summations in equation [ trate ] to integrals and rewrite the transition rate as @xmath38 where @xmath39 is the overlap integral or the matrix element . @xmath40 is the wave vector of electrons in their initial state and @xmath41 is the wave vector of the emitted electron . if the work function of the emission surface is @xmath42 , the energy of the final state , @xmath30 , can be written as @xmath43 where @xmath44 is the wave number of the emitted electron in the transverse direction ( @xmath45-@xmath46 plane ) . the delta function in equation [ trate1 ] can then be written as @xmath47 where @xmath48 , @xmath49 . the qe can simply be calculated as @xmath50 as will be shown in the following section , the qe is independent of @xmath51 as the matrix element is proportional to @xmath52 owing to the normalization of the wave functions within the bounding box . the transverse momentum spread or the rms transverse momentum can be calculated as @xmath53^{\frac{1}{2}}\label{rmsmom}\ ] ] one can obtain the qe and the rms transverse momentum by calculating the matrix elements @xmath54 and evaluating the integrals in equation [ trate1 ] . calculation of the matrix elements can not be generalized further and requires the knowledge of the band structure , wave functions and the orientation of the photo - emitting surface . in the next section , we calculate the matrix elements and perform the integrals to obtain the qe and the rms transverse momentum for the ag(111 ) surface as an example . in order to calculate the matrix elements , one needs to obtain the polarization vector ( @xmath13 ) for incident light inside the solid surface . expressions to obtain @xmath13 can be found in born and wolf s principles of optics@xcite or any other standard text on electromagnetic waves . however , we state them here for the sake of completion . we assume @xmath45-@xmath2 plane to be the plane of incidence . the complex angle of transmission is given by snell s law as @xmath55 where @xmath56 is the complex index of refraction , @xmath19 is the angle of incidence , the angle of the light wave vector inside the metal with respect to the @xmath2 axis can be given by @xmath57\ ] ] and the optical decay length for the fields can be given by @xmath58 where @xmath59 ^{2}+\left [ \frac{2n_rn_i}{\left ( n_r^{2}+n_i^{2}\right ) ^{2}}% \sin ^{2}\theta _ { i}\right ] ^{2}\right]^{\frac{1}{4}}\ ] ] and @xmath60 for @xmath61-polarized light the polarization vector of the vector potential is @xmath62 where @xmath63 for @xmath64-polarized light the polariztion vector of the vector potential is @xmath65 where @xmath66 in this section , we demonstrate the use of the formalism developed above to obtain analytic expressions for the qe and rms transverse momentum from a ag(111 ) surface . the calculated qe matches the experimental values showing the effectiveness of the formalism developed above . we use a two - band fit to the nearly free - electron like ag @xmath67 band dispersion model around the @xmath51 point@xcite . the total energy ( @xmath68 ) can be divided into the longitudinal part ( @xmath69 ) and the transverse part ( @xmath70 ) and can be written as @xmath71 direction . the pink shaded region is the lower @xmath67 bands filled with electrons . the region shaded in blue is the upper @xmath67 bands which is unoccupied . the solid red line is the surface state ( ss ) . the solid blue curves are contours of constant @xmath69 or correspondingly constant @xmath72 . the green curve is the free electron parabola corresponding to a photon energy of @xmath73 . the conservation of energy and transverse momentum allow only the occupied states above this parabola to be emitted . [ fig1],width=307 ] within the framework of the nearly free electron model , the dispersion relations for the two bands in the longitudinal direction ( [ 111 ] or @xmath2 direction ) are given by@xcite @xmath74 where @xmath69 are the longitudinal energies of the electrons in the lower and upper @xmath67 bands respectively , @xmath75 is the valance band maximum , @xmath76 is the absolute value of the pseudo potential form factor and equals one - half of the gap at the zone boundary , @xmath61 is the magnitude of the wave vector at @xmath51 point and is equal to @xmath77 ( @xmath78 being the lattice constant ) and @xmath79 are the effective mass parameters of the lower and upper @xmath67 bands respectively . it should be noted that @xmath79 represent higher order corrections from multi band effects and do no correspond to the curvature of the dispersion relations . the subscripts @xmath80 and @xmath81 represent the lower and upper @xmath67 bands respectively . the fermi level is assumed to be 0 . the scale for @xmath72 is chosen such that the zero lies at the @xmath51 point . the dispersion relation in the transverse directions ( @xmath45-@xmath46 plane ) is assumed to be cylindrically symmetric and can be modeled by nearly parabolic bands given by @xmath82 where @xmath70 are the transverse energies of the electrons in the lower and upper @xmath67 bands respectively , @xmath83 and @xmath84 are the transverse effective masses of the lower and upper @xmath67 bands respectively . the coefficients @xmath85 are higher order correction coefficient obtained by fitting the band structure of silver@xcite . figure [ fig1 ] shows the band structure of ag projected along the [ 111 ] direction . the pink shaded region is the lower @xmath67 bands filled with electrons . the states of this band that extend beyond the fermi level are unoccupied and not shown . the region shaded in blue is the upper @xmath67 bands which is unoccupied . the solid blue curves are contours of constant @xmath69 or correspondingly constant @xmath72 . the shape of these contours is nearly parabolic and given by equation [ trans_bstruct ] with an offset of @xmath69 . values of all parameters used for modeling the bulk band structure are given in table [ tab1 ] and were obtained by fitting the band structure of silver@xcite . ag(111 ) exhibits a shockley surface state@xcite within the band gap at the @xmath51 point with energy @xmath86 . the surface state has @xmath87 . since the surface state is located within the band gap , @xmath88 obtained from equation [ dispersion ] is imaginary for the surface state@xcite . the dispersion relation in the transverse direction is parabolic and given by @xmath89 the effective mass of the surface state has been measured to be @xmath90@xcite . the energy of the surface state @xmath86 can change significantly with the sample and surface preparation methods and is sensitive to the strain in the crystal . at room temperature it has been reported to range between -20 mev to -120 mev@xcite . here , we use it as a fitting parameter and obtain the best fit for qe at @xmath91 mev . | p.15 | p.70 | p.15 | @xmath92 & @xmath93 & @xmath94 + @xmath95 & longitudinal effective mass parameter of the lower @xmath67 band & @xmath96 + @xmath97 & longitudinal effective mass parameter of the upper @xmath67 band & @xmath98 + @xmath99 & transverse effective mass of lower @xmath67 band & @xmath100 + @xmath101 & transverse effective mass of upper @xmath67 band & @xmath102 + @xmath103 & effective mass of surface state & @xmath104 + @xmath105 & third order correction coefficient for lower @xmath67 band&@xmath106 evnm@xmath107 + @xmath108 & fourth order correction coefficient for lower @xmath67 band&@xmath109 evnm@xmath110 + @xmath111 & fifth order correction coefficient for lower @xmath67 band&@xmath112 evnm@xmath113 + @xmath114 & sixth order correction coefficient for lower @xmath67 band&@xmath115 evnm@xmath116 + @xmath117 & third order correction coefficient for upper @xmath67 band&@xmath118 evnm@xmath107 + @xmath119 & fourth order correction coefficient for upper @xmath67 band&@xmath120 evnm@xmath110 + @xmath121 & fifth order correction coefficient for upper @xmath67 band&@xmath122 evnm@xmath113 + @xmath123 & sixth order correction coefficient for upper @xmath67 band&@xmath124 evnm@xmath116 + @xmath86 & energy of surface state&@xmath125 mev + @xmath75 & valance band maximum at l point & @xmath126 mev + @xmath76 & pseudo potential form factor ( equals one half band gap at l point ) & @xmath127 ev + @xmath42 & work function of ag(111 ) & @xmath128 ev + @xmath78 & unit cell length & @xmath129 nm + close to the @xmath51 point , the @xmath45 and @xmath46 dependent part of the initial and final wave functions can be expressed as plane waves . thus the initial and final wave functions can be expanded as @xmath130 and @xmath131 respectively . in order to match the transverse part of the final wave functions at the boundary we require @xmath132 and @xmath133 . below we give the @xmath2 dependent parts of the wave functions . the @xmath2 dependent part of the initial wave functions for the bulk states inside the ag(111 ) surface can be given by the combination of two bloch states ( @xmath134 and @xmath135 ) of the lower @xmath67 band and outside the surface can be given by an exponential decay@xcite . thus for @xmath136 @xmath137 \bigg\ } \end{split}\ ] ] and for @xmath138 @xmath139 where @xmath140 . the normalization constant @xmath141 can be obtained by normalizing the wave function . constants @xmath142 and @xmath143 are obtained by matching the wave function and its derivative at @xmath21 . the expressions for @xmath144 , @xmath141 , @xmath142 and @xmath143 are given in the appendix . for the surface state @xmath87 , @xmath145 , @xmath146 and @xmath147 . the energy of the surface state ( @xmath86 ) lies withing the @xmath51 gap . hence , from equation [ dispersion ] we see that the value of @xmath148 is imaginary causing the surface state to decay into the bulk . for @xmath136 the wave function of the surface state ( ss ) is given by @xmath149 \end{split}\ ] ] and for @xmath150 it is given by @xmath151 where @xmath152 . @xmath153 is the normalization constant and can be obtained by normalizing the wave function . @xmath154 and @xmath15 can be obtained by matching the wave function and its derivative at @xmath21 . expressions for @xmath153 and @xmath154 are given in the appendix . an explicit expression can not be obtained for @xmath15 and its value needs to be calculated numerically to satisfy the continuity conditions of the wave functions as given in the appendix . the final state wave functions are not the free electron wave functions of the emitted electron , but are time reversed leed states as required by the one step photoemission theory@xcite . inside the ag(111 ) surface they can be given by the combination of two bloch states(@xmath134 and @xmath135 ) of the upper @xmath67 band along with an exponential decay to account for the various scattering mechanisms that prevent emission of excited electrons . the final wave functions outside the surface are plane waves . thus for @xmath136 @xmath155 z } + \psi _ { f}e^{\left [ -i\left ( k_{zf}-p\right ) + k_{d}\right ] z } \right\}\sqrt{\frac{2}{l}}\ ] ] where @xmath156 is the exponential decay constant that takes into account the scattering mechanisms that prevent emission of excited electrons . @xmath157 is the electron - electron scattering length , which is the dominant scattering mechanism in metals . the scattering parameter @xmath64 is used as a fitting parameter in the calculation . for @xmath138 @xmath158\sqrt{\frac{2}{l}}\ ] ] where @xmath159 is the momentum of the emitted electron in the @xmath2 direction . constants @xmath160 and @xmath161 are obtained by matching the wave function and its derivative at @xmath21 . expressions for @xmath162 , @xmath160 and @xmath161 are given in the appendix . note that the normalization of the final states is such that the out going plane wave representing the emitted photoelectron is normalized to unity . figure [ fig2 ] shows an example of the real component of the @xmath2 dependent part of the initial bulk , surface and final wave functions . for @xmath61 polarized light , using the hamiltonian from equation [ hamiltonian ] , the matrix elements @xmath39 from equation [ trate1 ] can be written as @xmath163 where @xmath164 using the wave functions given in the previous section and integrating over the box one can calculate @xmath165 for @xmath61 polarized light as @xmath166 where @xmath167 and @xmath168 ^ 2 $ ] . note that in the limit @xmath169 , @xmath170 . @xmath171 , @xmath172 and @xmath173 are given as follows @xmath174 the integrals @xmath171 and @xmath173 can be evaluated analytically and the expressions are given in the appendix . owing to the appropriate normalization of the wave functions , @xmath171,@xmath172 and @xmath173 are independent of @xmath51 when @xmath169 . note that the matrix element given in equation [ m2_p ] is asymmetric in @xmath175 . this can lead to an asymmetric photoemission where the number of electrons emitted with momentum @xmath176 is different from electrons with number of electrons with @xmath45 direction momentum @xmath177 . the matrix element for @xmath64 polarized light can be given by @xmath178 the total qe can be written as sum of qe contribution from the bulk states ( @xmath179 ) and the surface state ( @xmath180 ) @xmath181 @xmath179 can be given by @xmath182 where @xmath183 can be calculated from equation [ trate1 ] with the integrations being carried over all possible initial bulk and final states . @xmath184 is the incident photon flux per unit area given by equation [ flux ] . @xmath180 can be given by @xmath185 @xmath186 can be calculated by an expression similar to equation [ trate1 ] with the difference that the integration over the initial state have to be performed over @xmath175 and @xmath187 only , due to the 2-d nature of the surface state . @xmath186 can be written as @xmath188 using equations [ flux ] , [ trate1 ] , [ delta ] , [ m2_p ] and [ qe_bulk ] , @xmath179 for @xmath61 polarized light can be written as @xmath189 where @xmath190 and @xmath191 is the fine structure constant . note that we require @xmath132 and @xmath133 in order to match the transverse part of the final wave functions at the boundary . in the limit as @xmath169 , @xmath170 . taking the limit as @xmath169 and integrating over the final states we obtain @xmath192 note that @xmath173 , @xmath171 , @xmath172 , @xmath193 and @xmath29 are functions of @xmath194 and @xmath195 . integrating the @xmath31 functions in equation [ qe_expr1 ] we get @xmath196 , @xmath197 and @xmath198 . these @xmath31 functions enforce the conservation of transverse momentum and energy during photoemission . similarly , @xmath180 can be obtained using [ qe_ss ] as @xmath199 the normalization constant for the surface state ( @xmath153 ) is not dependent on @xmath51 . hence , for the surface state @xmath173 , @xmath171 and @xmath172 are proportional to @xmath200 even as @xmath169 . thus , the surface state qe as given in equation [ qe_expr2 ] remains independent of @xmath51 as @xmath169 . after writing @xmath175 and @xmath187 in cylindrical co - ordinates as @xmath201 and @xmath202 ; then integrating over @xmath203 the above expressions for @xmath179 and @xmath180 can be written as @xmath204\label{qe_expr3 } \end{split}\ ] ] and @xmath205\label{qe_expr4 } \end{split}\ ] ] respectively . the qe for @xmath64 polarized light can be similarly calculated by using the appropriate matrix elements . the 3-d momentum distributions and the rms transverse momentum can also be calculated easily as shown in equation [ rmsmom ] . figure [ fig3 ] compares the spectral response measured from an ag(111 ) surface to the result obtained from the photoemission model presented above , for @xmath61 polarized light at various angles of incidence . ) in @xmath61 polarized light . the error bar on the experimental measurement is @xmath206 10% [ fig3],width=307 ] in order to measure the qe , a commercially bought@xcite single crystal ag(111 ) sample was prepared in an ultra - high vacuum chamber with base pressure in the low @xmath207 torr range . several cycles of ar ion bombardment and annealing to @xmath208c were performed until a sharp hexagonal leed pattern was observed . the surface cleanliness was verified using auger electron spectroscopy . the qe was obtained by measuring the photocurrent and the power of light incident on the sample surface . a laser based plasma lamp with a monochromator@xcite was used as a light source for the qe measurement . the spectral width of the light source was 2 nm fwhm . all constants used for modeling the band structure to calculate the qe are given in table [ tab1 ] . the optical constants ( @xmath209 and @xmath210 ) for silver as a function of wavelength are well known@xcite . the surface constant @xmath211 and the electron - electron scattering length @xmath157 were obtained as a function of photon energy by extrapolating the values of @xmath211 and @xmath157 obtained from pes measurements@xcite . the scattering parameter @xmath64 is set to 12.5 to obtain a good match to the experimental data . figure [ fig3 ] shows that the calculated qe explains the experimental data , both qualitatively and quantitatively . with the exception of the scattering parameter @xmath64 , this photoemission model calculates the qe accurately without the use of any ad hoc coefficients or scaling factors . it is seen that the qe increases with the angle of incidence for @xmath61 polarized light ( vectorial photoelectric effect ) . the knee observed in the spectral response for higher angles of incidence at @xmath206 4.55 ev is caused due to the surface state . this becomes clear from figure [ fig4 ] which shows the contributions to the qe from the bulk and surface states . the sections below discuss the effect of the scattering parameter and the vectorial photoelectric effect respectively . and @xmath180 for @xmath212 and @xmath213 . @xmath180 is not very sensitive to @xmath64 because the surface state electrons are localized at the surface and do not need to travel inside the metal to get emitted . [ fig4],width=307 ] the decay constant of the final wave function @xmath214 takes into account the electrons that were excited by light but were unable to escape due to various scattering mechanisms while travelling towards the surface . the inelastic electron - electron scattering is the dominant scattering mechanism of excited electrons in metals . hence we write @xmath156 , where @xmath157 is the electron - electron scattering mean free path and @xmath64 is adjusted to match the calculated qe to the experimental value . figure [ fig4 ] shows @xmath179 and @xmath180 for @xmath212 and @xmath213 . we can see that @xmath180 does not change significantly with @xmath64 . the surface state is localized at the metal - vacuum interface . hence the electrons excited from the surface state do not need to travel inside the metal to get emitted . this causes @xmath180 to be insensitive to @xmath64 or @xmath214 . in order to match the experimental data , @xmath64 needs to be set to a particularly large value of 12.5 . this implies a much higher effective scattering rate than set by the electron - electron scattering lengths obtained from uv - pes data@xcite . the reason for this increased scattering is not clear . matches the experimental data measured at photon energy of 4.57 ev [ fig5],width=307 ] vectorial photoelecric effect is the variation of qe with the angle of incidence and polarization of incident light . the qe for @xmath61 polarized light is given by equations [ qe_expr3 ] and [ qe_expr4 ] . in these equations , the term @xmath215 corresponds to the qe contribution of the @xmath2 component of the polarization vector and the term @xmath216 corresponds to the qe contribution of the @xmath45 component of the polarization vector . for the band structure and wave functions used here , @xmath217 . as a result , the photoemission from the ag(111 ) surface is dominated by the @xmath2 component of the polarization vector ( i.e component perpendicular to the surface ) . neglecting the contribution of the @xmath45 ( parallel to surface ) component , the qe can be written as @xmath218 where @xmath219 is a constant independent of the angle of incidence . note that both @xmath220 and @xmath221 are dependent on the angle of incidence . figure [ fig5 ] shows that the experimentally measured angular dependence of qe for @xmath61 polarized light matches this calculation . this dependence is similar to the angular dependence of qe measured for several materials@xcite . the spectral response calculated by the model at @xmath222 angle of incidence is much smaller than the experimental value ( see figure [ fig3 ] ) . at @xmath222 angle of incidence only the @xmath45 and @xmath46 components of the polarization vector exist . this implies that the experimentally observed contribution of the @xmath45 and @xmath46 components of of the polarization vector is larger than that calculated by the model . the assumption that the wave functions in the @xmath45 and @xmath46 directions are modeled by plane waves could be one possible culprit for this . emission from parts of the band structure not modeled by the nearly free electron representation , many body photoemission effects like the hole state lifetime induces energy spread@xcite and the breakdown of the sudden approximation@xcite are other effects which may be responsible for this discrepancy . they may also be responsible for the large effective scattering parameter . making it a much better electron source than the current state - of - art . [ fig6],width=307 ] figure [ fig6 ] shows the rms transverse momentum expected from the ag(111 ) surface . the rms transverse momentum has been calculated using equation [ rmsmom ] for angle of incidence equal to 60@xmath0 . it can be seen that the rms transvserse momentum initially decreases , reaches a minimum and then increases with increasing photon energy . at photon energies very close to threshold only electrons from the ring formed by the intersection of the surface state with the fermi level are emitted . these electrons have a relatively high transverse momentum . at higher photon energies , electrons from the surface state with lower transverse momentum and lower energy can also be emitted along with the electrons from the surface state ring at the fermi level . this causes the rms transverse momentum to initially reduce with increasing photon energy . this decline continues till the photon energy is sufficiently high to allow emission from the entire surface state . at this photon energy the rms momentum reaches a minimum . at higher photon energies , the electrons from bulk states which are located near the fermi level and have a much higher transverse momentum are allowed to be emitted causing the rms transverse momentum to increase again . the smallest rms momentum measured from polycrystalline metal cathodes ( which are typically used as electron sources ) is limited by the room temperature to a value of @xmath223 ev / c@xcite at the photoemission threshold . however , at the photoemission threshold the qe is also very low ( in the @xmath224-@xmath225 ) range making polycrystalline cathodes unusable in this regime . at higher photon energies , the qe increases but so does the rms transverse momentum . in order to obtain a desirable qe greater than 10@xmath226 the photon energy used has to be several 100 mev above threshold . this sets the rms transverse momentum practically obtained in electron guns to @xmath227 ev / c@xcite . according to our calculations , ag(111 ) when operated at an angle of incidence of 60@xmath0 in @xmath61 polarized light at 4.57 ev can act as a cathode with rms transverse momentum lower than 180 ev / c and a qe as high as @xmath1 . this shows that ag(111 ) can act as a better photocathode the polycrystalline metals currently used as electron sources . we have presented a 3-d one step model that allows us to calculate photoemission properties like qe and rms transverse momentum of emitted electrons from single crystal surfaces . optimizing these photoemission properties can greatly improve the performance of electron source applications like fels and ued . using the example of photoemission from ag(111 ) we show that not only can this model calculate the spectral response from surface state without the use of any ad hoc parameters , but also explains the photoemission phenomena of the vectorial photoelectric effect accurately . we also calculate the rms transverse momentum from an ag(111 ) surface and show that in @xmath61 polarized light with a high angle of incidence , the ag(111 ) surface can exhibit high qe along with a small rms transverse momentum , making it a much better cathode than the currently used polycrystalline metals . upon integrating with other band structure and wave function calculation techniques like density functional theory , this methodology can be used to calculate the electron source relevant photoemission properties from any single crystal surface in order to identify ideal electron emitters from first principle calculations@xcite . such a methodology is essential to screen for materials to identify good electron emitters . the authors s. karkare and w. wan have contributed equally to this work . the authors would like to thank dr . t. miller for stimulating discussions . this work was supported by the director , office of science , office of basic energy sciences of the u. s. department of energy , under contract nos . kc0407-alsjnt - i0013 and de - ac02 - 05ch11231 ( w. w. , s. k. , j. f. , h. a. p. ) and the u.s . national science foundation under grant no . dmr 13 - 05583 ( t .- c . the analytic expressions to calculate several of the coefficients used in the wave function calculations are given below @xmath228e^{i\left(k_{zi}+p\right)z_0}+\left[\kappa_i+i\left(k_{zi}-p\right)\right]\psi_ie^{i\left(k_{zi}-p\right)z_0}}{\left[-\kappa_i+i\left(k_{zi}+p\right)\right]e^{-i\left(k_{zi}+p\right)z_0}+\left[-\kappa_i+i\left(k_{zi}-p\right)\right]\psi_ie^{-i\left(k_{zi}-p\right)z_0}}\\ c_2&=&2i\frac{\left(k_{zi}+p\right)+\left(k_{zi}-p\right)\psi_i^2 + 2k_{zi}\psi_i\cos\left(2pz_0\right)}{\left[\kappa_i+i\left(k_{zi}+p\right)\right]e^{-i\left(k_{zi}+p\right)z_0}+\left[-\kappa_i+i\left(k_{zi}-p\right)\right]e^{-i\left(k_{zi}-p\right)z_0}}\\ t_{pk}&=&\frac{2ik_z}{\left[i\left(k_{zf}+p+k_z\right)+k_d\right]e^{\left[i\left(k_{zf}+p\right)+k_d\right]z_0}+\left[i\left(k_{zf}-p+k_z\right)+k_d\right]\psi_fe^{\left[i\left(k_{zf}-p\right)+k_d\right]z_0}}\\ 1+r_{pk}&=&\frac{2ik_z\left[e^{\left[i\left(k_{zf}+p\right)+k_d\right]z_0}+\psi_fe^{\left[i\left(k_{zf}-p\right)+k_d\right]z_0}\right]}{\left[i\left(k_{zf}+p+k_z\right)+k_d\right]e^{\left[i\left(k_{zf}+p\right)+k_d\right]z_0}+\left[i\left(k_{zf}-p+k_z\right)+k_d\right]\psi_fe^{\left[i\left(k_{zf}-p\right)+k_d\right]z_0}}\\ n&=&\frac{\sqrt(2)}{\sqrt{\left(\frac{l}{2}\right)\left(1+\psi_i^2\right)\left(1+\left\vert c_1\right\vert^2\right)}}\left(\mathrm{assuming } l\rightarrow\infty\right)\\ c_s&= & e^{i\left(k_{zs}+p\right)z_0}+\psi_se^{i\left(k_{zs}-p\right)z_0}\\ n_s&=&\sqrt(2)\left[\frac{e^{2k_{zs}z_0}}{2k_{zs}}\left(1+\left\vert \psi_s\right\vert^2\right)+\mathrm{re}\left[\psi_s^{\ast}\frac{e^{2\left(ip+k_{zs}\right)}}{ip+k_{zs}}\right]+\frac{\left\vert c_s\right\vert^2}{2\kappa_s}\right]^{-\frac{1}{2 } } \\\end{aligned}\ ] ] note that the wave functions have been normalized to 2 in order to account for the electrons emitted from the equivalent @xmath51 point at @xmath229 . @xmath15 can be obtained by solving the following equation numerically @xmath230 @xmath231 z_{0}}}{i\left ( -k_{zf}+k_{zi}\right ) + k_{d}}+\phi _ { i}\frac{i\left ( k_{zi}-p\right ) e^{\left [ i\left ( -k_{zf}+k_{zi}-2p\right ) + k_{d}\right ] z_{0}}}{i\left ( -k_{zf}+k_{zi}-2p\right ) + k_{d}}\right . \\ & & -c_{1}\frac{i\left ( k_{zi}+p\right ) e^{\left [ -i\left ( k_{zf}+k_{zi}+2p\right ) + k_{d}\right ] z_{0}}}{-i\left ( k_{zf}+k_{zi}+2p\right ) + k_{d}}-c_{1}\phi _ { i}\frac{i\left ( k_{zi}-p\right ) e^{\left [ -i\left ( k_{zf}+k_{zi}\right ) + k_{d}\right ] z_{0}}}{-i\left ( k_{zf}+k_{zi}\right ) + k_{d } } \\ & & + \phi _ { f}\frac{i\left ( k_{zi}+p\right ) e^{\left [ i\left ( -k_{zf}+k_{zi}+2p\right ) + k_{d}\right ] z_{0}}}{i\left ( -k_{zf}+k_{zi}+2p\right ) + k_{d}}+\phi _ { i}\phi _ { f}\frac{i\left ( k_{zi}-p\right ) e^{\left [ i\left ( -k_{zf}+k_{zi}\right ) + k_{d}\right ] z_{0}}% } { i\left ( -k_{zf}+k_{zi}\right ) + k_{d } } \\ & & \left . -c_{1}\phi _ { f}\frac{i\left ( k_{zi}+p\right ) e^{\left [ -i\left ( k_{zf}+k_{zi}\right ) + k_{d}\right ] z_{0}}}{-i\left ( k_{zf}+k_{zi}\right ) + k_{d}}-c_{1}\phi _ { i}\frac{i\left ( k_{zi}-p\right ) e^{\left [ -i\left ( k_{zf}+k_{zi}-2p\right ) + k_{d}\right ] z_{0}}}{-i\left ( k_{zf}+k_{zi}-2p\right ) + k_{d}}\right ] \\ i=\int dz\phi _ { zf}^{\ast } \phi _ { zi } & = & nc_{2}\left ( \frac{2ik_{z}}{\kappa _ { i}^{2}+k_{z}^{2}}+\frac{1+r_{pk}^{\ast } } { -\kappa _ { i}+ik_{z}}\right ) \\ & & + nt_{pk}^{\ast } \left [ \frac{e^{\left [ i\left ( -k_{zf}+k_{zi}\right ) + k_{d}% \right ] z_{0}}}{i\left ( -k_{zf}+k_{zi}\right ) + k_{d}}+\phi _ { i}\frac{e^{% \left [ i\left ( -k_{zf}+k_{zi}-2p\right ) + k_{d}\right ] z_{0}}}{i\left ( -k_{zf}+k_{zi}-2p\right ) + k_{d}}\right . \\ & & -c_{1}\frac{e^{\left [ -i\left ( k_{zf}+k_{zi}+2p\right ) + k_{d}\right ] z_{0}}% } { -i\left ( k_{zf}+k_{zi}+2p\right ) + k_{d}}-c_{1}\phi _ { i}\frac{e^{\left [ -i\left ( k_{zf}+k_{zi}\right ) + k_{d}\right ] z_{0}}}{-i\left ( k_{zf}+k_{zi}\right ) + k_{d } } \\ & & + \phi _ { f}\frac{e^{\left [ i\left ( -k_{zf}+k_{zi}+2p\right ) + k_{d}\right ] z_{0}}}{i\left ( -k_{zf}+k_{zi}+2p\right ) + k_{d}}+\phi _ { i}\phi _ { f}\frac{e^{% \left [ i\left ( -k_{zf}+k_{zi}\right ) + k_{d}\right ] z_{0}}}{i\left ( -k_{zf}+k_{zi}\right ) + k_{d } } \\ & & \left . -c_{1}\phi _ { f}\frac{e^{\left [ -i\left ( k_{zf}+k_{zi}\right ) + k_{d}% \right ] z_{0}}}{-i\left ( k_{zf}+k_{zi}\right ) + k_{d}}-c_{1}\phi _ { i}\frac{e^{% \left [ -i\left ( k_{zf}+k_{zi}-2p\right ) + k_{d}\right ] z_{0}}}{-i\left ( k_{zf}+k_{zi}-2p\right ) + k_{d}}\right ] , \end{aligned}\ ] ] 35ifxundefined [ 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 in @noop _ _ , ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) in link:\doibase 10.1109/pac.2007.4441068 [ _ _ ] ( ) pp . http://dx.doi.org/10.1016/j.nima.2016.08.032 [ * * ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) `` , '' ( , , ) chap . @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' @noop * * , ( ) @noop * * , ( ) @noop `` , '' link:\doibase 10.1007/bf02451658 [ * * , ( ) ] link:\doibase 10.1364/josa.54.000216 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( )
in this paper , we present a 3-d one step photoemission model that can be used to calculate the quantum efficiency and momentum distributions of electrons photoemitted from ordered single crystal surfaces close to the photoemission threshold . using ag(111 ) as an example , we show that the model can not only calculate the quantum efficiency from the surface state accurately without using any ad hoc parameters , but also provides a theoretical quantitative explanation of the vectorial photoelectric effect . this model in conjunction with other band structure and wave function calculation techniques can be effectively used to screen single crystal photoemitters for use as electron sources for particle accelerator and ultrafast electron diffraction applications .
this paper gives a complete description of the solutions of the one dimensional ginzburg - landau equations which model superconductivity phenomena in infinite slabs . one of our main goals here is to determine the number , symmetry and stability of solutions . we give a complete numerical investigation of these issues over the entire range of physically important parameters : @xmath0 the size of the slab , @xmath1 the ginzburg - landau parameter , and @xmath2 , the exterior magnetic field . our experiments , which are summarized in figure [ figsummary ] , reveal the existence of two key - points in parameter space which play a central role in the way superconductivity is nucleated . indeed , according to the values of @xmath0 and @xmath1 , superconductivity appears either in the volume of the sample if the slab is thin , or on superconducting sheaths near the boundary if the slab is large , and in the form of two - dimensionnal vortex patterns if the slab is of intermediate size . this is why all values of @xmath0 and @xmath1 are of interest . as we shall see , sheaths and vortices can be described in terms of asymmetric solutions of the one dimensional problem , and each of the key - points gives rise to diverse pattern formation by means of various bifurcation phenomena . in addition , our results provide new insight into the problem of differentiating between the types of superconductors in terms of the parameters . finally , our global description allows us to understand better many interesting physical phenomena , to classify results previously obtained for each regime of the parameters and also to derive new open problems . in order to properly describe our results we begin with a brief summary of the model . the superconductivity of certain metals is characterized at very low temperatures by the loss of electrical resistance and the expulsion of the exterior magnetic field @xmath3 . in the model derived by ginzburg and landau in 1950 ( see @xcite ) , the electromagnetic properties of the material are completely described by the magnetic potential vector @xmath4 ( @xmath5 being the magnetic field ) and the complex - valued order parameter @xmath6 . in fact , @xmath6 is an averaged wave function of the superconducting electrons and its modulus corresponds to the density of superconducting carriers . when the sample is wholly normal , @xmath7 and the magnetic field inside the material @xmath8 is equal to the exterior magnetic field @xmath3 . on the other hand , when the sample is perfectly superconducting , @xmath9 and the magnetic field @xmath8 is identically 0 . furthermore , in the ginzburg - landau theory , the state of the sample is completely determined by the minimum of an energy depending on @xmath6 and @xmath4 . for a more precise description of the general theory , one may refer to @xcite , @xcite , @xcite , @xcite , @xcite , @xcite or to @xcite , @xcite . in the special case when the sample is an infinite slab of constant thickness , between the planes @xmath10 and @xmath11 , it is usual to assume that both @xmath6 and * a * are uniform in the @xmath12 and @xmath13 directions , and that the exterior magnetic field is tangential to the slab , that is @xmath14(0,0,@xmath2 ) . a suitable gauge can then be chosen so that @xmath15 is a real function , and @xmath16 , where * e*@xmath17 is the unit vector along the @xmath12 direction ( see @xcite for more details ) . in this case , the nondimensionalized form of the ginzburg - landau energy is given by : [ ekappa ] e_(f , q)=_-a^a ( 1 f^2+f^2q^2 + 12f^4-f^2+(q- h_0)^2 ) dx . the nondimensionalized parameter @xmath1 is called the ginzburg - landau parameter . it is the ratio of @xmath18 , the penetration depth of the magnetic field , to @xmath19 , the coherence length , which is the characteristic length of variation of @xmath20 . the value of @xmath1 determines the type of superconductor according to the type of phase transition which takes place between the normal phase and the superconducting phase : @xmath1 small describes what is known as a type i superconductor and @xmath1 large as a type ii . more precisely , for a type i superconductor , there is a critical magnetic field @xmath21 such that if @xmath22 , the material is entirely superconducting and the magnetic field is expelled from the sample apart from a boundary layer of size @xmath18 . this is called the meissner effect . if @xmath23 , superconductivity is destroyed and the material is in the normal state , that is @xmath24 and @xmath25 . for a type ii superconductor , the phase transition is different and there are two critical fields @xmath26 and @xmath27 : for @xmath28 , the exterior magnetic field is expelled from the sample and there is a meissner effect as for type i superconductors . but as @xmath2 is increased above @xmath26 , superconductivity is not destroyed straight away , since the superconducting and the normal phase coexist under the form of filaments or vortices : the vortex is a zone of diameter @xmath19 , at the center of which the order parameter @xmath20 vanishes . as @xmath2 increases further , the vortices become more numerous until the critical value @xmath27 is reached at which superconductivity is destroyed . for @xmath29 , there is no superconductivity and the material is in the normal state . the way superconductivity is nucleated is highly dependent on @xmath0 and @xmath1 , as we will see later in section 4 , where we will introduce a third critical field @xmath30 corresponding to nucleation of surface superconductivity . we refer to tinkham @xcite for a detailed explanation . the vortex phenomena in superconductivity have been widely studied in the literature . see for instance @xcite or @xcite and the references therein . the critical value of @xmath1 usually given to separate type i and type ii superconductors is @xmath31 . in the next sections , we will describe how this value is computed in the limiting case @xmath32 . however , as @xmath0 is decreased from infinity , we will see that the demarcation between type i and ii behaviours is no longer the constant @xmath33 . instead , we find that there is a well defined curve in the @xmath34 plane which separates the two types of behaviours . this result will be fully discussed in section 3 . similarly in section 4 , we find an extra curve which separates different type i superconductors according to surface effects . for a mathematical analysis of the problem , it is natural to assume that @xmath35 and @xmath36 . it then follows from standard variational arguments that there exists a minimizer of @xmath37 , and that the minimizer is a solution of @xmath38 notice that @xmath24 and @xmath39 is always a solution for any real @xmath40 . from now on , we will call this a _ normal solution_. regularity properties of minimizers yield that either @xmath20 is a normal solution , or @xmath20 does not change sign , hence we will study the case @xmath41 . an easy calculation shows that the energy @xmath37 is zero along the normal solution . thus a global minimizer can not have positive energy . the aim of this paper is to give a complete description ( number , symmetry and stability ) of the solutions of the system @xmath42 for which @xmath41 on @xmath43 $ ] , according to the values of the parameters @xmath0 , @xmath1 and @xmath2 . this work is based on numerical simulations obtained with auto , a software developed by doedel et al . ( see @xcite , @xcite , @xcite ) which computes bifurcation diagrams for systems of ode s . in order to properly describe our results , it is necessary that we first give a brief summary of previous studies of the @xmath42 system . this is done in the next section . then , in sections 3 and 4 we give the details of our numerical investigation for symmetric solutions ( section 3 ) and asymmetric solutions ( section 4 ) , together with their physical implications and suggestions for future research . our results are all summarized in figure [ figsummary ] . section 5 is devoted to the stability analysis . let us first recall the basic properties of solutions . [ summary ] if @xmath44 is a solution of @xmath42 and if f is not identically zero then * @xmath45 in @xmath46 , * @xmath47 has a unique zero @xmath48 in @xmath46 , @xmath47 is increasing on @xmath46 , @xmath49 is decreasing on @xmath50 and increasing on @xmath51 . * there exist @xmath52 and @xmath53 with @xmath54 and @xmath55 $ ] such that @xmath56 is increasing on @xmath57 \cup [ x_2,a]$ ] and decreasing on @xmath58 $ ] , @xmath20 is increasing on @xmath59 $ ] and decreasing on @xmath60 $ ] . the proof of @xmath61 and @xmath62 can be found for instance in @xcite and of @xmath63 in @xcite . there are two types of physically important solutions of @xmath42 : symmetric solutions and asymmetric solutions . we define a _ symmetric solution _ to be a solution of @xmath42 such that @xmath41 , @xmath20 is even and @xmath47 is odd on @xmath43 $ ] . thus , a symmetric solution satisfies the following problem : @xmath64 for @xmath65 and @xmath66 . we need to choose @xmath67 and @xmath68 such that @xmath69 . then @xmath44 will be a solution of @xmath42 with @xmath70 . notice that @xmath68 is the amplitude of @xmath20 for a symmetric solution . we define an _ asymmetric solution _ to be a solution of @xmath42 which satisfies @xmath41 on @xmath43 $ ] , yet which is not symmetric . that is @xmath71 or @xmath72 . for the existence of symmetric solutions , kwong @xcite has proved the following important result . [ kwong ] ( kwong @xcite ) for each @xmath68 in @xmath73 , there exists a unique @xmath74 such that the solution @xmath44 of @xmath75 satisfies @xmath69 . moreover , @xmath67 is a continuous , decreasing function of @xmath68 , [ kwongalpha ] _ for this choice of @xmath76 , and the corresponding solution @xmath44 of @xmath75 , let @xmath77 . then @xmath78 is well - defined , continuous and [ kwongh ] _ 0h()=h_s>0_1h()=0 . in the following the notation @xmath79 of ( [ kwongh ] ) plays an important role in the discussion of our results . there are three possible behaviours of the curve @xmath80 defined in kwong s theorem , and these are shown in figures [ figs1 ] , [ figs2 ] and [ figs3 ] . notice that instead of graphing @xmath80 vs @xmath68 , we have put @xmath68 on the vertical axis and @xmath78 on the horizontal axis , so that we keep the convention originally adopted by ginzburg ( * ? ? ? * figures 3 and 5 ) . @xmath81 @xmath82 @xmath83 * figure [ figs1 ] : for @xmath84 , @xmath80 is a decreasing function of @xmath68 : if @xmath85 , there is exactly one symmetric solution of @xmath42 , and if @xmath86 , no such solution . * figure [ figs2 ] : for @xmath84 , @xmath80 is increasing until @xmath78 reaches a maximum value @xmath87 ( here @xmath88 ) , and then decreasing to 0 as @xmath68 goes to 1 . thus , for @xmath89 , @xmath42 has a unique symmetric solution , for @xmath90 , two symmetric solutions , and for @xmath91 no such solution . * figure [ figs3 ] : for @xmath84 , @xmath80 is decreasing until @xmath78 reaches a local minimum value @xmath92 ( here @xmath93 ) , increasing until @xmath78 reaches a local maximum value @xmath87 ( here @xmath94 ) , and then decreasing to 0 as @xmath68 goes to 1 . there thus , for @xmath96 , @xmath42 has a unique symmetric solution , for @xmath97 , three symmetric solutions , for @xmath98 , two symmetric solutions , and for @xmath91 no symmetric solution . + if @xmath99 , then for @xmath96 , @xmath42 has a unique symmetric solution , for @xmath100 , three symmetric solutions , for @xmath101 , a unique symmetric solution , and for @xmath102 no symmetric solution . note that in figure [ figs2 ] and [ figs3 ] , there are points on the curve where the resultant slope is vertical , when @xmath103 or @xmath87 . we shall refer to such points as _ folds _ throughout the paper . these figures indicate ( and it has been verified experimentally ) that if the applied field @xmath2 is sufficiently strong , superconductivity is destroyed . this has been proved by kwong @xcite . [ meissner ] ( kwong @xcite ) for any @xmath0 and @xmath1 , there exists @xmath21 such that for @xmath23 , the only solution of @xmath42 is the normal solution @xmath24 , @xmath25 . a weaker result had been obtained earlier in @xcite and @xcite . as each of the figures indicates , as one decreases @xmath2 from infinity , the material remains in the normal state until a critical value of @xmath2 is reached at which there is a bifurcation of nontrivial solutions from the normal state . so for small @xmath104 , a nontrivial curve @xmath105 of solutions of @xmath42 starting from a normal solution @xmath106 is sought , with the following asymptotic development : [ bifurcatedsol ] . ll f(x,)=f_0(x)+^3 f_1(x ) + o(^3 ) h^2(-a , a ) , + q(x,)=q_0(x)+^2 q_1(x ) + o(^2 ) h^2(-a , a ) , + h()=h_0+^2 h_1+^4 h_2+o(^4 ) , . where @xmath107 . it is important to note that when @xmath108 , the branch gives rise to symmetric solutions and when @xmath109 to asymmetric solutions . furthermore , it can be proved ( see @xcite or @xcite for more details ) that for small @xmath104 , the energy of the bifurcated superconducting solution , @xmath110 , has the same sign as @xmath111 . this leads us to the following definition . [ bifur ] a curve of the form ( [ bifurcatedsol ] ) is said to result from a supercritical bifurcation if @xmath112 . then , for small @xmath104 , the bifurcated solutions have lower energy than the normal state . a bifurcation is said to be subcritical if @xmath113 . here the bifurcated solutions have larger energy than the normal state . this means that for figures [ figs1 ] and [ figs3 ] , the bifurcation is supercritical and for figure [ figs2 ] subcritical . we now summarize the main results previously obtained concerning these figures describing symmetric solutions . we note that ginzburg @xcite had investigated the case @xmath1 small and found that @xmath80 behaves as in figure [ figs1 ] for small @xmath0 and , as @xmath0 is increased through a critical value , the graph of @xmath80 changes from figure [ figs1 ] to [ figs2 ] . he explained the type of behaviour described by figure [ figs2 ] in terms of superheating and supercooling . more precisely , when @xmath2 is large , superconductivity does not occur and the material is in the normal state . as @xmath2 is decreased , the material stays in the normal state down to @xmath79 , even though there is a range of @xmath2 where the normal solution is only a local minimizer and the global minimizer is a superconducting solution ( see ( * ? ? ? * figure 2 ) ) . if @xmath2 is decreased further , there is a jump in the maximum of @xmath20 and the material becomes superconducting , the solution being given by the symmetric branch . in this case , @xmath79 is called the supercooling field . now , on the contrary , start from @xmath114 where the superconducting state @xmath115 is the global minimizer and increase @xmath2 . the material will remain superconducting until @xmath87 is reached , though for fields slightly less then @xmath87 , it is only a local minimizer and the global minimizer is the normal solution . for @xmath2 above @xmath87 , there is a jump in the maximum of @xmath20 and the material reverts to the normal state . this is the superheating phenomenon . these two phenomena give rise to a hysteresis loop as described in @xcite or later in @xcite . we will make further remarks about these stability properties in section 5 . ginzburg and landau @xcite had also noticed that if @xmath0 large is fixed , then there occurs a symmetric supercritical bifurcation of superconducting solutions from the normal state as @xmath1 is increased through a critical value , but had no special explanation for this , since at that time only superconductors with small @xmath1 were known . however , as we shall see in the next section , as @xmath1 is increased through this critical value , our studies indicate that the behaviour seen in figure [ figs2 ] changes into that seen in figure [ figs3 ] . chapman @xcite has studied the case @xmath32 , and showed that there is a change of bifurcation from subcritical to supercritical that takes place for @xmath31 , which is the critical value between type i and ii superconductors . moreover , in @xcite , a linear stability analysis through the time dependent equations yield that the value @xmath31 is also the one for which stability of the normal solution switches . in @xcite , he has investigated the behaviour of the superheating field @xmath87 in the limit @xmath32 . following these works which are mainly based on formal computations , bolley and helffer have extensively studied the phenomenon of bifurcation of solutions from the normal solution in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite . in particular , they have given rigorous proofs of properties of bifurcating branches and asymptotic formula for the superheating and supercooling fields . although their results are mainly local , they have made a first attempt in @xcite to give a global stability picture of the solutions . we will mention the details of their contributions relevant to the values of the parameters in the course of the discussion , and indicate how their results contribute to the broader global picture . recently , hastings , kwong and troy @xcite have further investigated the multiplicity of symmetric solutions . they have proved that if @xmath116 , and @xmath0 is sufficiently large , then the behaviour of @xmath80 is described by figure [ figs2 ] , and if @xmath117 and @xmath0 sufficiently large , by figure [ figs3 ] . seydel @xcite @xcite was the first to give numerical evidence that there is a range of parameters for which asymmetric solutions and multiple symmetric solutions coexist . one of the major goals of this paper is to extend the results described above and completely determine the multiplicity and stability of symmetric and asymmetric solutions for the entire range of values of @xmath1 and @xmath0 . in the next two sections , we present the detail of our numerical investigations for symmetric solutions ( section 3 ) and asymmetric solutions ( section 4 ) , together with their physical implications and suggestions for future research . in particular in section 4 , we explain how asymmetric solutions of @xmath42 give rise to nucleation of sheaths and vortices . the last section is devoted to stability properties . as we stated in the last section , one of the main goals of this paper is to gain a complete understanding of the important properties of symmetric solutions in terms of the parameters @xmath1 and @xmath0 . as we shall see , these global properties give us new insight into an important problem , that of demarcating the @xmath118 plane into regions corresponding to type i and type ii behaviours . the results of our numerical investigations concerning the existence and multiplicity of symmetric solutions are shown graphically in figure [ figkappaasym ] . they indicate that the @xmath118 plane is the union of three connected sets @xmath119 , @xmath120 and @xmath121 . in @xmath119 , the behaviour of @xmath80 of figure [ figs1 ] holds . likewise , @xmath120 and @xmath121 reflect the behaviour of @xmath80 of figures [ figs2 ] and [ figs3 ] respectively . in the following , we explain the results shown in figure [ figkappaasym ] and make several conjectures . autoka.pstex_t [ main ] there exist three continuous functions @xmath122 , @xmath123 and @xmath124 which separate the @xmath118 plane into three connected regions @xmath119 , @xmath120 and @xmath121 , as shown in figure [ figkappaasym ] . there exists exactly one point @xmath125 called the triple point ( with approximate value 1.60 ) such that @xmath126 . moreover , @xmath122 is defined and monotone increasing on @xmath127 $ ] , @xmath124 is defined and monotone increasing on @xmath128 and @xmath123 is defined and monotone decreasing on @xmath128 , [ kappa(a ) ] _ a5 /2_1(a)=0 , _ a_2(a)=1/2,_a_3(a)=+ . * region @xmath129 . if @xmath118 is in @xmath119 , the behaviour of the corresponding curve @xmath80 is given by figure [ figs1 ] and there is at most one symmetric solution of @xmath42 . * region @xmath130 . if @xmath118 is in @xmath120 , the behaviour of the corresponding curve @xmath80 is given by figure [ figs2 ] and there are at most two symmetric solutions of @xmath42 . + let us call @xmath131 the point where @xmath132 . then as @xmath118 tends to a point @xmath133 on the curve @xmath122 , @xmath134 tends to @xmath135 . * region @xmath136 . if @xmath118 is in @xmath121 , the behaviour of the corresponding curve @xmath80 is given by figure [ figs3 ] and there are at most three symmetric solutions of @xmath42 . + let us call @xmath137 the point where @xmath138 . then as @xmath118 tends to a point @xmath139 on the curve @xmath123 , @xmath140 tends to @xmath135 . as @xmath118 tends to a point @xmath141 on the curve @xmath124 , @xmath140 and @xmath134 have the same limit @xmath142 with @xmath143 . moreover , @xmath144 if and only if @xmath145 . note that in @xmath120 , @xmath79 is sometimes called the supercooling field and in @xmath120 and @xmath121 , @xmath87 refers to the superheating field . we will come back to this in section 5 , since this is linked to local stability . our physical interpretation of the 3 regions is the following : @xmath119 corresponds to what is usually called thin films : when the size of the slab is sufficiently small , there is no difference between type i and type ii superconductors . the meissner effect is observed , that is the magnetic field is expelled from the material and there is no superheating nor supercooling phenomenon . we refer to @xcite for more details . in particular , our curves @xmath122 and @xmath124 give a quantitative estimate of how small @xmath0 should be depending on @xmath1 so that only one dimensionnal phenomena happen . we claim that the curve @xmath123 for @xmath146 can be considered as the separating curve between type i and type ii behaviours . indeed , it is known ( see tinkham @xcite for instance for details ) that when @xmath2 is decreased from infinity , type i materials become superconducting through a first order phase transition and type ii materials through a second order phase transition . more precisely , type i materials supercool , that is remain normal until @xmath147 , where nucleation occurs , followed by a discontinuous and irreversible jump in the maximum of @xmath20 . this is due to the fact that the transition between the normal phase and the superconducting phase is accompanied by positive energy ( the superconducting solution has a positive energy , that is bigger than the energy of the normal solution ) , which corresponds to a bifurcation curve of locally unstable solutions ( subcritical bifurcation ) . on the contrary , for type ii materials , @xmath20 varies continuously at the transition because the superconducting solution has a negative energy , that is lower than the energy of the normal solution , so that the bifurcated curve of superconducting solutions is locally stable . therefore , we can take as a definition of type i ( resp . type ii ) behaviours the fact that the bifurcation from the normal solution is subcritical ( resp . supercritical ) . hence , the curve @xmath123 separates the two regimes for finite @xmath146 and we have @xmath148 : region @xmath120 corresponds to type i behaviours , where the bifurcation is subcritical , and region @xmath121 to type ii , where the bifurcation is supercritical . in the limiting case @xmath32 , the formal computations of chapman @xcite indicate that the bifurcation curve of symmetric solutions changes from subcritical to supercritical as @xmath1 passes through @xmath149 from below . below , we give further discussions and interpretations of the regions and the curves depicted in figure [ figkappaasym ] . for @xmath150 , ginzburg @xcite obtained the critical value @xmath151 : if @xmath152 , @xmath80 is decreasing ( figure [ figs1 ] ) so that no supercooling nor superheating is possible . for @xmath153 , @xmath80 has the behaviour described by figure [ figs2 ] and displays both supercooling and superheating phenomena . our study indicates that as @xmath118 passes through @xmath122 , the bifurcation at @xmath79 changes from supercritical in @xmath119 to subcritical in @xmath120 . along the curve @xmath122 , the function @xmath80 is decreasing and we believe that @xmath154 , or equivalently , the term @xmath155 in the development of @xmath78 in ( [ bifurcatedsol ] ) . this means that as @xmath118 enters @xmath120 , the fold in the @xmath80 curve is going to appear near @xmath156 . the fact that @xmath154 on the curve @xmath122 is directly linked to ginzburg s observation that at the transition , the specific heat ( which can be related to the inverse of a derivative of the energy ) discontinuity is infinite . further insight into the nature of this bifurcation could be obtained if one were to analyze a formula indicating the direction of bifurcation . such a computation is directly linked to the evaluation of the term @xmath111 in ( [ bifurcatedsol ] ) . we refer to millman - keller @xcite for a formula for @xmath111 . this is the limiting curve separating type i and type ii behaviours as we have explained earlier . our study indicates that as @xmath118 passes from @xmath120 to @xmath121 across @xmath123 , there is a change from subcritical bifurcation in @xmath120 to supercritical in @xmath121 . moreover , along the curve @xmath123 , for @xmath157 the function @xmath80 has the graph indicated in figure [ figs2 ] . additionnally @xmath154 , since @xmath158 changes sign near @xmath156 . this is similar to what is happening on @xmath122 . again by computing @xmath111 exactly as described above for @xmath122 , one can obtain further mathematical results about the nature of the bifurcation and the location of @xmath123 . the following additional information has been obtained in @xcite about the asymptotic behaviour of the curve when @xmath0 is large : [ kappalargebhs3 ] ( bolley - helffer @xcite ) there exists a constant @xmath159 such that for @xmath160 and for @xmath2 the eigenvalue of a spectral problem , then the bifurcated solution ( [ bifurcatedsol ] ) bifurcating from the normal solution @xmath161 satisfies @xmath162 . moreover , there exists a constant @xmath163 such that @xmath164 can be chosen such that @xmath165 our study indicates that as @xmath118 passes across @xmath124 , the two folds at @xmath134 and @xmath140 which exist for a solution in @xmath121 tend to the same limit @xmath166 . as a result , the function @xmath80 is decreasing along @xmath124 , but there is an interior point @xmath167 with infinite slope , that is @xmath168 : this is where the curve gains its s shape as @xmath118 enters @xmath121 ( see figure [ figs3zoom ] ) . @xmath169 at the left end of @xmath124 , that is at the triple point @xmath170 , we have @xmath144 . it is an interesting open mathematical problem to define the curve @xmath124 by means of a spectral problem in terms of @xmath171 for some @xmath172 . as described earlier , our numerical experiments indicate the existence of a unique point @xmath173 for which @xmath174 . thus , in figure [ figkappaasym ] we see that this ` triple point ' @xmath170 provides the common intersection of the three curves . to our knowledge , the existence of this important point has not previously been reported in the literature . at the point itself , we observe that @xmath175 , @xmath80 decreases on @xmath73 and we believe that @xmath176 , that is @xmath177 in ( [ bifurcatedsol ] ) . however , if we let @xmath118 vary from the triple point along rays leading into the interior of @xmath119 , @xmath120 and @xmath121 , we expect to see entirely different behaviours . along a ray leading into @xmath119 from the triple point , the bifurcation diagram for symmetric solutions is the same as that shown in figure [ figs1 ] . similarly , along rays leading into @xmath120 and @xmath121 , the bifurcation diagrams for symmetric solutions are the same as those shown in figure [ figs2 ] and [ figs3 ] , respectively . clearly , a careful bifurcation analysis needs to be done to expose the behaviour of solutions in a neighbourhood of the point @xmath178 , @xmath179 , @xmath180 . recall that the behaviour of @xmath80 is described by figure [ figs1 ] . this regime was originally studied by ginzburg @xcite in the case @xmath150 . he uses an approximate model in which @xmath20 is held constant , i.e. @xmath181 . with this approximation , he observes that for @xmath182 , the curve @xmath80 is decreasing , as described in figure [ figs1 ] . he computes @xmath183 and claims that the approximate model remains valid in describing the properties of the solutions for small @xmath1 . for the approximate model studied by ginzburg , his computations have been made rigorous by bolley - helffer @xcite-@xcite . for the full system @xmath42 , there is only the following local result concerning the bifurcation of symmetric solutions from the normal solution . [ kappaasmallbh ] ( bolley - helffer @xcite ) for any @xmath184 , there exists @xmath185 such that for @xmath186 , @xmath187 and @xmath2 the eigenvalue of a spectral problem , then the bifurcated solution ( [ bifurcatedsol ] ) starting from the normal solution @xmath161 satisfies @xmath162 . in particular , the bifurcation is supercritical . rigorous treatment of the global problem remains open . recently , this regime has received an increasing amount of attention since high temperature superconductors typically have large @xmath1 . for mathematical results , see berestycki , bonnet , chapman @xcite or aftalion @xcite . [ klarge ] let @xmath0 be fixed . when @xmath1 tends to @xmath188 , the limiting behaviour of the curve @xmath80 is the union of a horizontal line at @xmath189 from @xmath190 to @xmath191 , and a curve joining @xmath192 to @xmath193 , where @xmath191 is given by @xmath194 as explained in @xcite , the limiting problem when @xmath1 tends to @xmath188 is @xmath195 it is proved in @xcite that there is a unique solution @xmath196 of @xmath197 for @xmath198 and @xmath199 so that @xmath200 . this solution is symmetric , i.e. @xmath201 and @xmath202 . the critical field @xmath203 is determined by the extra condition @xmath204 . moreover , any minimizer of @xmath37 converges to @xmath196 as @xmath1 tends to @xmath188 . thus , it is natural to expect that the minimizer for finite @xmath1 is symmetric and gives the horizontal line of the @xmath80 diagram . note that @xmath205 , which will be the same limit later in @xmath121 for @xmath32 , @xmath1 large . further study is needed to resolve the behaviour of solutions in this regime . recall that the behaviour of @xmath80 is described by figure [ figs2 ] . this regime was also studied by ginzburg @xcite with the approximate model where @xmath20 is constant . he observes that for @xmath206 , the curve @xmath80 has the same behaviour as the one described by figure [ figs2 ] . again , only the computations for the approximate model have been made rigorous by bolley - helffer @xcite . in particular , they prove that , for small @xmath1 , the curve @xmath80 has a local maximum @xmath87 , for which they give asymptotic formula consistent with physicistsresults . the only result to date for the full system @xmath42 concerns the bifurcation of symmetric solutions from the normal solution : [ kappaasmallbhs2 ] ( bolley - helffer @xcite ) for any @xmath184 , there exists @xmath185 such that for @xmath186 , @xmath207 and @xmath2 the eigenvalue of a spectral problem , then the bifurcated solution ( [ bifurcatedsol ] ) starting from the normal solution @xmath161 satisfies @xmath208 . in particular , the bifurcation is subcritical . here the first rigorous multiplicity result is the following : [ hkts2 ] ( hastings , kwong @xmath209 troy , @xcite ) let @xmath116 be fixed . then for a large enough , there is a @xmath210 in @xmath211 such that @xmath212 . in particular , this means that @xmath213 and that there are values of @xmath2 such that @xmath42 has two symmetric solutions . from @xcite , we also have the following limit , which was originally predicted by ginzburg - landau @xcite : [ hsinfty ] ( hastings , kwong @xmath209 troy , @xcite ) @xmath214 . the only bifurcation result available at the moment is the following : [ kappaalargebhs2 ] ( bolley - helffer @xcite ) for any @xmath184 , there exists @xmath185 such that for @xmath215 , @xmath216 and @xmath2 the eigenvalue of a spectral problem , then the bifurcated solution ( [ bifurcatedsol ] ) starting from the normal solution @xmath161 satisfies @xmath208 . in particular , the bifurcation is subcritical . for the limiting behaviour of the curve @xmath80 when @xmath0 is large , our experiments lead us to the following . [ ainfty ] let @xmath217 be fixed . when @xmath0 tends to @xmath188 , the limiting behaviour of the curve @xmath80 is the union of a horizontal line at @xmath189 from @xmath190 to @xmath218 , and a curve joining @xmath219 to @xmath220 . the schematic limiting shape of the diagram can be found in figure [ figainfk.3 ] or [ figainfk.5 ] at the end of section 4 . the curve ` sym ' corresponds to symmetric solutions . it means that the fold is getting close to the line @xmath189 and the knee is getting very thin , as illustrated by figure [ figa10k.3 ] in section 4 , where the extra curve corresponding to asymmetric solutions has been added . to help resolve this conjecture , one should investigate two relevant boundary value problems which we now describe : the first one is for the curve from @xmath221 to @xmath193 in figure [ figainfk.5 ] . it is derived by letting @xmath0 tend to @xmath188 in @xmath75 : [ s2sym1]\ { ll 1 ^2f=f(f^2+q^2 - 1 ) ( 0 , ) , + f(0)=,f(0)=0 , + q=qf^2 ( 0 , ) , + q(0)=0,q(0)= , . [ conjs2infty ] let @xmath222 be fixed . for each @xmath68 in @xmath73 , there exists a unique @xmath67 such that the solution of ( [ s2sym1 ] ) satisfies @xmath223 . we define @xmath224 . then @xmath80 is a continuous increasing function of @xmath68 , [ s2hinfty ] _ 0 h()= _ 1 h()=12 . [ 1sqrt2 ] let @xmath31 . for each @xmath68 in @xmath73 , there exists an @xmath67 such that the solution of ( [ s2sym1 ] ) satisfies @xmath223 . moreover , @xmath225 this is proved by chapman in @xcite using the special decoupling of @xmath42 for @xmath31 mentionned in @xcite : \ { ll s = f+fq/2 , + f = q+(f^2 - 1)/2 . . chapman observes in @xcite that a symmetric solution of @xmath42 leads to @xmath226 @xmath227 . it follows from uniqueness of solutions of a first order system that @xmath228 and therefore the full @xmath42 system reduces to \ { ll f=-fq/ 2 , + q=(1-f^2)/ 2 . . the second bvp relevant to conjecture [ ainfty ] corresponds to the horizontal line @xmath189 in the bifurcation diagram of figure [ figainfk.5 ] . it comes from a boundary layer analysis : the extremal part of the slab is sent to 0 and the symmetry plane to @xmath188 , so that we have [ s2sym2]\ { ll 1 ^2f=f(f^2+q^2 - 1 ) ( 0 , ) , + f(0)=0,_xf(x)=1 , + q=qf^2 ( 0 , ) , + q(0)=h,_xq(x)=0 . . this problem and the boundary layer analysis were originally derived by ginzburg @xcite , and ( * ? ? ? * figure 5 ) leads to the following conjecture : [ conj2s2infty ] let @xmath229 be fixed . there exists @xmath230 such that ( [ s2sym2 ] ) has a unique solution for @xmath231 and two solutions for @xmath232 . this explains why the curve going down from @xmath189 to @xmath156 in figure [ figainfk.5 ] starts at @xmath233 . for this conjecture , partial results have been provided by bolley - helffer : [ superheatingbh ] ( bolley - helffer @xcite ) for any @xmath229 , there exists @xmath234 such that there is no solution of ( [ s2sym2 ] ) if @xmath235 , and at least a solution if @xmath236 . moreover , [ degennes ] _ 0_^2()=24 . this formula was first derived by de gennes . recall that the behaviour of @xmath80 is described by figure [ figs3 ] . here the only rigorous multiplicity result is the following : [ hkts3 ] ( hastings , kwong @xmath209 troy , @xcite ) let @xmath237 be fixed . then for a large enough , there are @xmath210 and @xmath238 in @xmath211 such that @xmath239 , @xmath240 and @xmath241 . in particular , this means that the set @xmath242 and that there are values of @xmath2 such that @xmath42 has at least 3 solutions . [ ainftys3 ] let @xmath243 be fixed . when @xmath0 tends to @xmath188 , the limiting behaviour of the curve @xmath80 is the union of a horizontal line at @xmath189 from @xmath190 to @xmath218 , and a curve joining @xmath219 to @xmath220 . notice that the bifurcation curve for @xmath0 large has the same limiting shape as in @xmath120 , but this time the curve from @xmath244 to @xmath193 has negative slope . this is illustrated by figure [ figainfk.9 ] below with the curve ` sym ' corresponding to symmetric solutions . the part of the diagram corresponding to the line @xmath189 is described by the same limiting problem as in @xmath120 . thus , the field @xmath245 is the value derived from conjecture [ conj2s2infty ] . [ conjs3infty ] let @xmath246 be fixed . for each @xmath68 in @xmath73 , there exists a unique @xmath67 such that the solution of ( [ s2sym1 ] ) satisfies @xmath247 . we define @xmath224 . then @xmath80 is a continuous decreasing function of @xmath68 , and [ s3hinfty ] _ 0 h()= _ 1 h()=12 . the asymptotic behaviour of the upper bound for the superheating field @xmath234 in @xmath121 has been proved in the following : [ superheatingbhs3 ] ( bolley - helffer @xcite ) the field @xmath234 defined in theorem [ superheatingbh ] satisfies [ hinfty ] _ _ ( ) = 1 . this proves part of the formal asymptotic development derived by chapman @xcite : @xmath248 as @xmath1 tends to @xmath188 . notice that this means that the ` superheating part ' of the curve is getting small ( recall conjecture [ conjs3infty ] ) and is consistent with the fact that we reach the boundary of @xmath119 . we find the same limit as @xmath249 . the only result so far concerning the bifurcation of symmetric solutions in this regime is the following [ kappaalargebhs3 ] ( bolley - helffer @xcite ) for any @xmath184 , there exists @xmath185 such that for @xmath215 , @xmath251 and @xmath2 the eigenvalue of a spectral problem , then the bifurcated solution ( [ bifurcatedsol ] ) starting from the normal solution @xmath161 satisfies @xmath162 . in particular , the bifurcation is supercritical . recall from section 2 that @xmath44 is an asymmetric solution of @xmath42 if it is not symmetric . notice that if @xmath252 @xmath253 is a solution , then so is @xmath254 , but in the following we will count this as one solution instead of two and assume that the maximum of @xmath20 is reached in @xmath255 $ ] . below , we summarize the results of our numerical investigation of asymmetric solutions shown in figures [ figa3k.35 ] and [ figa3 ] : the vertical axis has been labeled @xmath256 . for the symmetric branch , @xmath256 refers to @xmath257 as in figures [ figs1 ] , [ figs2 ] and [ figs3 ] . for the asymmetric branch , @xmath258 , which is attained at some point @xmath259 different from zero . for simplicity , we will sometimes refer to the complete bifurcation diagram as @xmath80 . in figures [ figa3k.35 ] and [ figa3 ] , the curve @xmath80 for symmetric solutions is the typical curve in @xmath120 . on these curves , there is a point @xmath260 denoted by a square : it is the branching point of the branch of asymmetric solutions . more precisely , for @xmath261 close to @xmath262 , the asymmetric solution is nearly symmetric . the branch of asymmetric solutions leads from the branch of symmetric solutions to the branch of normal solutions . if the branch of asymmetric solutions crosses the branch of symmetric solutions without any square indicated , it just means that for this special @xmath2 , @xmath42 has a symmetric solution and an asymmetric solution having the same maximum value of @xmath20 . we have observed the following : [ xbeta ] let @xmath259 be the point where @xmath20 reaches its maximum . then @xmath259 is a decreasing function of @xmath68 and @xmath263 @xmath264 @xmath265 autoakappaa.pstex_t we now describe the two possible behaviours of the _ asymmetric curve _ @xmath80 . * figure [ figa3k.35 ] : for @xmath266 , @xmath80 is an increasing function of @xmath68 and @xmath267 , @xmath268 , with @xmath269 . if @xmath270 , there is a unique asymmetric solution of @xmath42 and if @xmath271 or @xmath272 , no such solution . * figure [ figa3 ] : for @xmath266 , @xmath80 is a decreasing function of @xmath68 until @xmath78 reaches a minimum value @xmath273 and then increasing to @xmath274 as @xmath68 goes to @xmath262 . thus , for @xmath275 , there are two asymmetric solution of @xmath42 , for @xmath276 , a unique asymmetric solution and for @xmath277 or @xmath272 , no such solution . this implies in particular that for figure [ figa3k.35 ] the bifurcation of asymmetric solutions from the normal solution is subcritical and for figure [ figa3 ] supercritical . in figure [ figkappaaasym ] , we have computed two new curves , @xmath278 and @xmath279 which play an essential role in the study of existence and multiplicity of asymmetric solutions . [ asym ] there exist two continuous functions @xmath278 and @xmath279 separating the @xmath118 plane into three connected regions @xmath280 , @xmath281 and @xmath282 . there exists exactly one point @xmath283 called the quintuple point ( with approximate value 1.23 ) such that @xmath284 and @xmath285 . moreover , @xmath286 with c approximately equal to 0.90 , and @xmath279 is defined and monotone decreasing on @xmath287 with @xmath288 . * region @xmath289 . there are no asymmetric solutions of @xmath42 . thus in the @xmath80 plane , we only see curves as shown in figure [ figs1 ] and [ figs2 ] . * region @xmath290 . if @xmath118 is in @xmath281 , the bifurcation diagram for asymmetric solutions is given by figure [ figa3k.35 ] , it has no fold and there is at most one asymmetric solution of @xmath42 . + moreover , when @xmath118 tends a point @xmath291 on the curve @xmath278 , then @xmath292 and @xmath274 tend to @xmath79 . when @xmath118 tends to some @xmath293 , then @xmath292 tends to @xmath79 . * region @xmath294 . if @xmath118 is in @xmath282 , the bifurcation diagram for asymmetric solutions is given by figure [ figa3 ] , it has a unique fold and there are at most two asymmetric solutions of @xmath42 . + moreover , when @xmath118 tends to a point @xmath291 on @xmath278 , then @xmath292 and @xmath274 tend to @xmath79 . when @xmath118 tends to a point @xmath293 on @xmath279 , then @xmath292 tends to @xmath79 and @xmath273 to @xmath274 . of particular physical interest is the following : [ conjhas ] let @xmath295 . if @xmath296 then @xmath297 . thus , as @xmath2 decreases from infinity , the material is in the normal state until @xmath2 reaches the value @xmath292 where the bifurcation of asymmetric solutions occurs . if @xmath298 and @xmath299 , or if @xmath300 and @xmath296 , then the bifurcating asymmetric branch is supercritical and plays an important role in the onset of superconductivity in the material . in the following we explain the implications of conjecture [ conjhas ] . in the physics literature , @xmath292 is defined as @xmath30 and @xmath79 as @xmath27 . it is well known to physicists that when @xmath1 is fixed bigger than @xmath301 and @xmath0 is taken large enough ( which corresponds to @xmath1 being above @xmath302 ) , when the exterior magnetic field @xmath2 is decreased from infinity , superconductivity is not nucleated first in the volume of the sample , which would give rise to symmetric solutions , but rather in a sheath near the surface , due to the existence of asymmetric solutions . this is called surface superconductivity . if the slab is very thick , the two surface solutions @xmath303 and @xmath304 do not interact . in this region , superconductivity is first nucleated in surface layers of size @xmath305 near the boundaries , and the middle part of the material is normal . now , if the slab is of intermediate size , the solutions @xmath303 and @xmath304 interfere to create vortices . indeed , recall that the original ginzburg - landau energy is gauge invariant so that a solution @xmath306 has the same energy as @xmath307 for any constant @xmath308 . thus , when the sample is not too large , and @xmath78 decreases below @xmath292 , the linear combination of the asymmetric solutions @xmath303 and @xmath304 create two dimensionnal vortices along the mid - plane @xmath309 . this is reflected in the formula : [ psi ] = k y ( f(x)+f(-x))+ik y ( f(x)-f(-x ) ) further discussion of vortex formation and the details of the derivation of formula ( [ psi ] ) are given in tinkham @xcite . in particular , the distance between two vortices along the @xmath309 plane is proportionnal to @xmath310 . indeed , when moving up the asymmetric branch @xmath80 , we find that @xmath259 tends to zero as @xmath68 approaches the branching point on the symmetric branch . this means that the distance between the two vortices tends to infinity . at the limit , the material is perfectly superconducting because @xmath20 is symmetric . in fact , as described in @xcite , @xmath79 and @xmath292 can both be measured : @xmath292 by nucleation of superconductivity as we have seen and @xmath79 by magnetic transition , which needs bulk superconductivity to take place . boeck and chapman @xcite have studied surface superconductivity in detail and they have determined the regimes of @xmath118 where the asymmetric solution gives rise to a surface sheath or to vortex solutions , through formula ( [ psi ] ) . now recall that the surface sheath is of size @xmath311 ( that is of order of the coherence length @xmath19 ) , so it is consistent that a condition for surface superconductivity to exist should be @xmath312 . as mentionned by tinkham @xcite , physicists had found that @xmath313 but did not know how to predict the details of changeover of behaviour on this curve . we believe that our numerical results answer this open question . a curve similar to @xmath278 was also mentionned by boeck and chapman @xcite . the global results of our numerical investigations are shown graphically in figure [ figsummary ] where for each region of interest , we refer to the relevant figure describing the bifurcation diagram of both symmetric and asymmetric solutions . autokafig.pstex_t @xmath314 @xmath315 recall that a fold in the symmetric branch correponds to a superheating phenomenon , a subcritical bifurcation at the highest nucleation field to a supercooling phenomenon and a supercritical bifurcation of asymmetric solution to surface superconductivity . in particular , when surface superconductivity occurs there is no supercooling . thus , our physical interpretation of our new regions in the light of the discussion given by tinkham @xcite is the following : @xmath316 or @xmath317 and @xmath300 , that is @xmath318 ( figure 1 ) : thin films with type i behaviour . there is no supercooling nor superheating . @xmath319 and @xmath320 and @xmath321 , that is @xmath322 ( figure 10 ) : thin film with type ii behaviour . there is no supercooling nor superheating , but there is surface superconductivity due to the presence of a supercritical branch of asymmetric solutions . @xmath323 and @xmath324 , that is @xmath325 and @xmath326 ( figures 2 and 6 ) : type i behaviour . these superconductors supercool , that is the bifurcated branches from the normal solution are subcritical and superconductivity is nucleated in the bulk . @xmath327 and @xmath328 , that is @xmath329 ( figure 7 ) : the material has type i characteristics because at @xmath79 the bifurcation of symmetric solutions is subcritical . however , the bifurcation of asymmetric solutions is supercritical , and this results in the formation of surface sheaths of superconductivity . thus , in this regime , tinkham @xcite has given the name type i 1/2 to the superconductor . @xmath330 , that is @xmath331 ( figure 9 ) : type ii behaviour , with superheating and surface superconductivity . thus , we claim that the curve @xmath279 is the limiting curve separating type i behaviours and type i 1/2 . note that surface sheath effects can be inhibited if the superconducting surface is coated with a normal metal . in this situation , the boundary condition @xmath332 needs to be changed and replaced by @xmath333 as introduced by de gennes @xcite . this is why the type of superconductor is determined by the behaviour of the bifurcated branch at @xmath79 and not at @xmath292 . to our knowledge , the global behaviour of the asymmetric bifurcation curves have never been mentionned in the literature . in light of the discussion given above the following open problems have physical importance . prove how asymmetric solutions arise from bifurcation from symmetric solutions ; prove that as soon as the asymmetric curve arises from a subcritical bifurcation from the normal solution , the curve has no fold ; prove that as soon as the asymmetric curve arises from a supercritical bifurcation from the normal solution , the curve has a unique fold . as mentionned earlier , our experiments reveal the existence of a second key - point , in addition to the triple point defined in section 3 . to our knowledge , the existence of this point has not previously been reported in the literature . for our discussion of the properties of this point , we refer the reader to figure [ figsummary ] . first , we let @xmath283 be the unique value such that @xmath334 . we have @xmath335 . then our investigation shows that @xmath336 is on the curve @xmath278 and is the end point of @xmath279 . at the point itself , we observe that @xmath80 is decreasing to @xmath337 and there are no asymmetric solutions . however , if we let @xmath118 vary from this point , we expect to see 5 different behaviours of the bifurcation diagram . along a ray leading into @xmath338 , the bifurcation diagram is the same as shown in figure [ figs1 ] and there is no asymmetric solution . if @xmath118 lies on a ray leading into @xmath339 , the bifurcation diagram is the same as shown in figure [ figa1k1.2 ] : the asymmetric solutions bifurcate from @xmath340 and the asymmetric branch has a fold , though it is not easy to see it on the picture . next , let @xmath118 lie on a ray leading into @xmath341 ; the bifurcation diagram is as in figure [ figs2 ] so that the fold in the symmetric branch results from a subcritical bifurcation . if the ray leads into @xmath342 , the bifurcation diagram is as in figure [ figa3k.35 ] so that there is a fold in the symmetric branch resulting from a subcritical bifurcation at @xmath79 and there is also an asymmetric branch , with no fold , bifurcating from @xmath292 . both @xmath79 and @xmath292 are very close to @xmath340 . finally , if @xmath118 lies in @xmath343 , the diagram is as in figure [ figa3 ] , this time the branch of asymmetric solutions has a fold . as we have seen in the discussion following conjecture [ conjhas ] , the asymmetric solutions which appear play an important role in 2d vortex formation for intermediate values of @xmath0 larger than @xmath283 . in fact , our investigation indicates that their existence is due to a complicated bifurcation phenomenon which occurs at the quintuple point . thus , a complete mathematical investigation of bifurcation phenomena occuring at the quintuple point needs to be carried out . along the curve @xmath278 , there are only symmetric solutions . as @xmath118 passes from @xmath280 to @xmath281 or @xmath282 across @xmath278 , the asymmetric solutions bifurcate from @xmath156 , @xmath344 and @xmath292 tends to @xmath79 as @xmath118 approaches a point on @xmath278 . the nonexistence of asymmetric solutions bifurcating from the normal solution for @xmath250 small was obtained by bolley - helffer @xcite . [ bhunderk4 ] ( bolley - helffer , @xcite ) there is a constant @xmath185 such that for @xmath186 , there exists a unique nontrivial @xmath345 curve @xmath105 of solutions of @xmath42 bifurcating from a normal solution @xmath106 . these solutions bifurcate from the particular normal solutions @xmath346 where @xmath2 is the eigenvalue of a spectral problem , and therefore are symmetric solutions . the existence of asymmetric solutions bifurcating from the normal solution for @xmath250 large was obtained by bolley - helffer @xcite . [ bhabovek4 ] ( bolley - helffer , @xcite ) there exists a constant @xmath347 such that for @xmath348 , there exist exactly three nontrivial @xmath345 curves @xmath349 @xmath350 of solutions of @xmath42 bifurcating from normal solutions @xmath106 , where @xmath2 is an eigenvalue of a spectral problem . one curve of solutions starts from the particular normal solutions @xmath351 @xmath352 , and is a curve of symmetric solutions . the two others correspond to @xmath353 and @xmath354 and are asymmetric solutions . these two theorems only give local behaviour of the bifurcation curve near the normal solutions , but this scaling is consistent with the behaviour of @xmath278 which is @xmath355 . a global treatment of the region below @xmath278 for the full ginzburg - landau system remains an important open problem . as we have seen , @xmath279 is an extremely important curve . our study indicates that as @xmath118 passes from @xmath281 to @xmath282 across @xmath279 , there is a change of bifurcation for the asymmetric branch from subcritical in @xmath281 to supercritical in @xmath282 . this behaviour of the asymmetric branch is similar to what happens to the symmetric branch across @xmath122 and @xmath123 . the following additionnal information has been obtained in @xcite about the asymptotic behaviour of @xmath279 when @xmath0 is large . [ k5bh ] ( bolley - helffer , @xcite ) there exists a constant @xmath356 approximately equal to 0.4 such that for any @xmath184 there exists @xmath357 such that for @xmath358 , @xmath359 , @xmath360 and @xmath2 the eigenvalue of a spectral problem , then the solution of the form ( [ bifurcatedsol ] ) bifurcating from the normal solution @xmath106 satisfies @xmath361 . in particular , if @xmath362 , the bifurcation is subcritical , and if @xmath363 the bifurcation is supercritical . notice that @xmath356 is what we call @xmath364 . in @xcite , @xmath356 is given in terms of two integrals . this value already appears in physics papers ( see @xcite for instance ) and it is known to separate different behaviours of type i superconductors . the only rigorous result concerning the behaviour of the asymmetric branch is the following , which can be illustrated by figure [ figa3k.9asym ] . [ theoht](hastings and troy @xcite ) let @xmath365 be fixed . then for @xmath0 large enough , there is a range of @xmath2 for which there is no symmetric solution and yet there is an asymmetric solution . this theorem implies in particular that in the @xmath118 range mentionned , we have @xmath366 and @xmath297 . this gives a partial resolution to conjecture [ conjhas ] . the physical importance of this result will be discussed in section 5 . figure [ figa10k.3 ] illustrates the behaviour of the bifurcation diagram when @xmath0 is large . this is the limiting case of the one shown in figure [ figa3k.35 ] when @xmath0 gets large . @xmath367 we see that the branching point of the asymmetric curve has the same limit as the fold of the symmetric curve . thus , for the limiting behaviour of the asymmetric curve when @xmath0 is large , our experiment lead us to the following . [ conjasymalarge ] let @xmath1 be fixed . when @xmath0 tends to @xmath188 , the limiting behaviour of the bifurcation curve of the asymmetric solutions is the union of a horizontal line at @xmath189 from @xmath368 to @xmath369 ( defined in conjecture [ conj2s2infty ] ) and a curve joining @xmath219 to @xmath370 , where @xmath371 is approximately 0.59 , and is defined as the minimum over @xmath67 of the first eigenvalue of the neumann problem of the harmonic oscillator in @xmath372 . in particular , [ limhb ] _ ah_b= _ _ a_as=1/ 2 . this implies that @xmath373 : when @xmath374 , the curve from @xmath156 to @xmath189 has positive slope resulting from a subcritical bifurcation ( see figure [ figainfk.3 ] ) and when @xmath375 , this curve has negative slope resulting from a supercritical bifurcation ( see figures [ figainfk.5 ] and [ figainfk.9 ] ) . autoainfk.3.pstex_t autoainfk.5.pstex_t autoainfk.9.pstex_t the value of @xmath376 and the corresponding spectral problem were derived by bolley - helffer : [ bhainftyasym ] ( bolley - helffer , @xcite ) let @xmath1 be fixed . then [ hasainfty ] _ a h_as=. this is consistent with the experimental value found by st - james and de gennes @xcite which is @xmath377 . for fixed @xmath1 , chapman , howinson , mcleod and ockendon @xcite prove the existence and uniqueness of the solution of [ asym2]\ { ll 1 ^2f=f(f^2+q^2 - 1 ) ( - , ) , + _ x - f(x)=1,_xf(x)=0 , + q=qf^2 ( - , ) , + _ xq(x)=0,_xq(x)=12 , this gives the point @xmath189 , @xmath233 in the bifurcation diagram , but further study is needed to resolve conjecture [ conjasymalarge ] . we define @xmath44 to be locally stable if it is a local minimizer of @xmath37 in @xmath378 and unstable otherwise . local and global stability properties of both symmetric and asymmetric solutions have received recent attention in the literature . as a result of our investigations , we make the following conjecture , which is to be read in the light of figure [ figsummary ] . in @xmath341 and in @xmath342 , the symmetric solution is locally stable for @xmath379 , the normal solution is locally stable for @xmath86 , and in @xmath342 the asymmetric solution is unstable . moreover , there exists @xmath380 such that the symmetric solution is a global minimizer for @xmath381 and the normal solution is a global minimizer for @xmath382 . in @xmath339 the symmetric solution is locally stable for @xmath383 , the asymmetric solution is locally stable for @xmath384 and the normal solution is locally stable for @xmath385 . moreover , there exists @xmath386 such that the symmetric solution is a global minimizer for @xmath381 , the asymmetric solution is a global minimizer for @xmath387 and the normal solution is a global minimizer for @xmath388 . in @xmath343 the symmetric solution is locally stable for @xmath379 , the asymmetric solution is locally stable for @xmath384 and the normal solution is locally stable for @xmath385 . moreover , there exists @xmath380 such that the symmetric solution is a global minimizer for @xmath381 , the asymmetric solution is a global minimizer for @xmath387 and the normal solution is a global minimizer for @xmath385 . in @xmath389 the symmetric solution is locally stable for @xmath379 and locally unstable for @xmath390 , the asymmetric solution is locally stable for @xmath384 and the normal solution is locally stable for @xmath385 . moreover , there exists @xmath380 such that the upper symmetric solution is a global minimizer for @xmath381 , the asymmetric solution is a global minimizer for @xmath387 and the normal solution is a global minimizer for @xmath385 . this conjecture is consistent with physical observations of superheating and supercooling . the critical field @xmath87 is often called superheating field because when starting from a symmetric superconducting solution near @xmath190 , the material will remain in this locally stable state until @xmath87 is reached . then as @xmath2 is increased above @xmath87 , the maximum of @xmath20 has a discontinuity and bulk superconductivity is destroyed . now , on the contrary , starting from normal solution , when @xmath2 is decreased from infinity , the material remains in the normal state until the nucleation field ( @xmath79 or @xmath292 ) is reached . if the branch of bifurcating solutions is unstable , this field is called supercooling field because nucleation occurs through a jump in the maximum of @xmath20 : the material gets into the symmetric state which is the global minimizer . these two phenomena give rise to hysteresis loops . for all the cases they have studied , bolley - helffer @xcite have established the local stability or instability of the branches bifurcating from normal solutions . in fact , when they proved the bifucation to be subcritical , they also proved local instability of the branch , and stability for supercritical branches , except in @xmath389 and @xmath339 . note that in @xmath389 and @xmath339 , though the bifurcation is supercritical , they proved that the symmetric solution is unstable . for @xmath0 fixed and @xmath391 defined earlier in section 3.5.2 , aftalion @xcite proved that the global minimizer tends to a symmetric solution when @xmath1 tends to infinity . indeed , in this range corresponding to @xmath339 , there are no asymmetric solutions for @xmath391 , but for @xmath392 , it is an interesting open problem to prove that both symmetric and asymmetric solutions coexist and the asymmetric solution is the global minimizer . as we pointed out in section 4 , in @xmath389 and @xmath343 , the formation of sheaths and vortices is related to the existence of stable asymmetric solutions . the hastings - troy result @xcite gives a first step in the mathematical analysis of this phenomenon : hastings and troy @xcite gave a first step in the mathematical analysis of this phenomenon since they proved that for fixed @xmath393 and large @xmath0 , there is a range of @xmath78 values for which the asymmetric solution is the global minimizer . as we have said before in section 4 , when the asymmetric solution is a global minimizer , it is not known for which regime of @xmath118 , it gives rise to surface sheath or to vortices . we have provided a comprehensive study of the structure of symmetric and asymmetric solutions for the entire range @xmath394 . throughout we have tied together our new observations with previously known results . at the core of our new results are the existence of the triple point and the quintuple point . these points provide the birth of families of asymmetric and symmetric solutions via bifurcation . already , one of the authors , together with s.j . chapman , has begun further investigations of the structure of both of these points and this has resulted in two new papers @xcite , @xcite . 99 1997 _ on the minimizers of the ginzburg - landau energy for high kappa : the one - dimensional case . _ european j. appl . math . * 8 * , 331 - 345 . 1998 _ asymptotic analysis of the bifurcation diagram for symmetric one dimensionnal solutions of the ginzburg - landau equations . _ preprint lmens 98 - 34 . 1998 _ asymptotic analysis of a secondary bifurcation of the one dimensionnal ginzburg - landau equationsof superconductivity . _ preprint lmens 98 - 36 . 1994 _ a semi - elliptic system arising in the theory of superconductivity . _ nonlinear anal . * 1 * , 1 - 21 . 1994 _ ginzburg - landau vortices . _ birkhuser . 1997 _ bifurcation to vortex solutions in superconducting films . _ european j. appl . math . * 8 * , 125 - 148 . 1991 _ familles de branches de bifurcation dans les quations de ginzburg - landau . _ @xmath395 * 25 * , 307 - 335 . 1993 _ an application of semi - classical analysis to the asymptotic study of the supercooling field of a superconducting material . henri poincar ( phys . thorique ) * 58 * , 189 - 233 . 1996 _ rigorous results on ginzburg - landau models in a film submitted to an exterior parallel magnetic field i. _ nonlinear stud . * 3 * , 1 - 29 . 1996 _ rigorous results on ginzburg - landau models in a film submitted to an exterior parallel magnetic field ii . _ nonlinear stud . * 3 * , 121 - 152 . 1996 _ rigorous results for the ginzburg - landau equations associated to a superconducting film in the weak @xmath1 limit . _ rev . * 8 * , 43 - 83 . 1997 _ the ginzburg - landau equations in a semi - infinite superconducting film in the large @xmath1 limit . _ european j. appl . math . * 8 * , 347 - 367 . 1998 _ stability of bifurcating solutions for the ginzburg - landau equations . * 10 * 579 - 626 . 1994 _ nucleation of superconductivity in decreasing fields i. _ europ . j. appl . math . * 5 * , 449 - 468 . 1994 _ nucleation of superconductivity in decreasing fields ii . j. appl . math . * 5 * , 469 - 494 . 1995 _ superheating field of type - ii superconductors . _ siam j. appl . math . * 55 * , 1233 - 1258 . 1991 _ normal superconducting transitions in landau - ginzburg theory . royal soc . edinburgh , * 119 * 117 - 124 . 1992 _ macroscopic models of superconductivity . _ siam review , * 34 * 4 , 529 - 560 . 1992 _ introduction to superconductivity and high-@xmath396 materials . _ world scientific . 1997 _ auto97 : continuation and bifurcation software for ordinary differential equations . _ available by ftp from ftp.cs.concordia.ca/pub/doedel/auto . 1981 _ auto : a program for the automatic bifurcation analysis of autonomous systems . 10th manitoba conf . on num . math . and comp . , univ . of manitoba , winnipeg , canada ) * 30 * , 265 - 284 . 1991 _ numerical analysis and control of bifurcation problems . j. bifurcation and chaos , * 1 * , part i , 493 - 520 ; part ii , 745 - 772 . 1992 _ analysis and approximation of the ginzburg - landau model of superconductivity . _ siam review , * 34 * , 1 , 54 - 81 . 1966 _ superconductivity of metals and alloys . _ addison wesley . 1958 _ on the destruction and onset of superconductivity in a magnetic field . _ soviet phys . jetp , * 34 * , 78 - 87 . 1950 _ on the theory of superconductivity . _ soviet phys . jetp , * 20 * , 1064 - 1082 . english translation : _ men of physics : l.d.landau_ ed . d.ter harr , pergamon , oxford ( 1965 ) 138 - 167 . 1996 _ the existence of multiple solutions for a ginzburg - landau type model of superconductivity . . j. appl . math . * 7 * , 559 - 574 . 1997 _ on the existence of asymmetrics minimizer in the one - dimensional ginzburg - landau model of superconductivity _ , siam , in honor of i. stakgold . 1997 _ there are asymmetric minimizers for the one dimensionnal ginzburg - landau model of superconductivity . _ to appear in siam j. of math . 1995 _ on the one - dimensional ginzburg - landau bvps . equations * 8 * , 1395 - 1405 . 1969 _ perturbation theory of nonlinear boundary - value problems . _ j. math . * 10 * , 2 , 342 . 1963 _ onset of superconductivity in decreasing fields . * 7 * 306 - 307 . 1969 _ type ii superconductivity . _ pergamon , oxford . 1988 _ from equilibrium to chaos ; practical bifurcation and stability analysis . _ elsevier publ . co. 1983 _ branch switching in bifurcation problems in ordinary differential equations . _ numer . math . * 41 * , 93 - 116 . 1970 _ superfluidity and superconductivity . _ adam hilger , bristol and new york . 1975 _ introduction to superconductivity . _ mcgraw - hill , new - york . 1992 _ symmetric superconducting states in thin films . _ siam j.appl.math . * 52 * , 614 - 629 . 1990 _ boundary value problems of the ginzburg - landau equations . soc . of edin . * 114a * , 355 - 365 .
this paper gives a complete description of the solutions of the one dimensional ginzburg - landau equations which model superconductivity phenomena in infinite slabs . we investigate this problem over the entire range of physically important parameters : @xmath0 the size of the slab , @xmath1 the ginzburg - landau parameter , and @xmath2 , the exterior magnetic field . we do extensive numerical computations using the software auto , and determine the number , symmetry and stability of solutions for all values of the parameters . in particular , our experiments reveal the existence of two key - points in parameter space which play a central role in the formation of the complicated patterns by means of bifurcation phenomena . our global description also allows us to separate the various physically important regimes , to classify previous results in each regime according to the values of the parameters and to derive new open problems . in addition , our investigation provides new insight into the problem of differentiating between the types of superconductors in terms of the parameters . * pacs numbers : * 2.60.lj , 5.70.fh , 74.20.de * keywords : * superconductivity , bifurcation , symmetry , surface effects .
laser cooling and multiphoton spectroscopy of atoms and molecules commonly require the use of narrow - band frequency stabilised lasers working at various wavelengths . single - frequency cw ti : sapphire lasers with ring cavities are advantageous for laser spectroscopy due to their high power in single frequency and tuneability over a broad spectral range . a variety of techniques have been developed for laser frequency stabilisation needed in many applications in optics and spectroscopy . typically , the laser output is locked to a highly stable reference cavity or to an atomic transition by means of fringe - side locking @xcite or most widely used pound - drever - hall ( pdh ) technique @xcite . highly - stable fabry - prot cavities commonly use ultra - low expansion ( ule ) glass spacers and are temperature - stabilised and kept in vacuum in order to avoid drifts of the resonant frequency induced by temperature and air pressure variations @xcite . in such systems , the mirrors can not be placed on piezo - ceramic transducers ( pzt ) , which feature large frequency drifts of tens of mhz per hour @xcite . therefore , the laser frequency can only be stabilized at a series of fixed values , separated by the free spectral range of the cavity . to overcome this obstacle it is possible to shift the laser frequency using an external acousto - optical modulator or sideband locking technique with electro - optical modulator @xcite . locking the laser to an atomic transition via saturation spectroscopy is commonly used in experiments on atomic spectroscopy and laser cooling . however , this technique is difficult to apply in the case when the desirable laser wavelength does not match the wavelength of available atomic transitions . the same difficulty occurs when the reference cavity is locked to a highly stable reference laser . another recently developed method is based on the use of a high - precision wavelength meter , which also requires additional reference laser to achieve high accuracy @xcite . one of the common experimental demands is multistep excitation of ultracold atoms which requires locking of the laser to a transition between excited atomic states . for two - step excitation it is possible to use the resonances of electromagnetically - induced transparency in a gas cell to lock the lasers @xcite . however , generalization of this scheme for three - step laser excitation is technically challenging due to small dipole moments of transitions to rydberg states and necessity to combine three laser beams in a vapor cell . we have developed a relatively simple and inexpensive method relying on the digital measurement of the frequency difference between the output of a ti : sapphire laser and that of a highly stable diode laser using an auxiliary fabry - prot interferometer . the diode lasers which are locked to an atomic transition between the ground and first excited states of alkali - metal atoms are commonly used in laser cooling experiments . it is important that a single reference laser can be used to lock other lasers to the arbitrary wavelengths . essentially , in our locking system the time delay is measured between emergence of transmission peaks coming from the radiation of the ti : sapphire laser and from that of the highly stable diode laser as the fabry - prot cavity is scanned . this method has been successfully implemented in several previous works @xcite . a rather complicated analogue electronic detection of transmission peaks in the scanning cavity was first implemented in ref . a digital integrator was used to provide the feedback signal to the laser . the frequency drift of the laser did not exceed 1 mhz compared to the reference frequency - stabilised he - ne laser . in the later works , the electronics for laser stabilisation was substantially simplified by introduction of adc / dac modules with a computer control @xcite . in ref . @xcite the interference filters were used to individually measure the transmission of multiple laser beams through the cavity , so that several lasers could be locked simultaneously . in ref . @xcite the scanning rate was increased to 1 khz compared to 200 hz in ref . @xcite and the transmission signal was sampled at 4 mhz . increased scanning rate permits frequency locking in a wider bandwidth , but requires faster and more expensive adc conversion and data analysis . in ref . @xcite the scanning frequency was further raised to 3 khz and the feedback signal was generated using an analogue peak detector and a fast programmable microcontroller . ( color online ) . scheme of the experimental setup . the ti : sapphire laser is locked to cavity 2 using side - fringe locking technique . cavity 1 is scanned at 200 hz . the transmission peaks from two lasers are detected on pd1 , and recorded using adc , and analysed on a computer . the feedback signal from dac is sent to the pzt of cavity 2 . the part of the experimental setup which represents the commercial teknoscan laser lock is shown in the rectangular box . ] our approach relies on two confocal cavities for line narrowing and long - term frequency stabilisation of the laser . the confocal reference cavity with the mirrors separated by a 10-cm invar spacer is more stable than a long ( approximately 70 cm ) laser resonator . we use fast analogue locking electronics manufactured by the tekhnoscan company ( novosibirsk , russia ) to lock the laser on the reference cavity , and then we only compensate for the long - term drift of the optical length of the reference cavity . this technique allows us to avoid the necessity of using high - speed electronics for peak detection and data analysis , comparing to the previous works @xcite . we use a relatively cheap zetlab@xmath0 zet210 adc / dac module ( russia ) sampled at 400 khz for data acquisition and generation of the feedback output voltage , and national instruments labview@xmath0 for data analysis . the scheme of our experimental setup is shown in fig . [ scheme ] . the radiation of the ti : sapphire laser is split by semi - transparent mirror m1 and directed to confocal cavities 1 and 2 . the intensity of the light transmitted through the cavities is measured with photodiodes pd1 and pd2 . semi - transparent mirror m2 is used to create a reference laser beam for the side - fringe locking of the ti : sapphire laser to a mode of cavity 2 by measuring the difference between the intensities on photodiodes pd2 and pd3 and sending the feedback signal to the laser . this is a part of the commercial tekhnoscan laser lock . cavity 2 is a confocal fabry - prot resonator with a 10-cm invar spacer and one of the mirrors mounted on a pzt . the commercially available locking electronics by tekhnoscan reduces the short - term linewidth of laser radiation to less than 10 khz . by applying an offset voltage to the pzt of cavity 2 , it is possible to tune the laser frequency within the free spectral range of the cavity , which is 750 mhz . the laser frequency drift caused by temperature and air pressure variations , as well as the drift of the pzt , is more than 30 mhz per hour even for the cavity with temperature stabilisation . this does not meet our requirements for experiments with cold rydberg atoms where resonance widths at laser excitation are less than 5 mhz , and we need to keep the laser on resonance with the atomic transition for tens of minutes . in order to compensate for this drift , we combine on semi - transparent mirror m4 the radiation of the ti : sapphire laser , which is already locked to cavity 2 , with the radiation of the reference external - cavity diode laser with wavelength 780 nm , locked to an atomic transition in a rubidium vapour cell via pound - drever - hall technique @xcite . the linewidth of this laser was measured by observing the beat spectrum of two identical lasers averaged during 100 ms , and was found to be less than 1 mhz . the radiation of both lasers is send to cavity 1 which is scanned by a triangular signal at 200 hz from a gw instek dds function generator sfg-2004@xmath0 . low - voltage output of the function generator is amplified with a home - built high - voltage amplifier , which drives the pzt . this amplifier also provides additional offset dc voltage to the pzt . the scanning cavity 1 is a confocal fabry - prot resonator with a 12.5-cm invar spacer corresponding to a free spectral range of @xmath1 mhz . the cavity can be temperature stabilised , however this does not significantly change the performance of the system . the signal from photodiode pd1 , which has a bandwidth of 10 mhz , is measured using a commercial zetlab zet 210@xmath0 adc / dac module with maximum sampling frequency of 400 khz . ( color online ) . ( a ) timing diagram of the measurement . the ramp signal is used for synchronisation . the first minimum of the ramp signal and the measurement time bin are separated by offset time @xmath2 ; ( b ) time - dependent photoelectric signal on pd1 . ] as the signal from photodiode pd1 can only be measured continuously during the period defined by the adc buffer size of 4096 bytes , we have to use the signal from the function generator , which is measured by a second adc channel , for synchronization of the data acquisition process . we read the data from the adc buffer , which contains at least two periods of the synchronisation signal . then we find the time of the first minimum of the synchronization signal @xmath3 and extract data from a time bin defined by the specified width and offset @xmath2 as shown in fig . [ timing](a ) . adjustment of the dc offset voltage at the pzt of cavity 1 can be used to tune the positions of the peaks within the window without interrupting the frequency locking process . this is necessary to move the transmission peaks away from the turning points of the ramp signal and to compensate for temperature drift of the cavity in order to keep all the peaks within the measurement window . the measured time - dependent photoelectric signal on pd1 , which is proportional to the transmission of cavity 1 , is plotted in fig . [ timing](b ) . we measure the time intervals @xmath4 and @xmath5 between two peaks from the reference laser and the peak from the ti : sapphire laser using the peak detect virtual instrument implemented in national instruments labview@xmath0 , which uses quadratic fit approximation for accurate determination of the peak centers . then we calculate the ratio : @xmath6 this ratio depends linearly on the frequency of the ti : sapphire laser and ranges between 0 and 1 depending on the relation between the frequencies of the reference laser and ti : sapphire laser . we apply a 3-point average filter to an array of measured ratios . initially , the ti : sapphire laser is locked to cavity 2 . by changing the offset voltage on the pzt of this cavity we can tune the laser to any desired wavelength which define the initial ratio @xmath7 of time intervals between the transmission peaks of cavity 1 . for long - term stability , we need to minimize the variation @xmath8 . this is achieved by applying an additional voltage to the pzt of cavity 2 . the feedback signal is calculated using national instruments labview@xmath0 pid toolkit and then converted to voltage via the dac of the zetlab zet 210@xmath0 module . to tune the laser locked to cavity 1 , we can manually change the initially set value @xmath7 in the control program during the experiment . the locking system then drives the laser to a newly defined locking point . ( color online ) . ( a ) measured time trace of ratio @xmath9 of time intervals between the transmission peaks ; ( b ) time trace of dac output voltage ; ( c ) measured laser frequency drift with the ti : sapphire laser locked to cavity 2 only ( without temperature stablization of the cavity ) ; ( d ) measured laser frequency drift with the ti : sapphire laser locked to both cavity 1 and cavity 2 . the absolute long - term accuracy of the wavemeter is 200 mhz , but the relative accuracy is much better . ] the results of measurement of the ratio of the time intervals _ r _ during approximately 20 minutes are presented in fig . [ performance](a ) . the standard deviation of @xmath9 was 0.00115 which corresponds to the error of 0.7 mhz in the determination of the laser frequency . this result is consistent with the previous measurements @xcite . the output voltage of dac , shown in fig . [ performance](b ) , was automatically increased by the locking system from zero to around 0.7 v during the measurement due to the temperature and air pressure variations which affect the optical length of cavity 2 , and due to the drift of the pzt of cavity 2 . we have also measured the wavelength of ti : sapphire laser output using high finesse@xmath0 ws6 wavemeter . although this model has limited long - term accuracy ( 200 mhz error in absolute value ) , we have used it to study the short - term fluctuations and drifts of the laser frequency . in the mode when the laser was locked to cavity 2 only without temperature stabilisation of the cavity the drift was around 3 mhz per minute [ see fig . [ performance](c ) ] . when the laser was also locked to cavity 1 the measured drifts were reduced to 200 khz per minute [ see fig . [ performance](d ) ] which can be attributed to the temperature drift of the wavemeter . another possible source of error could be the drift of the reference laser , but we have not observed it in the beat spectrum of two identical reference lasers locked to 780 nm . our approach is close to the laser locking system described in ref . @xcite where low - frequency data acquisition at 400 khz had been used along with fully digital data analysis . however , it has been noted in ref . @xcite that their system was unable to compensate for acoustic noises , and it has been proposed to use an additional stable cavity and analog electronics for linewidth narrowing . in our work the laser is locked to cavity 2 by analog electronics via fringe - side locking , which removes the acoustic noises and provides less than 10 khz rms laser linewidth relative to cavity 2 . our digital locking system compensates only for slow drifts of cavity 2 . these are the main differences from the previous work @xcite . the limiting factors for the system performance are rather low scanning rate ( 200 hz ) , finesse of cavity 2 ( @xmath10 ) , and stability of the reference laser . higher scanning rate requires faster sampling , which was not possible with the adc we used . however , the peak positions fluctuate from scan to scan even for a stabilised laser due to the nonlinearities and hysteresis of the pzt @xcite . therefore we believe that increase of the scanning rate will not substantially improve the performance of our system , and we need to compensate only for slow temperature drifts of cavity 2 and air pressure variations . there is no need to further increase the bandwidth of the locking system by use of high - speed digital electronics , in contrast to previous works @xcite . besides that , for a faster lock , the fluctuations of the peak positions can substantially increase the linewidth of the laser . the system performance was good enough to compensate for the long - term drift of cavity 2 . we have used this system to lock the ti : sapphire laser at 743 nm for three - step laser excitation of cold rubidium atoms into the rydberg states , as shown in fig . [ rydberg](a ) @xcite . we used a cooling 780 nm laser diode as a reference source . the typical spectrum of laser excitation of 37@xmath11 rydberg rubidium atoms obtained by scanning of the ti : sapphire laser by linear ramp voltage applied to cavity 2 is shown in fig . [ rydberg](b ) . the width of the atomic resonance in our experiment was around 5 mhz . therefore the stability of our laser lock was good enough for rydberg excitation of ultracold rubidium atoms . ( color online ) . ( a ) scheme of the three - photon excitation of rubidium rydberg atoms ; ( b ) a typical spectrum of laser excitation of cold rubidium 37@xmath11 rydberg atoms obtained by scanning of the ti : sapphire laser working at 743 nm . ] y. y. jiang , a. d. ludlow , n. d. lemke , r. w. fox , j. a. sherman , l .- s . ma , c. w. oates , `` making optical atomic clocks more stable with @xmath12-level laser stabilization , '' nature photonics * 5 * , 158 ( 2011 ) . t. kessler , c. hagemann , c. grebing , t. legero , u. sterr , f. riehle , m. j. martin , l. chen , j. ye , `` a sub-40-mhz - linewidth laser based on a silicon single - crystal optical cavity , '' nature photonics * 6 * , 687 ( 2012 ) . r. p. abel , a. k. mohapatra , m. g. bason , j. d. pritchard , k. j. weatherill , u. raitzsch and c. s. adams `` laser frequency stabilization to highly excited state transitions using electromagnetically induced transparency in a cascade system '' appl . 94 , 071107 ( 2009 ) . k. matsubara , s. uetake , h. ito , y. li , k. hayasaka , and m. hosokawa , `` precise frequency - drift measurement of extended - cavity diode laser stabilized with scanning transfer cavity , '' japanese journal of applied physics * 44 * , 229 ( 2005 ) . d. b. tretyakov , v. m. entin , e. a. yakshina , i. i. beterov , c. andreeva , and i. i. ryabtsev , `` controlling the interactions of a few cold rb rydberg atoms by radiofrequency - assisted frster resonances , '' phys . rev . a 90 , 041403(r ) ( 2014 ) . v. m. entin , e. a. yakshina , d. b. tretyakov , i. i. beterov , and i. i. ryabtsev , `` spectroscopy of the three - photon laser excitation of cold rubidium rydberg atoms in a magneto - optical trap , '' j. exp . theor . phys . * 116 * , 721 ( 2013 ) . * ilya i. beterov * is a senior researcher at the rzhanov institute of semiconductor physics , siberian branch of russian academy of sciences and lecturer at novosibirsk state university . he received ph.d . degree from the institute of automation and electrometry siberian branch of russian academy of sciences in 2008 . his research area is quantum information with rydberg atoms . he is the author of more than 30 publications in scientific journals . * asparuh g. markovski * works at the technical university of sofia , bulgaria , at the faculty of automation . he received his ph . d. from the same university in 2004 . his major research interests are in the area of robust control , low - cost means for automation , optimization in the control theory . * sergey kobtsev * is a head of division of laser physics and innovative technologies at novosibirsk state university . he received phd ( 1992 ) and doctor of sciences ( 2010 ) degrees from the institute of automation and electrometry , siberian branch of the russian academy of sciences . his research areas include quantum electronics , fiber and nonlinear optics , high - precision and tunable laser systems . he is the author of more than 200 publications in scientific journals , conference proceedings , and books . * vasily m. entin * is a senior researcher at the rzhanov institute of semiconductor physics , siberian branch of russian academy of sciences . he received his ph.d . from the institute of automation and electrometry , siberian branch of the russian academy of sciences in 2006 . his major research interests are laser cooling and laser spectroscopy . he is the author of more than 30 publications in scientific journals . * denis b. tretyakov * is a researcher at the rzhanov institute of semiconductor physics , siberian branch of russian academy of sciences . he received his ph.d . from the institute of automation and electrometry , siberian branch of the russian academy of sciences in 2008 . his major research interests are quantum information with rydberg atoms and long - range atomic interactions . he is the author of more than 30 publications in scientific journals . * vladimir baraulya * graduated in 1982 from novosibirsk state university in quantum optics . his research activity was focused upon such areas of high importance as systems for laser cooling of atoms and molecules , frequency doublers , laser physics . * igor i. ryabtsev * is a head of laboratory at rzhanov institute of semiconductor physics , siberian branch of russian academy of sciences and lecturer at novosibirsk state university . he received ph.d . ( 1992 ) and doctor of sciences ( 2005 ) degrees from institute of automation and electrometry , siberian branch of the russian academy of sciences in 2005 . his major research interests are quantum information with rydberg atoms and long - range atomic interactions . he is the author of more than 90 publications in scientific journals .
we have implemented a simple digital system for long - term frequency stabilisation and locking to an arbitrary wavelength of the single - frequency ring cw ti : sapphire laser . this system is built using two confocal fabry - prot cavities , one of which is used to narrow short - term linewidth of the laser and the other to improve long - term stability of the laser frequency . the length of the second cavity is stabilised using the radiation from an external - cavity diode laser locked to an atomic transition . our system is an improvement of a commercial tekhnoscan laser lock . this system has been successfully used in our experiments on high - resolution laser spectroscopy of ultracold rubidium rydberg atoms . : ilya i. beterov , rzhanov institute of semiconductor physics sb ras , prosp . lavrentyeva , 13 630090 novosibirsk , russia ; tel : + 7 383 - 333 - 24 - 08 ; fax : + 7 383 - 333 - 27 - 71 ; e - mail : 2
prions are self - propagating , usually amyloid - like protein aggregates@xcite that are responsible for transmissible diseases . examples of prion diseases are scrapie in sheep@xcite , bovine spongiform encephalopathy@xcite(bse ) in cattle and new variant creutzfeldt - jakob disease@xcite in man . in contrast to these disease - related cases , several proteins displaying prion properties are well known in yeast and fungi@xcite . the het - s prion protein produced by the filamentous fungus _ podospora anserina _ is thought to be involved in a specific function . the switching on of its prion form triggers the programmed - cell - death phenomenon named `` heterokaryon incompatibility''@xcite which can prevent different forms of parasitism , by inducing the death of the heterokaryon formed by cell fusion of different fungal strains . the proteinase k - resistant core of the prion fibrils formed by the c - terminal residues 218 - 289 ( pfd : prion forming domain ) is unstructured in solution and forms infectious fibrils in vitro@xcite . earlier work showed that het - s pfd fibrils consist of four @xmath0-strands forming two windings of a @xmath0-solenoid@xcite however without giving further information about the details of the intra- and inter - molecular @xmath0-sheet architecture . recently , the structure of het - s pfd has been determined@xcite on the basis of nmr - derived intra and intermolecular distance restraints@xcite . this is the only atomic - resolution structural model of an infectious fibrillar state reported to date . on the basis of 134 intramolecular and intermolecular experimental distance restraints , het - s pfd forms a left - handed @xmath0-solenoid , with each molecule forming two helical windings , a compact hydrophobic core , at least 23 hydrogen bonds , three salt bridges and two asparagine ladders ( see fig . 1a and 1b ) . the model is supported by electronic diffraction and scanning transmission electronic microscopy@xcite . despite the tremendous economical and social relevance of diseases related to prion infections and protein aggregation and the enormous quantity of research devoted to their study , the mechanisms that rule and trigger fibril formation and elongation remain mostly elusive : the nature of protein aggregation , its limited structural order , its insolubility in water and its involvement with the cell membrane render its experimental study extremely difficult . for these reasons relevant breakthroughs in understanding the principles of amyloid formation and fibril growth might come from numerical simulations . recent advances in hardware and methodology have allowed for realistic atomic resolution molecular dynamics ( md ) simulations with physics - based potentials of small fibrils consisting of a few monomer units that make possible to span time scales of hundreds of nanoseconds @xcite and even of @xmath1 for smaller systems @xcite . these results were mainly obtained by studying the amyloid @xmath0-peptide ( a@xmath0 ) , related to the alzheimer s disease@xcite , or its mutants and fragments . in this context , experimental observations had first originated the proposal that incoming a@xmath0 monomers associate to the elongating fibril through a two stages `` dock and lock '' kinetic mechanism @xcite . in the first stage , an unstructured monomer docks onto the fibril while maintaining some degree of conformational freedom . in the second stage , the monomer locks into the final fibril state . such a mechanism was later confirmed and refined , also in the context of a@xmath0-peptide oligomerization , by other experimental and simulation studies @xcite . the dock and lock mechanism was further employed , for the a@xmath0-peptide , to describe fibril elongation within a thermodynamic context , by means of all - atom md simulations @xcite . a continuous docking stage was observed to occur over a wide temperature range without free energy barriers or intermediates , whereas the locking stage at lower temperatures was characterized by many competing free energy minima in a rugged landscape . in the case of the fibrillar het - s pfd protein , it was possible with md atomistic simulation to probe the stability of the nmr structure on a 10 ns timescale and to predict the behaviour of the salt bridge network@xcite . on the other hand , typical elongation times for amyloid fibrils formed by het - s pfd are of the order of hours@xcite , so that coarse grained approaches , in which protein chains and amino - acid interactions are modeled in a simplified way , are mandatory to investigate such longer time scales . indeed , despite the difficulty of finding reliable energy functions , these approaches has been successfully used in outlining general aspects of the full phase diagram of a generic aggregating polypeptide system@xcite , to emphasize the importance of the contribution of hydrophobic interactions and hydrogen bonding in the aggregation process of the a@xmath0 peptide peptide@xcite and even to study the mechanisms of monomer addition for the a@xmath0 peptide and some of its mutants@xcite . in this paper , in order to describe the fibril elongation mechanisms of a relatively long protein domain such as het - s pfd , we prefer to employ a still different coarse - grained approach since in this case there is the unique advantage of knowing from experiments the fibril structure . at a general level , our strategy falls in the class of approaches used in protein folding that builds on the importance of the native state topology in steering the folding process @xcite . in its simplest example , the formation of contacts is favoured only for pairs of amino acids that are found interacting in the native state , but non - native sequence dependent interactions could be introduced as well . despite the semplicity of this scheme , in the past few years , an increasing number of experimental and theoretical studies have confirmed the utility of the go - like approach in the characterization of various aspect of protein folding and binding processes@xcite . the study of protein aggregation has been already tackled by using go - like models@xcite , but due to the absence of experimental information on the fibrillar structure , hypothetical aggregate conformations had to be introduced to build the go - energy function driving aggregation . moreover , topologically based models , with reduced effect of non - local interactions , correspond to funneled energy landscapes and therefore their application should be limited to situations in which proteins are evolutionarily designed to follow the principle of minimal frustration@xcite , which results in a faster search through the many possible alternatives . in general , a funneled energy landscape is not expected in the case of non - functional fibril formation . therefore , the accurate knowledge of its structure , the complex intra - chain topology , and the plausible involvement in a functional process makes het - s pfd a suitable and , at the moment , unique candidate to exploit successfully go - like models for studying amyloid formation and fibril elongation mechanisms . within this approach , implemented through a monte - carlo procedure combined with replica exchange methods , we analyze the full thermodynamic properties of the fibril elongation process , e.g. of the association of a free chain to the already formed fibrillar structure of het - s pfd under different concentration conditions . the behaviour of both energies and heat capacities shows that the association process becomes more cooperative for concentrations in the range ( @xmath2 ) of standard in - vitro experiment@xcite . a careful study of the association process shows that fibril elongation is triggered by the docking of the free chain onto the fibril in a concentration dependent mechanism that involves the formation of both inter- and intra - chain hydrogen bonds stabilizing the longest @xmath0-strands , rapidly followed by the assembly of the full domain . this behaviour is similar to the `` dock and lock '' mechanism proposed for the amyloid a-@xmath0 fibril formation @xcite . another interesting aspect emerges clearly : elongation proceeds differently according to which side of the fibril ( see fig 1a and 1b ) is used as the growing end . elongation from one side is clearly favoured with respect to the other side , implying a strong polarity in the growth of het - s pfd fibrils , which may even lead to unidirectional elongation . polarity in fibril growth is a feature already discussed in the literature for other amyloid - forming proteins , in both experiments@xcite and numerical simulations@xcite . our data suggest that growth polarity can be explained for het - s pfd on the basis of the complex topological properties of its fibrillar structure . a key role is played by the behaviour of one long loop connecting two @xmath0-strands in consecutive rungs of the fibrillar structure . depending on the elongation side , this loop may help or not the winding of the c - terminal part of the attaching chain into the fibrillar form . since it is known that the prion loses its infectivity upon partial deletion of that loop@xcite , we argue that this topological mechanism may be important for functional fibril growth . chains a and b were selected from the nmr structure of het - s pfd ( pdb code : 2rnm ) , where chain b stays on top of chain a along the fibril . we keep one chain frozen whereas the other one is free . in the top elongation mode , chain b is free and chain a is frozen , whereas in the bottom elongation mode , chain a is free and chain b is frozen . the free chain is allowed to move in a hemisphere of radius @xmath3 defined by @xmath4 , @xmath5 ( @xmath6 ) for top ( bottom ) elongation . the frozen chain is placed with center - of - mass ( i.e. average c@xmath7 ) coordinates @xmath8 ( i.e. on the hemisphere axis ) and @xmath9 ( @xmath10 ) for top ( bottom ) elongation , rotated in such a way that the fibril axis is parallel to the hemisphere axis . the center - of - mass position along the hemisphere axis is chosen in order to expose only one ` sticky ' end of the frozen chain to the free chain , by placing the other end roughly on the hemisphere base @xmath11 . this allows a smaller computational effort , at the expense of prohibiting conformations that we do not expect to affect in a relevant way the binding of the free chain on the exposed end of the full fibril , here represented by the frozen chain . the portion of het chain that we simulate goes from position 217 to position 289 , which includes 73 aminoacids ( e.g. in the simulation we include also the met aminoacids at position 217 used to obtain the nmr structure ) . we thus simulate a system with 146 aminoacids . in order to perform extensive simulations , we use a coarse - grained representation of the protein chain coupled with an energy function based on the knowledge of the pdb fibril structure . in the spirit of go - like approaches widely used for globular proteins @xcite , the energy function is built in such a way to have its minimum for the pdb structure . inspired by ref . @xcite , each aminoacid is represented by an effective spherical atom located at the position of the corresponding c@xmath7 atom . the virtual bond angle associated with three consecutive @xmath12 atoms is constrained between @xmath13 and @xmath14 . virtual bond lengths are kept constant and equal to the native values from the pdb file . to implement steric constraints we require that no two non - adjacent @xmath15 atoms are allowed to be at a distance closer than @xmath16 . we assign an energy @xmath17 to each hydrogen bond that can be formed between two residues that form it in the pdb fibril structure , and we disregard any other attractive interaction ( i.e. hydrogen bonds can not be formed by two residues that do not form it in the pdb fibril structure and we do not consider any other type of pairwise interaction except for the excluded volume constraints ) . only @xmath0-sheet stabilizing hydrogen bonds can therefore be formed in our simulation , and in order to identify them within a @xmath12-representation , we use the geometrical rules for non - local hydrogen bonds introduced in ref . @xcite . in order to recognize the hydrogen bonds present in the pdb fibril structures the lower bound on the scalar product between binormal and connecting vectors was decreased to @xmath18 ( @xmath19 in the original formulation , see table 1 and fig . 1 in ref . @xcite for the precise listing of all hydrogen bond rules and the definition of binormal and connecting vectors ) . in this way , we find in the pdb fibril structure ( see fig . 2 and fig . 3 ) two long parallel @xmath0-strands ( @xmath20 and @xmath21 ) , connected by 9 hydrogen bonds and four shorter strands : @xmath22 parallel to @xmath23 ( linked together by 4 hydrogen bonds ) and @xmath24 parallel to @xmath25 ( 2 hydrogen bonds ) . those strands alternate within the fibrillar structure in pairs which form hydrogen bonds within the same chain ( intra - chain bonds ) and pairs which form hydrogen bonds between neighbouring chains ( inter - chain bonds ) . each strand in the core of the fibrillar structure forms intra - chain bonds on one side and inter - chain bonds on the other side . in the `` top '' side of the fibril ( see fig . 2 and fig . 3 ) , the exposed strands in the ` sticky ' end are @xmath21 , @xmath23 @xmath25 , whereas in the `` bottom '' side the exposed strands are @xmath20 , @xmath22 @xmath24 . since we keep fixed one chain , the ground state has energy -30 ( 15 intra - chain hydrogen bonds plus 15 inter - chain hydrogen bonds - we are not counting the intra - chain bonds of the frozen chain ) . hence , energy can range from @xmath26 ( unbound chains ) to @xmath27 ( fully bound chains in the fibrillar state ) . in order to fix a realistic temperature scale , the effective value of hydrogen bond energy in our go - like energy function was given the value @xmath28 , so that the unit temperature of our simulations corresponds to @xmath29 k. we have simulated the elongation of the fibril from both sides . for each side we simulated chains confined in a hemisphere centred in the origin and of radius @xmath30 , @xmath31 , and @xmath32 corresponding to concentrations @xmath33 , @xmath34 and @xmath35 . the latter value is close to typical concentrations used in vitro aggregation experiment @xcite . fibril elongation is simulated by means of a monte carlo procedure . multiple markov processes @xcite , each replicating the same system of a fixed chain and a free chain attaching to it described above , are run in parallel at different temperatures , with possible swaps of replicas , in order to sample efficiently the conformational space from high ( @xmath36 @xmath37c ) to low temperatures ( @xmath38 @xmath37c ) . within the same replica , conformations are evolved using trial moves , which either pivot a part of the chain from a randomly chosen residue to its end , or rotate a chain portion between two residues along the direction joining them . in the latter case the two residues are either chosen randomly or chosen to be next - nearest neighbours along the chain . trial moves are accepted or rejected according to the metropolis test @xcite . we use 20 different replicas , chosing their temperatures to sample more accurately the transition region ( @xmath39 @xmath37c @xmath40 @xmath37c ) and to provide reasonable overlap of energy histograms between neighbouring pairs . roughly @xmath41 monte carlo steps ( @xmath42 is the number of residues of the free chain ) are performed independently for each replica before one replica swap is attempted among a randomly chosen neighbouring pair . overall , roughly @xmath43 replica swaps are attempted for each simulation and the acceptance rate of replica swaps is in all cases above @xmath44% . the convergence to the equilibrium regime is assessed by looking at the evolution of system energy as a function of monte carlo steps . in order to compute equilibrium thermodynamic averages , the simulation portion corresponding to the initial @xmath45 swaps is discarded from the collected data , with @xmath45 ranging from @xmath46 to @xmath47 , depending on concentration . data from all temperatures are elaborated with the multiple - histogram method @xcite . in fig . 4 some snapshots of the simulations are represented . to characterize the thermodynamic properties of the het - s fibril growth process we study the behaviour of the energy and of the heat capacity of the system as a function of the temperature for three different protein concentrations ( see fig . 5 ) and by considering elongation both from the top and from the bottom side ( see methods ) . the peaks in the heat capacity curve signal the occurrence of large conformational rearrangements ( that strictly speaking would become transitions only in the thermodynamic limit of the number of system component going to infinite ) related to the process of the deposition of the free het - s pfd chain to the one that is kept fixed to represent the sticky end of the already formed fibril . as expected , the first association temperature , related to the high temperature heat capacity peak , decreases when lowering the concentration . it reaches values close to room temperature for the concentration @xmath48 . interestingly , the cooperativity of the transition increases for lower concentrations : this is signalled by a more sigmoidal behaviour of the energy profiles , by the sharpening of heat capacity peaks and by the almost complete merging in one single narrow peak of the otherwise complex peak structure . despite at a first glance the growth processes from the two different sides look similar , one can notice that at low concentration a residual peak remains at low temperature for the top elongation case . to elucidate further these behaviours and to understand the nature of the conformational rearrangements related to heat capacity peaks we computed the formation probability of hydrogen bonds for the different strands as a function of temperature ( shown in fig . we define the stabilization temperature of a group of hydrogen bonds to correspond to their average formation probability at thermodynamic equilibrium being equal to 0.5 . we compute stabilization temperatures for 6 possible groups of hydrogen bonds , corresponding to the different strand pairings shown as separate white / black bands in fig . the resulting stabilization temperatures are summarized in table [ t1 ] . we will use the term ` fibrillar ' for inter - chain hydrogen bonds . as an example , strand @xmath20 of the mobile chain couples with strand @xmath21 of the fixed chain , in the case of elongation from the top side ; whereas strand @xmath21 of the mobile chain couples with strand @xmath20 of the fixed chain in the case of elongation from the bottom side , and similarly for other strand pairs ( see fig . 6 caption for a detailed explanation of how intra - chain and fibrillar hydrogen bonds are represented in the figure ) . for all concentrations and for both elongation sides , the first hydrogen bonds which become stable are the inter - chain ones formed between the long strands @xmath20 and @xmath21 , at a temperature @xmath49 varying from @xmath50 @xmath37c for top elongation at concentration @xmath51 , to @xmath52 @xmath37c for bottom elongation at @xmath53 ( see table [ t1 ] ) . the stabilization temperature @xmath49 of fibrillar hydrogen bonds between the long strands is higher for higher concentrations and , at the same concentration , for the case of top side elongation . this first stabilization process is followed at a lower temperature @xmath54 ( varying from @xmath55 @xmath37c for top elongation at concentration @xmath51 , to @xmath56 @xmath37c for bottom elongation at @xmath53 , see table [ t1 ] ) by the stabilization of intra - chain hydrogen bonds formed between the two long strands , @xmath20 and @xmath21 , of the mobile chain . this second stabilization temperature is again higher for the top elongation case , at the same concentration , and increases with concentration . with a further decrease of the temperature the fibrillar hydrogen bonds formed between the short strands ( @xmath22 with @xmath23 and @xmath24 with @xmath25 ) become stable . the stabilization temperature @xmath57 for inter - chain strand pair @xmath22 and @xmath23 does not depend on concentration but only on the elongation side , being higher , @xmath58 @xmath37c , for top elongation with respect to @xmath59 @xmath37c for bottom elongation . the stabilization temperature @xmath60 for inter - chain strand pair @xmath24 and @xmath25 is roughly the same , @xmath61 @xmath37c , in all cases . the last step involves the stabilization of the intra - chain hydrogen bonds between the short strands and do not depend as well on concentration . at this stage the more significant difference between the two elongation sides emerges . for top side elongation , the two short strand pairs are stabilized roughly together at a temperature much lower with respect to the previous step ( @xmath62 @xmath37c corresponding to the small peak in the heat capacity curve and @xmath63 @xmath37c ) . for bottom side elongation , the stabilization of the two short intra - chain strand pairs takes place at quite different temperatures : hydrogen bonds between strands @xmath22 and @xmath23 are stabilized at a even slightly higher temperature than their inter - chain counterpart ( @xmath64 @xmath37c ) , whereas the intra - chain strand pair @xmath24 and @xmath25 is stabilized at a much lower temperature ( @xmath65 @xmath37c ) . upon assuming that the order in hydrogen bond stabilization mirrors a similar order in the kinetics of the elongation process , we can extract a general message from these results by stating that the attaching of a mobile chain onto the elongating fibril is triggered by the formation of the inter - chain hydrogen bonds of the longest strand ( @xmath20 or @xmath21 ) followed by a first partial folding of the chain through the formation of the long intra - chain strand pair ( between @xmath20 and @xmath21 ) and by the successive formation of all other inter- and intra - chain hydrogen bonds . the mechanism of addition of a soluble unstructured monomer to a preformed ordered amyloid fibril is a complex process : the deposition involves an association of the unstructured monomer to the fibril surface ( docking ) followed by a conformational rearrangement leading to the incorporation onto the underlying fibril lattice ( locking ) @xcite . we identify the docking stage with the formation of both inter- and intra - chain hydrogen bonds between the long strands @xmath20 and @xmath21 , as in both cases the stabilization temperatures @xmath49 and @xmath54 depends on concentration ( see table [ t1 ] ) . one indeed expects that in a denser regime it is easier for the mobile chain to dock onto the fibril end , while the locking into the @xmath0-helix structure necessary for fibril propagation should not be affected by concentration changes . therefore , the locking stage involves the formation of both inter- and intra - chain hydrogen bonds between the remaining shorter strands , since we observe that their stabilization temperatures do not depend on concentration . the dependence of the intra - chain hydrogen bond stabilization temperature @xmath54 upon concentration is non - trivial and is triggered by the strong concentration dependence of the formation probability of the fibrillar hydrogen bonds between the long strands . the higher the concentration , the more probable the fibrillar bonds to be formed . consequently , the more easily the intra - chain bonds are stabilized . from the data shown in fig . 6 and in table [ t1 ] , another general feature can be picked out : the temperature range in which the docking stage and thus the full elongation process take place decreases at lower concentrations . this is consistent with the increment of cooperativity as it appears from thermodynamic quantities ( e.g. sharpness of heat capacity peaks in fig . 5 ) when concentration diminishes . the most cooperative behavior , as shown by the presence of a single sharp peak in the heat capacity curve , is obtained at concentration @xmath53 in the case of bottom side elongation , whereas a second peak is clearly visible at low temperature for top side elongation at the same concentration . the above analysis of fig . 6 data reveals the crucial role of the formation of intra - chain hydrogen bonds between strands @xmath22 and @xmath23 in this respect . the stabilization temperature @xmath66 of these hydrogen bonds constitutes the most relevant difference between the two elongation modes in the first place . moreover , it does correspond closely in both cases to significant features in the heat capacity curve , namely the small peak at @xmath67c for top elongation and the small shoulder in the main peak at @xmath68c for bottom elongation . in order to gain a further understanding of the role played by intra - chain @xmath22-@xmath23 hydrogen bonds we computed for different temperatures the equilibrium occupation probabilities of macroscopic conformational states , which are defined according to the number of formed intra- or inter - chain hydrogen bonds . in fig . 7 and fig . 8 the results are shown for concentration @xmath53 and for the two different growth modes . occupation probabilities are shown in logarithmic scale , so that the resulting data could be interpreted as ( the opposite of ) effective free energies or mean force potentials . at high temperature , the bottom left corner is mostly populated , corresponding to conformations with very few or none intra- and inter - chain hydrogen bonds formed , typical of a mobile chain not yet attached to the fibril end . on the other hand , at very low temperature , the opposite top right corner is populated , describing structures with almost all the hydrogen bonds formed that correspond to mobile chains found already completely associated to the fibril end with a significant probability . consistently with the previous analysis , based on fig 6 data , the elongation process is complex , taking place in several stages . macrostates with only intra - chain hydrogen bonds are found to be populated to some extent at high temperatures , hinting to the possibility of a pre - structuring of the mobile chain before docking to the fibril tip , yet the pathway more significantly visited involves the population of first the 9 fibrillar hydrogen bonds between strand @xmath20 and @xmath21 and then the analogous intra - chain bonds ( see @xmath69c in fig . 7 and fig . 8) . after this first stage , that we already identified with docking , two different scenarios emerge again clearly , depending on the growth mode . the overall process is visibly more cooperative for bottom side elongation ( fig . 8) with respect to top side elongation ( fig . 7 ) , since the spreading of significantly visited macrostates is restricted to a narrower temperature range in the former case . moreover , around room temperature ( @xmath70c ) , the most populated state for bottom side elongation has 15 fibrillar and 13 intra - chain hydrogen bonds formed corresponding to a chain almost completely attached to the fibril end ( only strand @xmath24 is left wiggling a bit ) . instead , the most populated state for top side elongation has all 15 inter- but only 9 intra - chain hydrogen bonds locked into the fibrillar conformation , signalling again that the stability of intra - chain @xmath22-@xmath23 hydrogen bonds is strongly weakened with respect to bottom side elongation . in order to appreciate more easily the variations with temperature in the population of the different macrostates , we computed unidimensional free energy profiles as a function of the number of either intra- or inter - chain hydrogen bonds . the results are shown in fig . 9 for concentration @xmath53 and for the two different growth modes . fig . 9 pictures confirm the multistage nature of the association process with different macrostates that become the global free energy minimum at different temperatures . the free energy profiles as a function of the number of fibrillar hydrogen bonds are similar in both elongation modes . the main difference is the value of the temperature below which the free state ( none fibrillar hydrogen bond is formed ) is not the free energy minimum anymore : @xmath71c for top elongation and @xmath72c for bottom elongation , consistently with table [ t1 ] . the free energy of the free state and the free energy barrier that separates it from the competing minimum with @xmath73 fibrillar bonds ( the ones formed between the @xmath0-strands @xmath20 and @xmath21 that are the first to be stabilized in the association process ) do not basically depend on temperature . this is a signature of their entropic origin ( free energies in fig . 9 are plotted in @xmath74 units ) , as they are both dominated by the roto - translational entropy of the free chain . on the other hand , the free energy profiles as a function of the number of intra - chain hydrogen bonds display a relevant difference between the two elongation modes , consistently with previous observations . for top elongation , the free energy minimum at @xmath75c is the macrostate with only @xmath73 intra - chain hydrogen bonds ( i.e. the @xmath0-strands @xmath22 and @xmath23 are not yet paired ) , whereas for bottom elongation the free energy minimum at @xmath75c is the macrostate with @xmath76 intra - chain hydrogen bonds ( i.e. the @xmath0-strands @xmath22 and @xmath23 are already paired ) . interestingly , the above difference is due to the macrostate with @xmath73 intra - chain hydrogen bonds being entropically stabilized for top elongation with respect to bottom elongation . indeed , the free energies for @xmath77 intra - chain hydrogen bonds do not differ in the two elongation modes . moreover , the free energy difference between the two elongation modes for @xmath73 intra - chain hydrogen bonds has to be entropic , since the energy of that macrostate is the same in both cases , being given by the @xmath73 intra - chain plus the @xmath78 inter - chain hydrogen bonds ( the latter is the global free energy minimum at @xmath79c for both elongation modes ) . our analysis clearly established that fibril growth exhibits a deeply different thermodynamic behaviour depending on the side from which elongation proceeds : at room temperature only bottom side elongation is thermodynamically stable . what is the physical reason for the existence of such a strong growth polarity ? since our simulation study is based only on the knowledge of the fibril structure and not on the specificity of the amino - acid sequence , we can expect that growth polarity is a consequence of the topological properties of the structure . there is indeed a deep topological difference between the deposition mechanisms of het - s pfd on the two different sides of the fibril . in the first docking stage , common to both elongation modes , the formation of both inter- and intra - chain hydrogen bonds between @xmath20 and @xmath21 implies the positioning of the latter strands into a conformation already compatible with the final fibrillar structure . the remaining strands , then , yet to be positioned , acquire distinct topological roles , since @xmath22 and @xmath24 are in a loop between the two chain portions already pinned down in the fibrillar structure , whereas @xmath23 and @xmath25 are in the c - terminal tail of the chain . one can then predict that , for entropic reasons , the former pair can be accomodated into the final fibrillar structure more easily than the latter pair . nevertheless , different elongation modes may change this hierarchy . when elongation proceeds from the top side , the attaching chain wraps up onto the fibril tip starting with its n - terminal part ( see fig . the ` loop ' strands @xmath22 and @xmath24 ( fibrillar ) need to be positioned before the ` tail ' ones @xmath23 and @xmath25 ( intra - chain ) , consistently with the topological order suggested above . indeed we do observe this hierarchy for top side elongation ; even at low temperature intra - chain hydrogen bonds between the short strand pairs are not yet stable . when elongation proceeds from the bottom side , the attaching chain wraps up onto the fibril tip starting with the c - terminal part ( see fig . the ` tail ' strands @xmath23 and @xmath25 ( fibrillar ) need now to be positioned before the ` loop ' ones @xmath22 and @xmath24 ( intra - chain ) . elongation order takes over the topological order so that the difficult positioning of ` tail ' strands is assisted by the easier positioning of ` loop ' strands and both are stabilized at room temperature ( with the partial exception of the shortest strand @xmath24 ) . in this work we have used monte - carlo simulations of a coarse - grained representation of het - s pfd domain in order to get information about the elongation of the corresponding amyloid fibril by attaching of a mobile chain to a pre - fixed fibrillar structure . our approach , based on the knowledge of the fibrillar structure , relies on the currently well accepted assumption that protein topology plays a pivotal factor in determining unimolecular folding and protein assembly . at variance with other go - like studies @xcite , based on a hypothetical structure , the reliability of our study is justified by the knowledge of a high - resolution nmr structure of a plausibly functional amyloid fibril . there are two main results of our thermodynamic study . first , we observe that fibril elongation is driven by the formation of inter - chain and intra - chain hydrogen bonds between the long strands @xmath20 and @xmath21 , followed by the positioning of the rest of the attaching chain onto the growing fibril . this mechanism is known as _ dock and lock mechanism _ @xcite . we identify the docking stage as that part of the association process whose onset temperature displays a concentration dependance . a similar feature , i.e. the docking temperature range varies with concentration , was previously found in a thermodynamic study of a@xmath0-peptide fibril growth @xcite . within our go - like approach , we find a complex multistage association process , where both the docking and the locking stages proceed in several steps in a free energy landscape characterized by several different minima separated by barriers . in the case of a@xmath0-peptide fibril elongation , it was found on the contrary that docking is continuous and proceeds without the presence of intermediate or free energy barriers @xcite . our finding of a multistage docking behaviour might be due to the non - trivial intra - chain topology of het - s pfd monomer , lacking in the a@xmath0-peptide case . alternatively , it might be an artifact caused by our neglecting of non - native interactions . secondly , we predict that one side of the structure is more suitable to substain the growth of the fibril . the predicted fibril growth polarity can be rationalized by analyzing elongation topological properties , which turn out to be intrinsically different from the two fibril sides . the argument is based on the complex tertiary structure of the monomeric unit within het - s fibril , resulting into alternating intra- and inter - chain pairs of @xmath0-strands . after the first hydrogen bonds have been formed in the initial docking onto the growing fibril , the portion of the attaching chain which is going to acquire @xmath0-structure in the following locking stage , is partitioned into a ` loop ' and a ` tail ' part ( see fig . 10 ) . as a consequence , the entropy loss of the two parts upon locking is different , implying a ` topological ' hierarchy . the latter may or may not be affected by the ` winding ' hierarchy dictated by the choice of the growth side , thus resulting in a strong growth polarity . bottom side elongation is more sustainable because the ` loop ' part may assist the ` tail ' part to lock . the entropic origin of the growth polarity observed in our results is further confirmed by the free energy profiles shown in fig . 9 . as discussed in section [ entr ] , the macrostate populated after docking and before locking ( with @xmath78 inter - chain and @xmath73 intra - chain hydrogen bonds ) is entropically stabilized in the top elongation mode with respect to bottom elongation . being based on a topological argument , we believe our prediction to be robust against both variations of the details in the implementation of our go - like approach ( changing the coarse - graining of protein chain representation , employing in the energy function general pairwise contacts and not only hydrogen bonds , using different hydrogen bond rules ) and relaxation of other simplifying assumptions that we made , namely the fibril tip is represented by just one chain and kept completely frozen . this latter point is further motivated by the experimental observation @xcite that the het - s fibril accomodates incoming prion monomers without a substantial disorganization of its structure . this behaviour is quite different from those of a@xmath0 fibrils where it was experimentally observed @xcite an entropy gain in the elongation reaction which was related to an unfolding of the organized fibril end to accomodate the addition of the incoming monomers . we thus suggest our prediction may be accessible to experimental validation , for instance using the single fiber growth assay employed in @xcite for sup35 yeast prion , based on two variants of the prion domain that can be differentially labelled and distinguished by atomic force microscopy . the same topological argument can not be used for simpler amyloid structure such as the solid state nmr - model suggested for the alzheimer s @xmath80-peptide @xcite , in the absence of intra - chain hydrogen bonds . in fact , experimental evidence shows bidirectional fibril growth with no clear signs of growth polarity@xcite . interestingly , the presence of asymmetrical topological properties of fibril ends was indeed suggested for the @xmath80-peptide @xcite , depending on the different possible quaternary arrangements of the fibril @xcite . on the other hand , a clear evidence of growth polarity was demonstrated for sup35 fibrils @xcite , a yeast prion protein believed to have a functional role , similarly to het - s prion . there is no high - resolution information on the structure of sup35 fibrils and different conflicting structural models have been recently proposed @xcite , yet some of the available data suggests the existence of a complex intra - chain structure @xcite , which could justify the observed growth polarity within a framework similar to the one we propose here for het - s . a peculiar feature of het - s pfd is the presence of a long loop in the fibrillar structure that connects strands @xmath24 and @xmath21 ( aa 246 - 260 ) , which then contributes to increasing the length of the chain portion partitioned as ` loop ' in the topological argument discussed above ( in blue in fig . an interesting question is whether increasing such length favours or disfavours growth polarity according to the mechanism suggested here . one could argue that a longer loop may further decrease the fluctuations available to the ` tail ' part , thus enhancing its assisted locking for bottom side elongation . indeed , it is known that a large deletion in such loop ( 248 - 256 ) led to loss of het - s function and infectivity@xcite . even though no evidence is known about the possible biological relevance of fibril growth polarity in fungal and yeast prions , it is tempting to speculate that the role of the loop might be to enhance growth polarity as a way to control the elongation process more thoroughly in the context of the propagation of a functional prion . numerical simulations performed within the same go - like approach , based on a model structure to be built on the basis of het - s pfd but with a shorter loop , will shed further light on this hypothesis . our coarse - grained approach turned out to be effective in studying structural rearrangements which could not be tackled using more detailed protein chain representations . on the other hand , we are neglecting factors which may play an important role in the elongation process , such as the possible presence of oligomeric conformers of het - s pfd in the non - fibrillar soluble state @xcite . in order to capture similar effects , one needs to increase the accuracy of the energy function by introducing amino - acid specificity . this can be done by using coarse - grained potentials which take into account the different ability of each pair of amino - acids in forming hydrogen bonds in @xmath0-strands @xcite , coupled with similar potentials describing residue pairwise interactions @xcite or local conformational biases @xcite . such potentials might be used to modulate the go - like energy function presented in this work . the use of a more realistic description of the protein chain involving side chain atoms may also cause the amino - acid sequence to affect the elongation process in a side - depending manner by imposing chiral stereochemical constraints . the latter are not present in our c@xmath7-based representation , thus reinforcing the topological origin of growth polarity in our results . finally , the approach presented here could be also used to study the nucleation process of het - s pfd , and further extended to study the aggregation of the full het - s protein . the het - s n - terminal domain in the non - fibrillar soluble form is structured into a globular protein fold whose high - resolution structure has been very recently released@xcite ( pdb code 2wvn.pdb ) . such structure is known to lose partially its order upon ordering and aggregation of het - s pfd into the unsoluble fibrillar form @xcite . a go - like approach would be especially suited to study the resulting competition between the ordered structures of the two domains in the two different forms @xcite . we thank f. chiti and r. riek for enlightening discussions . we acknowledge financial support from university of padua through progetto di ateneo n. cpda083702 . 73 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , * * , ( ) . , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , , , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , * * , , , , , , * * , ( ) . , * * , ( ) . , , , , , , , , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , , * * , ( ) . , , , * * , ( ) . @xmath81 & @xmath82 & @xmath83 & @xmath84 & @xmath85 + + @xmath51 & 115 & 85 & 64 & 43 & 12 & -1 + @xmath86 & 88 & 71 & 63 & 44 & 11 & -2 + @xmath53 & 75 & 70 & 62 & 43 & 8 & -3 + + @xmath51 & 108 & 80 & 50 & 42 & 45 & 6 + @xmath86 & 85 & 68 & 47 & 40 & 43 & 7 + @xmath53 & 65 & 63 & 50 & 41 & 44 & 6 +
the prion - forming c - terminal domain of the fungal prion het - s forms infectious amyloid fibrils at physiological ph . the conformational switch from the non - prion soluble form to the prion fibrillar form is believed to have a functional role , since het - s in its prion form participates in a recognition process of different fungal strains . based on the knowledge of the high - resolution structure of het - s(218 - 289 ) ( the prion forming - domain ) in its fibrillar form , we here present a numerical simulation of the fibril growth process which emphasizes the role of the topological properties of the fibrillar structure . an accurate thermodynamic analysis of the way an intervening het - s chain is recruited to the tip of the growing fibril suggests that elongation proceeds through a dock and lock mechanism . first , the chain docks onto the fibril by forming the longest @xmath0-strands . then , the re - arrangement in the fibrillar form of all the rest of molecule takes place . interestingly , we predict also that one side of the het - s fibril is more suitable for substaining its growth with respect to the other . the resulting strong polarity of fibril growth is a consequence of the complex topology of het - s fibrillar structure , since the central loop of the intervening chain plays a crucially different role in favouring or not the attachment of the c - terminus tail to the fibril , depending on the growth side .
it has become clear in recent years that galaxy colors show a bimodal distribution ( strateva et al . 2001 , hogg et al . 2002 , blanton et al . 2003 ) : in a rest - frame color - magnitude diagram galaxies tend to segregate between a `` red sequence '' ( similar to , but less tight than that observed for cluster galaxies ) and a `` blue cloud '' . the red sequence is mostly composed by spheroidal galaxies , a class of objects of special interest as it includes the most massive galaxies , adding up to represent about half the stellar mass in the universe ( bell et al . 2004a , b ; baldry et al . 2004 , hogg et al . this behaviour seems to be already established at fairly high redshifts ( bell et al . 2004a ; giallongo et al . 2005 ; cucciati et al . 2006 ; franzetti et al . the sdss data ( hogg et al . 2004 ) show that the blue cloud contains mainly late ( spiral ) morphological types , while the bulk of the red sequence consists of early - type galaxies . hogg et al . ( 2004 ) find also that galaxies with high srsic indices ( indicating an early - type morphology ) dominate the red sequence , almost independently of local density . balogh et al . ( 2004 ) , using a color criterion to describe galaxy populations , also find that the colors of the red and the blue populations depend weakly on the environment . a dependence of galaxy colors on the environment would seem to be a natural expectation of a hierarchical scenario : the most dense regions are also those that collapse first , and therefore those appearing as most evolved at any epoch . if the baryonic component simply follows the evolution of dark matter haloes , high density regions should contain galaxies that are older and redder than the low - density field population . how these properties of the stellar population are inter - laced with the construction of the morphological type and its relation to the environment is not obvious , given the impossibility of semi - analytical models to predict galaxy morphologies , a part from adopting simple recipes . thus , attempts to explain the observed bi - modality in the rest - frame color - magnitude distribution seem to be succesfull in reproducing the gross color features of the population ( menci et al . 2005 ; dekel et al . 2005 ) , but obviously do not tell us much on the specific morphological differentiation . investigation of the actual morphological composition of the red sequence at high redshift has been performed so far by bell et al . ( 2004b ) , using the gems survey , and by weiner et al . ( 2005 ) , using the smaller deep-1 survey . both works consistently show that the red sequence also contains galaxies of type later than sa for a percentage between 20% and 25% . they do not explore , however , the dependence of the morphological composition of the color sequences on the environment . the cosmic evolution survey ( cosmos , scoville et al . 2007a ) provides us with an ideal combination of area , high - resolution hst imaging and multi - band information to perform such study . in this paper we present a first pilot study based on a sub - sample of the cosmos catalogue , with mean redshift @xmath10 . this sample is centered around a large - scale structure discovered by an adaptive - filter search in redshift space ( scoville et al . 2007b ) . a parallel analysis ( guzzo et al . 2007 , paper i hereafter ) investigates in detail the local environment of galaxies in this sample using projected densities and x - ray and weak - lensing measurements . a comparison with the total parent cosmos sample within the same redshift range shows this subsample to be representative with respect to galaxy color , luminosity and mass . while we are already working on extending our analysis to the whole area using improved spectroscopic and photometric redshifts , the current uncertainties on galaxy distances and consequently on measured local densities have suggested us this first exploratory work in an area where these quantities are better under control ( see paper i ) . the paper is organized as follows : in [ data ] we give a brief description of the data used here ; in [ env_col_mor ] we present the sample selection , we review the techniques used to estimate local densities and we present the tool used to assign morphological types ; in [ cmr ] we describe the color - magnitude relation , and its morphological content ; in [ a1 ] we analyze in more detail the properties of red late - type and blue early - type populations ; in [ cmassr ] , we introduce the color - mass relation with the aim of understanding the color - magnitude diagram ; we discuss these findings in [ discussion ] and summarize paper results in [ end ] . we adopt throughout the paper a `` concordance '' cosmological model , with @xmath11 km s@xmath12 mpc@xmath12 , @xmath13 , @xmath14 . however , when needed , we quote lengths and densities in units of @xmath15 . the cosmos _ treasury _ project is centered upon a complete 2-degree survey in the near - infrared ( @xmath16 ) f814w band using the advanced camera for surveys ( acs ) on board hst . the field is centered at @xmath17(j2000 ) = @xmath18 and @xmath19(j2000 ) = @xmath20 ( scoville et al . 2007a , 2007c ) . the acs camera , with a field of view of 203 arcsec side , has covered the whole field with a mosaic of 590 tiles , corresponding to one orbit each , split into 4 exposures of 507 seconds dithered in a 4-point box pattern . the final 590 images , obtained with the `` multi - drizzled '' software ( koekemoer et al . 2002 ) using the latest improved geometric distortion corrections for acs , are @xmath21 pixels each , with 0.05 arcsec pixels and an absolute astrometric accuracy of better than 0.1 arcsec . more details and a full description of the acs data processing and products are provided in koekemoer et al . ( 2007 ) . the cosmos 2-square degree field has been also fully covered by optical ground - based observations ( taniguchi et al . 2007 ) , using suprime - cam on the 8.2 m subaru telescope on mauna kea ( kaifu et al . the suprime - cam panoramic camera consists of @xmath22 ccds of 2k @xmath23 4k pixels , with a pixel scale of @xmath24 arcsec pixel@xmath12 ( miyazaki et al . 2002 ) . during two observing runs in january and february 2004 , @xmath25 , @xmath26 , @xmath27 , @xmath28 , and @xmath5 band images of the whole cosmos field were obtained . these data were complemented with further deep imaging in @xmath29 and @xmath30 from cfht , and in @xmath31 from kpno / ctio , uh88 and ukirt . the @xmath31-band data are essentially complete to @xmath32 . the detection of the sources has been performed on a combination of the subaru @xmath28 and chft @xmath30 original psf images . the photometry , on the other hand , has been measured within 3 apertures in psf omogenized images . in pratice , each image is smoothed to achieve the same @xmath33 of the image with the largest psf ( the @xmath31-band ) , and then the photometry is measured in a 3 aperture . this procedure reduces the effects of the psf variation from band to band and has the advantage that the total correction factors are identical in all bands . the resulting global photometric catalogue contains around 440,000 objects to @xmath34 . more details on the observations , data reduction and assembly of the catalogue are given in taniguchi et al . ( 2007 ) and capak et al . ( 2007a ) . photometric redshifts were derived from this catalogue using a bayesian photometric redshift method ( bpz ) ( benitez 2000 ; mobasher et al . 2007 ) , which uses six basic sed types , together with a loose prior distribution for galaxy magnitudes . the output of the code includes , in addition to the best estimate of the redshift , the 68 and 95% confidence intervals , the galaxy sed type , its absolute magnitude in several bands , and stellar masses . based on nearly 1200 objects for which spectroscopic observations are available in the cosmos area ( z - cosmos , lilly et al . 2007 ) , the difference between spectroscopic and photometric redshifts has a typical _ rms _ value @xmath35 ( not considering catastrophic failures ) . the quality of the photometric redshift however does not strongly depend on the sed type . as discussed in detail in mobasher et al . ( 2007 ) , we have @xmath36 for early - type ( e / s0 ) galaxies , @xmath37 for spiral galaxies and @xmath38 for starbursts , with a number of catastrophic failures below 2% . these figures are consistent with the restricted statistical sample of spectroscopic redshifts available within the large - scale structure at @xmath39 studied here ( paper i ) . to improve typical redshift errors , for this analysis we shall restrict the photometric redshift sample to @xmath2 . stellar masses were estimated from the same sed fits and rely , in particular , on the available k - band photometry . indeed , at the moderate redshifts of our sample galaxies , the k - band effectively probes the low - mass stellar content dominating the galactic mass budget ( while higher - z determinations will need the longer - wavelength spitzer photometry ) . more detailed information on the quality of the photometric redshifts and stellar mass estimates can be found in mobasher et al . scoville et al . ( 2007b ) use an adaptive - smoothing algorithm to search for large - scale structures within the cosmos field , by exploiting photometric redshift to divide the survey volume in `` slices '' . this analysis provides evidence among a number of systems at different redshifts for a conspicuous overdensity peaking in the redshift interval @xmath40 $ ] . the reality of this structure , with a mean redshift @xmath41 , has been spectroscopically confirmed by early redshift measurements of 14 galaxies in this area ( comastri et al . , priv . comm . ) and , more recently , by the first data from the vimos spectroscopic survey of the field ( lilly et al . the higher - density peaks within these structures are detected both in the xmm x - ray data ( finoguenov et al . 2007 ) and in the weak - lensing mass - reconstruction map ( massey et al . 2007 ; see paper i for comparisons ) . we concentrate here ( as in paper i ) on a sample of galaxies selected from an area of @xmath42 arcmin@xmath43 that roughly encloses the main part of the detected structure . this area is covered by 30 acs tiles of the overall cosmos acs survey . following dressler ( 1980 ) , galaxy projected densities are estimated as @xmath44 , where @xmath45 is the projected comoving distance ( in h@xmath12 mpc ) to the @xmath46 nearest galaxy , and then corrected for background contamination . the robustness of our local density estimates has been extensively tested using simulations ( paper i , appendix ) . in particular , we have applied our technique to a mock catalog generated from the millennium simulation ( kitzbichler & white 2006 ) , and we have showed that projected densities correlate well with real 3-d ones above @xmath47 . capak et al . ( 2007 ) , using a different technique , find similar results . as thoroughly discussed in paper i , we find that our estimate is fairly robust with respect to varying the redshift slice thickness , and on this basis adopt as standard for our analysis the redshift interval @xmath48 $ ] corresponding roughly to @xmath49 around the structure mean redshift . the sample includes 2041 galaxies to @xmath50 . the size of the cosmos sample naturally suggests the use of an automatic and objective morphological classification technique . our aim in this paper is to separate galaxies into early - types ( including ellipticals and s0 ) and late types ( spirals of any b / d , irregulars and mergers ) . we have therefore based our morphological estimates on non - parametric measurements of the galaxy light distribution , extending the work of cassata et al . we have made use of a parameter set including concentration @xmath51 , asymmetry @xmath52 , clumpiness @xmath53 , gini index @xmath54 and moment @xmath55 ( conselice 2003 , abraham et al . 2003 , lotz et al . we estimated these parameters for all galaxies in the 30 acs / hst images covering the @xmath56 structure . at this redshift , the observed f814w band approximately corresponds to rest - frame @xmath25 band . the thickness of our redshift slice is small enough to ensure that morphological k - correction and cosmological dimming effects can be neglected inside it . most of the results discussed here are based on internal comparisons ( e.g. for different density regimes ) ; when reference to local ( e.g. sdss ) morphological types is made , these effects are in general small for our two simple classes of early- and late - type galaxies ( see brinchmann et al . 1998 ) and in any case well within the differences in morphological definitions among different authors . the technique used is directly derived from cassata et al . ( 2005 ) , to which we refer for all technical details about the methods for measuring concentration , asymmetry and clumpiness . in this paper we have added two further estimators , namely the _ gini index _ and _ m20 _ parameter . the gini coefficient is a concentration parameter , measuring how fairly the light is distribuited among the galaxy pixels . in pratice , it is high when the light is concentrated in a small number of pixels , wherever they are located inside the galaxy . on the contrary , it is low when the galaxy light is spread over a large number of pixels . m20 is the moment of the brightest @xmath57 of galaxy flux : it is another concentration parameter , that however is more sensitive than concentration to any off - axis light clump . thus , it correlates with concentration for normal galaxies , and is expected to diverge from that for non - symmetric objects with clumps of light far from the center . parameters are defined and measured as in lotz et al . the only difference is in the segmentation technique used to assign image pixels to galaxies . in this work , we select an area around each galaxy 4 times larger than the isphotal area measured by sextractor , centered on the galaxy centroid , and then consider only pixels brighter than 1.5 times the rms of the background . in fact , given that we are dealing here with galaxies almost at the same redshift , we do not need to use petrosian apertures , that have the advantage of including the same part of galaxy image independently of redshift , minimizing surface brightness dimming effects . this latter approach is being applied in the extension of our work to the whole cosmos area and to different redshifts , which is in preparation . stars were removed from the sample using _ sextractor _ ( bertin & arnouts 1996 ) , according to their class - star and flux - best parameters measured on the acs data , for which this method proves to be reliable ( much more than on ground - based images ) . as shown by cassata et al . ( 2005 ) , there is a precise locus in the cas parameter space where early - type galaxies typically fall . to re - define this region for the cosmos acs data , adding the new two parameters , we used a control sample of 211 galaxies , extracted from two acs / hst tiles including the core of the main structure , for which we performed a visual classification into early - types ( ellipticals and s0 ) and late - types ( spirals , irregulars and mergers ) . in figure [ casgm ] we show the distribution of these galaxies in the space defined by concentration , asymmetry , clumpiness , gini and m20 parameters . using the following empirical selection rules : @xmath58 we are able to recover 92% of the galaxies visually classified as early - types in the control sample , with a contamination of only 6% from the late - type class ( mainly bulge dominated spirals ) . the contours in the different panels of fig . [ casgm ] describe the areas where 90% of the total galaxy sample lies , showing how the control sample is a representative random selection of the parent population . we therefore adopt eq . [ eqcas ] to classify early - type galaxies in the full sample , ending up with 1614 late - type and 427 early - type objects . figure [ col - mag ] plots the observed @xmath4 color versus @xmath59band magnitude for early- and late - type galaxies in our sample , for different intervals of local density @xmath60 . although we are working with _ observed _ colors , the limited redshift range covered ( @xmath61 ) makes this plot grossly equivalent to a rest - frame diagram . using observed rather than rest - frame k - corrected colors reduces the impact of redshift errors and avoids further uncertainties that are introduced by the sed fitting procedure ( see franzetti et al . the choice of the @xmath4 colour , among the several available , optimally brackets the 4000 break at @xmath62 $ ] , maximizing the color contrast of red galaxies . even if some contamination by foreground and background objects not belonging to our redshift slice is certainly present , because of the uncertainties in photometric redshifts ( see paper i for detailed discussion ) , the bimodal nature of the color distribution for the whole population is evident . the morphological information contribuites to separate the two sequences more clearly : a quite well defined red sequence , composed in large majority by early - type galaxies and with a small color scatter ; and a bluer , somewhat more dispersed one , dominated by late - type morphologies . the blue sequence has a larger scatter and a slightly steeper slope than that of the red sequence . however , in comparison to other analyses based on rest - frame colors ( e.g. bell et al . 2004a , tanaka et al . 2005 , franzetti et al . 2006 ) it is remarkably well - defined . also , as shown by the contour plots in the right bottom panel of figure [ col - mag ] , the two sequences keep their identity independent of density . bi - dimensional gaussian fits to the two sequences show the following dependences of the average colors on magnitude : @xmath63 for the red and @xmath64 for the blue sequence . we have verified that these fits do not depend on the magnitude limit chosen ( @xmath65 ) , by building a similar plot for all galaxies brighter than @xmath66 , and , within @xmath67 , recovering the same best - fit values of eqs . [ eq1 ] and [ eq2 ] . although being dominated by spheroidal and late - type galaxies , respectively , both red and blue sequences include a mixture of morphological types . we will discuss extensively in [ a1 ] the origin of these two classes . the solid lines in the color magnitude ( cm ) diagrams of figure [ col - mag ] show the locus for the red sequence predicted by the model of kodama & arimoto ( 1997 ) , for a population of early - type galaxies formed at @xmath68 and observed at @xmath69 . in this model galaxies are coeval and described by a single burst of star formation , after which they evolve passively . the inclination of the red sequence is produced in this scenario by a metallicity gradient ( see [ discussion ] for more details ) . one notes that while the slope of the data is reproduced fairly well , the model corresponding to the structure mean redshift seems to be systematically redder by about 0.1 magnitudes . we also plot ( dotted lines ) the red sequences corresponding to our assumed upper and lower redshift boundaries , @xmath70 and @xmath71 . the scatter is comparable to the region enclosed between the two extreme model sequences . the effect of the environment on the the color distribution of the galaxies in this sample is summarized in the histograms of figure [ col - dens ] , corresponding to the ordinate - axis marginal distributions of the 2d plots of figure [ col - mag ] ( see also kodama et al . 2001 and tanaka et al . 2005 , for similar analyses around clusters at different redshifts ) . the upper panels show the color distributions for the two morphological classes separately , the bottom - left one for the total sample . the vertical long tick - marks on the x - axis with different line styles give the median values for the corresponding distributions . the color distribution of the global sample ( bottom - left panel ) is bimodal in all environments , although the distribution of galaxies between the two main sequences does clearly depend on local density , with the red sequence being less evident in the lowest density bin , as already noticed in the color - magnitude diagram . when looking at the color distribution of early - types only , an interesting excess of blue galaxies , independently of the environment , can be noted . at the same time , the color distribution of late - types show a tail of galaxies with @xmath722 . for the rest , the histograms confirm that the color distributions of spheroidal galaxies , as well as that of late - types , are only weakly dependent on the environment : the shape of the histograms for the two classes separately are not sensitive to local density , with median values rather close to each other , and only a slight tendency for having redder colors at the highest projected densities . this is summarized in the bottom right panel of figure [ col - dens ] , that reports the median color as a function of local density for the whole population and for the early- and late - types separately . when considered as a whole , galaxies display a clear direct dependence of the median color on local density , in agreement with the color - density relation recently observed also at these redshifts by cucciati et al . ( 2006 ) and cooper et al . ( 2006a ) . our analysis specifically shows how this relation behaves for morphologically homogeneous classes . in fact , the median colors of early- and late - type galaxies separately do not seem to depend strongly on local density . we conclude that the bulk of the observed color - density relation is a by - product of the varying fraction of the two basic morphological types with environment : the median color for each class alone shows a mild dependence on local density , but when they are combined , the median color is dominated by blue spirals / irregulars at low densities and by red ellipticals / s0s at high densities , thus producing the observed overall effect . the objects in the color - magnitude plot of figs . [ col - mag ] with `` anomalous '' colors and morphologies , either `` blue early - types '' or `` red late - types '' , bear a potential for understanding the relationship of the two galaxy categories and whether they might evolve into each other . moreover , it is important to quantify the number of late - type populating the red sequence . when in fact a morphological analysis is unavailable , due to the lack of high resolution imaging , it is common to use a simple color criterion to segregate passive early - type galaxies from star - forming late - type galaxies ( e.g. bell et al . 2004a ) . we have 78 early- and 184 late - type objects with anomalous colors in our sample . the nature of these galaxies is unveiled in figure [ mosaic1 ] and figure [ mosaic2 ] , where we have collected postage - stamp acs images for 25 early - type - classified galaxies with `` blue '' colors ( @xmath74 ) and 25 late - type galaxies with `` red '' colors ( @xmath75 ) , randomly drawn from the total population in each class . for each galaxy , we report the values of the mass , @xmath4 color and local projected density @xmath60 . the first encouraging result from these figures is that the large majority of the displayed cases are not failures of our automatic morphological classification . in fact , in figure [ mosaic1 ] all objects are morphologically dominated by an evident spheroid , while in figure [ mosaic2 ] the prevalent morphology is disk - dominated or irregular . as evident from figure [ mosaic2 ] , virtually all the red late - type objects are edge - on or very inclined spirals , showing red colors because of strong internal reddening due to their geometrical orientation . weiner et al . ( 2005 ) , studying the relation between colors and morphologies for galaxies observed in the deep1 survey , also find a significant number of late - type galaxies showing red rest - frame colors . they claim in particular that only 75% of red galaxies are ellipticals or bulge dominated galaxies . similarly , bell et al . ( 2004b ) , using the gems survey , find that at @xmath76 80% of the @xmath25-band luminosity density in red galaxies is contributed by ellipticals or bulge - dominated galaxies . franzetti et al . ( 2006 ) , via spectroscopic measurements , claim that 35% of the red galaxies in the vvds survey show emission lines due to star formation ( with a very small contamination by agn ) . as in our case , the majority of those galaxies in deep1 and gems are edge - on spirals , in which galaxy colors are reddened by dust absorption , or bulge - dominated spirals . to quantify the excess of highly inclined disks observed in fig . [ mosaic2 ] , and to verify the hypothesis that a particular geometrical orientation of spiral galaxies can explain their extreme red colors , we studied the distribution of the ellipticities @xmath77 for late - type galaxies as a function of their color . the axial ratio @xmath78 is related to the galaxy disk inclination @xmath30 ( the angle between the perpendicular to the plane of the disk and the plane of the sky ) by the holmberg formula : @xmath79 where @xmath80 is the intrinsic ratio between the height and the radius of the disk ( a typical value for @xmath80 is 0.2 , tully & pierce 2000 ) . in fig . [ dist_ell ] we compare the distribution of ellipticities that we have measured for late - type galaxies lying in the blue and red sequences . the dotted line , in particular , shows the ellipticity for blue late - type galaxies : the excess of intermediate ellipticity values ( @xmath81 ) is very probably due to the large number of irregular / merging objects , that are mixed to spiral galaxies in the late - type class . for red galaxies , instead , the distribution is significantly skewed towards higher ellipticity values , peaking at @xmath82 . a kolmogorov - smirnov test shows that the probability that the two ellipticity distributions are drawn from the same intrinsic population is less than @xmath83 . this confirms that the inclination of the disk is the dominant effect in determining the measured color for late - types that are observed well redward of the blue sequence . consequently , a simple color criterion to identify passive elliptical galaxies will always suffer of a certain degree of contamination . based on our 184 `` red spirals '' out of 533 galaxies , this contamination amounts to @xmath8433% . these red late - type galaxies contribute to 20% of the luminosity density on the red sequence , a figure that confirms the result by bell et al . ( 2004b ) , weiner et al . ( 2005 ) and franzetti et al . ( 2006 ) . remarkably , our percentage of red galaxies with an early - type morphology is comparable to that measured in the local universe for sdss galaxies ( 58% , see renzini 2006 ) . also , a parallel study of the evolution of the global early - type population in the cosmos field ( scarlata et al . 2007 ) is in very good agreement with our estimate at @xmath0 , showing also a similar fraction between @xmath85 and @xmath86 . taking into account the differences in the selection criteria , these results indicate that the fraction of red galaxies with a relaxed morphology remains substantially unchanged between @xmath86 and the current epoch . whereas the red late - type galaxy population is clearly the result of a geometrical effect , the colors of the _ bonafide _ blue early - types could be indicative of recent or ongoing star - formation activity , placing them outside of the red sequence . these are interesting objects , as they may represent a transition stage , in the process of migrating onto the red sequence . evidence for blue colors or star - formation signatures in the spectra of early - type galaxies has been reported by various studies in the past ( schade et al . 1999 ; i m et al . 2001 ; menanteau , abraham & ellis 2001 ; menanteau et al . 2004 ; cross et al . 2004 ; cassata et al . 2005 ; treu et al . 2005a ; ilbert et al . schade et al . ( 1999 ) , in their study of cfrs galaxies find a percentage of about 30% of morphologically- selected spheroidals that exhibit [ oii]@xmath87 emission . cassata et al . ( 2005 ) , studying the morphological properties of the @xmath88 sample ( cimatti et al . 2002 ) , recognize a class of `` peculiar ellipticals '' , i.e. spheroidal galaxies with some distortion of the isophotes or with signs of a recent interaction . the majority of these galaxies show evidence for recent star formation in their spectra . the morphologies of the blue early - type galaxies found here are quite compatible with these properties . a direct check of their spectroscopic properties will soon become possible thanks to the ongoing z - cosmos redshift survey at the vlt ( lilly et al . 2007 ) . in our sample , we have 78 blue objects among the 427 spheroidals ( @xmath8918% ) . this percentage is lower than that measured at similar redshifts by various authors ( i m et al . 2001 , cross et al . 2004 , ilbert et al . 2001 ) , who indicate fractions around 30% . this is not surprising , as our sample is centred on a large - scale structure that is clearly over - abundant in high - density regions , including the very massive x - ray / lensing cluster ( guzzo et al . 2007 ) , making it possibly richer in early - type objects than an average volume at the same redshift . the blue early - types concur to 5% of the mass in ellipticals and to less than 3% of the total mass in galaxies . this means that if these will eventually migrate to the red sequence , they are not able to increase significantly its mass content . figure [ frac - col ] investigates a possible relation between the blue early - type population and the environment . specifically , we plot the fraction of objects as a function of local density separately for all early - type galaxies and for those with blue colours only . while the fraction of all early - types rises with local density , ( see paper i for a full discussion of the morphology - density relation at this redshift ) the percentage of blue spheroidals remains constant as a function of the environment , about @xmath90 . a kolmogorov - smirnov test confirms that the two distributions are statistically distinct , as the probability that they are drawn from the same original population is less than @xmath83 . could these objects be precursors of some of today s red early - type galaxies ? ferreras et al . ( 2005 ) suggest that a robust criterion to isolate the progenitors of early - type galaxies at different redshifts should exclude those objects that , even though showing a clear elliptical morphology , will not plausibly evolve into the local kormendy relation at the current epoch ( see also scarlata et al . the kormendy relation relates the effective radius @xmath91 ( the radius , in physical units , containing half of the total light ) with @xmath92 , the mean surface brightness within @xmath9 ( in arcsec ) . in this plane , local bright elliptical galaxies show a relatively tight scaling relation , for sizes between @xmath891 and 100 kpc ( kormendy 1977 ) . this is nothing else than a projection of the well - known fundamental plane of elliptical galaxies , connecting these two quantities to the galaxy velocity dispersion . capaccioli et al . ( 1992 ) show that more in general galaxies form 2 distinct families in the @xmath92 vs @xmath91 plane : while luminous ellipticals follow the classical _ kormendy relation _ , some s0s , spiral bulges and dwarf ellipticals populate a narrow nearly vertical strip with sizes typically smaller than @xmath93 kpc and a wide range of surface brightnesses . we have therefore computed the surface brightness in the @xmath25 rest - frame band , @xmath8 , and the corresponding @xmath91 from the acs / f814w band , for the early - type galaxies in our sample . the result is plotted in fig . [ kormendy ] . to obtain the plotted quantities , one has to take into account that due to the cosmological dimming the observed surface brightness @xmath94 is related to the rest - frame intrinsic surface brightness @xmath95 as @xmath96\end{aligned}\ ] ] where @xmath95 is the rest - frame surface brightness , @xmath97 reproduces the redshifting of the bandwidth , the ratio between @xmath98 and @xmath99 accounts for the difference in flux between the observed and emitted bands , and @xmath100 is the @xmath31-correction . for the redshift range we are exploring ( @xmath62 $ ] ) , the observed @xmath101 band well matches the rest - frame @xmath25 band . hence , to first approximation , @xmath102 , and the @xmath31-correction reduces to @xmath103 , without any evolutionary correction . the main uncertanties on the measured quantities are due to the scatter in the photometric redshifts , that affect the determination of distances and the surface brightness dimming correction . we estimate a typical error of less than 10% on @xmath91 , and @xmath890.2 mags on @xmath8 . for comparison , we also plot in fig . [ kormendy ] the points corresponding to early - type galaxies in the coma cluster ( jorgensen et al . 1995 ) , together with the @xmath104 corridor around the best fit ( grey / yellow band ) . we can notice that , in our sample , blue early - type galaxies have typically @xmath105kpc , while early - type galaxies larger than this size have normally red colors . also , for a fixed @xmath91 blue galaxies have fainter @xmath8 than red ones . on the contrary , the most luminous red galaxies follow a flatter relation , virtually parallel to the @xmath106 kormendy relation , with sizes in the range 3 kpc @xmath107 kpc . thus , blue elliptical galaxies seem to form a different population with respect to the most luminous red ones . simple evolutionary considerations can be used to establish which of the blue and red galaxies can evolve onto the local kormendy relation , and can thus be considered progenitors of nowadays spheroidals . purely passive evolution models ( bruzual & charlot 2003 ) with an initial burst of 1 gyr and a salpeter imf , predict a dimming of 0.7 - 1 magnitudes between @xmath108 and @xmath106 , for a redshift of formation between 2 and 5 . if this is the case , the majority of our red early - type objects will move onto the yellow / grey strip indicating the local relation for coma elliptical galaxies . it seems thus reasonable to conclude that the large majority of our red early - type galaxies could well be progenitors of local ellipticals , in terms of their surface - brightness / radius properties . we should also consider that the @xmath109 kormendy relation shown in fig . [ kormendy ] strictly holds for early - type galaxies in high density environments , being based on coma - cluster galaxies . in our sample , instead , only @xmath110 elliptical galaxies among 427 live in high density regions ( @xmath111 ) . these galaxies are plotted as filled circles in fig . [ kormendy ] , showing that they do not seem to differ from the overall population of red early - types in this diagram . a fading of 0.7 to 1 magnitude is actually a lower limit to the luminosity decrease of any galaxy between @xmath108 to @xmath106 ( see also gebhardt et al . 2003 , van dokkum & stanford 2003 , treu et al . blue objects will in general fade by an even larger amount . a reasonable estimate can be obtained by a model which continuously forms stars at a constant rate from the redshift of formation , down to the redshift of observation ( @xmath69 in our case ) ; at that point star formation is switched off . implementing this recipe using the bruzual & charlot ( 2003 ) code and considering formation redshifts between 2 and 5 , we obtain an expected fading of 2.5 magnitudes between z=0.73 and today . the decay in magnitude will be even larger than this if these galaxies are experiencing a short - lived star - burst triggered for example by an interaction / merger . even conceiving a fading as small as 1.5 magnitudes for our blue sample ( which is unrealistic if these galaxies are starbursting ) , then , just the 26 objects lying above the yellow / grey region ( i.e. @xmath112 of the total sample of blue spheroidals ) , would end up onto the local kormendy relation . more in general , those with surface brightness fainter than @xmath8919.5 mag arcsec@xmath113 should eventually move below the local relation , occupying the region with @xmath105kpc where capaccioli et al . ( 1992 ) find mainly bulges of spirals , s0 galaxies and dwarf spheroidals . could our blue spheroidals be a mixture of these three classes ? one one side , looking carefully at the morphologies shown in fig . [ mosaic1 ] ( and to the other galaxies in the same sub - sample ) , we see that only about 10% of the objects show signs for a faint disk component . we can not exclude that , additionally , some very low - surface brightness disks are in fact below the sb limit of the acs data . however , previous studies based on deeper , multi - band hst imaging ( as goods , cassata et al . 2004 ) do not find evidence for a general population of low - surface brightness disks around similar objects . also , resolved color analysis of similar objects shows evidence of blue cores rather than blue disks ( menanteau et al . 2004 ) . at the same time it seems also difficult to think of these objects as precursors of s0 s , for the very simple reason that we do not detect any correlation with density of their relative fraction , that is instead observed locally for s0 ( e.g. dressler 1980 ) . finally , if the fading is comparable to that computed from our simplified star - forming model ( i.e. of the order of 2.5 magnitudes ) , we see that only a fraction of these objects might become as faint as to be classified as dwarf ellipticals . considering a blue elliptical in our sample with @xmath8=20 and @xmath91=1.5 kpc , this implies a total magnitude @xmath114=21.4 and thus an absolute magnitude @xmath115 . assuming as a definition of a dwarf ellipticals a magnitude limit @xmath116 - 18 , we see that a fading of more than 3.5 magnitudes would be required . this is not implausible , but would require that our blue spheroidals are typically experimenting a strong burst of star formation . it is interesting to recast the cm diagrams of figure [ col - mag ] in terms of stellar mass . this is shown in figure [ col - mass ] . also in this case , two well - defined red and blue sequences , respectively dominated by spheroidal and late - type galaxies , are evident . the red sequence is more clearly defined as the projected density increases , whereas in the lowest density bin it is nearly absent , similarly to the cm plot . with this noted exception , in all other density regimes the red and blue sequences , again , keep remarkably invariant . the bottom - right panel summarizes these results by plotting the density contour plots in the color - mass plane for the red early - type and blue late - type galaxies ( that is galaxies that satisfy both morphological and color criteria ) in various density regimes . the remarkable consistency of the contours confirms the substantial indipendence of the color - mass relation _ within each morphological class _ on the environment . the thick continuous lines correspond to the average dependences of red and blue galaxies found with a 2d gaussian fitting . while the slopes of the two sequences in the cm diagrams of figure [ col - mag ] were fairly similar ( just slightly steeper in eq . [ eq2 ] than in eq . [ eq1 ] ) , in the current plots the blue sequence is steeper than the red one : @xmath117 @xmath118 we have tested ( as for eq . [ eq1 ] and [ eq2 ] ) that these fits do not change by including galaxies with @xmath119 . in this paper we have detailed the morphological composition of the blue and red parts of the cm diagram at @xmath0 as a function of the environment , extending previous results from combined hst - ground - based surveys ( bell et al . 2004b , weiner et al . our results show at @xmath76 the substantial invariance of the main features in this diagram of galaxies with respect to local density , similarly to what was found by hogg et al . ( 2004 ) at @xmath109 . the properties of the red sequence at moderate / high redshifts have so far been so far studied in particular in clusters ( see e.g. gladders et al . 1998 , kodama et al . 2001 , blakeslee et al . 2003 , yee et al . 2005 , tanaka et al . 2005 ) , or using rest - frame colors , for large comprehensive galaxy surveys ( as sdss , e.g. hogg et al . 2004 , combo-17 , bell et al . 2004a , deep-2 , cooper et al . 2006a , b and vvds , cucciati et al . 2006 and franzetti et al . 2006 ) . on the other hand , the evidence for a `` blue sequence '' when considering galaxy colors at a fixed redshift emerges here more clearly than in previous works . as it is evident from figure [ col - mag ] , the separation in morphological classes helps to better separate the two color populations . these patterns become only slightly more confused confused when other colors , not bracketing the 4000@xmath120 break , are used instead of the @xmath121 . this suggests that they are not simply the result of a population - dependent balmer jump , but derive from more general properties of the uv - optical spectrum . we have shown in fig . [ dist_ell ] that a significant fraction of highly inclined spiral galaxies are observed to have red colors only because of internal extinction . correcting for this would clearly have the effect to move them down in the cm diagram . when photometrically estimated stellar masses are plotted instead of galaxy luminosities ( figure [ col - mass ] ) , the galaxy distribution patterns keep some similarities , but also show differences , compared to the cm plots . in particular , the two red and blue color sequences show steeper dependences on galaxy mass ( eqs . [ eq3 ] and [ eq4 ] ) than on magnitude . another difference is that , while the slopes of the red and blue sequences in the cm plot appear to be similar ( cf . bottom panel of figure [ col - mag ] ) , the blue color - mass sequence is significantly steeper than the red one . figure [ col - mass ] shows that galaxies with colors @xmath122 to @xmath123 and masses in the range @xmath124 to @xmath125 are dominated by late - type morphologies , while the locus with colors from @xmath126 to @xmath127 is dominated by spheroidal systems with @xmath128 . these transition scales in mass are comparable to those shown at z@xmath890 by scodeggio et al . ( 2002 ) and kauffmann et al . ( 2004 ) and agree with the trend of an evolving star formation quenching mass suggested by bundy et al . ( 2006 ) . the models by kodama & arimoto ( 1997 ) , that we have superimposed on top of our observed red sequence distributions in figures [ col - mag ] and [ col - mass ] , have been originally developed to explain the tight red sequence observed in galaxy clusters . in these models , galaxies are co - eval ( with formation redshift @xmath68 in our case , that give , for the same redshift , a slightly redder sequence . ] ) and are described by a classical single burst of star formation and subsequent passive evolution in color and luminosity down to the time of observed galaxy redshift . the slope of the red sequence is explained as a differential metallicity effect : galaxies with smaller mass have more difficulty to retain the metals ejected by supernova explosions , and thus look bluer than their more massive companions with the same age . on the other hand , the blue / late - type sequence shows a significantly steeper dependence of color on mass , and can not be explained in terms of metallicity , especially at such blue colors . it seems more physically plausible to associate the color of blue galaxies to the timescale of star formation ( searle , sargent & bagnuolo 1973 , larson & tinsley 1978 ) . thus , the observed blue sequence can be interpreted as a relationship between the actual specific star formation , evidenced by the galaxy color , and its mass . in this picture , the observed steep sequence would indicate a gradual fading of star formation , with bluer hot stars gradually disappearing for late - type galaxies of larger and larger mass . this shows in quite an explicit way , at a fixed epoch , the so - called `` downsizing '' effect , now observed in several galaxy surveys ( e.g. cowie et al . 1996 , gavazzi et al . 1996 , franceschini et al . 1998 , treu et al . 2005a , bundy et al . this picture is reinforced by the observation that the slope of the blue sequence in the color - mass plane does not depend on the value of the local density ( figure [ col - mass ] ) , indicating that the specific amount of young stars produced per unit mass in late - type galaxies does not depend on the environment and is essentially regulated by internal processes . what is dependent on the environment is the relative number of early- and late - type galaxies populating the two sequences , i.e. the well - known morphology - density relation , still present at these redshifts ( see paper i and references therein ) . in this respect , our results agree very well with those recently obtained at the same redshift by the vvds ( cucciati et al . 2006 ) and deep-2 ( cooper et al . 2006b ) . the vvds work , in particular , explcitly claims that _ the color - magnitude distribution is not universal but strongly depends upon environment _ , a statement which might seem in apparent contraddiction with the findings of our work . this apparent confusion needs to be clarified . what this sentence and the overall cucciati et al . work imply , is simply that there exist a color - density relation observed at virtually any redshift ( with a tendency to disappear and possibly even invert its trend for @xmath129 , which is the main result of that paper ) , and thus that the density of points occupying the red and blue parts of the diagram is a function of environment . again , this is simply the color - density or morphology - density relation that we observe also here . what we mean here by stating that _ the color sequences do not depend on the environment _ , is that the location and slope of the two main clouds is independent of the local density , as shown by fig . [ col - mag ] . this is also implicitly noted by cucciati et al . , that find that _ the location of the colour gap between the red and blue peaks appears to be roughly constant and insensitive to the environment at all redshifts_. in their fig . 10 , in fact , while the blue sequence is too compressed to define a slope , it is very clear that the red sequence position and slope is rather insensitive , at all redshifts , to local density . the main results we have obtained in this paper can be summarized as follows . * the @xmath130 color - magnitude diagram at @xmath0 for our sub - sample of the cosmos catalogue shows a clear red sequence compatible to the expectations of the passive models of kodama & arimoto ( 1997 ) . rather than forming a `` blue cloud '' , blue galaxies also tend to aggregate along a similar , bluer sequence with increasing @xmath130 average color with luminosity . the distinction and `` sharpness '' of the two sequences is enhanced by separating galaxies into morphological types , with early ( late ) types dominating the red ( blue ) sequence . * the distribution function of galaxy colors is a function of the environment , as locally measured by balogh et al . 2004 , using sdss , with a distribution more skewed towards redder colors in in high - density regions . this confirms the positive color - density relation measured at similar redshift by cucciati et al . ( 2006 ) and cooper et al . ( 2006a ) . however , when early- and late - type galaxies are separated , the distributions become nearly independent of the environment . * we also study in detail the population of `` anomalous '' galaxies , with contrasting colors and morphologies . `` red late - types '' are found to be mostly edge - on spiral galaxies , for which colors are dominated by internal reddening by dust . their ellipticity distribution in fact is dominated by a peak around @xmath82 . in a simple color - selected sample , these galaxies would represent a 33% contamination over the population of truly red , passive spheroidals . at the same time , red late - type galaxies are less luminous than red early - types , thus contributing only to the 20% of the luminosity density of red galaxies ( similarly to bell et al . 2004b ) . * we also identified an interesting population of `` blue spheroidals '' , that show a flat morphology density relation , at variance with that of red early - types . they are also found to have smaller physical half light radii and fainter surface brightnesses , and contribute to less than 3% of the total galaxy mass content . simple considerations about the evolution of their stellar populations lead us to conclude that at least 70% of them will not evolve into the @xmath106 kormendy relation for bright ellipticals , but will occupy a region in the @xmath131 plane where nowadays we find s0 galaxies , spiral bulges and dwarf ellipticals . none of these three classes seems to be an obvious candidate for the descendents of our blue spheroidals . * in a color - mass diagram color sequences are even better defined , with red galaxies covering in general a wider range of masses at nearly constant color , and blue galaxies showing a more pronounced dependence of color on mass . we interpret the blue sequence as a specific star formation sequence , thus with the less massive late - type galaxies showing a younger average age of their stellar population . notably , the slope of this sequence has no dependence on the environment , indicating that the specific amount of young stars produced per unit mass does not depend on local density , but seems to be regulated more by the total galaxy mass itself . what the environment does , is to regulate the relative number of galaxies between the red and blue sequences , which at this redshift we still find ( once cleaned of spurious effects as internal extinction due to inclination in spirals ) , is nearly synonimous of being morphologically an early - type or late - type galaxy , respectively . this could be obtained via _ ab initio _ processes ( e.g. by simply forming more massive galaxies , which form stars and become redder faster , in higher - density regions , as it is natural in hierarchical models ) . alternatively , rapid bursts of star formation and/or gas depletion could be induced later in the life of the galaxy , as e.g. due to interactions / mergers in groups or ram - pressure stripping during infall of the galaxy into clusters or cluster - cluster mergers ( see evidence for this in paper i ) . both classes of mechanisms would seem to be able to rapidly move the galaxy away from the ` quiet life ' of the blue sequence ( where things develop driven simply by the galaxy mass ) and place it into the passive repository of the red sequence . using the full - fledged cosmos data set we hope to be able in the near future to be able to disentangle the respective role of these mechanisms , perhaps including also the possibly crucial contribution , not considered here , of switching - on an agn in the galaxy nucleus . pc thanks the osservatorio astronomico di brera for hospitality during the development of this work . lg thanks c. firmani and a. boselli for useful discussions . we thank t. kodama for providing us with his model red sequences in electronic form . pc thanks p. franzetti for providing bruzual & charlot models . we thank marcella carollo and simon lilly for suggesting the kormendy relation analysis . abraham , r. g. , et al . , 2003 , , 588 , 218 baldry , i k. , glazebrook , k. , brinkmann , j. , et al . , 2004 , , 600 , 681 balogh , m. l. , et al . , 2004 , , 615 , 101l bell , e. f. , et al . , 2004a , , 608 , 752 bell , e. f. , et al . , 2004b , , 600 , 11l benitez , n. , 2000 , , 536 , 571 bertin , e. , & arnouts , 1996 , a&as , 125 , 1882 blakeslee , j. p. , et al . , , 596 , 143 blanton , m. r. , et al . , , 594 , 186 bower , r. g. , et al . , 1998 , mnras , 299 , 1193 brinchmann , j. , et al . , 1998 , , 499 , 112 bruzual , g. , & charlot , s. , 2003 , mnras , 344 , 1000 bundy , k. , et al . , 2006 , , 651 , 120 capaccioli , m. , caon , n. , & donofrio , m. , 1992 , mnras , 259 , 323 capak , p. , et al . , 2006 , , this volume capak , p. , et al . , 2006 , , this volume cassata , p. , et al . , 2005 , mnras , 357 , 903 cimatti , a. , et al . , 2002 , a&a , 392 , 395 conselice , c. j. , 2003 , , 147 , 1 cooper , m. c. , et al . , 2006a , mnras , 370 , 198 cooper , m. c. , et al . , 2006b , submitted to mnras , ( astro - ph/0607512 ) cowie , l. , songaila , a. , hu , e.m . , cohen , j.g . , 1996 , aj , 112 , 839 cross , n j. g. , et al . , 2004 , aj , 128 , 1990 cucciati , o. , et al . , 2006 , a&a , 458 , 39 dekel , a. , et al . , 2005 , nature , 437 , 707 dressler , a. , 1980 , , 236 , 351 ferreras i. , lisker t. , carollo c. m. , et al . , 2005 , , 635 , 243 finoguenov , a. , et al . , 2007 , , this volume franceschini a. , silva l. , fasano , g. , et al . , 1998 , , 506 , 600 franzetti , p. , et al . ( the vvds team ) , 2006 , a&a , submitted gavazzi , g. , pierini , d. , & boselli , a. , 1996 , a&a , 312 , 397 gebhardt , k. , et al . , 2003 , , 597 , 239 guzzo , l. , et al . , 2007 , , this volume ( paper i ) hogg , d. w. , et al . , 2003 , , 585 , l5 giallongo , e. , et al . , 2005 , , 622 , 116 gladders , m. d. , lopez - cruz , o. , yee , h. k. c. , $ kodama , t. , 1998 , , 501 , 571 hogg , d. w. , et al . , 2002 , aj , 124 , 646 hogg , d. w. , et al . , 2004 , , 601 , l29 ilbert , o. , et al . , 2006 , a&a , 453 , 809 i m , m. , et al . , 2001 , , 122 , 750 jorgensen , i. , franx , m. , & kjaergaard , p. , 1995 , mnras , 273 , 1097 kaifu , n. , et al . 2000 , pasj , 52 , 1 kauffmann , g. , et al . , 2004 , mnras , 353 , 713 kitzbichler , m. g. , & white , s. d. m. , 2006 , submitted to mnras , ( astro - ph/0609636 ) kodama , t . , & arimoto , n. , 1997 , a&a , 320 , 41 kodama , t . , et al . , 2001 , , 562 , l9 koekemoer , a. , et al . , 2007 , , this volume kormendy . , j. , 1977 , , 218 , 333 larson . , r. b. , & tinsley , b. m. , 1978 , 219 , 46 lilly , s. , et al . , 2007 , , this volume lotz . , j. m. , et al . , 2004 , 128 , 163 massey , r. , et al . , 2007 , , this volume menanteau , f. , abraham , r. g. , & ellis , r. s. , 2001 , mnras , 322 , 1 menanteau , f. , et al . , 2004 , , 612 , 202 menci , n. , et al . , 2005 , , 632 , 49 mobasher , b. , capak , p. , et al . , 2007 , , this volume yagi m. , kashikawa , n. , sekiguchi , m. , doi , m. , yasuda , n. , shimasaku , k. and okamura , s. , 2002 , aj , 123 , 66 miyazaki , s. , et al . 2002 , pasj , 54 , 833 renzini , a. , 2006 , annual reviews of astronomy & astrophysics , ( astro - ph/0603479 ) scodeggio , m. , et al . , 2002 , a&a , 384 , 812 scarlata , c. , et al . , 2007 , , this volume scoville , n. z. , et al . , 2007a , , this volume scoville , n. z. , et al . , 2007b , , this volume scoville , n. z. , et al . , 2007c , , this volume schade , d. , et al . , 1999 , , 525 , 31 searle , l. , sargent , w. l. w. , & bagnuolo , w. g. , 1978 , , 179 , 427 strateva , i. , et al . , 2001 , aj , 122 , 1861 tanaka , m. , et al . , 2005 , _ mnras _ , 362 , 268 taniguchi , y. , et al . , 2007 , , this volume treu , t. , et al . , 2005a , , 622 , l5 treu , t. , et al . , 2005b , , 633 , 174 tully , r. b. , & pierce , m. j. , 2000 , , 533 , 744 van dokkum , p. g. , & stanford , s. a. , 2003 , , 585 , 78 yee , h. k. c. , hsieh , b. c. , lin , h. , & gladders , m. d. , 2005 , , 629 , 77 weiner , b. j. , et al . , 2005 , , 620 , 617
we study the enviromental dependence and the morphological composition of the galaxy color - magnitude diagram at @xmath0 , using a pilot sub - sample of the cosmos survey . the sample includes @xmath1 galaxies with @xmath2 and photometric redshift within @xmath3 , covering an area of 270 square - arcmin . galaxy morphologies are estimated via a non - parametric automatic technique . the @xmath4 vs. @xmath5 color - magnitude diagram shows a clear red - sequence dominated by early - type galaxies and also a remarkably well - defined `` blue sequence '' described by late - type objects . while the percentage of objects populating the two sequences is a function of environment , following a clear morphology / color - density relation also at this redshift , we establish that their normalization and slope are independent of local density . we identify and study a number of objects with `` anomalous '' colors , given their morphology , polluting the two sequences . red late - type galaxies are found to be mostly highly - inclined or edge - on spiral galaxies , for which colors are dominated by internal reddening by dust . in a sample of color - selected red galaxies , these would represent a 33% contamination with respect to truly passive spheroidals . conversely , the population of blue early - type galaxies is composed by objects of moderate luminosity and mass , concurring to only @xmath6 of the mass in spheroidal galaxies . the majority of them ( @xmath7 ) occupy a position in the @xmath8-@xmath9 plane not consistent with them being precursors of current epoch elliptical galaxies . their fraction with respect to the whole galaxy population does not depend on the environment , at variance with the general early - type class . in a color - mass diagram , color sequences are even better defined , with red galaxies covering in general a wider range of masses at nearly constant color , and blue galaxies showing a more pronounced dependence of color on mass . while the red sequence is adequately reproduced by models of passive evolution , the blue sequence is better interpreted as a specific star - formation sequence . the substantial invariance of its slope and normalization with respect to local density suggests that the overall , `` secular '' star formation is driven more by galaxy mass than by environment .
parity nonconservation ( pnc ) is caused by the weak interaction . according to the standard model this interaction is described in terms of charged and neutral currents . the charged currents play a dominant role in nuclei , e.g. in @xmath7-decay . the neutral currents lead to the pnc electron - nuclear interaction and can be observed in atomic experiments @xcite . in this paper we propose that enhanced pnc effects can be seen in electron recombination of multiply charged ions ( mci ) . the first suggestions and estimates of pnc effects in mci were made in 1974 by @xcite . a successful observation of pnc effects in optical experiments with heavy neutral atoms ( see the recent review @xcite for references ) has renewed the interest in pnc effects in mci @xcite . the obvious advantage of mci is the @xmath8 scaling of the pnc matrix elements with the nuclear charge @xmath9 , as opposed to @xmath10 scaling in neutral atoms @xcite . however , this advantage is usually compensated by larger energy differences between the levels of opposite parity . indeed , pnc effects in atoms and ions appear because of the mixing of the levels of opposite parity . this mixing leads , for example , to an admixture of a negative - parity state @xmath11 to a positive - parity state @xmath12 due to the parity nonconserving weak interaction @xmath13 , @xmath14 , as determined by the first - order perturbation expression @xmath15 the mixing coefficient @xmath16 is real when the level width @xmath17 is negligible compared to the level spacing @xmath18 . in neutral atoms the valence energies are roughly independent of @xmath9 , and @xmath19 scales as @xmath10 . in mci the level energies @xmath20 are proportional to @xmath21 and a typical pnc mixing @xmath19 again scales as @xmath10 . in some special cases the levels of opposite parity in mci can be anomalously close . for example , levels of the configurations @xmath22 and @xmath23 in he - like ions cross several times as @xmath9 varies @xcite . their proximity leads to a strong enhancement of the pnc effects . at the crossing point ( @xmath24 ) the maximal size of the mixing parameter is limited by the level widths and can be estimated as : @xmath25 according to ref . @xcite , a crossing of the @xmath26 and @xmath27 levels takes place at @xmath28 . because of the difference in the total electronic angular momentum , these levels can only be mixed by the nuclear - spin - dependent ( nsd ) part of the pnc interaction @xcite . two opposite - parity levels with the total angular momentum @xmath29 , @xmath30 and @xmath31 , cross twice at larger @xmath9 , around 65 and 90 @xcite . for such ions one can expect enhanced nuclear - spin - independent ( nsi ) pnc effects . in both cases the detection schemes involve radiative transitions . in this paper we propose to study pnc mixing in he - like ions by looking at the parity - violating asymmetry in kll dielectronic recombination ( dr ) of electrons with h - like ions . here the pnc interaction manifests itself as a difference between the recombination cross sections for electrons with positive and negative helicities . the observation of such difference means a correlation between the spin and momentum of the incident electron of the form @xmath32 , which does violate parity , since @xmath33 is a vector and @xmath34 is a pseudovector ( @xmath35 , while @xmath36 under spatial inversion ) . the pnc interaction in dr mixes the intermediate doubly excited @xmath37 and @xmath38 states of the he - like ion , which decay by the emission of a photon . in this respect pnc effects in dr are similar to those in neutron scattering from heavy nuclei . pnc asymmetries of up to 10% have been observed in nuclei by tuning the neutron energy to the @xmath39-wave compound nuclear resonances . this enhancement over the typical size of the nuclear weak interaction ( @xmath40 ) is caused by the proximity of @xmath41- and @xmath39-wave resonances , and by the large ratio of the @xmath41- to @xmath39-wave neutron capture amplitudes ( see , e.g. , review @xcite and references therein ) . kll dielectronic recombination can be observed in experiments with electron beam ion traps ( ebits ) or ion storage rings ( see , e.g. , @xcite ) . however , these device do not have a polarized ion target or electron beam , as is required . furthermore , the present generation of devices do not achieve the sensitivity required to observe the pnc effect . criteria for the sensitivity requirements are outlined in section [ discuss_sect ] from which the feasibility can be established for any future experimental devices . the doubly excited @xmath1 configurations contain a larger number of closely spaced levels than the singly excited @xmath42 configurations . in sec . [ energy ] we calculate the energies of the doubly excited @xmath1 states for @xmath43 and identify crossings between levels of opposite parity with @xmath44 and 1 . we then estimate the widths of the close levels of opposite parity and the pnc mixing coefficients @xmath19 . in sec . [ sigma ] we evaluate the dr cross section and pnc asymmetries . the paper concludes with a short feasibility analysis of pnc measurements in recombination of mci . the main aim of our work is to present the first analysis of pnc effects in the dr on h - like ions , to obtain a reliable estimate of the size of the pnc effects and to find the resonances and nuclear charges where these effects are largest . at the next stage it should be possible to improve significantly the accuracy of the calculations by using the well developed theory of the h - like and he - like ions . the energies of the @xmath1 states are determined by diagonalization of the effective hamiltonian in the @xmath45 subspace . the eigenstates are obtained as latexmath:[$\sum c_{lj , l'j ' } the hydrogen - like orbitals with @xmath45 . the single - electron part of this hamiltonian includes hydrogenic dirac orbital energies and the lamb shift . the two - electron part of the hamiltonian matrix for the configurations @xmath37 , @xmath47 , and @xmath38 is taken from ref . this work presents it as a double expansion in parameters @xmath48 and @xmath49 and we use three terms of this expansion of order @xmath9 , @xmath50 , and @xmath51 ( atomic units are used throughout the paper and @xmath52 is the fine structure constant ) . in particular , the first term accounts for the coulomb interaction between the electrons . the term @xmath50 accounts for the breit interaction and for the relativistic corrections to the wave functions . the last term ( @xmath51 ) corresponds to the second order in the coulomb interaction . , @xmath53 , and @xmath54 orbitals of the h - like ions . solid circles show the values of @xmath55 from eq . ( [ e1 ] ) calculated by johnson and soff @xcite . the curves correspond to 5-parameter fits used in our calculations for he - like ions.,title="fig : " ] , @xmath53 , and @xmath54 orbitals of the h - like ions . solid circles show the values of @xmath55 from eq . ( [ e1 ] ) calculated by johnson and soff @xcite . the curves correspond to 5-parameter fits used in our calculations for he - like ions.,title="fig : " ] the lamb shift appears in the higher order , @xmath56 , but it is known to be essential for the level crossings within the @xmath57 manifold @xcite . below we show that this is also true for the @xmath1 states . by factoring out the main dependence on @xmath9 and the principal quantum number @xmath58 , the lamb shift for the hydrogenic orbital @xmath59 is written as @xmath60 the values of @xmath55 calculated for @xmath61 and @xmath9 up to 95 by @xcite are shown in fig . [ fig_lamb ] . they account for the self - energy correction , vacuum polarization , and finite nuclear size effects . the results of the diagonalization of the hamiltonian matrix are shown in fig . [ fig_e2 ] . the eigenstates are labeled as @xmath62 and additional superscripts @xmath63 are added to distinguish levels with identical quantum numbers . one can see two crossings of the levels of opposite parity : a pair of levels with @xmath64 cross at @xmath65 and another pair with @xmath66 cross at @xmath4 . the latter crossing between @xmath67 and @xmath68 levels is entirely due to the lamb shift . this crossing disappears if the lamb shift is neglected . instead , another crossing with @xmath64 appears near @xmath69 between the levels @xmath70 and @xmath71 . as seen in fig . [ fig_e2 ] , their energies are very close for @xmath72 . and @xmath47 configurations relative to the @xmath67 level for he - like ions . all energies are divided by @xmath21 to account for the general scaling of energies in mci . the levels are labeled as @xmath62 and additional superscripts @xmath63 are added to distinguish levels with identical electron configurations and @xmath73 . ] of course , the exact position of the level crossings may depend on the higher terms of the @xmath49 expansion , which are neglected here . the first crossing at @xmath74 is rather sharp and takes place at relatively small @xmath9 , where radiative corrections are still small . therefore , this crossing is known rather accurately . the crossing between the levels @xmath67 and @xmath68 is more sensitive . however , it should not disappear when higher - order terms are included because of the @xmath75 scaling of the lamb shift . finally , the relative position of the @xmath70 and @xmath71 levels is most sensitive to the higher order terms and they may still cross at high @xmath9 . the crossing at @xmath65 leads to enhanced nsd pnc effects . the crossing at @xmath4 is favorable for the observation of the nsi part of the pnc interaction . here much larger pnc effects can be expected due to their scaling with @xmath9 . in addition , nsi interaction is roughly @xmath9 times stronger than nsd interaction , see sec . [ pnch ] . because of that we focus on the effects due to pnc mixing of the @xmath67 and @xmath68 levels . according to eq . ( [ i2 ] ) the pnc mixing near the crossing points depends on the line widths . for autoinizing states the total width is the sum of the radiative and autoionizing widths , @xmath76 . in the non - relativistic hydrogenic approximation the radiative widths of the states @xmath77 and @xmath78 are given by : @xmath79 where the coefficient @xmath80 defines contribution of the configuration @xmath37 to the state @xmath81 : @xmath82 the autoionizing widths are evaluated in appendix [ width ] . they depend weakly on @xmath9 and for @xmath83 are smaller than the radiative widths . hence , for @xmath83 the size of the pnc mixing is limited largely by the radiative width , @xmath84 . this means that the exact position of the level crossing is not very important for calculation of the pnc effect . the hamiltonian of the pnc interaction consists of the nsi and nsd parts and in relativistic notation has the form @xcite : @xmath85 where @xmath86 a.u . is the fermi constant of the weak interaction , @xmath87 are the dirac matrices , @xmath88 is the nuclear spin , and @xmath89 is the nuclear density normalized as @xmath90 . the dimensionless constants @xmath91 and @xmath92 characterize the strength of the nsi and nsd parts , respectively . @xmath93 is known as the weak charge of the nucleus . in the lowest order the standard model yields : @xmath94 where @xmath95 is the number of neutrons and @xmath96 is the weinberg angle , @xmath97 . radiative corrections to eq . ( [ p2 ] ) change @xmath91 by few percent @xcite . the constant @xmath92 includes contributions from the anapole moment @xmath98 and from the electron - nucleon neutral currents @xmath99 @xmath100 . flambaum and khriplovich showed that @xmath101 , where @xmath102 is the number of nucleons , and for heavy nuclei dominates over the constant @xmath99 @xcite . one more contribution to the constant @xmath92 was calculated by @xcite and by @xcite . except for the very heavy nuclei , this contribution is significantly smaller than that of the anapole moment . weak charges of the nuclei @xmath103tl and @xmath104cs were measured with high accuracy by @xcite and by @xcite . these measurements played an important role in low - energy tests of the standard model ( see review @xcite ) . up to now the only measurement of the nsd pnc amplitude was made for @xmath104cs @xcite . a detailed discussion of this matter and a complete list of references can be found in the recent review @xcite . because of the short - range nature of the interaction in hamiltonian it effectively mixes only one - electron states with @xmath105 , i.e. @xmath106 and @xmath107 . for a point - like nucleus the corresponding matrix element turns to infinity because of the singular behavior of the dirac orbitals at the origin . for a finite nucleus of the radius @xmath108 this matrix element can be approximately given by the following expression @xcite : @xmath109 where @xmath110 is the relativistic enhancement factor : @xmath111^{1/2 } , \label{me3a}\end{aligned}\ ] ] and the following approximation can be used for the nuclear radius : @xmath112 the accuracy of these expressions is a few percent , at least for the nsi part . a more accurate calculation can easily be done using dirac orbitals for the finite nucleus . at 1% level of accuracy the details of the nuclear structure and radiative corrections become important ( see @xcite and references therein ) . let us examine the enhancement of the pnc mixing due to the proximity of the levels of opposite parity in he - like ions . the mixing parameter @xmath19 is estimated with the help of eqs . and . the nsi part of the pnc hamiltonian mixes only states with @xmath66 . figure [ fig_e2 ] shows that there is only one pair of close levels of opposite parity which meets this requirement , namely @xmath67 and @xmath68 . the first of these states has admixtures of the configurations @xmath113 and @xmath114 [ see eq . ( [ e4 ] ) ] . for example , for @xmath115 the weights of these configurations are 0.19 and 0.02 , respectively . thus , the interaction between configurations @xmath37 and @xmath47 should be taken into account . using eqs . and , we obtain : @xmath116 [ cols="<,>,>,>,>,>,>,>,>,>,>,>",options="header " , ] results of the calculation of the pnc mixing for different @xmath9 are presented in table [ tab1 ] . the resonant enhancement at the level crossing is not very pronounced . firstly , the level crossing is not sharp . secondly , for @xmath117 the radiative width , which grows as @xmath75 , becomes greater than the autoionizing width and for @xmath72 it exceeds the level spacing . as a result , the absolute value of the pnc mixing grows steadily with @xmath9 . however , the enhancement at the level crossing is clearly seen when we consider the mixing strength divided by the @xmath118 pnc scaling parameter . the real part of the mixing changes sign at the resonance , where @xmath16 is equal to : @xmath119 describes the electron with momentum @xmath33 and helicity @xmath120 incident on the h - like ion in the @xmath121 ground state . the double lines 1 and 2 correspond to the doubly excited states of the he - like ion which decay by emission of a photon @xmath122 to the final state @xmath58 . the cross denotes pnc mixing of the states 1 and 2 . ] the formalism we use to calculate the pnc asymmetry in dr is similar to that developed for pnc effects in neutron scattering ( see , e.g. , review by @xcite ) . dr is described by the diagrams in fig . [ fig_sigma ] , where we assume that the incident electron @xmath123 has the energy @xmath124 , which is close to the transition energy between the ground state of the h - like ion and the levels @xmath125 of the configuration @xmath126 of the corresponding he - like ion . then the contribution of resonance 1 to the dr amplitude @xmath127 is : @xmath128 where @xmath129 and @xmath122 define the polarization and frequency of the photon , @xmath130 describes the initial state of the target with spin projection @xmath131 , and @xmath132 is the coulomb interaction . we use dipole approximation for radiative transition and @xmath133 is the quantization volume for the electromagnetic field @xcite . the total width of the doubly excited states 1 and 2 is given by the sum of the radiative and autoionizing widths , @xmath134 . as we have seen in sec . [ pncm ] , the strongest pnc effect is expected for the crossing of two levels with @xmath135 , which simplifies the derivation . hence we assume that the states 1 and 2 in fig . [ fig_sigma ] correspond to the levels @xmath136 discussed in sec . [ pncm ] . the total dr cross section is given by the sum over the final states of the ion and polarization of the photon : @xmath137 where @xmath138 and we neglect the square of the small pnc amplitude . when we substitute eq . ( [ t1 ] ) in , both pc and pnc terms appear to be proportional to the radiative width @xmath139 : @xmath140 the last expression is standard ( see @xcite ) and was used to calculate @xmath141 in eqs . and . the pc cross section now reads : @xmath142 the remaining coulomb matrix element determines the autoionizing width @xmath143 : @xmath144 where we take into account that @xmath145 . note that the sum in does not depend on the electron helicity @xmath146 , or on the direction of its momentum @xmath33 , while the individual matrix elements do depend on @xmath146 and @xmath131 . introducing the branching ratio @xmath147 for autoionization into channel @xmath148 , we can rewrite in the final form : @xmath149 similarly , the pnc contribution to the cross section becomes : @xmath150 further simplification of eq . ( [ t7 ] ) requires an explicit form of the coulomb matrix elements . let us expand the incident electron state in partial waves , @xmath151 where @xmath152 and @xmath153 are spherical and ordinary spinors and @xmath154 is the scattering phase shift . wave function is normalized so that @xmath155 , and the radial functions are normalized to the delta function of energy , @xmath156 . if we direct the quantization axis along @xmath157 , the spinor matrix element in can be written explicitly : @xmath158 where @xmath159 is the clebsh - gordon coefficient and @xmath160 is the spherical harmonic . when we use expansion to calculate the coulomb matrix elements in eq . ( [ t7 ] ) , the angular and parity selection rules leave only one term of this expansion with @xmath161 and @xmath162 , or 1 depending on the parity @xmath163 of the intermediate state @xmath164 : @xmath165 : @xmath166 where @xmath167 . substituting and in we obtain the following expression for the autoionizing width : @xmath168 we can also use to find the branching ratio @xmath147 in : @xmath169 this expression is valid only if the quantization axis for the angular momentum of the target ion coincides with the direction of the momentum of the incident electron . averaging over polarizations of the beam and the target gives @xmath170 and eq . ( [ t6 ] ) transforms into the standard breit - wigner expression @xcite : @xmath171 intermediate levels 1 and 2 in eq . ( [ t7 ] ) have different parity leading to different partial wave contributing to the matrix element : @xmath172 . the corresponding spherical spinors are related by ( see , e.g. , ref . @xcite ) : @xmath173 where @xmath174 . the partial matrix elements in are real and we can write : @xmath175 the factor @xmath176 in depends on the signs of the partial matrix elements in . therefore , the pnc cross section takes the form : @xmath177 eq . ( [ t13 ] ) is valid for the polarized as well as unpolarized target . in the first case one should use eq . ( [ t6 ] ) , while in the second case eq . ( [ t6a ] ) applies . for the unpolarized electron beam @xmath178 must be averaged over the helicity and eq . ( [ t13 ] ) gives zero for the unpolarized target . however , for the polarized target @xmath178 is not zero , because @xmath179 in eq . ( [ t6 ] ) selects the helicity through the branching ratio . in fact we can substitute @xmath180 in with @xmath181 , where @xmath182 is the spin of the ion . now we apply the formalism developed in the previous sections to calculate the pnc effect in the dr cross section at the energies near the @xmath183 resonances in the he - like ions . in the diagrams in fig . [ fig_sigma ] and in corresponding equations and the states 1 and 2 can be either @xmath67 and @xmath68 , or vice versa . these two contributions lead to the final states @xmath58 with different parities and we sum the corresponding cross sections , @xmath184 . equations , , and show that the phase factor @xmath185 in is equal to 1 . the incident electron energy scales as @xmath21 and the coulomb phase shifts in the non - relativistic approximation are independent of @xmath9 , @xmath186 . taking this into account and using and , we obtain the following total pnc cross section : @xmath187 , \label{pr1}\end{aligned}\ ] ] where @xmath188 , @xmath189 is given by , and again we can substitute @xmath190 with @xmath181 for a polarized target rather than a polarized electron beam . figure [ fig_pnc ] presents the plots of @xmath179 , @xmath178 and the pnc asymmetry , @xmath191 where @xmath192 are the cross sections for positive and negative helicity . the peak magnitude of the asymmetry increases from @xmath193 for @xmath194 to @xmath195 for @xmath196 , i.e. at the crossing point . it practically does not grow at larger @xmath9 . figure [ fig_pnc ] also shows that for @xmath197 the two resonances overlap . for @xmath117 the radiative width dominates over the autoionizing width and second term in square brackets in is suppressed . the first term changes sign between the resonances . therefore , the pnc cross section and asymmetry also change sign . consequently , the net asymmetry integrated over energy is suppressed . growth of the pnc matrix element for higher @xmath9 is compensated by the increase in the width and pnc asymmetry stops growing . it is useful to estimate the feasibility of measuring the pnc asymmetry in kll recombination calculated above and from this estimate derive the sensitivity requirements for an experimental apparatus capable of observing the pnc using the scheme proposed . the number of counts in an experiment with a fully polarized electron beam with positive helicity is given by : @xmath198 where @xmath199 is the electron flux , @xmath200 is the number of target ions , @xmath201 is the acquisition time , and @xmath202 is the detection efficiency . the number of counts for negative helicity is @xmath203 . for a beam or target with polarization @xmath204 , to detect the pnc asymmetry , the difference between the counts , @xmath205 should be greater than statistical error , @xmath206 , which gives : @xmath207 where @xmath208 is the magnitude of any background occurring through direct radiative recombination or as an experimental artifact ( e.g. detector dark counts ) . we express this as a cross section for convenience although for experimental artifact background signal , this is the effective cross section to which the apparatus background would correspond . for the rest of this analysis we consider the ideal limit , @xmath209 , @xmath210 . equation is valid for a mono - energetic electron beam . if the electron energy spread in the beam is greater than the resonance spacing and widths , then the flux @xmath199 in should be replaced by the flux density @xmath211 . the counts @xmath212 are obtained by integrating over the electron energy and the effect can be detected if @xmath213 the first integral above is equal to @xmath214 , where @xmath215 is the strength of resonance @xmath164 . the integral @xmath216 in eq . ( [ d3 ] ) can be written as @xmath217 , where @xmath218 } \nonumber\\ & \times\left[(e_+-e_-)\cos\delta_{sp } -\tfrac12(\gamma_++\gamma_- ) \sin\delta_{sp } \right ] , \nonumber\end{aligned}\ ] ] is the pnc strength of the two resonances . thus , eq . ( [ d3 ] ) reads : @xmath219 equations and show that for the two limiting cases of narrow and wide energy distribution in the beam the feasibility of the experiment on ions with nuclear charge @xmath9 depends on the functions : @xmath220 , \label{d7}\\ f_{\rm av } & = \int(\sigma^+ + \sigma^- ) d \veps \left[\int ( \sigma^+ - \sigma^- ) d \veps\right]^{-2}\!\ ! , \label{d8}\end{aligned}\ ] ] where minimum is taken with respect to the energy of the beam . these functions are shown in fig . [ fig_fred ] , where cross sections are in barns ( 10@xmath221 @xmath222 ) and energy in ev . for mono - energetic beam the measurements become more feasible at large @xmath9 , where the pnc asymmetry is greater . the strongest effect can be seen for @xmath223 , i.e. near the level crossing . the averaged pnc effect is strongly suppressed by the factor @xmath224 in eq . ( [ d5 ] ) , which monotonically decreases with @xmath9 , as @xmath225 . this suppression is caused by almost antisymmetric shape of the pnc signal along the resonances . we conclude that observation of the pnc effects in dr is much easier with mono - energetic beam . in b@xmath226 ( solid line ) and @xmath227 in b@xmath226ev@xmath226 ( dashed line ) . shallow minimum of @xmath228 at @xmath223 approximately corresponds to the level crossing . ] it is worth noting that @xmath229 and @xmath230 have the same dimensions as a cross section and a resonance strength respectively . indeed , these quantities are useful for estimating the feasibility of any future experiment designed to observe the pnc effect in kll dielectronic recombination resonances . for an experiment to be able to observe the effect predicted , it would have to be able to detect a cross section as small as @xmath229 or a resonance strength as small as @xmath230 , i.e. about @xmath231 b or @xmath232 b ev , in the absence of background . of course , this is extremely demanding , but it is worth remarking that the level crossing considered here , gives rise to an enhancement of eight orders of magnitude compared to the basic strength of the weak interaction in atoms . let us compare the present scheme with other proposals for measuring pnc in ions . @xcite suggested to observe pnc effect in the auger emission from the he - like uranium . he considered the mixing of the same states , @xmath37 and @xmath38 with @xmath29 , and obtained asymmetries of about @xmath40 , which is comparable to our results . however , that estimate neglected the radiative widths of the levels , which for @xmath233 exceed the level spacing . other proposals were based on the observations of pnc asymmetries in radiative transitions in he - like ions . @xcite focused on the two - photon @xmath234-@xmath235 transition between two metastable levels , @xmath236 , separated by 1 ev in u@xmath237 . they showed that the pnc mixing is @xmath238 , and concluded that lasers with intensities above @xmath239 w/@xmath222 would be required to observe it . in refs . @xcite two - photon and hyperfine - quenched transitions @xmath240 were examined . here the mixing between @xmath241 and @xmath242 levels leads to circular polarization of the photons ( up to @xmath243 ) or to an asymmetry in the photon angular distribution ( @xmath244 for gd@xmath245 for a fully polarized ion beam ) . although these values seem large , there is a number of associated problems : low counting rates for the highly forbidden transitions involved , photon background , detection of the circular polarization of gamma quanta , and creation of the polarized ion beam . as a result , the number of events necessary to measure the effect is large , e.g. @xmath246 @xcite . authors are thankful to v. shabaev , who pointed at the sign error in the numerator of eq . ( [ d5 ] ) in the published version of this paper @xcite . in this corrected version of the e - print we traced the error back to eq . ( [ t12 ] ) and made respective changes to the text and figures . authors are grateful to a. korol , l. labzowsky , and a. nefiodov for valuable discussions and to d. budker for reading the manuscript . this work is supported in part by russian foundation for basic research , grant no . 05 - 02 - 16914 . one of us ( mk ) thanks international research centre for experimental physics in belfast for the dvf fellowship . the autoionizing widths of the doubly excited states with @xmath29 are given by eq . ( [ t9a ] ) . their wave functions are linear combinations of the two - electron states of the form : @xmath248 where parity @xmath249 . in the initial state the incident electron is described by the wave function and the h - like ion is in the ground state @xmath250 . the coulomb matrix elements on the right - hand - side of , after substituting , are reduced to the two - electron matrix elements : @xmath251 \sum_k \left(\!\begin{array}{ccc } j_i & j_b & k \\ -m_i & m_b & q \\ ! \label{w2}\\ & \times \left(\!\begin{array}{ccc } j_c & \tfrac12 & k \\ -m_c & m & q \\ ! \left(\!\begin{array}{ccc } j_i & j_b & k \\ \tfrac12 & -\tfrac12 & 0 \\ \end{array}\ ! \right)\ ! \left(\!\begin{array}{ccc } j_c & \tfrac12 & k \\ \tfrac12 & -\tfrac12 & 0 \\ \end{array}\ ! \right)\ ! r^k_{b , c , i,1s } , \nonumber\end{aligned}\ ] ] where @xmath252 \equiv [ ( 2j_a+1)\dots]^{1/2}$ ] and @xmath253 is the coulomb radial integral . it is nonzero for even @xmath254 and @xmath255 . equations and allow one to calculate the matrix elements in . neglecting the dependence of the radial integrals on @xmath256 we have for the odd state , @xmath257 where @xmath258 . for the even state we obtain @xmath259 and for the eigenstate we arrive at @xmath260 to estimate the widths @xmath261 we use non - relativistic hydrogenic radial coulomb integrals , which do not depend on @xmath9 : @xmath262 equations , , and give @xmath263 for all @xmath9 . @xmath264 is somewhat smaller than @xmath265 and weakly depends on @xmath9 via the coefficients @xmath266 in eq . ( [ w5 ] ) . for pure @xmath267 state eqs . and give @xmath268 . this value is in agreement with ref . @xcite . the same nonrelativistic hydrogenic approximation for the radiative transitions was used in eqs . and for the radiative widths @xmath141 . again the negative parity state has larger width . comparison of eqs . and shows that the radiative width becomes equal to the autoionizing width for @xmath269 and dominates near the level crossing at @xmath4 .
we discuss a parity nonconserving asymmetry in the cross section of kll dielectronic recombination of polarized electrons on the hydrogen - like ions with @xmath0 . this effect is strongly enhanced because of the near - degeneracy of doubly - excited @xmath1 states of opposite parity in he - like ions . for ions with @xmath2 the asymmetry is of the order of @xmath3 . for @xmath4 a level crossing takes place , leading to the pnc asymmetry of @xmath5 , which is @xmath6 times greater than the basic strength of the weak interaction in atoms .
zadeh @xcite has introduced the concept of fuzzy sets and fuzzy set operations . subsequently several authors have discussed various aspects of its theory and applications such as fuzzy topological spaces , similarity relations and fuzzy orderings , fuzzy measures of fuzzy events , fuzzy mathematical programming etc . + in recent years , there has been an increasing interest in various mathematical aspects of operations defined on fuzzy sets . based on these , sequences of fuzzy numbers have been introduced by several authors and they have obtained many important properties . matloka @xcite has studied the bounded and convergent sequences of fuzzy numbers where it is shown that every convergent sequence is bounded . the class of all @xmath0-summable convergent sequences of fuzzy numbers is introduced by nanda @xcite . later on , talo and basar @xcite , @xcite determined the dual of classical sets of sequences of fuzzy numbers and related matrix transformations . recently bounded variation for fuzzy numbers is studied by tripathy et al in @xcite@xcite . + as the set of all real numbers can be embedded in the set of all fuzzy numbers , many results in reals can be considered as a special case of those fuzzy numbers . however , since the set of all fuzzy numbers is partially ordered and does not carry a group structure , most of the facts known for the sequences of real numbers may not be valid in fuzzy setting . therefore , this theory is not a trivial extension of what has been known in real case . + recently , polat et al @xcite have defined sequence spaces of real numbers using generalized weighted mean . this has been extended by ojha and srivastava @xcite by introducing new sets of sequences @xmath1 , @xmath2 of fuzzy numbers . various properties such as quasilinearity , completeness of these sequences are investigated by them . in the present paper , we have defined a more general class of sequences of fuzzy numbers using modulus function over the class @xmath1 . a fuzzy real number @xmath3 $ ] is a fuzzy set satisfying the following conditions : a. @xmath4 is normal i.e. there exists an @xmath5 such that @xmath6 . b. @xmath4 is fuzzy convex i.e. @xmath7\geq \min \{x(t),x(s)\}$ ] for all @xmath8 and for all @xmath9 $ ] . c. @xmath4 is upper semi - continuous . d. @xmath10 is compact . clearly , @xmath11 is embedded in @xmath12 , the set of all fuzzy numbers , in this way : for each @xmath13 , @xmath14 is defined as , @xmath15 for , @xmath16 , @xmath17-cut of a fuzzy number @xmath4 is defined by @xmath18 . the set @xmath19 is a closed , bounded and non - empty interval for each @xmath20 $ ] . + also , we denote the triangular fuzzy numbers as @xmath21 $ ] i.e. its membership function is defined as , @xmath22 now for any two fuzzy numbers @xmath23 , matloka @xcite has introduced a metric , @xmath24 where @xmath25 and @xmath26 are the lower and upper bound of the @xmath17-cut and showed that @xmath27 is a complete metric space . + a partial ordering relation on @xmath12 is defined as follows : @xmath28 @xmath29 $ ] . the absolute value of @xmath30 is defined by , @xmath31 ( representation theorem ) @xcite + let @xmath32 $ ] for @xmath30 and for each @xmath33 $ ] . then the following statements hold : a. @xmath34 is a bounded and non - decreasing left continuous function on ( 0 , 1 ] . b. @xmath35 is a bounded and non - increasing left continuous function on ( 0 , 1 ] . c. the functions @xmath34 and @xmath35 are right continuous at the point @xmath36 . d. @xmath37 . conversely , if the pair of functions @xmath38 and @xmath39 satisfies the conditions ( i)-(iv ) , then there exists a unique @xmath30 such that @xmath40 $ ] for each @xmath41 $ ] . the fuzzy number @xmath4 corresponding to the pair of functions @xmath38 and @xmath39 is defined by @xmath42,\ x(t ) = \sup\{\alpha : p(\alpha)\leq t \leq q(\alpha)\}$ ] . from the above theorem , it is clear that for any @xmath30 , @xmath43 . let @xmath44 be the set of sequences all fuzzy numbers . let @xmath45 . then the set @xmath46 is said to be 1 . solid if @xmath47 , whenever @xmath48 and @xmath49 , for all @xmath50 . 2 . symmetric if @xmath51 , whenever @xmath48 , where @xmath52 is a permutation on @xmath53 . convergence free if @xmath47 , whenever @xmath48 and @xmath54 implies @xmath55 . throughout this paper , @xmath56 be the set of all real sequences @xmath57 , @xmath58 such that for all @xmath59 , @xmath60 . let @xmath61 and consider the matrix @xmath62 as @xmath63 for all @xmath64 where @xmath65 depends only on @xmath59 and @xmath66 depends only on @xmath67 . this matrix @xmath68 is called generalized weighted mean or factorable matrix . many authors such as altay and basar @xcite , polat et al . @xcite etc constructed real sequence spaces by using this generalized weighted mean . recently in 2014 , ojha and srivastava @xcite used this matrix to introduce a new class of sequences of fuzzy numbers . + in 2014 , ojha and srivastava @xcite introduced a new class defined as , @xmath69 where @xmath70 and @xmath61 . in the present paper , we have defined the set of sequences of fuzzy numbers by using modulus function and generalized weighted mean over @xmath71 and investigated various properties of it . as special cases , we get cesaro mean , weighted mean of fuzzy numbers etc . a. for @xmath72 and @xmath73 for all @xmath50 , we get @xmath74 which is cesaro sequence of fuzzy numbers . b. if we take @xmath75 s as positive numbers and @xmath76 , then @xmath77 is the weighted mean over @xmath27 . ruckle @xcite introduced the concept of modulus function and also defined a new sequence spaces by using modulus function of real or complex numbers . a modulus function @xmath78 is a function @xmath79 such that , a. @xmath80 iff @xmath81 . b. @xmath82 . c. @xmath78 is increasing . d. @xmath78 is continuous from the right at 0 . it follows that @xmath78 must be continuous everywhere on @xmath83 . using the modulus function , maddox @xcite@xcite introduced the generalizations of the classical spaces of strongly summable sequences by defining certain sequence spaces using a modulus function @xmath78 . + let @xmath84 be a bounded sequence of strictly positive real numbers . if @xmath85 , then for any real or complex numbers @xmath86 , we have @xmath87 where @xmath88 and @xmath89 is any complex number . let @xmath61 and @xmath84 is a bounded sequence of strictly positive real numbers . we define a new set of sequences of fuzzy numbers using the modulus function @xmath78 as follows : @xmath90 = \big\{x=(x_k)\in w^f:\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(x_i,\bar{0})\big|\big)\big]^{r_k}\in \lambda\big\}\ ] ] for @xmath91 and in case of @xmath92 , we define the class as follows : @xmath93 = \big\{x=(x_k)\in w^f:\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(x_i , x_0)\big|\big)\big]^{r_k}\to 0 \ \mbox{for}\\ \mbox{some}\ x_0\in l(r)\big\}\end{aligned}\ ] ] since @xmath96\subset l(r)$ ] , so all the properties as listed in @xcite trivially follows from the property of @xmath12 . so it is enough to show that @xmath96 $ ] is closed under addition and scalar multiplication which follows from the properties of the modulus function and the metric @xmath98 on @xmath12 . therefore , we omit the details of the proof . the set @xmath96 $ ] for is a complete metric space under the following metric , @xmath99^{r_k / m}\ \mbox{for $ \lambda = l_\infty , c , c_0$}\\ d_p(x , y ) & = & \big\{\sum\limits_{k=0}^{\infty}\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(x_i , y_i)\big|\big)\big]^{r_k}\big\}^{1/m}\ \mbox{for $ \lambda = l_p$ $ ( 0<p<\infty)$}\end{aligned}\ ] ] where @xmath100 and @xmath101 $ ] . we shall give the proof for @xmath102 . rest will follow the same lines . it is easy to verify that @xmath103 is a metric on @xmath104 $ ] . to prove the completeness property , let @xmath105 be a cauchy sequences in @xmath104 $ ] . then , @xmath106^{r_k / m}\to 0 \ \mbox{as}\ n , m\to\infty \\ \rightarrow \ f\big(\big|u_k\sum\limits_{i=0}^k v_id(x_i ^n , x_i ^m)\big|\big)\to 0 \ \mbox{as}\ n , m\to\infty\ \mbox{and for all}\ k\end{aligned}\ ] ] this follows because each term is non - negative and @xmath107 is a bounded sequence of non - negative real numbers . since @xmath78 is modulus function , so for all @xmath67 , we have @xmath108 which shows that @xmath109 is a cauchy sequence in @xmath12 , hence it is convergent . suppose @xmath110 for all @xmath111 as @xmath112 . so , @xmath113 . now , it is remained to show that @xmath114 $ ] . @xmath115^{r_k } & \leq & b\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(x_i , x_i ^n)\big|\big)\big]^{r_k } \\ & & + b\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(x_i ^n,\bar{0})\big|\big)\big]^{r_k}\end{aligned}\ ] ] where @xmath116 is a constant as defined earlier . since @xmath117 $ ] and @xmath118 under the metric @xmath103 , so @xmath114 $ ] i.e. @xmath104 $ ] is a complete metric space . it is enough to show it under the metric @xmath103 . others will follow similar lines . + here we have to show that the projection operator @xmath121\to l(r)\ \mbox{given by } \ p_k(x)=x_k,\ x\in \lambda^f [ u , v , f , r]\ ] ] is continuous for each @xmath50 i.e. it is enough to show that if @xmath122 be a sequence in @xmath96 $ ] such that @xmath123 in @xmath96 $ ] , then @xmath124 in @xmath12 @xmath125 . let , @xmath123 in @xmath96 $ ] , then we have @xmath126^{r_k / m}\to 0\end{aligned}\ ] ] @xmath127 . which gives that , @xmath128 for each @xmath67 as @xmath112 . proceeding with similar arguments as before , we get @xmath129 so , @xmath124 in @xmath12 i.e. @xmath96 $ ] is a @xmath119-space . assume @xmath131 $ ] for the above choice of @xmath132 and @xmath133 is a sequence of fuzzy numbers such that @xmath134 . then , @xmath135 thus we have , @xmath136 $ ] whether @xmath102 or @xmath137 . so , @xmath138 $ ] is solid . @xmath139 $ ] is not solid . to show this , consider the following example . + assume @xmath94 and @xmath140 for all @xmath141 . let us take the sequences @xmath142 and @xmath143 as , for all @xmath144 , @xmath145 then it is clear that @xmath146 and @xmath147 $ ] but @xmath148 $ ] . then , @xmath150 \leq \frac{1}{\sqrt{k } } \to 0\\ \big|\sum\limits_{i=1}^k u_kv_id(y_i,\bar{0})\big|=\frac{1}{k } \frac{k}{2 } = \frac{1}{2}\nrightarrow 0\end{aligned}\ ] ] where @xmath151 $ ] is the box function . therefore @xmath152 $ ] whereas @xmath153 $ ] . + in a similar way , we can show for @xmath154 also . so , @xmath138 $ ] is not symmetric . to verify this for @xmath155 , let us assume @xmath156 and @xmath95 for all @xmath141 and . also let @xmath78 be a identity function on @xmath83 to @xmath83 . let @xmath157 be a sequence defined as , @xmath158 $ ] . then @xmath159\end{aligned}\ ] ] since @xmath160 @xmath161 $ ] . + let @xmath133 be an another sequence defined by @xmath162 $ ] . then @xmath163 by similar arguments , @xmath164 $ ] . + now , @xmath165 $ ] and therefore @xmath166\end{aligned}\ ] ] since , @xmath167 . so , @xmath168 $ ] . therefore @xmath138 $ ] does not form a sequence algebra . + rests can be proved in similar way . @xmath172 , & k \ \mbox{is odd } \\ \bar{0 } , & k\ \mbox{is even}\\ \end{array } \right.\ \mbox{and}\ y_k = \left\ { \begin{array}{c l } [ -\frac{1}{k } , 0 , \frac{1}{k } ] , & k \ \mbox{is odd } \\ \bar{0 } , & k \ \mbox{is even}\\ \end{array } \right.$ ] @xmath173\leq\frac{1}{2k}\to 0 \ \mbox{as $ k\to\infty$ } \\ \mbox{and } u_k\sum\limits_{i=1 } ^k v_id(y_i,\bar{0 } ) & = & \left\ { \begin{array}{c l } \frac{1}{k^2}[1 + 3 + 5+\cdots+(k-1 ) ] , & k \ \mbox{is even } \\ \frac{1}{k^2}[1 + 3 + 5+\cdots+k ] , & k\ \mbox{is odd}\\ \end{array } \right.\\ & = & \left\ { \begin{array}{c l } \frac{1}{k^2}(k-1)^2 , & k \ \mbox{is even } \\ \frac{1}{k^2}k^2 , & k\ \mbox{is odd}\\ \end{array } \right.\end{aligned}\ ] ] a. let @xmath170 and @xmath73 for all @xmath67 . also take @xmath179 for all @xmath144 . then @xmath180 . so , it is easy to see @xmath175 $ ] but @xmath181 $ ] . b. @xmath182 for all @xmath67 and let @xmath183 , & k \ \mbox{is odd } \\ \bar{0 } , & k \ \mbox{is even}\\ \end{array } \right.$ ] + then by the previous example , we can show that @xmath184 as it converges to 1 as @xmath174 but @xmath185 c. let @xmath186 and @xmath187 for all @xmath67 . then @xmath188 or 2 according to @xmath67 is even or odd . so , @xmath189 but @xmath190 . @xmath199^{q_k } & = & b\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(y_i,\bar{0})\big|\big)\big]^{q_k}\\ & & + b\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(z_i,\bar{0})\big|\big)\big]^{q_k}\\ & & \mbox{where $ b$ is a constant.}\\ & \leq & b\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(y_i,\bar{0})\big|\big)\big]^{r_k}\\ & & + b\big\{\big[f\big(\big|u_k\sum\limits_{i=0}^k v_id(z_i,\bar{0})\big|\big)\big]^{r_k}\big\}^{q_k/ r_k}\\ & \leq & b[m+m^{q_k / r_k}]\end{aligned}\ ] ] a. if @xmath202 for all @xmath203 , then @xmath204\subseteq \lambda^f [ u , v , f , r]$ ] for @xmath205 . b. @xmath96\cap \lambda^f [ u , v , g , r]\subset\lambda^f [ u , v , f+g , r]$ ] for @xmath205 . c. @xmath96\subset \lambda^f [ u , v , g\circ f , r]$ ] for @xmath206 . d. if @xmath207 , then @xmath208 for some positive integer @xmath59 for @xmath206 . a. is easy to check . so we omit it . b. for any two functions @xmath201 , we have , @xmath209 which easily gives @xmath96\cap \lambda^f [ u , v , g , r]\subset\lambda^f [ u , v , f+g , r]$ ] . c. from the continuity of @xmath210 , we have for a pre - assigned positive @xmath211 , @xmath212 such that @xmath213 . now assume @xmath214 $ ] . then @xmath215 for all @xmath216 for some fixed @xmath217 . thus we get @xmath218 so , @xmath219 $ ] . similarly , we can prove it for @xmath92 also . d. putting @xmath220 , identity function on @xmath83 and @xmath221 for some positive integer @xmath59 in ( iii ) , we got @xmath222\subseteq \lambda^f [ u , v , f^n , r]$ ] . to prove the converse , let @xmath223 . then from @xcite , @xmath224 . so there is a @xmath225 such that @xmath226 for all @xmath227 . + so @xmath228 . generalizing we get @xmath229 for some positive @xmath59 . assume @xmath230 . now putting @xmath231 , @xmath232^{r_k}\\ & \leq & \max\{1,\beta^{-nh}\ } \big[f^n\big(\big|u_k\sum\limits_{i=0}^k v_id(x_i,\bar{0})\big|\big)\big]^{r_k}\\ \mbox{i.e.}\ ( x_k ) & \in & \lambda^f ( u , v , r)\end{aligned}\ ] ] combining we get @xmath208 . a. in theorem 2.9(iii ) , if @xmath210 be bounded , then the inclusion relation holds for @xmath102 also . b. if @xmath78 be bounded in theorem 2.9(iv ) , then the inclusion relation is true for @xmath102 also . ojha s , srivastava p. d. : some new sets of sequences of fuzzy numbers by using generalized weighted mean , _ proceedings of international conference on mathematical sciences ( icms-2014 ) _ , paper - 04 fuzzy logic , 407 - 410 , ( 2014 ) . talo ozer , basar feyzi : determination of the duals of clasical sets of sequences of fuzzy numbers and related matrix transformations , _ computers and mathematics with applications _ , * 58 * , 717 - 733 , ( 2009 ) .
in this paper , we have introduced a new class of sequences of fuzzy numbers defined by using modulus function and generalized weighted mean over the class defined in @xcite . we have proved that this class form a quasilinear complete metric space under a suitable metric . various inclusion relations and some properties such as solidness , symmetry are investigated . + * keywords :* sequences of fuzzy numbers ; modulus function ; generalized weighted mean . + * ams subject classification :* 46s40 ; 03e72 .
polarization of starlight by aligned interstellar dust grains was discovered nearly half a century ago ( hiltner 1949a , b ; hall 1949 ; hall & mikesell 1949 ) , but the processes responsible for the observed alignment remain uncertain . davis & greenstein ( 1951 ) observed that interstellar grains were expected to be rotating rapidly due to brownian motion , and proposed that these spinning grains could be aligned with the local magnetic field by paramagnetic dissipation . however , further study of the statistical mechanics of grain alignment ( jones & spitzer 1967 ; purcell & spitzer 1971 ) raised questions about the ability of the interstellar magnetic field to achieve the observed degree of alignment , since random gas - grain collisions would tend to oppose the alignment process . purcell ( 1975 , 1979 ) first recognized that interstellar grains were expected to have superthermal rotational velocities as the result of systematic torques , the most important of which appeared to be due to the process of @xmath0 formation on the grain surface . as discussed by purcell , superthermal rotation due to torques which are fixed in body coordinates can enhance the degree of grain alignment , since the `` thermal '' torques due to random collisions with gas atoms now have little effect on the direction of the grain angular momentum , allowing paramagnetic dissipation to inexorably bring the angular momentum into alignment with the galactic field . the systematic torques considered by purcell were due to processes taking place at the grain surface @xmath0 formation , photoelectric emission , and inelastic collisions with gas atoms but the grain surface may be altered due to contamination or erosion on relatively short time scales . the resulting changes in direction of the systematic torque can cause the grain to occasionally undergo short periods when its rotation is `` spun down '' ; during these `` crossover '' episodes the grain may become disaligned ( spitzer & mcglynn 1979 ) . because of this disorientation during `` crossover '' , it was not clear whether ordinary paramagnetic dissipation plus superthermal rotation driven by `` purcell torques '' can account for the observed grain alignment . in addition , dust grains were observed to be aligned in some dense molecular regions where `` purcell torques '' were expected to be ineffective due to a low h / h@xmath7 ratio , attenuation of the ultraviolet radiation required for photoelectric emission , and near - equality of gas and grain temperatures . as a result , there has been renewed interest in alternatives , including the possibility that grains may be superparamagnetic ( jones & spitzer 1967 ; duley 1978 ; martin 1995 ; goodman & whittet 1995 ) , or that grain alignment is due to gas - grain streaming ( gold 1952 ; lazarian 1994 , 1995a ; roberge , hanany , & messinger 1995 ) . lazarian ( 1995b ) has emphasized the possible importance of grain `` helicity '' , since helical grains can be driven to superthermal rotational velocities , and possibly aligned , when exposed to either streaming gas atoms or anisotropic radiation . harwit ( 1970a , b ) suggested that the quantized angular momentum of the photon could lead to rapid rotation of a grain following absorption and emission of many photons , and proposed that the anisotropy of starlight could result in a tendency for interstellar grains to spin with their angular momentum vectors parallel to the galactic plane . dolginov ( 1972 ) observed that interstellar grains might have different absorption and scattering cross sections for left- and right - handed circularly polarized light , so that the grain angular momentum could be changed if illuminated by unpolarized but anisotropic radiation . this effect was further discussed by dolginov & mytrophanov ( 1976 ) for particles in the rayleigh limit , and by dolginov & silantev ( 1976 ) for larger particles but with refractive index close to unity so that the rayleigh - gans approximation could be used . dolginov & mytrophanov noted that this process could lead to both rapid rotation and possible alignment . unfortunately , dolginov and collaborators were unable to calculate the torques on grains with realistic compositions and sizes . that irregular interstellar grains should be subject to radiative torques is not surprising . we consider two macroscopic examples for illustration . [ fig : targs]a shows a four - fold symmetric target with square top and bottom , with each of its four rectangular sides divided into perfectly absorbing and perfectly reflecting halves . in the geometric optics limit the normal component of the radiation pressure force will be twice as large on the reflecting sections as on the absorbing sections ; as a result , an isotropic radiation field illuminating this target will produce a positive torque along axis @xmath8 . however , incident radiation which is either parallel or antiparallel to @xmath8 will not produce any torque on this target . fig . [ fig : targs]b shows a target obtained by starting with the shape of fig.[fig : targs]a and removing four wedges from the top and four from the bottom , with the resulting shape being symmetric under reflection through the centroid . fig.[fig : targs]b is an example of a shape with `` helicity '' . from symmetry it is clear that an isotropic radiation field will produce no torque on this target : whatever torque is exerted on the `` top '' half of the grain will be cancelled by an opposite torque on the `` bottom '' half . however , anisotropic illumination can produce a torque on this target . if , for example , the surfaces are all perfectly reflecting , then incident radiation which is antiparallel to @xmath8 will produce a torque parallel to @xmath8 , while radiation parallel to @xmath8 will produce a torque antiparallel to @xmath8 . these two examples show that macroscopic objects will be subject to radiative torques unless they are highly symmetric ; thus irregular targets should generally be subject to radiative torques . interstellar grains are , of course , comparable to or smaller than the wavelength of the illuminating radiation , but one does not expect the radiative torques to vanish when target geometries such as fig . [ fig : targs ] are reduced to sizes comparable to the wavelength . in this paper we discuss the forces and torques on grains of arbitrary shape and with sizes which are neither large nor small compared to the wavelength of the incident radiation . we show how these forces and torques can be calculated using the discrete dipole approximation . quantitative results are obtained for one particular irregular grain shape . the first part of this paper , [ sec : dda]-[sec : target_orient ] , is devoted to development of the theory of scattering by irregular particles to enable efficient computation of radiative torques . this is carried out within the conceptual ( and computational ) framework of the discrete dipole approximation . in [ sec : dda_results ] we report results of extensive computations for one specific irregular grain geometry and composition , and three different sizes , @xmath9 , 0.05 , and @xmath10 ( @xmath2 is the radius of a sphere of equal volume ) . readers primarily interested in the implications for grain rotation may elect to skip [ sec : dda]-[sec : dda_results ] , and proceed directly to [ sec : interstellar ] , where we discuss the effects of starlight and gas drag under realistic conditions in interstellar diffuse clouds , and [ sec : superthermal ] , where the resulting superthermal rotation is evaluated . it is seen in [ sec : diffuse ] that moderately anisotropic starlight can torque @xmath3 grains in diffuse clouds up to extremely large rotational velocities , exceeding even the superthermal rotation due to @xmath0 formation . smaller ( @xmath11 ) grains , on the other hand , are only weakly affected by radiative torques . we also examine the importance of radiative torques within quiescent dark clouds ( [ sec : dark ] ) , and within active star - forming regions ( [ sec : starforming ] ) . our results are summarized in [ sec : summary ] . our principal result is that the torques exerted on interstellar grains by background starlight are dynamically very important for grains with @xmath5 . these torques will produce extreme superthermal rotation of interstellar grains in both diffuse clouds and star - forming clouds . in a separate paper ( draine & weingartner 1996 ) we examine the role of these radiative torques in the alignment of interstellar grains with the galactic magnetic field . we adopt the `` discrete dipole approximation '' ( draine & flatau 1994 ) , and represent the target by an array of polarizable points , at locations @xmath12 , with electric polarizabilities @xmath13 . the space between the points is vacuum . we consider the response of this array to monochromatic illumination , and adopt the usual complex representation for all time - dependent variables which are first - order ( i.e. , @xmath14 and @xmath15 fields , and dipole moments @xmath16 ) . second - order variables ( i.e. , forces @xmath17 and torques @xmath18 ) will be taken to be purely real . we assume an incident plane wave _ inc = _ inc,0 ( i - i _ 0 t ) _ inc=_inc,0 ( i -i_0 t ) _ inc,0=_inc,0 where the unit vector @xmath19 , and @xmath20 . by suitable choice of complex @xmath21 we can represent general elliptical polarization . the dipole at location @xmath12 acquires a dipole moment _ j(t)=_j(0)e^-i_0 t=_j _ j where @xmath22 is the electric field at @xmath12 due to the incident electromagnetic field and all dipoles except dipole @xmath23 : _ j = _ inc , j + _ sca , j _ inc , j=_inc,0(i_j - i_0 t ) _ sca , j= e^-i_0 t _ lje^ikr_jlr_jl^3 \ { k^2_jl(_l_jl ) + ( 1-ikr_jl)r_jl^2 } , [ eq : esca ] where @xmath24 . similarly , the magnetic field at @xmath12 may be written ( see , e.g. , jackson 1975 ) . _ j = _ inc , j + _ sca , j _ inc , j= _ inc,0 ( i_j - i_0 t ) _ sca , j = e^-i_0 t _ lj k^2e^ikr_jlr_jl^2 ( _ jl_l)(1 - 1ikr_jl ) , [ eq : bsca ] the instantaneous force on point dipole @xmath23 is _ [ eq : force ] the first term in ( [ eq : force ] ) is due to gradients in the local electric field ; the second is the lorentz force on the currents associated with the oscillating dipole moments . direct evaluation of @xmath31 from ( [ eq : fsca ] ) , using ( [ eq : esca ] ) and ( [ eq : bsca ] ) , involves summing over @xmath32 terms , which for @xmath33 becomes computationally prohibitive . operations using fft techniques similar to those used for evaluation of the fields @xmath22 ( goodman , draine , & flatau 1991 ) , but this has not been implemented here . ] alternatively , we may evaluate the net rate of momentum transport to infinity by the scattered radiation field ; let this be denoted @xmath34 . overall momentum conservation requires that the momentum removed from the incident beam either be transferred to the grain or carried off by the outgoing scattered radiation : c_ext|_inc,0|^2= _ rad + _ out . thus we see that @xmath35 : @xmath36 can be thought of as simply the `` recoil '' of the grain due to the scattered radiation . we define a dimensionless radiation pressure efficiency vector @xmath41 such that the radiation pressure force _ rad= _ pra_eff^2|_inc,0|^2 8 , where the effective target radius @xmath42 , where @xmath43 is the volume of target material . the vector @xmath41 depends on both the orientation of the grain and the direction of propagation and polarization of the incident radiation ( e.g. , linearly or circularly polarized ) . it is clear that @xmath44 is just the quantity usually described as the `` radiation pressure efficiency factor '' @xmath45 , where @xmath46 and @xmath47 are the usual absorption and scattering efficiency factors , and @xmath48 is the mean value of @xmath49 for the scattered radiation , where @xmath50 is the scattering angle ( see , e.g. , bohren & huffman 1983 ) . it is obvious that @xmath51 for targets e.g. , spheres which are symmetric under rotation around an axis parallel to @xmath52 . we choose a coordinate system such that the grain center - of - mass is at @xmath53 . the instantaneous radiative torque on the grain is _ rad = _ j=1^n _ j + _ j=1^n ( _ j)(_j ) . as for the force calculation , we separate the radiative torque into `` incident '' and `` scattered '' components : _ rad = _ inc + _ sca ; @xmath54 is the torque exerted on the oscillating dipoles by @xmath55 and @xmath56 , and @xmath57 is the torque exerted on the oscillating dipoles by the scattered radiation field i.e. , the fields @xmath58 and @xmath59 due to the oscillating dipoles . the time - averaged value of @xmath54 is straightforward : _ inc= 12 . as was the case for @xmath36 , direct evaluation of @xmath57 would be time - consuming . by analogy to @xmath36 , we appeal to the fact that the time - averaged torque exerted by the scattered field plus the time - averaged rate of transport of angular momentum to infinity by this field must sum to zero : @xmath60 , where @xmath61 is the net rate of transport of angular momentum by the scattered radiation field across a fixed surface enclosing the target . in dyadic notation , the instantaneous angular momentum flux may be written ( jackson 1975 , problem 6.11 ) = 14 ( ( ) ( ) + ( ) ( ) - |()|^2+|()|^22 ) , where @xmath62 is the unit dyadic . integrating @xmath63 over a spherical surface of radius @xmath64 : _ out=_s r^3d4 . [ eq : gamout ] now , if @xmath65 , the field radiated by the oscillating dipoles has @xmath66 , @xmath67 , @xmath68 , and @xmath69 . thus we need retain only the leading - order terms in evaluating each of these quantities . expanding @xmath58 and @xmath59 as shown in appendix a , letting @xmath70 , _ sca= -_out= -k^48 d ( s_e^*_b + s_b^*_e ) , [ eq : gamout2 ] s_e _ j=1^n _ j(0 ) ( -ik_j ) , s_b _ j=1^n _ j(0)_j ( -ik_j ) , _ e _ j=1^n ( -ik_j ) , @xmath71 noting that for pure right circularly polarized radiation , the time - averaged angular momentum flux is just @xmath72 ( easily seen by noting that each photon carries angular momentum @xmath73 ) , it is convenient to define a dimensionless `` torque efficiency vector '' @xmath74 such that the time - averaged torque _ rad= _ inc+ _ sca= _ a_^2 |_inc,0|^2 8k = a_^2 u_rad 2 _ , where @xmath75 is the time - averaged energy density of the incident radiation . clearly we have _ , inc = 4ka_^2 |_inc,0|^2 , _ , sca = -k^5 a_^2 |_inc,0|^2 d ( s_e^ * _ b + s_b^ * _ the direction and magnitude of @xmath76 depends on the grain orientation , and on the direction of propagation and polarization of the incident radiation . it is clear that for right circularly polarized radiation with @xmath77 , we will have @xmath78 . it is necessary to specify the orientation of the target relative to the incident radiation . let the target contain two axes @xmath8 and @xmath79 ( with @xmath80 ) which are fixed relative to the target . let us further suppose that @xmath8 , @xmath79 , and @xmath81 are the principal axes of the target s moment of inertia tensor , with moments of inertia @xmath82 about these axes . the grain orientation is completely specified by the orientations of any two non - parallel fixed axes , in particular @xmath8 and @xmath79 . we define a `` scattering frame '' , defined by unit vectors @xmath83 , @xmath84 and @xmath85 . three angles are required to specify the target orientation . the orientation of @xmath8 in the scattering frame is described by the two angles @xmath86 $ ] and @xmath87 $ ] , where _ 1 = _ 1 + _ 2 + _ 3 a third angle @xmath88 $ ] describes rotations of the target axis @xmath79 around @xmath8 : _ 2 = -_1 + ( -)_2 + ( + ) _ 3 the target orientation in the scattering frame is illustrated in fig . [ fig : orient ] . we consider one specific grain shape . this shape , shown in fig . [ fig : shape ] , is not intended to be realistic , but is chosen to be ( 1 ) easily described , ( 2 ) asymmetric , and ( 3 ) well - suited to representation by a cubic array of point dipoles . the shape chosen is an assembly of thirteen identical cubes . the coordinates of these cubes ( in units of cube width ) are listed in table [ tab : cubes ] . the first eight cubes are arranged to form a single larger cube ; the remaining five cubes are attached to the central eight cubes . let the eigenvalues @xmath82 of the moment of inertia tensor be written i_j = 25_j m a_^2 = 815_j a_^5 ; where @xmath89 is the solid density , and @xmath13 are geometric factors . a sphere has @xmath90 ; a 2:2:1 rectangular solid ( `` brick '' ) has @xmath91 , @xmath92 . the principal axes @xmath93 and factors @xmath13 for our adopted grain geometry are given in table [ tab : axes ] . let @xmath94 be the grain temperature . purcell ( 1979 ) has shown that if the grain is spinning with angular momentum @xmath95 , and @xmath96 , then viscoelastic damping and barnett effect damping will rapidly bring the grain s principal axis @xmath8 into alignment with @xmath95 . in our discussion below we will therefore assume that @xmath8 is at all times perfectly aligned with the angular momentum @xmath95 of the grain . the grain rotation period will be short compared to all other dynamical time scales , so it is appropriate to average the forces and torques over the grain rotation angle @xmath97 , and henceforth we will discuss only the `` @xmath97-averaged '' efficiency vectors @xmath98 and @xmath99 . we have carried out scattering calculations for grains with the shape shown in fig . [ fig : shape ] , the refractive index of `` astronomical silicate '' ( draine & lee 1984 ) and effective radii @xmath100 , @xmath101 , and @xmath10 . the computations were carried out using the discrete dipole approximation program ddscat.5a . the targets were represented by arrays of @xmath102 , @xmath103 , and @xmath104 dipoles ; in each case , @xmath105 was taken to be large enough to satisfy the validity criterion @xmath106 ( draine & flatau 1994 ) , where @xmath107 is the complex refractive index , and @xmath108 is the interdipole spacing . lattice dispersion relation theory ( draine & goodman 1993 ) was used to determine the optimal choice of dipole polarizabilities at each wavelength . the computations employed fft techniques ( goodman , draine & flatau 1991 ) and a stabilized version of the bi - conjugate - gradients method ( flatau 1996 ) ; nevertheless the computational demands were substantial , because of the need to compute scattering for many different wavelengths and orientations . for example , determining the scattering properties for one orientation and two polarization states for the @xmath3 target at a single wavelength @xmath109 , with the target represented by @xmath110 dipoles , required 720 cpu - sec on a sun-10 computer with a 125mhz hypersparc cpu . for the @xmath3 grain , computations were carried out for 26 distinct wavelengths ( from @xmath111 to @xmath112 ) and 285 orientations ( 15 values of @xmath97 for each of 19 values of @xmath113 ) , requiring @xmath114 cpu - hr . consider a grain spinning around axis @xmath8 , with @xmath115 . in fig . [ fig : qpol_0.2 ] we show q_ext q_pol where @xmath29 is the extinction cross section averaged over grain rotation angle @xmath97 . we see that the polarization cross section for the @xmath3 grain peaks near @xmath116 . grains of approximately this size would therefore be suitable for producing the bulk of the observed interstellar polarization , which tends to peak near @xmath117 . the @xmath118 grain , on the other hand , is most effective at polarizing near @xmath119 ( see fig . [ fig : qpol_0.05 ] ) ; the observed wavelength - dependence of interstellar polarization indicates that grains of this size are _ not _ appreciably aligned in the interstellar medium ( kim & martin 1994b ) . we can restrict our calculations to @xmath120 ; for other values of @xmath121 , the efficiency vectors @xmath122 may be obtained by a simple rotation : ( , ) = ( , 0)_1_1 + ( , 0)_2(_2+_3 ) + ( , 0)_3(_3-_2 ) . let ( @xmath123 , @xmath124 ) and ( @xmath125 , @xmath126 ) be the radiation pressure and radiation torque efficiency vectors for polarization states 1 and 2 ( these can be be any two orthogonal polarization states , e.g. , two linear polarization states , or left- and right - handed circular polarization states ) . for unpolarized incident radiation ( i.e. , equal amounts of two orthogonal polarization states ) we have efficiency vectors @xmath127 henceforth in this paper we will be concerned only with unpolarized radiation , and we will omit the @xmath128 superscript . in fig . [ fig : qpr_kvslambda ] we show our results for the component of @xmath129 along @xmath52 , as a function of @xmath130 , for several values of @xmath113 . this is , of course , just the usual `` radiation pressure efficiency factor '' @xmath131 $ ] . we see the expected tendency for @xmath44 to be small when @xmath132 , and to be of order unity when @xmath133 ( i.e. , @xmath134 for @xmath3 , and @xmath135 for @xmath118 . in fig . [ fig : qgam_0.2_30 ] we show the three components of the radiation torque efficiency vector @xmath76 for the @xmath3 target oriented at @xmath136 . for @xmath137 results are shown for three different dipole arrays , with @xmath102 , 22464 , and 53248 dipoles , with interdipole separations @xmath138 , 114 , and @xmath139 . at @xmath140 we have @xmath141 , so that for the three dipole arrays at this wavelength we have @xmath142 , 0.98 , and 0.74 . we see that the dda calculations appear to have converged when @xmath143 . at the shortest wavelength @xmath144 we have @xmath145 , and for the @xmath146 array we have @xmath147 , marginally satisfying our validity criterion . figs . [ fig : qgam_a1vslambda_0.2 ] and [ fig : qgam_a1vslambda_0.05 ] show , as a function of @xmath130 , the component of @xmath76 along the principal axis @xmath8 about which we have assumed the grains to be spinning , for @xmath3 and @xmath148 grains . this component of @xmath76 will tend to spin up the grain if @xmath149 , or to spin it down if @xmath150 . the average interstellar radiation field ( `` isrf '' ) spectrum in the solar neighborhood has been estimated to be ( mezger , mathis , & panagia 1982 ; mathis , mezger , & panagia 1983 ) @xmath151 where the ultraviolet function @xmath152 the three blackbody components are indicated in table [ tab : radfield ] , where we also provide values of @xmath153 and @xmath154 for each of the four spectral components , where uu_d , . the total starlight energy density is @xmath155 . we expect appreciable anisotropy in the typical interstellar radiation field : generally , there will be more starlight from the direction of the galactic center , the nearest bright star or association may be important , and nearby dust may attenuate the starlight over one part of the sky . we define an anisotropy parameter @xmath156 , such that the net poynting flux is @xmath157 , where @xmath158 is the total energy density of radiation . we assume @xmath159 to be a characteristic value . in our discussions below we will represent the radiation field as an isotropic component with energy density @xmath160 plus a unidirectional component with energy density @xmath161 . for a grain subject to a unidirectional radiation field with energy density @xmath161 , the force and torque on the grain are @xmath162 where @xmath163 now denotes spectral averaging : ( , ) . in table [ tab : grain_props ] we present values of @xmath164 and @xmath164 for @xmath165 , with the spectral averages evaluated for each of the @xmath166 in table [ tab : radfield ] , for three grain sizes , @xmath167 , 0.05 , and @xmath168 . note that the torque efficiencies @xmath169 become very small as the grain size is decreased below @xmath170 . in figs . [ fig : qgam_a1vstheta_0.2 ] [ fig : qgam_alignvstheta_0.05 ] we show the three vector components of @xmath171 , for the @xmath9 and @xmath118 grains ; results are again shown for each of the @xmath166 in table [ tab : radfield ] . consider a grain spinning around axis @xmath8 . an isotropic radiation field will result in a force @xmath172 and torque @xmath173 on the grain which are each fixed in body coordinates . because of the grain rotation , in inertial coordinates the only component of the force or torque which will not average to zero will be the component parallel to the rotation axis @xmath8 : @xmath174 where q^_0^ d _ 1(,0)is the effective efficiency vector for a grain spinning around @xmath8 and exposed to isotropic radiation . values of @xmath175 and @xmath176 are given in table [ tab : grain_props ] . for a grain in neutral gas of hydrogen density @xmath177 and temperature @xmath178 , the gas drag torque on a grain with angular velocity @xmath179 around axis @xmath8 may be written _ drag , gas = -23 ( 1.2)(8m_kt)^1/2 a_^4 _ 1 . [ eq : t_drag ] where @xmath180 is the mass of an h atom . if all impinging atoms `` stick '' and then leave with negligible velocity relative to the local ( moving ) surface , then @xmath181 for a sphere , and @xmath182 for a 2:2:1 brick ; in general we expect @xmath183 . the factor 1.2 in eq . ( [ eq : t_drag ] ) allows for the effects of helium with @xmath184 . ) should be replaced by @xmath185 . ] the rotational damping time is _ drag , gas = _ 1a_3(2m_kt)^1/2 = 8.7410 ^ 4 _ 3 a_-5 t_2 ^ 1/2 ( 3000 ^ -3kt ) where @xmath186 , @xmath187 , and @xmath188 . in addition to collisions with gas atoms , there is rotational damping associated with absorption and emission of photons by the grain ( purcell & spitzer 1971 ; roberge , degraff & flaherty 1993 ) . if we assume that the grain is heated by starlight to a temperature @xmath189 , and that @xmath190 at wavelengths @xmath191 , then the damping time due to thermal emission may be written @xmath192 where @xmath193 denotes the riemann @xmath194-function , q_absu_q_abs()d , and we have assumed @xmath195 in eq.([eq : tau_drag_em ] ) , as expected for simple models ( draine & lee 1984 ) and as appears to be required by the far - infrared emission from dust in diffuse clouds ( draine 1994 ) . is insensitive to the precise value of @xmath97 : the factor @xmath196 decreases by only 22% if @xmath97 is reduced from @xmath195 to 1 . ] in eq . ( [ eq : tau_drag_em ] ) it is assumed that the radiated photons have angular momentum @xmath73 relative to the grain center of mass ; this will be true for radii @xmath197 . there is additional damping associated with absorption of starlight photons , but it is smaller than the damping due to thermal emission by a factor @xmath198 where @xmath199 is the color temperature of the radiation responsible for heating the grain ( purcell 1979 ) , and hence may be neglected . values of @xmath200 are given in table [ tab : grain_props ] . the rotational damping time @xmath201 is given by _ drag^-1=_drag , gas^-1+_drag , em^-1 . [ eq : taudrag ] the thermal rotation rate for a grain may be written _ t^2 = 158_1 kt a_^5 , assuming rotation around @xmath8 with kinetic energy @xmath202 . purcell ( 1979 ) pointed out that @xmath0 formation on the grain surface , taking place only at randomly - located `` active sites '' , would result in a large torque . the component of the torque perpendicular to the rotation axis @xmath8 would average to zero ; for a 2:2:1 rectangular solid purcell estimated the component parallel to the rotation axis _ _ 1 = 13 ( 3)^1/6 f n ( ) ( 2e kt)^1/2 a_^2 l p(t ) where @xmath203 is the density of h atoms in the gas , @xmath204 is the surface area per @xmath0 formation site on the grain surface , @xmath205 is the fraction of arriving h atoms which depart as @xmath0 , @xmath206 is the kinetic energy of the departing @xmath0 molecules , and @xmath207 is a random variable with time averages @xmath208 , @xmath209 , where @xmath210 is the `` lifetime '' of a surface recombination site . if the only torques acting on the grain are those due to gas drag and @xmath0 formation , the grain will attain a rotation rate @xmath211 ; the mean kinetic energy will exceed @xmath212 by a factor ( purcell 1979 ) ( _ _ t)^2=5_1 f^2216 ^ 2 ( n())^2 ekt a_l^2 m _ ( _ drag_drag , gas)^2 ( t_0t_0+_drag ) . in a diffuse cloud where the hydrogen is predominantly atomic , @xmath0 recombination on the grain surface will result in very rapid grain rotation . values of @xmath217 are given in table [ tab : diffuse ] for three grain sizes , and three possible values of the surface recombination site lifetime @xmath210 . a typical diffuse cloud may have @xmath218 ; the average interstellar radiation field ( isrf ) then corresponds to @xmath219 . we adopt an anisotropy factor @xmath220 as representative . as seen from figs . [ fig : qgam_a1vstheta_0.2 ] and [ fig : qgam_a1vstheta_0.05 ] , the spectrum - averaged @xmath221 varies considerably with @xmath113 , the angle between @xmath8 and the radiation flux . values of @xmath222 are given in table [ tab : diffuse ] for @xmath223 and @xmath224 , and for three grain sizes . for the @xmath3 grain at @xmath223 , we find that radiative torques result in extreme superthermal rotation with @xmath225 , greater than the rotation due to @xmath0 formation on the grain surface if @xmath226 . the @xmath118 grain , on the other hand , has a radiatively - driven rotation rate only slightly in excess of thermal , with @xmath227 ( for @xmath223 ) , far smaller than the rotation velocity expected from @xmath0 formation . the @xmath228 grain is essentially unaffected by radiative torques : the rotational kinetic energy resulting from these torques is small compared to @xmath229 . we therefore see that at least for the particular shape considered here radiative torques will drive the larger ( @xmath5 ) grains to highly superthermal rotation rates , while having only a slight effect on the smaller ( @xmath6 ) grains . it has been known for some time that the observed wavelength dependence of polarization requires that if the small grains and large grains have similar shapes , then the degree of alignment declines rapidly for @xmath230 ( mathis 1979 , 1986 ; kim & martin 1994a , b ) . this was somewhat unexpected : the standard davis - greenstein alignment time varies as @xmath231 , so one might naively expect that small grains would exhibit a _ higher _ degree of alignment than large grains . several ideas have been put forward to explain the low degree of alignment of small grains . mathis ( 1986 ) proposed that a grain would not be aligned unless it contained at least one superparamagnetic inclusion , and that small grains were statistically unlikely to contain such an inclusion . mathis obtained a good fit to the average polarization curve by assuming that a grain with @xmath232 had a 50% probability of containing one or more inclusions . lazarian ( 1994 ) argued that if grain alignment was due to the gold mechanism ( gas - grain streaming ) driven by alfvn waves , then only the larger grains ( @xmath5 ) were sufficiently inertial to be aligned . lazarian ( 1995c ) argued that @xmath0 formation on fractal grains would only be able to drive grains with @xmath233 to superthermal rotation . as pointed out by purcell ( 1975 , 1979 ) , superthermal rotation contributes to grain alignment by rendering unimportant the disaligning effects of random collisions with gas atoms . purcell considered three processes which could drive grains to superthermal rotation : @xmath0 formation on the grain surface , photoelectric emission , and variations in the effective accomodation coefficient over the grain surface ; of the three , @xmath0 formation appears to be the most powerful . as discussed in [ sec : superth_h2 ] , the effects of this torque depend on the `` correlation time '' @xmath210 for the torques driving superthermal rotation . if the only torques are due to @xmath0 formation and collisions with gas atoms , the grain will undergo spindown and `` crossover events '' with a frequency @xmath234 , where the time @xmath235 ( purcell 1979 ) . if @xmath236 , then grain alignment by paramagnetic dissipation will be suppressed , whereas if @xmath237 , then paramagnetic dissipation will result in a high degree of alignment . the three processes discussed by purcell depend upon surface properties ; if the grain surface is altered on a relatively short time scale , then substantial grain alignment will not occur unless the grains are superparamagnetic ( in which case @xmath238 is short ) . here we propose another mechanism for selective alignment of the larger grains . the radiative torques on the grain depend upon the global geometry of the grain , and will therefore have a correlation time of order the grain lifetime , @xmath239 . for the @xmath3 grains considered here , @xmath240 , and we may expect long - lived superthermal rotation . by itself , long - lived superthermal rotation would lead to grain alignment by paramagnetic dissipation on the davis - greenstein timescale @xmath1 , as expected for `` purcell torques '' which are fixed in body coordinates . however , when an anisotropic radiation field is present , the radiative torque components perpendicular to @xmath8 may act directly to change the grain alignment on time scales short compared to @xmath1 ; it appears that this may be the mechanism responsible for the observed alignment of @xmath5 dust grains with the interstellar magnetic field ( draine & weingartner 1996 ) . for the @xmath118 grains , on the other hand , the radiative torques are weak compared to the purcell torques due to @xmath0 formation . if the surface site lifetime @xmath241 , there will be frequent crossover events , with a resulting low degree of grain alignment . this may explain the observed preferential alignment of the larger @xmath5 interstellar grains , and the minimal alignment of @xmath6 grains . the dynamics of grain alignment including radiative torques will be examined in more detail in a later paper ( draine & weingartner 1996 ) . the alignment of grains in dense clouds has been problematic . recent studies of polarization of background stars indicate a low degree of grain alignment in the central regions of some dark clouds ( goodman et al . this has been attributed to lack of superthermal rotation in regions where the hydrogen is predominantly molecular , where the grains are shielded from ultraviolet photons capable of producing photoelectric emission , and where the gas and grain temperatures tend to be similar . however , observations of polarized fir emission from m17 , m42 ( orion ) , and w3 ( hildebrand et al . 1995 ; hildebrand 1996 ) shows that the grains responsible for the fir emission in these dark clouds are significantly aligned , and it has not been clear why grains in some dense molecular regions are aligned , while in others they are not . the dark cloud l1755 observed by goodman et al . ( 1995 ) has @xmath242 and @xmath243 , with a peak extinction through the cloud @xmath244 . we suppose the cloud to be externally illuminated by the average interstellar radiation field . at the cloud surface we have @xmath245 and @xmath246 , and at a depth @xmath247mag from the cloud surface , we estimate and is highly forward - scattering , then @xmath248 , @xmath249 , where @xmath250 and @xmath251 are exponential integrals . for @xmath252 , we would have @xmath253 , @xmath254 . ] @xmath255 , @xmath256 ; the radiation will of course be substantially reddened , but we neglect this here . table [ tab : dark ] lists values of @xmath222 for grains at this depth . we see that the weakened radiation field , and the higher gas density , make the radiative torques relatively unimportant in this environment . near the cloud surface , of course , conditions are closer to the diffuse cloud conditions of table [ tab : diffuse ] , and radiative torques will maintain superthermal rotation . thus we expect grain alignment near the cloud surface , but not at depths @xmath257mag . in star - forming clouds , there are obviously sources of starlight within the clouds . to estimate the characteristic starlight energy density to which the fir - emitting dust is exposed , we note that for grains with far - infrared emissivity @xmath258 , the power radiated by the grain @xmath259 , where @xmath189 is the grain temperature . if we assume that the grain is heated by starlight with an energy density @xmath158 , then , since the average interstellar radiation field @xmath260 heats grains to @xmath261 , we find u_rad ( t_d 18)^6 u _ . as an example , we consider one point on the m17 molecular cloud : @xmath262 . the @xmath263 emission at this point has a linear polarization of @xmath264 ( hildebrand et al . 1995 ) , indicating that the grains are appreciably aligned . the emission from the dust has a @xmath265 color temperature of @xmath266 ( gatley et al . 1979 ) ; for a @xmath267 emissivity this color temperature corresponds to a dust temperature @xmath268 . if heated radiatively , these grains must therefore be exposed to a radiation field , the @xmath269 optical depth at this position is @xmath270 , so that the grains must be heated by radiation at @xmath271 . ] with @xmath272 if the radiation heating the grains has a spectrum similar to interstellar starlight ( `` isrf '' ) . while the spectrum no doubt differs from the isrf spectrum , it is not obvious how : the embedded massive stars will tend to be bluer , but the spectrum will be reddened by propagation through the dust . for the present purposes , we will assume the isrf spectrum , with an anisotropy factor @xmath273 . the gas in m17 has @xmath274 ( cf . stutzki & gsten 1990 ) , and presumably has @xmath275 , close to the grain temperature . we use eq.([eq : omegaradoveromegat ] ) to obtain the @xmath214 values given in table [ tab : m17 ] . we see that @xmath3 grains will have @xmath276 ( for @xmath223 ) . evidently the radiation field within a star - forming region like m17 is sufficiently intense to drive extreme superthermal rotation of the @xmath5 grains . in addition to producing superthermal rotation , the components of the radiative torques perpendicular to @xmath8 may bring about the observed grain alignment ( draine & weingartner 1996 ) . the principal results in this paper are as follows : 1 . we show how forces and torques on an irregular grain may be calculated using the discrete dipole approximation . we report numerical results ( in fig . [ fig : qpr_kvslambda ] and table [ tab : grain_props ] ) for @xmath277 , the component of the radiation pressure efficiency vector parallel to the radiative flux for one particular irregular grain geometry , where the grain is assumed to be spinning around its principal axis @xmath8 of largest moment of inertia . we report numerical results for the `` radiation torque efficiency vector '' @xmath278 , ( in figs . [ fig : qgam_a1vslambda_0.2 ] [ fig : qgam_alignvstheta_0.05 ] ) for one irregular grain geometry . the torque depends upon the angle @xmath113 between the incident flux and the grain rotation axis @xmath8 . 4 . in interstellar diffuse clouds , radiative torques will drive @xmath5 grains to extreme superthermal rotation . radiative torques will dominate the torques due to @xmath0 formation for @xmath5 interstellar grains . for these grains , the superthermal rotation is expected to be very long - lived , since it depends upon global properties of the grain , rather than the relatively short - lived surface properties which determine superthermal rotation due to @xmath0 formation , photoelectric emission , or variations in surface accomodation coefficient . 6 . in an isotropic radiation field , the long - lived superthermal rotation of @xmath5 grains due to radiative torques would facilitate alignment of these grains with the galactic magnetic field by the davis - greenstein mechanism of paramagnetic relaxation . 7 . since the rotation of smaller ( @xmath279 ) grains is not dominated by radiative torques , alignment of these grains with the galactic magnetic field would be limited due to random variations in the torque associated with @xmath0 formation on the grain surface . this may explain why @xmath5 grains in diffuse clouds are aligned , while there is evidently minimal alignment of the @xmath6 grains which dominate the extreme ultraviolet extinction . radiative torques appear to be able to drive grains to superthermal rotation in star - forming regions , such as m17 , where the ratio of anisotropic radiation to gas pressure is relatively high , thereby potentially explaining the observed alignment of far - infrared emitting dust in m17 and other star - forming regions with warm dust . in cold dark clouds , on the other hand , radiative torques are unable to drive the grains to superthermal rotation , consistent with the observed lack of aligned grains deep in quiescent dark clouds . in addition to the torque component parallel to the grain rotation axis @xmath8 ( which drives superthermal rotation ) , anisotropy of the radiation field can result in radiative torque components perpendicular to the grain angular momentum ; these components can affect grain alignment directly . this research was supported in part by nsf grant ast-9219283 to btd , and by an nsf graduate research fellowship to jcw . we are grateful to r.h . lupton for the availability of the sm plotting package , and to a. lazarian , w.g . roberge , and l. spitzer for helpful comments . bohren , c.f . , & huffman , d.r . 1983 , _ absorption and scattering of light by small particles _ , ( new york : wiley ) davis , l. , & greenstein , j.l . 1951 , , 114 , 206 dolginov , a.z . 1972 , , 18 , 337 dolginov , a.z . , & mytrophanov , i.g . 1976 , 43 , 291 dolginov , a.z . , & silantev , n.a . 1976 , , 43 , 337 draine , b.t . 1988 , , 333 , 848 draine , b.t . 1994 , in _ the first symposium on the infrared cirrus and diffuse interstellar clouds _ , ed . cutri & w.b . a.s.p . conf . series _ , * 58 * , 227 draine , b.t . , & flatau , p.j . 1994 , _ josa a _ , 11 , 1491 draine , b.t . , & lee , h.m . 1984 , , 285 , 89 draine , b.t . , & weingartner , j.c . 1996 , in preparation duley , w.w . 1978 , , 219 , l129 gatley , i. , becklin , e.e . , sellgren , k. , & werner , m.w . 1979 , , 233 , 575 gold , t. 1952 , , 112 , 215 goodman , a.a . , jones , t.j . , lada , e.a . , & myers , p.c . 1995 , , 448 , 748 goodman , a.a . , & whittet , d.c.b . 1995 , , 455 , l181 goodman , j. , draine , b.t . , & flatau , p.j . optics letters _ , 16 , 1198 hall , j.s . 1949 , , 109 , 166 hall , j.s . , & mikesell , a.h . 1949 , , 54 , 187 harwit , m. 1970a , , 226 , 61 harwit , m. 1970b , bull . , 21 , 204 hildebrand , r.h . , dotson , j.l . , dowell , c.d . , platt , s.r . , schleuning , d. , davidson , j.a . & novak , g. 1995 , in _ airborne astronomy symposium on the galactic ecosystem _ , ed . haas , j.a . davidson , & e.f . erickson , _ a.s.p . , 73 , 97 hildebrand , r.h . 1996 , in _ polarimetry of the interstellar medium _ , eds . roberge & d.c.b . whittet , _ a.s.p . , 97 , 000 hiltner , w.a . 1949a , , 109 , 165 hiltner , w.a . 1949b , , 109 , 471 jackson , j.d . 1975 , _ classical electrodynamics _ , ( new york : wiley ) jones , r.v . , & spitzer , l. 1967 , , 147 , 943 kim , s .- h . , & martin , p.g . 1994a , , 431 , 783 kim , s .- h . , & martin , p.g . 1994b , , 444 , 293 lazarian , a. 1994 , , 268 , 713 lazarian , a. 1995a , , 451 , 660 lazarian , a. 1995b , , 277 , 1235 lazarian , a. 1995c , , 293 , 859 martin , p.g . 1995 , , 445 , l63 mathis , j.s . 1979 , , 232 , 747 mathis , j.s . 1986 , , 308 , 281 mathis , j.s . , mezger , p.g . , & panagia , n. 1983 , , 128 , 212 purcell , e.m . 1975 , in _ the dusty universe _ , ed . field & a.g.w . cameron ( new york : neal watson ) , p. 155 purcell , e.m . 1979 , , 231 , 404 purcell , e.m . , & spitzer , l. 1971 , , 167 , 31 roberge , w.g . , degraff , t.a . , & flaherty , j.e . 1993 , , 418 , 287 roberge , w.g . , hanany , s. , & messinger , d.w . 1995 , , 453 , 238 spitzer , l. , & mcglynn , t.a . 1979 , , 231 , 417 stutzki , j. , & gsten , r. 1990 , , 356 , 513 c c c uv [ eq.([eq : uvfield ] ) ] & @xmath286 & 0.1566@xmath287 , @xmath288 ( @xmath289 ) & @xmath290 & 0.7333@xmath291 , @xmath292 ( @xmath289 ) & @xmath293 & 1.3319@xmath294 , @xmath295 ( @xmath289 ) & @xmath296 & 1.7755isrf & @xmath297 & 1.2021 l c c c @xmath298 & 0.758 & 1.254 & 1.100@xmath299 & 0.0150 & 0.0479 & 0.340@xmath300 & 0.00855 & 0.0234 & 0.163@xmath301 & 0.00664 & 0.0181 & 0.111@xmath302 & 0.0716 & 0.130 & 0.273@xmath303 & @xmath304 & @xmath305 & @xmath306@xmath307 & @xmath308 & @xmath309 & @xmath310@xmath311 & & @xmath312 & @xmath313@xmath314 & & & @xmath315@xmath316 & @xmath317 & @xmath318 & @xmath319@xmath320 & @xmath321 & @xmath322 & @xmath323@xmath324 & @xmath325 & @xmath326 & @xmath327@xmath328 & @xmath329 & @xmath330 & @xmath331@xmath332 & @xmath333 & @xmath334 & @xmath335@xmath336 & @xmath337 & @xmath338 & @xmath339@xmath340 ( @xmath223 ) & 1.141 & 1.860 & 1.438@xmath341 ( @xmath223 ) & 0.0203 & 0.184 & 1.117@xmath342 ( @xmath223 ) & 0.00965 & 0.0454 & 0.651@xmath343 ( @xmath223 ) & 0.00724 & 0.0257 & 0.457@xmath344 ( @xmath223 ) & 0.105 & 0.226 & 0.784@xmath345 ( @xmath223 ) & @xmath346 & @xmath347 & @xmath348@xmath345 ( @xmath349 ) & @xmath350 & @xmath351 & @xmath352@xmath353 ( @xmath223 ) & @xmath354 & @xmath355 & @xmath356@xmath353 ( @xmath349 ) & @xmath357 & @xmath358 & @xmath359@xmath360 ( @xmath223 ) & @xmath361 & @xmath362 & @xmath363@xmath360 ( @xmath349 ) & @xmath364 & @xmath365 & @xmath366@xmath367 ( @xmath223 ) & @xmath368 & @xmath369 & @xmath370@xmath367 ( @xmath349 ) & @xmath371 & @xmath372 & @xmath370@xmath373 ( @xmath223 ) & @xmath374 & @xmath375 & @xmath376@xmath373 ( @xmath349 ) & @xmath377 & @xmath378 & @xmath379 c c c c @xmath380 ( yr ) & @xmath381 & @xmath382 & @xmath383@xmath384 ( yr ) & @xmath385 & @xmath386 & @xmath387@xmath201 ( yr ) & @xmath388 & @xmath389 & @xmath390@xmath391 ( rad @xmath392 ) & @xmath393 & @xmath394 & @xmath395@xmath396 for @xmath397 & @xmath398 & @xmath399 & @xmath400@xmath396 for @xmath401 & @xmath402 & @xmath403 & @xmath404@xmath396 for @xmath405 & @xmath406 & @xmath407 & @xmath408@xmath222 for @xmath223 & @xmath409 & @xmath410 & @xmath411 @xmath222 for @xmath349 & @xmath412 & 11.5 & @xmath413 c c c c @xmath380 ( yr ) & @xmath414 & @xmath415 & @xmath416@xmath384 ( yr ) & @xmath417 & @xmath418 & @xmath419@xmath201 ( yr ) & @xmath414 & @xmath415 & @xmath416@xmath391 ( rad @xmath392 ) & @xmath420 & @xmath421 & @xmath422@xmath396 for @xmath397 & @xmath423 & @xmath424 & @xmath425@xmath396 for @xmath426 & @xmath427 & @xmath428 & @xmath429@xmath222 for @xmath223 & @xmath430 & @xmath431 & @xmath432@xmath222 for @xmath349 & @xmath433 & @xmath434 & @xmath435 c c c c @xmath380 ( yr ) & @xmath436 & @xmath437 & @xmath438@xmath384 ( yr ) & @xmath439 & @xmath440 & @xmath441@xmath201 ( yr ) & @xmath442 & @xmath437 & @xmath438@xmath391 ( rad @xmath392 ) & @xmath443 & @xmath444 & @xmath445@xmath396 for @xmath446 & @xmath447 & @xmath448 & @xmath449@xmath222 for @xmath223 & @xmath450 & @xmath451 & @xmath452@xmath222 for @xmath349 & @xmath453 & @xmath454 & @xmath455
irregular dust grains are subject to radiative torques when irradiated by interstellar starlight . it is shown how these radiative torques may be calculated using the discrete dipole approximation . calculations are carried out for one irregular grain geometry , and three different grain sizes . it is shown that radiative torques can play an important dynamical role in spinup of interstellar dust grains , resulting in rotation rates which may exceed even those expected from @xmath0 formation on the grain surface . because the radiative torque on an interstellar grain is determined by the overall grain geometry rather than merely the condition of the grain surface , the resulting superthermal rotation is expected to be quite long - lived . by itself , long - lived superthermal rotation would permit grain alignment by normal paramagnetic dissipation on the `` davis - greenstein '' timescale @xmath1 . however , radiative torques arising from anisotropy of the starlight background can act directly to alter the grain alignment on times short compared to @xmath1 . radiative torques must therefore play a central role in the process of interstellar grain alignment . the radiative torques depend strongly on the grain size , measured by @xmath2 , the radius of a sphere of equal volume . in diffuse clouds , radiative torques dominate the torques due to @xmath0 formation for @xmath3 grains , but are relatively unimportant for @xmath4 grains . we argue that this may provide a natural explanation for the observation that @xmath5 grains in diffuse clouds are aligned , while there is very little alignment of @xmath6 grains . we show that radiative torques are ineffective at producing superthermal rotation within quiescent dark clouds , but can be very effective in star - forming regions such as the m17 molecular cloud .
according to the quantum theory non - locality of a quantum word manifests itself in a way that is very different from the intuitive classical views . at the very outset of quantum epoch reflections on that fact create a variety of paradoxes beginning from the eistein - podolski - rosen paradox and the famous neither dead nor alive schrdinger cat @xcite . only towards the end of the xx - th century after the advances of technology , when a manipulation with quantum coherency became reality , the pragmatic approach to the problem rises the question of a practical usage of quantum non - locality . time for the realization of quantum communications and creation of the quantum computer has to come @xcite . the difference between quantum and classical correlations has a very transparent mathematical background . one can already see its roots comparing the basic states of classical and quantum computers ; bits and qubits . while an arbitrary @xmath4-bit string can be transformed into other by the so - called `` local transformation '' , acting on its constituent bits , in quantum case this is true for one qubit states only . in other words the action of `` local transformation '' cease to be _ transitive _ for multiqubit systems @xcite . the action of local transformations splits the space of an arbitrary quantum system into the equivalence classes in way that each class is characterized by a different non - local properties @xcite . therefore the problem of classification of non - localities in a system of @xmath4-qubits reduces to the mathematical problem of description of orbits of `` local '' group action on the space of states @xcite . the corresponding orbit space , @xmath5 , is termed as _ `` entanglement space '' _ @xcite . for its characterization the mathematical formalism based on the classical theory of invariants ( c.f . @xcite ) is often applied . in this approach , in order to separate orbits , i.e. , to introduce coordinates on @xmath6 the polynomials in elements of the density matrices , which are invariant under the local transformation , are used . the entanglement space has highly nontrivial geometric and topological structure @xcite . especially , a complexity of @xmath5 is steeply rising with number of qubits growing up . this makes computations very tedious . however , for the lowest , 2-qubits system , the approach based on the classical theory of invariants allows to obtain a series of important algebraic results , clarifying the properties of @xmath7 @xcite . there is one further complication with the description of @xmath5 . in virtue of a physical requirement the density matrices should be positive semi - definite @xcite . therefore , the space of local group action is not a linear space , but represents a certain semi - algebraic variety @xmath8 . this circumstance should be taken into account applying the classical theory of invariants for construction of the orbit space . in the present article this problem is analyzed and detailed solution is given for the case of 2-qubits . with this aim the semi - definiteness of density matrices is formulated explicitly in the form of polynomial inequalities in the scalars of the adjoint action of the group @xmath9 apart from this , the integrity basis for the polynomial ring @xmath3^{\mathrm{su(2)\otimes su(2)}}\,$ ] , that includes the minimal number of elements subject to the above inequalities , will be presented . our plan is as follows . we start , in paragraphs 2 and 3 , with the brief review of necessary notions from quantum mechanics and put them into context suitable for the characterization of entanglement within the classical theory of invariants . further , in the 4-th paragraph the system of polynomial inequalities in the casimir operators of enveloping algebra of @xmath10 that describes the space @xmath8 is derived . regarding to these inequalities , in the last paragraph the integrity basis for the ring @xmath3^{\mathrm{su(2)\otimes su(2)}}\,$ ] is constructed . a generic mixed state of @xmath4-level quantum system is described by @xmath11 complex matrix , the density matrix @xmath12 @xcite , satisfying following conditions corresponds to the so - called _ pure states _ , whose description reduces to the usage of rays in a hilbert space . a mixed state is a mixture of pure states . ] : 1 . _ hermicity _ @xmath13 2 . _ finite trace _ @xmath14 3 . _ positive semi - definiteness _ @xmath15 . mixed states form the subspace @xmath8 , of the space of hermitian @xmath11 matrices . it is instructive , before considering a generic @xmath4-level system , to start with the simplest two - level quantum mechanical model . in quantum theory of information an abstract quantum mechanical model with two levels takes a special place and independently of its physical realization carries the universal name _ qubit_. the qubit state is given by a density matrix that coincides with the standard density matrix of the non - relativistic spin 1/2 : @xmath16 where @xmath17 - set of the pauli matrices - matrices is given below , in paragraph 5 , formulaes ( [ paulimatrices ] ) . ] and @xmath18 is defined as mathematical expectation : @xmath19 in the representation ( [ eq : qbit ] ) requirements ( i. ) and ( ii . ) are taken into account by construction . the condition ( iii . ) restricts the parameter space of mixed states by a unit ball @xmath20 while for the pure states of qubit the expectation @xmath18 lies on the bloch 2-sphere @xmath21 analogously to the qubit the special terminology for @xmath22 - level quantum system , a _ `` qudit '' _ , has been introduced . the generalization of representation ( [ eq : qbit ] ) to the case of qudits reads @xcite : @xmath23 where @xmath24 is @xmath25- dimensional bloch vector . in the expansion ( [ eq : qdit ] ) components of the vector @xmath26 represent the elements of the @xmath27 algebra normalized by conditions @xmath28 @xmath29 is the kronecker symbol . the constants @xmath30 and @xmath31 are the so - called totally symmetric and antisymmetric structure constants of the algebra : @xmath32\lambda_c),\ ] ] where @xmath33=\lambda_a \lambda_b-\lambda_b \lambda_a.\ ] ] as for the case of a qubit , the properties ( i. ) and ( ii . ) of the qudit s density matrix are already taken into account in the decomposition ( [ eq : qdit ] ) . the non - negativity requirement ( iii . ) imposes further , more subtle than ( [ eq : qball ] ) , restrictions . a complete characterization of qudit s bloch vector space , @xmath34 , in an arbitrary dimension , is an open problem . however , some general properties of this space is already known . particularly , it can be shown that @xmath34 is a convex subset of a @xmath35-dimensional unit ball @xmath36 it being know that all pure states are concentrated on its surface . more precisely , qudit s pure states are determined by the equation @xmath37 where @xmath38 from the standpoint of quantum information theory it is mostly interesting to consider states composed of several qubits . according to the quantum theory axiom on composite systems @xcite , the space of states of system , which is obtained by joining two systems @xmath39 and @xmath40 represents a subspace of the tensor product of their individual hilbert spaces @xmath41 and @xmath42 : @xmath43 the definition ( [ eq : comphil ] ) in conjunction with the superposition principle is a source of an existence of correlations in the joint system , which do not have any classical analog . if a mixed state @xmath44 , describing the joint @xmath45 system , admit ( not necessary in unique way ) representation of the form @xmath46 where @xmath47 and @xmath48 are density matrices of subsystems , then this joint state is called _ separable _ @xcite . for such a state correlations between subsystems are classically conceivable . but the states ( [ eq : sep ] ) are far to exhaust all possible states of combined system . the states that can not be written as ( [ eq : sep ] ) , are called _ entangled_. for a pair of @xmath49 and @xmath50-qudits it is useful to represent the density matrix in the so - called fano form @xcite : @xmath51 in ( [ rhobipartite ] ) matrices @xmath52 and @xmath53 are basis elements of the @xmath54 and @xmath55 algebras respectively . the real @xmath56 matrix @xmath57 and @xmath58 becomes clear after performing the partial trace operation @xcite : @xmath59 the vectors @xmath60 a @xmath61 are bloch vectors for subsystems whose states are describing by matrices @xmath62 and @xmath63 , respectively . the entanglement properties of density matrices ( [ rhobipartite ] ) as well as more general multipartite systems admit formulation in terms of invariants of the so - called local groups @xcite . in the next paragraph the corresponding notions will be introduced . on the space of density matrices of @xmath4-level system the group su(n ) acts in adjoint manner @xmath64 if a quantum system is obtained by combining @xmath49-subsystems with @xmath65 levels , the non - local properties of the resulting composite system can be put into a correspondence with a certain decomposition of the unitary operations in ( [ eq : untran ] ) . namely , from all unitary actions we separate the group of so - called _ local unitary transformations _ ( lut ) @xmath66 acting independently on the density matrices of each subsystems @xmath67 two states of composite system connected by the lut transformations ( [ eq : lut ] ) have the same non - local properties . the latter can be changed only by the rest of the unitary actions @xmath68 generating the class of non - local transformations . as it was mentioned in the introduction the action of lut is not transitive . the equivalence of states regarding the action ( [ eq : lut ] ) gives rise to a decomposition of the space of matrices into the equivalence classes ( orbits ) . the union of these classes , i.e. , the orbit space , is customary to call as the `` entanglement space '' @xmath69 the main motivation for studying of @xmath70 is necessity to work out qualitative criteria and quantitative measures for characterization of non - locality in composite systems @xcite . as it was mentioned above , a canonical method for description of the orbit space @xmath70 is the theory of classical invariants @xcite . within this approach , starting from the works by linden and popescu @xcite , series of interesting results , which clarify the mathematical contents of the entanglement phenomenon , has been obtained . a considerable progress was achieved for pure states . as an example , we refer here to the construction of hilbert series for a multipartite systems of qubits @xcite and classification of pure entangled states based on the theory of hyperdeterminants @xcite . analysis of the orbit space for systems in mixed states is much more vague . the generic questions of construction of basis for rings of local invariants for mixed states have been considered in @xcite . with this aim the algorithmic methods of computer algebra were used @xcite , @xcite . according to the theory of invariants @xcite , the ring of polynomial invariants @xmath71^g$ ] , of linear space @xmath72 over the complex numbers @xmath73 , under the action of a group @xmath74 represents the graded algebra @xmath75^g=\bigoplus_{k=1}^\infty a_k,\ ] ] where @xmath76 is the space of homogeneous g - invariant polynomials of degree @xmath77 . the special unitary groups @xmath78 belong to the reductive algebraic groups . their ring is finitely generated @xcite , and @xmath71^g$ ] is cohen - macaulay type @xcite . however , straightforward application of this construction to the problems of quantum entanglement is complicated by the fact that the space @xmath8 , on which the local group ( [ eq : lut ] ) acts is not a linear space . as it was already emphasized in the introduction , density matrices are the positive semi - definite and therefore the space of representations @xmath8 is nonlinear semi - definite algebraic manifold . bellow we suggest the trick how to overcome this difficulty , exemplifying the problem in details for a system of pair of qubits . let start with the construction of the ring @xmath79^{\mathrm{su(2)\otimes su(2)}}$ ] of the adjoint action invariants on the space of @xmath80 hermitian matrices @xmath81 . in order to define the ring @xmath3^{\mathrm{su(2)\otimes su(2)}}$ ] , note that the space of positive - definite matrices @xmath82 is subspace of @xmath83 which is invariant under the action of @xmath84 as we demonstrate below the subset @xmath82 admits representation via the set of polynomial inequalities similar to the given here , in @xcite . ] @xmath85 in three invariants , @xmath86 and @xmath87 of the enveloping algebra of @xmath88 group . from the other side , since @xmath89 are at the same time invariants of @xmath90 , then it is possible to construct in @xmath79^{\mathrm{su(2)\otimes su(2)}}$ ] such a basis that includes these invariants . as result , having this basis and taking into account the inequalities ( [ eq : inset ] ) , we will be able to characterize the ring @xmath3^{\mathrm{su(2)\otimes su(2)}}$ ] completely . according to the consideration given in the subsequent paragraphs a basis of the ring can be chosen in a way that only the primary invariants of degree 2 , 3 and one secondary invariant of degree 4 presented in the ring s hironaka decomposition @xcite are constrained by the above polynomial inequalities ( [ eq : inset ] ) . to succeed in our program of construction of an optimal homogeneous basis for the ring @xmath3^{\mathrm{su(2)\otimes su(2)}}$ ] let us start with the discussion of positive semi - definiteness of density matrices . below the requirement of non - negativity will be formulated in the form of inequalities in invariants of the adjoint action of @xmath78 group on @xmath91 a hermitian operator is positive semi - definite if and only if all its characteristic numbers are non - negative . the condition of non - negativity of hermitian operator can be formulated solely in terms of coefficients of its characteristic equation : @xmath92 the coefficients @xmath93 in ( [ eq : chareq ] ) are given as the sums of principal minors of @xmath77-th order : @xmath94 since the matrix @xmath44 is hermitian all its characteristic numbers @xmath95 are real roots of the characteristic equations ( [ eq : chareq ] ) . when @xmath95 are positive then all @xmath93 being the symmetric polynomials @xmath96 are non - negative real numbers . the inverse statement is true as well ; the non - negativity of the coefficients @xmath93 provides the non - negativity of roots @xmath97 the descartes theorem @xcite : a number of positive roots ( taking into account their multiplicity ) equals to the number of signs changes in the sequence of the coefficients of the polynomial equation , gives proof of this observation ( see e.g. @xcite ) . so , non - negativity of density matrices can be written in the invariant way as condition of non - negativity of the coefficients of its characteristic equation : @xmath98 we give here , for further use , the explicit form of a few first coefficients @xmath99 written in terms of @xmath4-dimensional bloch vector @xmath100 @xcite : @xmath101 besides from the restrictions ( [ eq : nonnegatcoef ] ) there are upper bounds on @xmath93 due to the normalization condition @xmath102 @xmath103 for @xmath104 note , that the equality fulfils for pure states and the maximal values of @xmath93 are achieved for the equal eigenvalues @xmath105 of density matrices . finally , the positive semi - definiteness and normalizability conditions for density matrices of @xmath4-level system can be written as the following set of inequalities @xmath106 coefficients @xmath107 of the characteristic equation are invariants under the adjoint action of @xmath108 group . they are algebraically independent and can be represented via polynomials in the casimir operators of the corresponding enveloping algebra . below , the case @xmath109 related to the system of 2-qubits , is considered in details and inequalities ( [ eq : ineqsyst ] ) are rewritten directly in terms of the casimir operators of the enveloping @xmath110 algebra . the group su(4 ) has three casimir operators whose expressions in terms of the components of 15-dimensional bloch vector @xmath100 ( see the decomposition ( [ eq : qdit ] ) ) can be written as @xmath111 because for an arbitrary 4-level system the coefficients @xmath112 and @xmath113 of characteristic equation of density matrix are expressible via these casimir operators @xmath114 the set ( [ eq : ineqsyst ] ) reduces to the following constraints on @xmath88 invariants @xmath115 in the space spanned by invariants @xmath116 and @xmath117 inequalities ( [ eq : ineqsyst4 ] ) define the bounded domain depicted on the figure [ fig ] . and @xmath118 , width=302 ] consider the density matrix of two qubits parameterized in the fano form : @xmath119\,,\ ] ] where 3-component vectors @xmath120 and @xmath121 are bloch vectors of constituent qubits , @xmath122 - the pauli matrices making up the basis of algebra @xmath123 : @xmath124 the correlation matrix @xmath125 of pair of qubits has 9 elements @xmath126 at first , following our program of the construction of the ring of invariants @xmath127^{\mathrm{su(2)\otimes su(2)}}\,$ ] outlined in paragraph [ sec : progrlut ] , we identify the space of parameters @xmath128 and @xmath129 with @xmath130 ignoring for a moment all restrictions due to the non - negativity of density matrices . apart from this we linearize the adjoint action ( [ eq : untran ] ) of the local group @xmath131 @xmath132 with @xmath133 matrix @xmath134 . so , our preliminary issue is to build up the ring of polynomial invariants of linear action of @xmath135 group on the linear space @xmath136 . note that the linearization ( [ eq : linlut ] ) allows to use the prompt from the molien formula for the generating function of invariants for @xmath137 representations of a compact group @xmath138 @xcite : @xmath139 where the integral is taken over the group @xmath138 with the haar measure @xmath140 the molien function provides information on the polynomial invariants ring s structure . firstly , its formal series in powers of parameter @xmath141 , the so - called hilbert - poincare series : @xmath142\,,\ ] ] points out the dimension , @xmath143 , of the space of homogeneous invariants of degree @xmath144 . secondarily , being a rational function , ( [ eq : molienfunc ] ) admits ( non - uniquely ) for @xmath145 the representation @xmath146 from this form of the molien function on can conclude on the number and order of the primary @xmath147 and secondary @xmath148 invariants of the cohen - macaulay algebra @xmath149^g=\bigoplus_{k=0}^{r } \,j_k\,\mathbb{c}[k_1 , k_2 , \ldots , k_n]\,.\ ] ] as computations show , the molien function for mixed states of two qubits can be written as @xcite : @xmath150 according to the result ( [ eq : molien2qbit ] ) , a basis of ring consists from 10 primary invariants of degree @xmath151 and 15 secondary invariants of degree @xmath152 . more detailed information on invariants dependence on coefficients of the decomposition ( [ eq : rho2qb ] ) can be extracted using the so - called method of many - parametric generating functions @xcite . in our case the many - parametric generating function depends not only on one parameter @xmath141 , but is a function of three arguments , @xmath153 . the contribution from variables @xmath128 and @xmath154 into the molien function now is taken with the weights determined by each independent parameter , @xmath155 and @xmath156 respectively . it is worth to note that the generating function @xmath153 was found already in the middle of 70-th of the last century @xcite , in connection with so - called problem of _ `` missing index '' _ which arose within issue of nuclei spectrum classification . the corresponding mathematical formulation and solution of the problem can be found , e.g. , in @xcite . further , in our presentation , we will mainly follow the article @xcite . consider the space of all polynomials in fifteen variables @xmath157 and @xmath154 @xmath158 . in virtue of the adjoint action of the local group the space of bloch s parameters is decomposed into the irreducible representations of the so(3 ) @xmath159 so(3 ) group . more precisely , the variables @xmath160 and @xmath161 are transformed according to the representations @xmath162 and @xmath163 correspondingly . since the subspace , @xmath164 $ ] , of homogeneous polynomials in variables @xmath165 of degree @xmath166 correspondingly , is invariant under the action @xmath167 all invariants @xmath168 can be classified according to their degrees of homogeneity , i.e. , @xmath169 . consider , following the construction suggested in @xcite , the set of invariants : 1 . 3 invariants of second degree @xmath170 2 . 2 invariants of third degree @xmath171 3 . 4 invariants of fourth degree @xmath172 4 . 1 invariant of fifth degree @xmath173 5 . 4 invariants of sixth degree @xmath174 6 . 2 invariants of seventh degree @xmath175 7 . 2 invariants of eighth degree @xmath176 8 . 2 invariants of ninth degree @xmath177 from these invariants the basis of @xmath3^{\mathrm{su(2)\otimes su(2)}}\,$ ] can be build . as the criterion of its construction we choose the principle of usage of basis with the minimal number of elements involved in the definition of @xmath91 having in mind this rule and noting that the space @xmath8 is defined in terms of the casimir operators of @xmath178 group ( [ c2casimir])-([c4casimir ] ) , we expand @xmath179 over the set of above introduced local invariants ( [ eq:2order])-([eq:4order ] ) : @xmath180\ , . \label{eq : kq3}\end{aligned}\ ] ] from equations ( [ eq : kq1])-([eq : kq3 ] ) it follows that one can consider the casimir operators @xmath89 as the basis elements instead of scalars @xmath181 and @xmath182 . bear in mind this observation and using the results of @xcite , where the ring @xmath183^{\mathrm{su(2)\otimes su(2)}}\,$ ] was described , we define the following set consisting from 10 _ primary invariants _ , including the casimir operators @xmath116 , @xmath184 and 15 _ secondary invariants _ including the casimir @xmath185 @xmath186 we conclude that the set of homogeneous invariants ( [ eq : primary])-([eq : secondary ] ) represents the basis for the ring @xmath127^{\mathrm{su(2)\otimes su(2)}}$ ] : @xmath187^{\mathrm{su(2)\otimes su(2)}}=\bigoplus_{k=0}^{15 } \,j_k\,\mathbb{c}[k_1 , k_2 , \ldots , k_{10}]\,,\ ] ] under the condition , that two primary invariants @xmath188 and one secondary invariant @xmath189 satisfy the inequalities ( [ eq : ineqsyst4 ] ) . an essential issue of the quantum theory of information is qualitative and quantitative characterization of purely quantum correlations caused by the entanglement of quantum states . theory of classical invariants provides tools for studies of the corresponding space of entanglement , i.e. , the orbit space of action of the group of local transformations on the space of states of composite systems . for the case we are interesting in , system of two qubits in a mixed state , the local transformations of the density matrices form the @xmath190 group . its adjoint action , on the space of the hermitian , unit trace matrices , identified with @xmath130 defines the principal orbit space @xmath191 with dimension @xmath192 however , the orbit space defined in such a way is not the space of entanglement @xmath193 due to the non - negativity of density matrices the space of physical states is @xmath194 in the present article we suggest the description of @xmath195 based on the polynomial inequalities in casimir operators of the enveloping algebra @xmath196 furthermore , we show how these restrictions can be effectively taken into account constructing the basis for the ring @xmath3^{\mathrm{su(2)\otimes su(2)}}\,,$ ] provided for the hironaka decomposition with only two primary invariants of degree 2 , 3 and one secondary invariant of degree 4 constrained by the polynomial inequalities ( [ eq : ineqsyst4 ] ) . concluding it is important to emphasize that without the inequalities ( [ eq : ineqsyst4 ] ) , the usage of local invariants for `` coordinatization '' of the space of entanglement @xmath197 is not correct . we leave for a future publications analysis of those constraints consequences on the geometry of @xmath198 . this work was supported in part the georgian national science foundation research grant gnsf / st08/4 - 405 and by the russian foundation for basic research ( grant no . 10 - 01 - 00200 ) and by the ministry of education and science of the russian federation ( grant no . 3810.2010.2 ) . 99 a. einstein , b. podolsky and n. rosen , can quantum - mechanical description of physical reality be considered complete ? * 47 * ( 1935 ) , 777780 . j. bell . _ speakable and unspeakable in quantum mechanics , _ cambridge university press , cambridge , 1987 . a. aspect . bell s theorem : the naiv view of an experimentalist . in : _ quantum [ un]speakables : from bell to quantum information _ , r.a . bertlmann and a. zeilinger ( eds . ) , springer verlag , berlin , 2002 . nielsen and i.l . chuang,_quantum computation and quantum information _ , camridge university press , 2000 . v.vedral , _ introduction to quantum information science _ , oxford university press , new york , 2006 . i. bengtsson and k. yczkowski , _ geometry of quantum states . an introduction to quantum entanglement , _ cambridge university press , 2006 . werner , quantum states with einstein - podolski - rosen correlations admitting a hidden - variable model , phys . rev . a * 40 * ( 1989 ) , 4277 . j. schlienz and g. mahler , description of entanglement , phys.rev . a * 52 * ( 1995 ) , 43964404 . n. linden and s. popescu , on multi - particle entanglement , fortschr . phys . * 46 * ( 1998 ) , 567578 . h. weyl , _ the classical groups : their invariants and representations _ , princeton university press , 1939 . v.l . popov and e.b . vinberg , _ invariant theory _ , in algebraic geometry iv , encycl . sci . , vol.55 , springer - verlag , 123 - 273 , 1994 . m. kus and k. yczkowski , geometry of entangled states , phys . a * 63 * ( 2001 ) , 032307 [ 13 pages ] . m. grassl , m. rtteler , and t. beth , computing local invariants of qubit systems , phys . rev . a 58 ( 1998 ) , 18531856 . king , t.a . welsh and p.d . jarvis , the mixed two - qubit system and the structure of its ring of local invariants , j. phys . a : math . theor . * 40 * ( 2007 ) , 1008310108 . j. von neumann , warscheinlichtkeitstheoritischer aufbau der quantemechanik , gttingen nachrichten , ( 1927 ) , 245272 . landau , das dmpfungsproblem in der wellenmechanik , zeitschrift fr physik , * 45 * ( 1927 ) , 430441 . k. blum , _ density matrix . theory and applications _ , plenum press , new york , 1981 . hioe and j.h . eberly , @xmath199level coherence vector and higher conservation laws in quantum optics and quantum mechanics . lett . * 47 * ( 1981 ) , 838841 . u. fano , description of states in quantum mechanics by density matrix and operator techniques . * 29 * ( 1957 ) , 7493 . u. fano , pairs of two - level systems . * 55 * ( 1983 ) , 855874 . luque and j .- y . thibon , the polynomial invariants of four qubits . a * 63 * ( 2003 ) , 042303 [ 5 pages ] . + j .- g . luque and j .- y . thibon , algebraic invariants of five qubits . j. phys . a : math * 39 * ( 2005 ) , 371377 . a. miyake , classification of multipartite entangled states by multidimensional determinants , phys . a * 67 * ( 2003 ) , 012108 [ 10 pages ] . b. sturmfels , _ algorithms in invariant theory_. springer - verlag , wien , 1993 . h. derksen and g. kemper , _ computational invariant theory _ , encyclopedia of mathematical sciences , vol . 130 , springer - verlag , berlin , 2002 . b. buchberger , grbner bases an algorithmic method in polynomial ideal theory , in : n.k . bose ( ed . ) : _ multidimensional systems theory _ , d. reidel , dordrecht , ( 1985 ) , 184232 . m. hochster and j. roberts , rings of invariants of reductive groups acting on regular rings are cohen - macaulay , advances in mathematics * 13 * ( 1974 ) , 125175 . g kimura , the bloch vector for n - level systems , phys . a * 314 * , ( 2003 ) , 339349 , m.s . byrd and n. khaneja , characterization of the positivity of the density matrix in terms of the coherence vector representation , phys . a*68 * ( 2003 ) , 062322 [ 13 pages ] . s. kryszewski and m. zachcial , positivity of the n@xmath200 n density matrix expressed in terms of polarization operators , j.phys.a : math . theor . * 39 * ( 2006 ) , 59215931 . vinberg , _ a course in algebra _ , ( graduate studies in mathematics , vol . 56 ) ams 2003 . judd , w. miller jr . , j. patera and p. winternitz , complete set of commuting operators and o(3 ) scalars in the enveloping algebra of su(3 ) , j. math . phys . , * 15 * ( 1974 ) , 17871799 . c. quesne , su(2)@xmath159 su(2 ) scalars in the enveloping algebra of su(4 ) , j. math . * 17 * ( 1976 ) , 14521467 .
the entanglement characteristics of two qubits are encoded in the invariants of the adjoint action of @xmath0 group on the space of density matrices @xmath1 defined as the space of @xmath2 positive semi - definite hermitian matrices . the corresponding ring @xmath3^{\mathrm{su(2)\otimes su(2)}}\,$ ] of polynomial invariants is studied . the special integrity basis for the ring @xmath3^{\mathrm{su(2)\otimes su(2)}}\,$ ] , is described and constraints on its elements due to the positive semi - definiteness of density matrices are given explicitly in the form of polynomial inequalities . the suggested basis is characterized by the property that only a minimal number of invariants , namely two primary invariants of degree 2 , 3 and one secondary invariant of degree 4 appearing in the hironaka decomposition of @xmath3^{\mathrm{su(2)\otimes su(2)}}\,,$ ] are subject to the polynomial inequalities .
dwarf irregulars galaxies ( dig ) and blue compact galaxies ( bcg ) are very interesting objects to study galaxy evolution , for they are easier to model because of their generally small sizes and simple structures . their stellar populations appear to be mostly young , their metallicity is low and their gas content is large . all these features indicate that these galaxies are poorly evolved objects , and may have undergone discontinuous or gasping star formation activity . in particular , bcg are the most luminous among dwarf irregular galaxies : their very low metal content , large gas fraction and very blue colours , suggest that star formation may have proceeded in short and intense bursts of activity ( searle et al . 1978 ) . on the other hand , dig seem to show a ` gasping ' activity of star formation ( tosi et al . 1992 ) . in the last few years a great deal of theoretical models for dig and bcg appeared in the literature . by means of analytical models , matteucci and chiosi ( 1983 ) first studied the effects of galactic winds on the chemical evolution of these galaxies , and suggested , among other hypotheses , that galactic winds of variable intensity could explain the observed spread in the physical properties of these galaxies . in the following years , by means of numerical models , matteucci and tosi ( 1985 ) , pilyugin ( 1992 , 1993 ) and marconi et al . ( 1994 ) , studied the effects of galactic winds on the chemical evolution of such galaxies and confirmed the previous results . in these papers the possibility of metal enriched winds was also considered and favoured relative to the normal galactic winds . this metal enriched wind hypothesis , in fact , allowed them to explain the observed he abundances versus metal abundances . however , none of these studies had taken into account either the energy injected by supernovae and stellar winds into the ism or the presence of dark matter halos which are of fundamental importance for the development of a galactic wind . the galactic winds were just assumed to occur every time the starburst was active . dark matter in dwarf irregular galaxies seems to be quite important as shown by recent data indicating an increasing amount of dark matter with decreasing luminosity ( skillman 1996 ) , and therefore it can not be ignored in chemical evolution models . kumai and tosa ( 1992 ) also explored the possibility of dark matter in these galaxies in a very simple way , and suggested that various amounts of dark matter could explain the observed spread in the @xmath0 relation . the energetics of supernovae and stellar winds and their interaction with the ism , is a difficult problem , still not quite understood . the only existing models of chemical evolution taking into account the stellar energetics relative to the occurrence of galactic winds , are those developed to study the chemical and photometric evolution of elliptical galaxies ( arimoto and yoshii 1987 ; matteucci and tornamb 1987 ; angeletti and giannone 1990 ; ferrini and poggianti 1994 ; bressan et al . 1994 ; gibson 1994 , 1997 ) . in these models , some specific assumptions relative to the amount of energy transferred into the ism are made . gibson ( 1994 ) has shown that the energetics from stellar winds is negligible in massive ellipticals whereas it could be important in smaller ones . for this reason it seems worthwhile to explore the effects of both supernovae and stellar winds on the evolution of dwarf irregulars and blue compact galaxies . the aim of this paper is to take into account both the stellar energetics and the presence of dark matter in modelling the chemical evolution of dwarf irregular galaxies . the development of a galactic wind will therefore be calculated in detail and the existence of metal enriched or differential winds will be taken into account . by differential wind we intend a wind carrying out only some heavy elements , in particular those produced by type ii supernovae which are the predominant supernovae during a starburst ( marconi et al . 1994 ) and eject material at much higher velocities than normal stellar winds . therefore , under this assumption elements such as nitrogen and helium , which are restored by low and intermediate mass stars through stellar winds , will not leave the star forming region whereas oxygen and the other @xmath1-elements , ejected during type ii supernova explosions are likely to be ejected outside the region and perhaps outside the galaxy . iron will only in part leave the star forming region since only a fraction of this element is produced by sne ii , whereas the bulk of it comes from sne ia . it should be again said that the differential wind assumption has proven to be the most viable solution to the helium problem in blue compact galaxies ( see pilyugin 1993 and marconi et al . 1994 ) . the paper is organized as follows : in section 2 . we will present the observational data concerning dwarf irregulars and blue compact galaxies , in section 3 . we will describe the chemical evolution model and the nucleosynthesis prescriptions , in section 4 . we will give our model results and finally in section 5 . we will draw some conclusions . the data we refer to are those presented in marconi et al . ( 1994 ) ( mainly pagel et al . 1992 and campbell 1992 ) and in skillmann ( 1996 ) ( mainly thuan 1995 and garnett 1990 ) . thuan at al . ( 1995 ) presented high quality spectrophotometric observations of 15 supergiant hii regions in 14 bcg . we compare our model results with their data relative to n , o , and fe . their data showed that none of the ratios n / o , fe / o , ne / o , ar / o , s / o , depends on the oxygen abundance indicating that abundance ratios in these galaxies show the pollution only from massive stars and that all the elements , including nitrogen , should have a primary origin ( see paragraph 3.1 ) . garnett et al . ( 1995a ) presented uv observations of 7 hii regions in low - luminosity dwarf irregular galaxies and the magellanic clouds ( hst data ) . this allowed them to measure the carbon abundances and discuss the evolution of the c / o ratio in these galaxies . they found that the c / o ratio increases continuously with increasing o / h and suggested that this can be due to the fact that in the most metal poor galaxies , massive star nucleosynthesis predominates ( oxygen production ) , while in more metallic objects the delayed contribution of intermediate mass stars to carbon is more evident . however , their comparison with chemical evolution models is not really appropriate since they have adopted , for such comparison , models relative to the solar neighbourhood , where the star formation mechanism and evolution are definitively different from those of dwarf irregular galaxies . to calculate the chemical evolution of the star forming region of dig and bcg , we used an improved version of the model presented in marconi et al . the main features of our chemical evolution model are the following : 1 . one - zone , with instantaneous and complete mixing of gas inside this zone ; 2 . no instantaneous recycling approximation , i.e. the stellar lifetimes are taken into account ; 3 . the evolution of several chemical elements ( he , c , n , o , fe ) due to stellar nucleosynthesis , stellar mass ejection , galactic wind powered by supernovae and stellar wind and infall of primordial gas , is followed in details . if @xmath2 is the fractional mass of the element _ i _ in the gas , its evolution is given by the equations : @xmath3 where @xmath4 is the gas mass in the form of an element _ i _ normalized to a total luminous mass @xmath5 . + the quantity @xmath6 represents the abundance by mass of an element _ i _ and by definition the summation over all the elements present in the gas mixture is equal to unity . the quantity @xmath7 is the total fractional mass of gas . the star formation rate we assume during a burst , @xmath8 , is defined as : @xmath9 where @xmath10 is the star formation efficiency ( expressed in units of @xmath11 ) , and represents the inverse of the timescale of star formation , namely the timescale necessary to consume all the gas in the star forming region . more precisely , @xmath10 can be expressed as @xmath12 where @xmath13 is the real star formation efficiency , and @xmath14 is a parameter taking into account the stochastic nature of the star formation processes ( gerola et al . 1980 ) as shown in matteucci and chiosi ( 1983 ) , where a detailed description can be found . the rate of gas loss via galactic winds for each element is assumed to be simply proportional to the amount of gas present at the time @xmath15 : @xmath16 where @xmath17 is the abundance of the element _ i _ in the wind and in the interstellar medium ( ism ) ; @xmath18 is a free parameter describing the efficiency of the galactic wind and expressed in @xmath11 . we have considered normal and differential winds : in the first case , all the elements are lost at the same efficiency ( @xmath19 ) . in the second case , the value of @xmath18 has been assumed to be different for different elements : in particular , the assumption has been made that only the elements produced by type ii sne ( mostly @xmath1-elements and one third of the iron ) can escape the star forming region . we have made this choice following the conclusions of marconi et al . ( 1994 ) , who showed that models with differential winds can better explain the observational constraints of blue compact galaxies in general . the conditions for the onset of the galactic wind are studied in detail and described in paragraph @xmath20 . the chemical evolution equations include also an accretion term : @xmath21 where @xmath22 is the abundance of the element _ i _ in the infalling gas , assumed to be primordial , @xmath23 is the time scale of mass accretion @xmath24 is the galactic lifetime , and @xmath25 is a constant obtained by imposing to reproduce @xmath26 at the present time @xmath24 . the parameter @xmath23 has been assumed to be the same for all dwarf irregulars and short enough to avoid unlikely high infall rates at the present time ( @xmath27 years ) . finally , the initial mass function ( imf ) by mass , @xmath28 , is expressed as a power law : @xmath29 we considered two cases for the imf : 1 . the exponent @xmath30 over the mass range @xmath31 ( salpeter 1955 imf ) , and 2 . @xmath30 over the mass range @xmath32 , and @xmath33 over the mass range @xmath34 ( scalo 1986 imf ) . @xmath35 is the normalisation constant , obtained with the following condition : @xmath36 the @xmath37 term of the equation ( 1 ) represents the stellar contribution to the enrichment of the ism , i.e. the rate at which the element _ i _ is restored into the ism from a stellar generation ( see marconi et al . this term contain all the nucleosynthesis prescriptions : 1 . for low and intermediate mass stars ( @xmath38 ) we have used renzini and voli s ( 1981 ) nucleosynthesis calculations for a value of the mass loss parameter @xmath39 ( reimers 1975 ) , and the mixing length @xmath40 . the standard value for @xmath41 is 8 @xmath42 ; 2 . for massive stars ( @xmath43 ) we have used woosley s ( 1987 ) nucleosynthesis computations but adopting the relationship between the initial mass @xmath44 and the he - core mass @xmath45 , from maeder ( 1989 ) . it is worth noting that the adopted @xmath46 ) relationship does not substantially differ from the original relationship given by arnett ( 1978 ) and from the new one by maeder ( 1992 ) based on models with overshooting and @xmath47 . these new models show instead a very different behaviour of @xmath46 ) for stars more massive than 25 @xmath42 and @xmath48 , but our galaxies never reach such a high metallicity ; 3 . for the explosive nucleosynthesis products , we have adopted the prescriptions by nomoto et al . ( 1986 ) and thielmann et al . ( 1993 ) , model w7 , for type ia sne , which we assume to originate from c - o white dwarfs in binary systems ( see again marconi et al . 1994 for details ) . as already said , nitrogen is a key element to understand the evolution of galaxies with few star forming events since it needs relatively long timescales as well as a relatively high underlying metallicity to be produced . the reason is that n is believed to be mostly a secondary element ( secondary elements are those synthesized from metals originally present in the star and not produced _ in situ _ , while primary elements are those synthesized directly from h and he ) . in fact , n is secondary in massive stars , and mostly secondary and probably partly primary in low and intermediate mass stars ( renzini and voli 1981 ) . however , some doubts exist at the moment on the amount of primary nitrogen which can be produced in intermediate mass stars due to the uncertanties related to the occurrence of the third _ dredge - up _ in asymptotic giant branch stars ( agb ) . in fact , if blcker and schoenberner ( 1991 ) calculations are correct , the third _ dredge - up _ in massive agb stars should not occur and therefore the amount of primary n produced in agb stars should be strongly reduced ( renzini , private communication ) . as a consequence , the only way left to produce a reasonable quantity of n during a short burst ( no longer than 20 @xmath49 ) is to require that massive stars produce a substantial amount of primary nitrogen . this claim was already made by matteucci ( 1987 ) in order to explain the [ n / o ] abundances in the solar neighbourhood . recently , woosley et al . ( 1994 ) have shown that massive stars can indeed produce primary nitrogen , and marconi et al . ( 1994 ) and kunth et al . ( 1995 ) have taken into account this possibility . in this paper we also have considered this possibility and , since the woosley et al . data are not yet available we have used the same parametrization adopted by marconi et al . ( 1994 ) and kunth et al . ( 1995 ) . our model presents a new formulation of galactic winds relative to previous published models of this type ( matteucci and tosi 1985 and marconi et al . in particular , we adopted the prescriptions developed for elliptical galaxies ( matteucci and tornamb 1987 ; gibson 1994 , 1997 ) : galaxies develop galactic winds when the gas thermal energy @xmath50 , exceeds its binding energy @xmath51 , i.e. when : @xmath52 a detailed treatment of the energetics of the ism is considered , in order to compute the gas thermal energy : @xmath53 we calculated the energy fraction deposited in the gas by stellar winds from massive stars @xmath54 , and by supernova explosions @xmath55 . here the supernova contribution is given by : @xmath56 where : @xmath57 @xmath58 @xmath59 @xmath60 and @xmath61 are the rates of supernova ( ii and ia ) explosion , @xmath28 is the imf and @xmath8 is the star formation rate . the type ia and ii sn rates are calculated according to matteucci and greggio ( 1986 ) . the energy injected and effectively thermalized into the ism from supernova explosions ( @xmath62 ) and stellar winds from massive stars ( @xmath63 ) , are given by the following formulas : @xmath64 @xmath65 where @xmath66 is the total energy released by a supernova explosion , @xmath67 is the energy injected into the ism by a typical massive star through stellar winds during all its lifetime , and @xmath68 and @xmath69 are the efficiencies of energy transfer from supernova and stellar winds into the ism , respectively . we adopted a formulation for @xmath62 and @xmath70 which is described in detail in appendix . our formulation refers to an ideal case characterized by an uniform ism , and no interaction with other supernova explosions or interstellar clouds . when a supernova explodes the stellar material is violently ejected into the ism and its expansion is gradually slowed down by the ism . the energy released by the supernova explosion is assumed to be @xmath71 , while the energy effectively transferred and thermalized into the ism is given by eq . ( 13 ) and depends on the assumed value for @xmath68 . the main parameters used to derive @xmath68 , as described in the appendix , are : the initial blast wave energy @xmath72 , the interstellar gas density @xmath73 , and the isothermal sound speed in the ism @xmath74 . for typical values of these parameters @xmath75 , @xmath76 and @xmath77 we obtained the following range of possible values for @xmath68 : @xmath78 in particular , we have adopted an intermediate value of @xmath79 which we assume to be the typical sn energy transfer efficiency . the energy injected into the ism by a typical massive star through stellar winds during all her lifetime is estimated to be : @xmath80 where @xmath81 is the time the star spends in main sequence and @xmath82 is the stellar wind luminosity . by using the arguments developed in the appendix , we obtained : @xmath83 if we assume a 20@xmath42 star as a typical massive star contributing to the stellar winds then @xmath84 , @xmath85 with @xmath76 and @xmath86 , we obtain : @xmath87 therefore , in our formulation we adopted : @xmath88 and @xmath89 it is worth noting that the assumed value for @xmath67 is in agreement with that calculated by gibson ( 1994 ) for a star of initial mass @xmath90 . the introduction of dark matter halos with variable amounts and concentrations of dark matter is considered when we compute the binding energy of interstellar gas , @xmath51 : @xmath91 the two terms on the right of the equation take in account the gravitational interaction between the gas mass @xmath92 , and the total luminous mass of the galaxy @xmath93 , and between the gas mass and the dark matter @xmath94 : @xmath95 @xmath96 where @xmath97 $ ] . + @xmath98 is the gravitational constant , and @xmath99 is the ratio between the effective radius of luminous matter and the effective radius of dark matter . these equations are taken from bertin et al . ( 1992 ) and are valid for @xmath100 defined in the range @xmath101 . + it is worth noting that the original formulation of bertin et al . ( 1992 ) was thought for massive elliptical galaxies , and that it is not necessarily the right one for dwarf irregulars . however , we used such a formulation since theoretical formulations for the binding energy of dwarf irregulars are not available . our study of the evolution of bcg and dig starts from the results obtained by marconi et al . ( 1994 ) : _ i ) _ the star formation is assumed to proceed in short and intense bursts of activity , which may induce galactic winds ; _ ii ) _ metal enriched and normal winds have been considered . in particular , for the metal enriched winds the assumption is made that galactic winds carry mostly the nucleosynthesis products of supernovae of type ii , with the consequence of removing elements such as oxygen but not elements such as helium and nitrogen which are mainly produced in low and intermediate mass stars , and ejected through stellar winds . the novelty is that _ i ) _ galactic winds are powered both by supernova explosions ( snii and snia ) and stellar winds from massive stars , and that _ ii ) _ we consider the presence of dark matter halos in these galaxies . in order to understand the observed distribution of n / o , c / o _ versus _ o / h , and of [ o / fe ] _ versus _ [ fe / h ] , we computed different galaxy models by varying some parameters such as the number of bursts , the duration of each burst , the star formation efficiency , the galactic wind efficiency , the dark matter mass and distribution , and finally , the imf exponent . we first considered three sets of _ standard models _ characterized by 1 , 3 , 5 , 7 and 10 bursts of a duration of 20 , 60 and 100 @xmath49 , whereas the other parameters were fixed by the results presented in marconi et al . in particular , the imf is the salpeter one , and the star formation efficiency is @xmath102 . the galactic wind is differential , namely @xmath18 is different from zero only for the @xmath1-elements ejected through type ii supernova explosions . in particular , for the elements studied here , @xmath103 is zero for n and he , whereas it is @xmath104 for o , ne , na , mg and si , is @xmath105 for c , since carbon is partly produced and ejected by low and intermediate mass stars through stellar winds , and is @xmath106 for fe since only @xmath107 of this element is produced by sne ii . the ratio between dark and luminous matter is assumed to be 10 and that between the luminous and dark core radius is @xmath108 for the three sets of standard models . normal galactic winds instead require @xmath109 to be the same for all the chemical elements . from now on we will indicate the models with differential winds by a capital @xmath110 , ( e.g. @xmath111 indicating the value for the @xmath1-elements ) . among our standard models , those with burst duration @xmath112 never developed galactic winds , and those with @xmath113 predicted the destruction of the galaxy ( i.e. when the thermal energy of gas is larger than the binding energy of the galaxy ) when a large number of bursts was assumed ( in particular when @xmath114 ) ; on the contrary models with @xmath115 never predicted galaxy destruction and developed galactic winds for @xmath116 . as a consequence of this , we chose to examine only models with @xmath115 . we started then from our standard models and varied several parameters such as the star formation efficiency , the wind efficiency , the amount and distribution of dark matter . we found that in order to reproduce the observed spread of @xmath117 and @xmath118 and @xmath119\,\ , versus \,\ , [ fe / h]$ ] , the most important parameter is the star formation efficiency . we computed two series of models characterized by two different imf : the salpeter ( 1955 ) and the scalo ( 1986 ) function . to reproduce the observed spread in all three diagrams , we needed a star formation efficiency ( @xmath10 ) between 0.1 and 7 @xmath11 when using the salpeter imf , and between 1 and 10 @xmath120 when using the scalo imf . the best value for the wind parameter was @xmath121 . consider first the salpeter imf case : we computed models with 1 , 2 , 3 , 5 and 7 bursts , and tested the importance of the presence of the dark matter . greater is the amount of dark matter , deeper is the galaxy potential and higher the gas binding energy . therefore , the thermal energy required to develop the galactic wind has also to be higher . we found that if the amount of dark matter is the same as that of the luminous matter ( @xmath122 ) any model predicts galaxy destruction during the first starburst if the star formation efficiency is high ( @xmath123 ) . this upper limit of @xmath10 depends of course on the number of bursts : models with many bursts , are characterized by a great consume of gas through star formation activity , so they must have lower star formation rates to keep bound , than models with just one or two bursts . we previously said that we need @xmath124 to explain the observed spread of the abundances ratios . so , let s consider the presence of a greater amount of dark matter : ten times the luminous matter ( @xmath125 ) . in this case none of the models characterized by 1 or 2 bursts and @xmath126 predicts galaxy destruction during starbursts , whereas those with 3 or more bursts sometimes do . considering an even greater amount of dark matter , fifty times the luminous matter ( @xmath127 ) , none of the models predicts galaxy destruction for any value of the star formation efficiency @xmath128 . in table [ t1 ] we summarize the range of variation of the star formation efficiency and the occurrence of the galactic wind for different values of the number of bursts and different amounts of dark matter for stable models ( i.e. the galaxy never blows up ) . in column 1 is presented the number of bursts ( @xmath129 1 , 2 , 3 , 5 , 7 ) , in column 2 the range of variation of the star formation efficiency ( @xmath10 ) and in column 3 we indicate if models develop the galactic wind or not . this same scheme is reproduced three times , for @xmath130 and 50 . .salpeter imf models characterized by @xmath108 , @xmath121 : we indicate the range of variation of the star formation efficiency ( @xmath10 , expressed in @xmath11 ) for different values of @xmath131 and @xmath132 , and the development of the galactic wind ( @xmath133 ) . [ cols="^,^,^,^,^,^,^ " , ] [ t2 ] in this paper we attempted to model the chemical evolution of dwarf irregulars and blue compact galaxies . in particular , we considered the presence of dark matter halos and of the different contributions to the interstellar gas energy due to supernovae ( ii and ia ) and stellar winds from massive stars . the comparison with the available data on the abundances and abundance ratios of elements such as he , n , c , o , and fe allowed us to conclude that dig and bcg must contain a substantial amount of dark matter in order to be gravitationally bound against the intense starbursts , that galactic winds powered by supernovae and stellar winds from massive stars are preferably enriched ( differential ) , and that the energy contribution from type ia supernovae to the total thermal energy of the gas is negligible relative to the type ii supernova component . we found also that the stellar wind contribution is negligible relative to that of the type ii supernovae . * a number of bursts @xmath134 ; * a star formation efficiency @xmath135 ; * differential galactic winds mostly powered by supernovae of type ii , with a wind efficiency parameter @xmath136 ; * a ratio between dark matter and luminous matter @xmath137 ; * a dark matter distribution given by the ratio between luminous effective radius and dark effective radius @xmath138 ; * salpeter imf is favoured relative to scalo imf : in fact the observed abundance ratios are better reproduced by salpeter models than by scalo models , in particular the spread of n / o abundance ratio . in order to calculate the energy gained by the ism after an isolated sn explosion we consider the expansion of the snr through an uniform medium of number density @xmath139 and isothermal sound speed @xmath140 cm s@xmath141 . initially the temperature of the remnant is very high and the energy radiated away is negligible as compared to the explosion energy @xmath142 ; this phase is called ` adiabatic ' and is described by the sedov solution , @xmath143 . as the remnant expands its temperature decreases and eventually the radiative losses from the swept - up dense shell become relevant at the cooling time given by ( cf . cioffi & shull 1991 ) : for @xmath147 most of the mass and the kinetic energy of the snr is contained by the thin , cold , massive expanding shell , while the interior cavity is filled with hot dilute gas that contains most of the thermal energy . at this stage the hot bubble has a very long cooling time and cools adiabatically as it expands pushing the outer shell . we thus assume that its thermal energy varies as @xmath148 and neglect a correction @xmath149 due to the effect of cooling ( cioffi , mckee & bertschinger , 1988 ) . this is compatible with the other simplifying assumptions made in our evaluation of the sn efficiency . the remnant therefore expands as @xmath150 in this ` snow plow ' phase . at a later time the pressure inside the cavity becomes equal to the external pressure and the remnant stalls . we define the stalling radius as the radius at which the shell velocity is equal to @xmath151 . as long as the snr expands supersonically , the ism outside the remnant is not aware of its presence and the fraction of the sn explosion energy which is not cooled off remains locked inside the cavity . at merger , the shell dissipates its remaining kinetic energy and its thermal energy ( we assume that the shell temperature does not become lower than the temperature of the unperturbed ism , so that its thermal energy is three times the kinetic energy at the stalling radius ) which add to the thermal content of the ism . the efficiency is thus given by the ratio of this shell energy to the explosion energy . with the assumptions above it is found : @xmath152 it may happen that the remnant pressure becomes lower than the external pressure before the merger . in this case the remnant enters in the ` momentum conserving ' phase , @xmath153 . spitzer ( 1968 ) has shown that , in the limiting case in which the ` snow plow ' phase is entirely skipped and the ` momentum conserving ' phase starts just after @xmath154 , the efficiency is given by @xmath155 , where @xmath156 is the remnant velocity at @xmath154 . in our notations we get : @xmath157 thus , the ` true ' efficiency will be intermediate between the two above . andersen & burkert ( submitted ) , in describing the evolution of the ism of the dwarf galaxies , assume @xmath158 independent of the environment , but find that the exact value is not of importance . before to explode , stars earlier than b0 ( @xmath159 ) suffer considerable mass loss ( @xmath160 @xmath161 yr@xmath141 ) through supersonic winds @xmath162 km s@xmath141 ) , carving large , hot bubbles ( weaver et al . , 1977 ) and injecting large amounts of energy into the ism . in fact , as the wind impinges on the ism , a reflected shock forms which thermalizes the wind itself . this rarefied , hot , pressurizzed gas fills the expanding cavity which is surrounded by a thin , cold , dense shell formed by the swept up gas of the ism . as the expansion procedes the radiative losses increase and the inner pressure decreases untill eventually the bubble stalls . as for the snr case , we calculate the kinetic energy of the dense shell pushed by the expanding bubble when it becomes sonic . to obtain the wind efficiency @xmath163 the shell energy is then divided by the total mechanical energy of the wind @xmath164 , where @xmath165 is the time spent by the star on the main sequence , and @xmath166 erg s@xmath141 is the wind mechanical luminosity . from the classical ` energy conserving ' solution of weaver et al . , @xmath167 , we get : to obtain the above relation we assumed @xmath169 ( mckee et al . 1984 ) . as in the case of snrs , the above efficiency represents an upper limit ; in fact , ` momentum conserving ' bubbles ( @xmath170 ) may be realized if a substantial enhancement of radiative losses occurs for some reason such as cloudy medium ( mckee et al . 1984 ) or non steady cooling ( e.g. dercole 1992 ) . in this case the hot gas quickly cools and the shell is pushed by the ram pressure of the wind that impinges directly on it . the efficiency then becomes : we finally point out that in principle the sn efficiency worked out above could not be valid for exploding stars more massive than 20 @xmath161 because the remnant interacts with the bubble rather than with the uniform ism . if however we consider 20 @xmath161 as representative of massive stars ( @xmath174 , then the maximum bubble radius is rather small ( @xmath175 pc ) , shorter of the radius where the snr becomes radiative ( @xmath176 pc ) and much shorter than the radius at which it stalls ( @xmath177 pc ) . if , moreover , heat conduction is active , the bubble tends to be replenished during the red supergiant phase of the star ( dercole 1992 ) . we thus consider the above estimate of @xmath178 rather accurate for our purposes .
a chemical evolution model following the evolution of the abundances of h , he , c , n , o and fe for dwarf irregular and blue compact galaxies is presented . this model takes into account detailed nucleosynthesis and computes in detail the rates of supernovae of type ii and i. the star formation is assumed to have proceeded in short but intense bursts . the novelty relative to previous models is that the development of a galactic wind is studied in detail by taking into account the energy injected into the interstellar medium ( ism ) from both supernovae and stellar winds from massive stars as well as the presence of dark matter halos . both metal enriched and normal winds have been considered . + our main conclusions are : _ i ) _ a substantial amount of dark matter ( from 1 to 50 times larger than the luminous matter ) is required in order to avoid the complete destruction of such galaxies during strong starbursts , and _ ii ) _ the energy injected by stellar winds and type ia supernovae into the ism is negligible relative to the total thermal energy , and in particular to the type ii supernovae , which in fact , dominate the energetics during starbursts . # 1to 0pt#1
energy spectrum of a quantum ring threaded by a magnetic flux is important since several interesting properties of the ring depend on it@xcite . the energy spectrum is periodic with the period equal to the quantum unit of flux @xmath0 , where the elementary charge @xmath1 . this effect is a consequence of the aharonov bohm effect . the angular momentum of an electron in the groundstate of a quantum ring in a magnetic field can be non zero , as can be seen in the energy spectrum of fig.[fig : spectrum](a ) . this fact is closely related to the fascinating physics of the persistent currents in mesoscopic rings@xcite . the change of groundstate angular momentum about the quantization axis from @xmath2 to @xmath3 is a decisive feature distinguishing quantum rings from quantum dots . such angular momentum transitions were observed in self assembled quantum rings@xcite . effects of spin - orbit terms have been studied in mesoscopic rings@xcite . currently there is renewed interest in them in semiconductor quantum rings . this is because electron spin may be controlled by spin - orbit terms . such a control would be valuable for spintronics , quantum information , and spin qubits@xcite . recently several effects of spin orbit coupling on optical@xcite and transport@xcite properties of semiconductor quantum rings have been investigated . a spin filter@xcite and a qubit@xcite have been also proposed . in this paper we investigate the energy degeneracy of self - assembled semiconductor rings on the magnetic flux in the presence of spin orbit interactions . we show how _ large gauge _ and _ time reversal _ transformations can be used to determine the degenerate properties of the single electron energy spectrum of such a system . before we give a summary of our main results we comment on some basic properties of large gauge transformations , spin orbit terms in semiconductors , and time reserval operation . consider an ideal one - dimensional ring without spin orbit coupling and with the radius much larger than the width of the ring . suppose it is threaded by a magnetic flux @xmath4 ( the direction of the magnetic field is chosen to be along the z - axis ) . it has rotational invariance about the z - axis that goes through the center of the ring . in addition it is invariant under a _ large _ gauge transformation , which transforms wavefunctions as follows . when the flux is increased from @xmath4 to @xmath5 the electron wavefunction changes as @xmath6 where _ only _ integer values of @xmath7 are allowed@xcite . the azimuthal angle of the cylindrical coordinate system is @xmath8 . although time reversal symmetry is broken when a flux threads the ring spin invariance is present . each energy level is thus doubly spin degenerate . however , when the dimensionless flux @xmath9 is a half - integer or integer each energy level can be 4 times degenerate , as shown in fig.[fig : spectrum](a ) . in semiconductors when the external confinement potential breaks the inversion symmetry the rashba spin orbit term is relevant . when the crystal potential itself breaks the inversion symmetry the dresselhaus spin orbit term is relevant . in ii - vi semiconductors the rashba term is expected to be larger than the the dresselhaus coupling . in iii - v semiconductors , such as gaas , the opposite is true@xcite . spin orbit terms break both spin and angular momentum symmetries and the double degeneracy of the ideal one - dimensional ring is broken . in the absence of a vector potential and the zeeman term the spin - orbit hamiltonian is invariant under time reversal symmetry : @xmath10 and @xmath11 , where @xmath12 and @xmath13 are the wavevector and spin of an electron . the time reversal operator is @xmath14 , where the operator @xmath15 stands for complex conjugation and @xmath16 is one of the pauli spin matrices @xmath17 . if the hamiltonian is invariant under time reversal operation @xmath18 then the wavefunctions @xmath19 and @xmath20 are degenerate and orthonormal . note that @xmath21 . this symmetry is the origin of kramers double degeneracy in quantum dots@xcite we explore the interplay between large gauge transforation , time reversal operation , and spin orbit coupling in semiconductor rings . we show in the presence of the rashba and/or dresselhaus terms that , although spin double degeneracy is lifted , all the energy levels at integer and half integer values of @xmath22 are _ doubly _ degenerate , see fig.[fig : spectrum](b ) . this is because at these particular values of @xmath22 a large gauge transformation followed by time reversal operation@xcite is a _ symmetry operation _ of the hamiltonian . this result is valid even in the presence of a disorder potential and in non - circular rings . the wavefunctions of a degenerate pair are related to each other by the symmetry operation . when only the rashba term is present analytical solutions of eigenstates are possible . in this case we find , as shown in fig.[fig : spectrum](b ) , that the energy curve @xmath23 of the ideal ring splits into curves @xmath24 and @xmath25 , where @xmath26 is an integer . as a function of @xmath22 the energy @xmath24 or @xmath25 describes a parabola . we find that degeneracy occurs when @xmath27 , which leads to @xmath28 since @xmath26 and @xmath29 are both integers @xmath22 must be half integers or integers . unlike the previous case of the ideal ring the total degeneracy is now only two . this relation between @xmath26 and @xmath29 can be understood in terms of time reversal and gauge transformations , as we demonstrate below . when the zeeman term is present some degenerate level anticross , as shown in fig.[fig : zeeman ] . before we analyse the narrow semiconductor rings it would be instructive to review the basic properties of the ideal one - dimensional ring@xcite in the absence of spin orbit coupling . the periodic nature of the energy spectrum of an ideal one - dimensional ring as a function of @xmath22 can be understood by considering invariance under large gauge transformations . consider an electron moving in an ideal one - dimensional ring with radius @xmath30 on the x - y plane . the hamiltonian is @xmath31 , where @xmath32 and @xmath33 are the effective mass of the electron and the vector potential . the electron is under the influence of the vector potential due to a solenoid at the center of the ring . the vector potential given by @xmath34 , where @xmath35 and @xmath4 is the total magnetic flux through the ring . without spin orbit terms the hamiltonian commutes with spin operator and each energy level is _ doubly spin degenerate_. the wavefunction @xmath36 satisfies @xmath37 the solutions @xmath36 are given by @xmath38 , where @xmath29 is an integer . a gauge transformation @xmath39 leads to the transformed wavefunction @xmath40 . when an integral multiple of quantum unit of flux is added to the ring the wavefunction satisfies @xmath41 the wavefunctions after and before this process are related by @xmath42 boundary condition @xmath43 implies that @xmath44 . since the wavefunction must be single valued _ not _ all gauge transformations are allowed . only those with integer @xmath45 are possible . it should be noted that gauge transformations that change the flux _ continuously _ from zero to @xmath46 do _ not _ exist . the only possible gauge transformations are those that add an integral multiple of @xmath46@xcite , i.e. , only _ large _ gauge transformations are possible . under a large gauge transformation the wavefunction transforms @xmath47 the angular momentum has decreased by @xmath48 . the gauge invariance requires that the energies before and after are equal : @xmath49 . this should be contrasted with the adiabatic addition of flux from zero to one unit of quantum flux , which changes the electron energy from @xmath50 to @xmath51 while the wavefunction remains unchanged . let us now consider semiconductor rings with spin orbit terms . the electron is under the influence of the vector potential due to a solenoid at the center of the ring . the total hamiltonian is @xmath52 . the first part is @xmath53 , where @xmath54 is the radial potential energy and @xmath55 is the rashba confinement potential . if the radius of the solenoid is equal or larger than the radius of the ring the electron experiences a magnetic field and the zeeman term should be included : @xmath56 , where @xmath57 , @xmath58 , and @xmath59 are , respectively , the effective g - factor , the bohr magneton , and the magnetic field . @xmath60 are the kinematic momentum operators with @xmath61 and @xmath62 . the x and y components of the vector potential are @xmath63 and @xmath64 . the rashba spin orbit term is @xmath65 the constant @xmath66 _ depends on the external electric field _ @xmath67 applied along the z - axis . the dresselhaus spin orbit term is @xmath68 there is another term of the form @xmath69 in the dresselhaus spin orbit term but it vanishes since the expectation value of the first subband wavefunction @xmath70 along z - axis @xmath71 . the constant @xmath72 represents breaking of inversion symmetry by the crystal in zinc blende structures . the confinement potentials along the z - axis and radial direction are assumed to be sufficiently strong that only the lowest energy subbands are relevant . the total electron wavefunction in a narrow semiconductor ring can be written as @xmath73 . the lowest subband wavefunction along the radial direction is @xmath74 . the conduction band bloch wavefunction is @xmath75 . in the following when the electron wavefunction is written only the azimuthal part @xmath36 will be shown and other wavefunctions will be suppressed . the azimuthal part of the wavefunction can be written as @xmath76 or in spinor notation @xmath77 , where @xmath78 and @xmath79 . under time reversal operation@xcite @xmath18 , i.e. , @xmath80 or @xmath81 we have suppressed the bloch wavefunction of the conduction band in applying the time reversal operator since it is unaffected by the operator @xmath82 . our wavefunctions are all effective mass wavefunctions and only the conduction band bloch wavefunction at @xmath83 is relevant . note that time reversal symmetry is _ absent _ when a vector potential is present . suppose that a solenoid with an integer or half integer fluxes , @xmath84 , threads the quantum ring . the vector potential of the solenoid is @xmath33 . consider a large gauge transformation that flips the direction of the vector potential : @xmath85 . this operation is equivalent to the subtraction of the flux @xmath86 . if the flux @xmath86 is _ subtracted _ first and time reversal operation is applied next we find that the wavefunction transforms as follows : @xmath87 note that this procedure is equivalent to applying time reversal operation first and _ adding _ the flux @xmath86 next : @xmath88 from the hamiltonian it can be seen that under the gauge transformation , @xmath89 , and the time reversal operation , @xmath10 and @xmath11 , the electron hamiltonian is invariant in the absence of the zeeman term . the angular momentum components of the wavefunctions before and after these operations are related as follows : @xmath90 and @xmath91 where @xmath92 and @xmath93 . these wavefunctions are degenerate . unlike the case of the ideal one - dimensional ring the total degeneracy is now only two . the result given in this section is valid in the absence of the zeeman term . however , a disorder potential and non - circular ring will not break the presence of degeneracies becuase the relevant potential @xmath94 of these systems is invariant under the gauge transformation , @xmath89 , and the time reversal operation , @xmath10 and @xmath11 . a narrow semiconductor ring with the rashba spin orbit term can be solved exactly even in the presence of the zeeman term . the radial wavefunction @xmath74 is taken to be a gaussian and the correct hamiltonian@xcite is @xmath95 here @xmath96 . note that the zeeman term can be written as @xmath97 , where @xmath98 and @xmath99 . the eigenstates are of the form@xcite @xmath100 the coefficients @xmath101 and @xmath102 satisfy the matrix equation @xmath103 with two eigenvalues @xmath104 where @xmath105 and @xmath106^{1/2}.\nonumber\\\end{aligned}\ ] ] the corresponding eigenvectors are @xmath107 where @xmath108 with the normalization factors @xmath109^{1/2}.\nonumber\\\end{aligned}\ ] ] when the zeeman term is zero the expression for the eigenvalues simplifies @xmath110 these energies are not smooth functions of @xmath22 . they can be combined into smooth functions @xmath24 and @xmath111 that are given by @xmath112 and @xmath113 where @xmath114 . note @xmath115 the corresponding eigenvectors are @xmath116 and @xmath117 when @xmath118 the constants @xmath119 and @xmath120 . as a function of @xmath22 the energies @xmath24 and @xmath25 describe parabola , as shown in fig.([fig : setup ] ) . the roots of @xmath24 are @xmath121 and @xmath122 . the difference between them is @xmath123 . the roots of @xmath25 are @xmath124 and @xmath125 . again the difference between them is @xmath123 . degeneracy at _ half integer or integer _ values of @xmath22 occurs when the condition @xmath27 is satisfied . this leads to @xmath28 at @xmath126 , for example , the pair represented by the quantum numbers @xmath127 is degenerate : @xmath128 . the corresponding degenerate wavefunctions are @xmath129 and @xmath130 where a and b are real constants . they are related to each other by the subtraction of one unit of quantum flux and the application of time - reversal operation : @xmath131 . note that in order to obtain this result we have multiplied the expression of @xmath132 , given by eq.([eigenvec ] ) , with a phase factor . suppose the electron experiences an external magnetic field along the z - axis with the vector potential @xmath133 . to illustrate the effect of the zeeman term we calculate the energy spectrum for @xmath134 with the physical parameters @xmath135 and @xmath136 . for this semiconductor the rashba term is the dominant spin orbit coupling . we assume that the radius is much larger than the ring width . for the radius of the ring @xmath137 the energy scale is @xmath138 and @xmath139=6.70 f$ ] . the zeeman term couples the degenerate states @xmath140 and @xmath141 since @xmath142 is non zero . but it does not couple @xmath143 and @xmath132 since @xmath144 is zero . this is the reason why there is only one anticrossing near , for example , @xmath126 , as can be see in fig.([fig : zeeman ] ) . the value of degenerate energy is @xmath145 in the absence of the zeeman term . note that the degeneracy between @xmath141 and @xmath143 at @xmath146 is lifted unlike the degeneracy between @xmath132 and @xmath140 at @xmath147 . we have argued that the electron hamiltonian of narrow semiconductor rings with the rashba and dresselhaus spin orbit terms is invariant under time - reversal operation followed by a large gauge transformation , provided that the zeeman effect is absent . we find that all the eigenstates are doubly degenerate when integer or half - integer quantum fluxes thread the quantum ring . the wavefunctions of a degenerate pair are related to each other by the symmetry operation . these qualitative results are valid both for ii - vi and iii - v semiconductor rings . a disorder potential and non - circular ring will not break the relevant symmetry and the presence of degeneracy . we have obtained analytical and simple expressions for the energy spectrum when the rashba term is dominant . some of degenerate energy levels anticross in the presence of the zeeman term . since the zeeman effect complicates the energy spectrum to some degree , as shown in fig.([fig : zeeman ] ) , semiconductor rings with a rather small zeeman effect , such as gaas rings , are more suitable for experimental study of the level degeneracy . however , in this case , a considerable numerical work is needed to compute a quantitatively correct energy spectrum due to the presence of the dresselhaus term . measurement of conductance oscillations@xcite or optical emission lines@xcite may reveal the level degeneracy . another suitable ring system to investigate is a superconductor and semiconductor hybrid system@xcite . each vortex of the abrikosov lattice of the superconductor has half of the quantum unit flux , and a semiconductor ring under the superconductor can trap this flux@xcite . the distance between the vortices must be larger than the diameter of the semiconductor ring , and this may be achieved by controlling the strength of the magnetic field . the advantage of such a system is that the zeeman effect is absent since only the central part of the ring is threaded with half integer flux . when an inas semiconductor is used , for example , the rashba term is dominant , and our analytical results should describe well the energy spectrum of such a ring . it is desirable to calculate numerically the energy spectrum quantitatively when both the rashba and dresselhaus terms are present in addition to the zeeman term . it may be worthwhile to investigate the effect of the finite width of the ring @xcite . the double degeneracy at an integer or half integer flux may be used to generate non - abelian berry phases , see ref.19 . this work was supported by grant no . r01 - 2005 - 000 - 10352 - 0 from the basic research program of the korea science and engineering foundation and by quantum functional semiconductor research center ( qsrc ) at dongguk university of the korea science and engineering foundation .
the electron hamiltonian of narrow semiconductor rings with the rashba and dresselhaus spin orbit terms is invariant under time - reversal operation followed by a large gauge transformation . we find that all the eigenstates are doubly degenerate when integer or half - integer quantum fluxes thread the quantum ring . the wavefunctions of a degenerate pair are related to each other by the symmetry operation . these results are valid even in the presence of a disorder potential . when the zeeman term is present only some of these degenerate levels anticross .
grand unified theories ( guts ) are considered to be among the most appealing scenarios for physics beyond the standard model . qualitatively they always predict ( i ) unification of gauge couplings of the standard model and ( ii ) proton decay . the first feature can not be directly probed since unification takes place at a very high energy scale the so - called gut scale . however the second feature _ can _ be probed and it offers the only realistic way of testing grand unification . it is thus important to single out and investigate viable models of grand unification ; where proton decay is not only well predicted , but also experimentally accessible in both current and future proton decay experiments . out of all grand unified theories the scenarios based on @xmath0 gauge symmetry are arguably the most predictive ones for proton decay . recall , of all simple gauge groups that allow sm embedding only @xmath0 has unique single - step symmetry breaking pattern . this allows rather accurate determination of high energy scales relevant for proton decay . and , in its non - supersymmetric version , @xmath0 gut avoids uncertainties pertaining to the exact nature as well as the relevant scale of supersymmetry breaking ; both of those features are yet to be established experimentally . moreover , it simplifies the discussion on the dominant source(s ) of proton decay . all these appealing properties single out non - supersymmetric @xmath0 as the theory for proton decay . we accordingly focus our attention on its simplest realistic realizations . our starting point is the @xmath0 model proposed long ago by h. georgi and s. glashow @xcite . their model offers partial matter unification by accommodating @xmath5th generation of matter fields in the @xmath6 and @xmath7 dimensional representations . the scalar sector is composed of a @xmath2 dimensional higgs representation and a @xmath1 dimensional higgs multiplet . the sm singlet in @xmath2 breaks @xmath0 symmetry down to the standard model , while the sm @xmath8 doublet in @xmath1 accomplishes electroweak symmetry breaking . the model , however , is not realistic ; the gauge couplings do not unify , neutrinos are massless and @xmath9 at the gut scale . there are two possible model building approaches that lead to simple yet realistic extensions of the georgi - glashow ( gg ) model in view of generation of realistic charged fermion masses . one approach is to allow for higher - dimensional operators which modify bad mass relations @xmath9 @xcite . that approach requires no additional higgs fields to be introduced to fix those relations . and , the strength of required corrections in the charged sector might allow one to place an upper bound on the scale where the uv completion of the unified theory takes place . of course , in order to improve unification more split representations need to be present . the other approach is to stick with renormalizable operators . the latter approach requires addition of a @xmath3 dimensional higgs multiplet . ( recall that the tensor product @xmath10 . ) within that approach , as far as the neutrino masses are concerned , one can either introduce some fermion singlets the right - handed neutrinos or a @xmath4 dimensional higgs representations ( or both ) . the simplest non - renormalizable model based on @xmath0 has been subject of a recent investigation @xcite . it has been shown that the model , with the higgs sector composed of @xmath1 , @xmath2 and @xmath4 , can be tested at future proton decay experiments and at future collider experiments , for example at lhc . the first possibility is due to existence of an upper bound on the proton decay lifetime . namely , @xmath11years , where @xmath12 is the nucleon matrix element . the second one is based on potential production of light leptoquarks @xcite . see reference @xcite for the study of several phenomenological and cosmological issues in this context . in this work we want to investigate the minimal extension of the georgi - glashow model within the renormalizable framework . namely , we want to study predictions of a model with the higgs sector made out of @xmath1 , @xmath3 and @xmath2 representations . we initially assume that the matter sector contains right - handed neutrinos to generate neutrino masses through the type - i see - saw mechanism @xcite . however , we also discuss the case when there is an extra @xmath4 dimensional representation that generates neutrino mass via type - ii see - saw @xcite . the paper is organized as follows : in section i we describe the minimal renormalizable @xmath0 and consequently study the unification constraints . in section ii we compare different scenarios based on @xmath0 and discuss possibility to test them . in the last section we summaries our results . the higgs sector of the minimal renormalizable @xmath0 is composed of the @xmath1 , @xmath2 , and @xmath3 dimensional representations . their sm @xmath13 decomposition is given by : @xmath14 where we also set our notation . the yukawa potential for charged fermions read as : @xmath15 where the field @xmath3 satisfies the following conditions : @xmath16 in this model the masses for charged fermions are given by : @xmath17 where @xmath18 . @xmath19 , @xmath20 , @xmath21 and @xmath22 are arbitrary @xmath23 matrices . ( note the georgi - jarlskog @xcite factor in eq . ( [ gj ] ) . ) clearly , there are enough parameters in the yukawa sector to fit all charged fermions masses . for previous studies in this context see @xcite . now , let us understand the unification constraints within this model . necessary conditions for the successful gauge coupling unification can be expressed via two equalities . ( see reference @xcite for details . ) these are @xmath24 where all experimentally measured quantities on the right - hand sides are to be taken at @xmath25 energy scale . the first one is the so - called `` b - test '' and the second one is the `` gut scale relation '' . in what follows we use @xcite @xmath26 , @xmath27 and @xmath28 to obtain : @xmath29 the left - hand sides , on the other hand , depend on particular particle content of the theory at hand and corresponding mass spectrum . more precisely , @xmath30 , where @xmath31 coefficients are given by : @xmath32 @xmath33 are the sm coefficients while @xmath34 are the one - loop coefficients of any additional particle @xmath35 of mass @xmath36 ( @xmath37 ) . ( recall , for the case of @xmath38 light higgs doublet fields @xmath39 , @xmath40 and @xmath41 . ) relevant @xmath42-coefficient contributions in our scenario are listed in table [ table1 ] . .[table1 ] contributions to the @xmath42 coefficients . the masses of the higgs doublets are taken to be at @xmath25 . [ cols="<,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] there are two fields in @xmath4 that can improve unification ; these are @xmath43 and @xmath44 . @xmath43 has very small contribution to @xmath45 and very large contribution to @xmath46 . this means that its impact on the gut scale is not significant . @xmath44 , on the other hand , has large impact on the gut scale relation but , in general , it mediates proton decay and it is probably better to keep it heavy . in any case , the @xmath47 contribution to the running of the gauge couplings is crucial to achieve high scale unification in agreement with experimental data . if its contribution to the decay of the proton is set to zero by additional symmetry the unification scale could be very large . therefore , since in that case the most important contributions to the decay of the proton are the gauge @xmath48 ones , we can conclude that proton could be _ stable _ for all practical purposes in the minimal renormalizable @xmath0 . as we discussed in previous sections there are two simple candidates for unification based on @xmath0 . in the first scenario the higgs sector is composed of @xmath1 , @xmath2 , and @xmath4 , and higher - dimensional operators are used to modify the relation @xmath9 let us call this model gut - i . the second scenario is the one discussed in this work the renormalizable model with a higgs sector composed of @xmath1 , @xmath2 , and @xmath3 . let us call this gut - ii . there are two possibilities to test the gut - i @xcite . one is through proton decay in the current and next generation of experiments and the other is through the production of light leptoquarks in future colliders . in this work we have concluded that gut - ii model can not be tested through proton decay since the lifetime of the nucleon could be very large . therefore , we can say that the only gut candidate based on @xmath0 which can be falsified in the near future is the model presented in reference @xcite . we have studied the possibility to achieve unification without supersymmetry in a minimal realistic grand unified theory based on @xmath0 , where the higgs sector is composed of the @xmath1 , @xmath2 , and @xmath3 dimensional representations . we have pointed out that the proton could be practically stable in this scenario . we have accordingly concluded that the best candidate to be tested is the gut model with @xmath1 , @xmath2 , and @xmath4 dimensional representations in the higgs sector . i. d. thanks stephen m. barr and bartol research institute for support and hospitality . the work of p. f. p has been supported by _ fundao para a cincia e a tecnologia _ ( fct , portugal ) through the project cftp , pocti - sfa-2 - 777 and a fellowship under project pocti / fnu/44409/2002 . we would like to thank goran senjanovi for discussions and comments . h. georgi and c. jarlskog , `` a new lepton - quark mass relation in a unified theory , '' phys . b * 86 * ( 1979 ) 297 . j. r. ellis and m. k. gaillard , `` fermion masses and higgs representations in su(5 ) , '' phys . b * 88 * ( 1979 ) 315 . i. dorsner and p. fileviez prez , `` unification without supersymmetry : neutrino mass , proton decay and light leptoquarks , '' nucl . b * 723 * ( 2005 ) 53 [ arxiv : hep - ph/0504276 ] . i. dorsner , p. fileviez prez and r. gonzalez felipe , `` phenomenological and cosmological aspects of a minimal gut scenario , '' arxiv : hep - ph/0512068 . to appear in nucl . b. h. georgi and c. jarlskog , `` a new lepton - quark mass relation in a unified theory , '' phys . b * 86 * ( 1979 ) 297 . + the @xmath49 was used for the first time in : p. h. frampton , s. nandi and j. j. g. scanio , `` estimate of flavor number from su(5 ) grand unification , '' phys . b * 85 * ( 1979 ) 225 . p. minkowski , `` mu @xmath50 e gamma at a rate of one out of 1-billion muon decays ? , '' phys . b * 67 * ( 1977 ) 421 ; t. yanagida , proceedings of the _ workshop on unified theories and baryon number in the universe _ , tsukuba , 1979 , eds . a. sawada , a. sugamoto , kek report no . 79 - 18 , tsukuba . s. glashow , in _ quarks and leptons , cargse 1979 _ , eds . m. l ' evy . et al . , ( plenum , 1980 , n ew york ) . m. gell - mann , p. ramond , r. slansky , proceedings of the _ supergravity stony brook workshop _ , new york , 1979 , eds . p. van niewenhuizen , d. freeman ( north - holland , amsterdam ) . r. mohapatra , g. senjanovi ' c , `` neutrino mass and spontaneous parity nonconservation , '' phys.rev.lett . * 44 * ( 1980 ) 912 g. lazarides , q. shafi and c. wetterich , `` proton lifetime and fermion masses in an so(10 ) model , '' nucl . b * 181 * ( 1981 ) 287 . + r. n. mohapatra and g. senjanovic , `` neutrino masses and mixings in gauge models with spontaneous parity violation , '' phys . d * 23 * ( 1981 ) 165 . a. giveon , l. j. hall and u. sarid , `` su(5 ) unification revisited , '' phys . b * 271 * ( 1991 ) 138 . s. eidelman _ et al . _ [ particle data group ] , `` review of particle physics , '' phys . lett . b * 592 * ( 2004 ) 1 . k. s. babu and e. ma , `` suppression of proton decay in su(5 ) grand unification , '' phys . b * 144 * ( 1984 ) 381 . p. kalyniak and j. n. ng , `` symmetry breaking patterns in su(5 ) with nonminimal higgs fields , '' phys . d * 26 * ( 1982 ) 890 . + p. eckert , j. m. gerard , h. ruegg and t. schucker , `` minimization of the su(5 ) invariant scalar potential for the fortyfive - dimensional representation , '' phys . b * 125 * ( 1983 ) 385 . i. dorsner and p. fileviez prez , `` how long could we live ? , '' phys . b * 625 * ( 2005 ) 88 [ arxiv : hep - ph/0410198 ] . nath and p. fileviez prez , `` proton stability in grand unified theories , in strings , and in branes , '' arxiv : hep - ph/0601023 . s. aoki _ et al . _ [ jlqcd collaboration ] , `` nucleon decay matrix elements from lattice qcd , '' phys . rev . d * 62 * ( 2000 ) 014506 [ arxiv : hep - lat/9911026 ] .
we investigate unification constraints in the simplest renormalizable non - supersymmetric @xmath0 framework . we show that in the scenario where the higgs sector is composed of the @xmath1 , @xmath2 , and @xmath3 dimensional representations the proton could be practically stable . we accordingly demonstrate that of all the @xmath0 scenarios only the non - renormalizable one with the @xmath1 , @xmath2 , and @xmath4 dimensional higgs multiplets can be verify if low - energy supersymmetry is not realized in nature .
ultrafast semiconductor spectroscopy provides a unique laboratory for the investigation of coherence and scattering effects in solids : @xcite a short laser pulse creates electron - hole pairs propagating in phase with the exciting laser ; this initial phase coherence is consecutively destroyed through mutual interactions among the photoexcited carriers and couplings to environmental degrees of freedom , e.g. phonons , thus leading to _ decoherence _ , @xcite which can be monitored again by means of ultrafast spectroscopy . recent years have seen spectacular examples of such ultrafast semiconductor spectroscopy , e.g. the observation of phonon quantum beats @xcite or the buildup of screening , @xcite and have revealed dephasing and relaxation times of the order of femto- and picoseconds for carrier - carrier and carrier - phonon interactions , respectively . in semiconductor quantum dots , @xcite sometimes referred to as _ artificial atoms _ , the strong confinement in all three spatial directions results in a substantial suppression of decoherence as compared to higher - dimensional semiconductors , i.e. , bulk , quantum - wells , or quantum wires . on the one hand , coulomb interactions among carriers captured in the dot do not result in scattering but only give rise to energy renormalizations of the electron - hole few - particle states . @xcite on the other hand , at low temperatures phonon - mediated dephasing of the electron - hole states of lowest energy is of only minor importance , @xcite and can become even negligible in comparison to spontaneous emission of photons which occurs on a nanosecond time scale . @xcite these remarkable features render quantum dots ideal candidates for the solid - state implementation of quantum engineering , @xcite e.g. , for the purpose of quantum computation , which has led to considerable research activities in recent years . @xcite observation of optical coherence effects in ensembles of dots is usually spoiled by inhomogeneous line broadening due to dot size fluctuations , with typical broadenings comparable to the level splittings themselves . to overcome this problem , within the last couple of years a number of experimental techniques were developed to allow the observation of single quantum dots , hereby establishing the rapidly growing field of _ single - dot spectroscopy_. @xcite such single - dot measurements are not only indispensable for the observation of quantum coherence and the implementation of quantum engineering , but also open the possibility to study decoherence and relaxation in _ single _ quantum systems . the theoretical analysis of single quantum system , however , is more cumbersome as compared to ensembles of corresponding systems . in the latter case , one usually employs the framework of quantum transport theory or quantum kinetics @xcite which describes the time evolution of macroscopic quantities , e.g. , the interband polarization , directly accessible to experiment ; here , the detailed evolution of one specific subsystem is completely irrelevant , which allows to submit the problem to the laws of statistical physics . on the other hand , for a single system one has to be more specific about the occurrence of single scatterings and their monitoring through a measurement apparatus . the question of how to theoretically describe such problems first arose almost two decades ago when it became possible to store single ions in a paul trap and to continuously monitor their resonance fluorescence , and led to the development of the celebrated _ quantum - jump approach_. @xcite this approach combines the usual master - equation approach with the rules of demolition quantum measurements , @xcite and provides a flexible tool for the description of single - system dynamics subject to continuous monitoring . in this paper we employ the quantum - jump approach to two representative examples for the observation of single scatterings in single semiconductor quantum dots . first , we study the cascade decay of a biexciton , which has recently attracted considerable interest : @xcite in a first step , the biexciton decays through emission of a photon to the spin - degenerate exciton states ; this final - state ambiguity leads to an _ entanglement _ of photon and exciton ; when subsequently the exciton decays , the photon - exciton entanglement is transferred to a two - photon entanglement . @xcite as shown below , for dots with an exciton fine - structure splitting @xcite the two - photon correlation function of this cascade decay exhibits an oscillatory behavior , provided that the photons are detected in a properly chosen polarization basis ( see ref . for a similar experiment ) . our analysis carefully examines the photon emission processes , and improves upon the simple - minded framework of fermi s golden rule to provide a proper description of the problem under concern . our second example is based on an experiment recently performed by zrenner _ et al . _ @xcite in which the authors used a quantum dot embedded in a field - effect structure to transform an optically excited exciton through tunneling into a photocurrent . here , we study the decay of a charged exciton , i.e. , of a coulomb - renormalized complex consisting of two electrons and a hole . @xcite through the tunneling process one electron with either spin - up or spin - down orientation is transferred from the dot to the reservoir ; since the electron in the dot has an opposite spin orientation , the spins of the reservoir- and dot - electron become entangled . @xcite in contrast to the optical biexciton decay additional difficulties arise because the tunnel - generated electron and hole do not propagate freely , as photons would in the corresponding scheme , but are subject to interactions in the contact . we develop a prototypical , though simplified , description scheme of the scattering channels in the reservoir , and demonstrate that this spin entanglement is robust against spin - unselective scatterings and is only deteriorated by spin - selective scatterings . we have organized our paper as follows . in sec . [ sec : two.photon ] we discuss the monitored biexciton cascade decay in a single quantum dot . we specify our model system , provide a short discussion of the master - equation and quantum - jump approach , and finally show how to obtain the system s time evolution through unraveling of the master equation . the charged - exciton decay of a dot embedded in a field - effect structure is discussed in sec . [ sec : spin.entanglement ] . we develop a theory accounting for the complete cascade process of the buildup of coherence through tunneling , the swapping of quantum coherence to the spin entanglement through dephasing and relaxation , and the process of disentanglement through spin - selective scatterings . finally , in sec . [ sec : conclusion ] we draw some conclusions . and @xmath0 , where @xmath1 ( @xmath0 ) is sensitive to photons from the 3@xmath2 ( @xmath20 ) decay . ( b ) level scheme as described in the text : @xmath3 is the groundstate , @xmath4 the spin - degenerate single - exciton states , and @xmath5 the biexciton one . the arrows indicate the optically allowed transitions and the symbols indicate @xmath6- or @xmath7-polarization for photons propagating along @xmath8 . ] we examine a prototypical level scheme depicted in fig . [ fig : biexciton]b , which consists of the groundstate @xmath3 ( no electron - hole pairs present ) , the spin - degenerate single - exciton states @xmath4 of lowest energy , and the biexciton state @xmath5 ; other dot states , such as excited excitons or biexcitons , are energetically well separated and can be safely neglected . the single - exciton energies @xmath9 are assumed to be split by a small amount @xmath10 ( typically a few tens of @xmath11ev ) because of the _ electron - hole exchange interaction _ , @xcite and the corresponding states to be polarized along @xmath6 and @xmath7 . finally , the biexciton energy @xmath12 is reduced by the _ biexciton binding energy _ ( typically a few mev ) because of coulomb correlations . @xcite within this scheme , the hamiltonian of the quantum - dot states reads : @xmath13 where energy zero is given by the groundstate . in what follows , we shall make use of the fact that the groundstate is optically connected to the biexciton state @xmath5 through two independent pathways , namely 013 and 023 , which allows to establish a _ quantum coherence _ between the single - exciton states directly observable in optical experiments . because of the important role of states @xmath4 we shall find it convenient to introduce the pauli matrices @xmath14 , @xmath15 , @xmath16 , which together with the unit matrix @xmath17 form a complete basis within the single - exciton subspace . the hamiltonian of eq . ( [ eq : level.biexciton ] ) can then be rewritten as @xmath18 in the linear basis the hamiltonian ( [ eq : level.biexciton.linear ] ) is diagonal . hence , a laser pulse propagating in @xmath8-direction with polarization along @xmath6 or @xmath7 allows to selectively excite one of the two single - exciton eigenstates . on the other hand , any other light polarization will induce a superposition state , e.g. , @xmath19 for circular polarization @xmath20 . further below we shall find it convenient to switch between the linear basis @xmath4 and the circular one , eq . ( [ eq : level.biexciton.circular ] ) , through a unitary transformation @xmath21 , where @xmath22 in other words , the energy splitting @xmath10 translates in the circular basis to a coupling between states @xmath23 . thus , if the system is initially prepared in , e.g. , state @xmath24 ( which is not an eigenstate of the hamiltonian ) , it will start to oscillate between @xmath25 where the oscillation period is given by @xmath26 . we describe the light - matter coupling within the usual dipole and rotating - wave approximations @xcite @xmath27 here , @xmath28 is the electric field of the light evolving with positive or negative frequency components , @xcite @xmath29 the _ interband polarization _ of the quantum dot , @xmath11 the dipole moment , @xmath30 a unit vector along @xmath6 or @xmath7 , and @xmath31 denotes the state with polarization perpendicular to @xmath2 . suppose that the quantum dot initially in its groundstate is excited by a short laser pulse . we assume that @xmath28 is sufficiently intense to be treated classically , and that the spectral width of the pulse is sufficiently broad to excite both 0@xmath2 transitions with comparable strength but narrow enough to inhibit transitions between @xmath2 and 3 . then , such a pulse with circular polarization @xmath32 will induce a superposition @xmath33 , where @xmath34 and @xmath35 are determined by the laser parameters . since @xmath24 is not an eigenstate but a _ coherent superposition _ of states @xmath4 , as time goes on the system will oscillate between @xmath36 . this optically induced quantum coherence corresponds to an oscillating interband polarization which can be directly monitored by a second , time - delayed laser pulse ( see , e.g. , refs . for the observation of such _ coherence beats _ ) . let us assume that the frequency of the second pulse is tuned to the @xmath23 transitions and that the delay time is short as compared to the decoherence times . then , the absorption of the second pulse will sensitively depend on the time delay between the two pulses , and will exhibit oscillations on a time scale of @xmath26 . in this paper we shall address the question whether it is possible to observe a similar effect by means of incoherent optics . we assume that initially the dot is in the biexciton state @xcite and consider the setup depicted in fig . [ fig : biexciton]a , where the optical decay of photons along @xmath37 is monitored by two photo detectors . for this geometry the usual optical selection rules ( to be discussed below ) apply . in addition , we assume that photo detector @xmath1 ( @xmath0 ) is sensitive to photons from the 3@xmath31 ( @xmath20 ) decay , @xcite which can be achieved through spectral filtering ( and restricts the time resolution to @xmath38 ) . suppose that at time @xmath39 a photon with polarization @xmath40 is detected from the biexciton decay . analogously to the coherent excitation - scenario described above , we might assume that after photon detection the system is prepared in state @xmath24 . since this is not an eigenstate of the hamiltonian , the system will oscillate between @xmath36 which could be detected by monitoring the probability distribution for the second photon emission . thus , time - resolved photon correlation measurements together with appropriate frequency and polarization filtering should provide a means for the measurement of quantum coherence in an incoherent optical experiment . at this point we encounter a conceptual problem . if we would describe the photon emissions through fermi s golden rule we would obtain scattering rates @xmath41 between the asymptotic states before , @xmath5 , and after , @xmath42 , the scattering . here , the photon correlations would strongly depend on the choice of the basis @xmath42 : in the circular basis , the detection of a @xmath40 photon would correspond to an optical decay from 3 to @xmath24 and would result in the aforementioned coherence beats . on the other hand , in the linear basis the detection of a @xmath40 photon had to be associated with equal probability to either the 31 or 32 decay , and the state after photoemission would be a mixture _ without any quantum coherence between @xmath4 _ ( since the nature of a scattering is intrinsically incoherent ) . here is the problem : within the framework of fermi s golden rule different predictions arise for different basis sets . this is a very unfortunate situation which calls for a more careful theoretical analysis . it is the purpose of the remainder of this section to reexamine the problem of photon correlation measurements from the biexciton decay and to provide a comprehensible answer . we describe the coupling between the quantum - dot states and photons through eq . ( [ eq : ham.op ] ) , where the electric field of a photon with wavevector @xmath43 and polarization @xmath44 @xmath45 is expressed in terms of the usual bosonic field operators @xmath46 . @xcite here , @xmath47 is the light frequency , @xmath48 the speed of light , @xmath49 the semiconductor dielectric constant , @xmath50 its refractive index , and @xmath51 the polarization vector ( see also appendix [ sec : wigner - weisskopf ] ) . with eq . ( [ eq : efield.photon ] ) the light - matter coupling ( [ eq : ham.op ] ) can be cast to @xmath52 with the coupling constant @xmath53 . additionally , in the following we will need the hamiltonian for free photons @xmath54 . next , we address the problem of how to describe the time evolution of a single quantum system in contact with its environment , i.e. , an _ open system_. @xcite suppose that the hamiltonian describing the system coupled to its environment can be split into the parts @xmath55 , where @xmath56 accounts for the uncoupled system and reservoir ( here @xmath57 ) and @xmath58 to the coupling ( here @xmath59 ) . because of the incoherent nature of such system - environment couplings we have to adopt a _ density - matrix description _ , where the density operator @xmath60 fully characterizes the system : its diagonal elements provide information about state occupancies whereas the off - diagonal elements account for quantum coherence . in lowest - order time - dependent perturbation theory and assuming that at time zero system and environment are effectively decoupled , the time evolution of @xmath60 is governed by : @xcite @xmath61\,]\,,\ ] ] where the superscript @xmath62 denotes the interaction representation according to @xmath56 . two points are worth emphasizing : first , eq . ( [ eq : open.system ] ) is of markovian form , i.e. , the time evolution of @xmath60 at time @xmath39 is completely determined by the density operator itself ; as discussed in detail in ref . , this approximation is appropriate for systems where higher - order contributions to eq . ( [ eq : open.system ] ) are negligible , or in other words , where scatterings occur seldomly on a time scale given by @xmath56 ( here @xmath63 ) ; secondly , the time integral in eq . ( [ eq : open.system ] ) extends to the past of the system and thus explicitly accounts for the buildup of scatterings . note that the limit @xmath64 in eq . ( [ eq : open.system ] ) , which we will _ not _ perform below , would recover the framework of fermi s golden rule and would lead to the aforementioned final - state ambiguities . below we will need the expression @xmath65\nonumber\\ & & - g_{\bm k,\sigma s}^ * a_{\bm k\sigma } \,[\,{|s\rangle\langle0| } \ , e^{-i(\omega_k-\epsilon_s)t'}+ { |3\rangle\langle\bar s|}\ , e^{-i(\omega_k+\epsilon_{\bar s}-\epsilon_3)t'}]\ , ) \,,\nonumber\\ \end{aligned}\ ] ] which accounts for the buildup of scatterings . the time integrals of the exponentials can be performed analytically , @xmath66 with @xmath67 , and we have neglected in the last expression on the right - hand side terms which only contribute to an energy renormalization but not to dephasing and relaxation . @xcite @xmath68 is a symmetric function of @xmath69 which in the limit @xmath64 gives dirac s delta function . finally , from eq . ( [ eq : hop.interaction ] ) and the definition for @xmath68 we obtain @xmath70\nonumber\\ & & - g_{\bm k,\sigma s}^ * a_{\bm k\sigma } \,[\,{|s\rangle\langle0| } \ , \gamma(\omega_k-\epsilon_s , t)+ { |3\rangle\langle\bar s|}\ , \gamma(\omega_k+\epsilon_{\bar s}-\epsilon_3,t ) ] \,)\,.\nonumber\\\end{aligned}\ ] ] this expression provides the starting point for our following discussion . we shall now be more specific about the question what happens to the emitted photon . quite generally , the photon can propagate into a direction not covered by the photo detectors , in which case it will not be observed , or it can be measured by one of the two photo detectors , in which case we will acquire additional information about the state of the system . let us first discuss the case of the _ unmonitored decay _ , i.e. , photon emission into a segment not covered by the photo detector . for simplicity , we only consider the biexciton decay monitored by photo detector @xmath1 , where analogous conclusions hold for the exciton decay monitored by @xmath0 . in the spirit of the usual framework for the description of open - system dynamics , @xcite we shall trace out the reservoir degrees of freedom . this procedure implies that we keep ignorant about the state of the reservoir . @xcite denoting with @xmath71 the photon modes not covered by the photo detector , relaxation and dephasing in the dot due to unmonitored decay are described by the expression @xmath72\,]\,,\ ] ] where @xmath73 denotes the _ reduced density operator _ of the quantum dot . ( [ eq : op.decay],[eq : unmonitored ] ) give us the prescription of how to compute the generalized scattering rates . since each system reservoir coupling @xmath58 comes with the bosonic field operators @xmath74 and @xmath75 , the resulting expressions contain expectation values of @xmath76 , @xmath77 , @xmath78 , and @xmath79 . we shall treat the reservoir ( photons ) as a thermal bath at zero temperature , i.e. , we assume the usual correlation functions @xcite [ eq : correlation.reservoir ] @xmath80 with these correlations it becomes possible to decompose eq . ( [ eq : unmonitored ] ) into two parts , which , in the spirit of the boltzmann equation , can be considered as generalized out- and in - scatterings . we start by evaluating the double commutator @xmath81=vv'\rho+\rho v'v - v\rho v'-v'\rho v$ ] . when tracing over the photon degrees of freedom , we realize that the only non - vanishing terms are proportional to @xmath82 , @xmath83 , and @xmath84 . here , the first two terms do not change the number of photons . correspondingly , we shall ascribe the contributions @xmath85 to _ generalized out - scatterings _ which lead to decoherence and relaxation in the dot . if only these terms would be considered in the time evolution of @xmath73 , the trace of the density operator , which describes the probability of finding the system in any of the quantum - dot states , would decrease . this lack of norm conservation is restored by the remaining terms @xmath86 which describe the effects of photon emissions , and may thus be interpreted as _ generalized in - scatterings_. as will be shown in sec . [ sec : unraveling ] , this decomposition into out- and in - scatterings allows for an efficient solution scheme through an _ unraveling of the master equation_. @xcite the case of a _ monitored decay _ has to be treated with slightly more care . quite generally , such monitoring requires to specify in more detail the quantum mechanical measurement process . to this end , in the following we shall employ the framework of the _ quantum - jump approach_. @xcite we assume that the photon counting can be approximately described by a series of repeated gedanken measurements at times @xmath87 . here , @xmath88 is assumed to be sufficiently short to allow for a time - resolved measurement of the system s dynamics , i.e. , @xmath89 . on the other hand , @xmath88 should be sufficiently large to enable the buildup of complete scatterings , such that the measurement process itself does not inhibit or enhance scatterings ( _ quantum zeno effect _ ) . @xcite suppose that at time @xmath39 the density operator @xmath90 corresponds to an uncoupled dot and reservoir . with the projector @xmath91 associated to the result of the measurement @xcite ( no photon count , @xmath92 , or photon count , @xmath93 ) , the density operator at a later time can be evaluated to @xcite @xmath94\,]\,\mathbb{p}_n^{\omega_s}\,.\ ] ] if within the time interval @xmath88 no photon is detected from the biexciton decay , either no photon has been emitted ( which corresponds to a projector - like measurement @xmath95 on the photon vacuum ) or a photon has been emitted into a segment @xmath96 . on the other hand , photon detection in @xmath88 corresponds to a projector - like measurement @xmath97 on the single - photon subspace . through the measurement we know that the biexciton has decayed , and accordingly we have to modify the density operator to account for this increase of information . we can again evaluate the double commutator @xmath81 $ ] in eq . ( [ eq : monitored ] ) and split the resulting expressions into generalized out- and in - scatterings ; however , within the context of monitoring these terms acquire a different interpretation : as regarding out - scatterings , a nonvanishing contribution only arises for projection on the photon vacuum , which becomes @xmath98 . this term alone is responsible for a decrease of @xmath99 associated to the diminishing probability of finding the system in the initial biexciton state . @xcite to restore this lack of norm conservation , we have to additionally consider in - scatterings . the corresponding contribution , which is associated to photo detection , is non - zero only if projected on @xmath97 , and becomes @xmath100 . this term gives us the prescription of how the density operator must change after photon detection . @xcite with the results of the previous subsections we are now in the position to improve upon the simple - minded fermi s golden rule framework . first , we note that for both unmonitored and monitored decay the resulting expressions are , e.g. , of the form @xmath101 , where the photon modes extend over a given segment @xmath102 . let us consider one specific contribution @xmath103 where we have explicitly evaluated the trace over the photon degrees of freedom and used the definition of eq . ( [ eq : ham.op.gamma ] ) for the coupling constants @xmath104 . the quantity of primary interest is the integral over @xmath105 , which can be rewritten as @xmath106 eq . ( [ eq : photon.spectral ] ) has a rather precise physical meaning : it accounts for the temporal buildup and decay of correlations between dot states and photons , and thus plays the role of a memory function for the reservoir . assume that @xmath107 which can be easily fulfilled on the time scale of the system s dynamics ( since @xmath108 ) . then @xmath109 is a function which is strongly peaked around @xmath110 . we thus can approximately replace in eq . ( [ eq : photon.spectral ] ) @xmath111 with @xmath112 . then , @xcite @xmath113 where in the last line we have extended the integral to @xmath114 . ( [ eq : photon.spectral.delta ] ) is the expression we were seeking for . it shows that generalized scattering rates can be computed without invoking the adiabatic approximation which would be problematic in view of the choice of asymptotic states . rather it suffices for the problem of our concern to assume that the reservoir memory is sufficiently short - lived to be approximately treated as a delta correlation on the time scale of the system dynamics . for the @xmath69-integrals in eq . ( [ eq : trace.photon ] ) we assume @xmath115 where two approximations have been employed : first , for the unmonitored decay we have extended the integral over the complete sphere and have accounted for the missing segment @xmath0 through a reduction factor @xmath116 . accordingly , we have introduced a factor @xmath35 for the monitored segment @xmath0 , and have assumed that within @xmath0 the polarization vectors are approximately @xmath117 , where @xmath118 . with these expressions we can cast the equation of motion for the reduced density - operator @xmath73 ( henceforth we will omit the superscript @xmath119 ) to a master equation of lindblad form , @xcite @xmath120 where the lindblad operators @xmath121 for the various scattering channels are listed in table [ table : lindblad ] and @xmath122 is a non - hermitian hamiltonian accounting for the dot states and out - scatterings , which can be simplified to @xmath123 . . lindblad operators for the unmonitored and monitored decay channels . we use @xmath124 and @xmath125 , with @xmath126 the wigner - weisskopf decay rate given in appendix [ sec : wigner - weisskopf ] . [ cols= " < , < , < " , ] within this scheme we can solve the master equation subject to the initial density operator @xmath127 through unraveling , @xmath128 where @xmath129 is the charged - exciton tunneling rate and @xmath130 the time evolution operator accounting for the propagation of the tunneled electron ( see below ) . the density operator after tunneling is @xmath131 , with @xmath132 eq . ( [ eq : psi.entangled ] ) is an important and non - trivial result . first , it demonstrates that despite the incoherent nature of tunneling and hole relaxation the electron system can be described in terms of wavefunctions ; note that the detection of the hole would even allow to purify this wavefunction , @xcite which might be of relevance when initially @xmath60 is not equal to @xmath133 . secondly , a closer inspection of eq . ( [ eq : psi.entangled ] ) reveals that the spin part @xmath134 is a _ maximally entangled state _ of the electrons in the dot and reservoir . we emphasize that this maximal entanglement is independent of the spin basis , which guarantees that our scheme is not deteriorated by possible polarization anisotropies of the dot states , e.g. , fine - structure splittings . after tunneling the electron in the reservoir propagates in presence of scatterings , as described by @xmath130 in eq . ( [ eq : rho.tunneled ] ) . quite generally , we assume that the orbital degrees of the reservoir electron are subject to much stronger interaction channels , e.g. , phonons , than the spin degrees , as evidenced by the long measured spin lifetimes ( @xmath135ns ) in @xmath136-doped semiconductors . @xcite for that reason , let us first consider an elastic electron scattering which does not depend on spin , i.e. , lindblad operators of the form @xmath137 with @xmath138 the scattering rate . hence , the effective hamiltonian reads @xmath139 . with @xmath140 the density operator after scattering and @xmath141 , the master equation accounting for elastic spin - unselective scatterings can be solved through unraveling . suppose that the tunneling has occurred at time @xmath142 , corresponding to one of the possible histories in eq . ( [ eq : rho.tunneled ] ) . then , @xmath143 is the corresponding conditional density operator at later time . here , @xmath144 is the density operator after the elastic scattering and @xmath145 accounts for the propagation of the scattered electron . in comparison to eq . ( [ eq : psi.entangled ] ) the density operator of eq . ( [ eq : rho.entangled ] ) is diagonal in @xmath146 , i.e. , the elastic scattering has led to a destruction of the phase coherence ( dephasing ) . however , the spin part still shows the same degree of entanglement , where similar conclusions would apply for inelastic but spin - independent scatterings . thus , the decay of an optically excited charged - exciton indeed generates a robust spin entanglement between the electrons in the dot and reservoir . we finally comment on the process of disentanglement . in fact , any scattering channel which couples with unequal strength to the spins or affects only one spin orientation is responsible for such entanglement decay . to simplify our following analysis , we introduce operators acting in the reservoir subspace @xmath147 which are averaged over the orbital degrees of freedom @xmath146 . without specifying the details of such spin - selective scattering , we introduce the generic lindblad operator @xmath148 where @xmath149 is a scattering rate and @xmath49 a factor determining the asymmetry of the coupling to spin - up and spin - down electrons . as briefly sketched in appendix [ sec : two - level ] , the corresponding master equation can be solved analytically and we arrive at @xmath150 here , the first term on the right - hand side accounts for the unscattered state , whose probability decays with time , and the second term for the disentangled state after scattering . because of our assumption for the lindblad operators , eq . ( [ eq : lindblad.disentangle ] ) , the density operator after disentanglement still exhibits a classical correlation where the spins of the electrons in the reservoir and dot are antiparallel ; apparently , it would require additional spin - flip processes to completely decouple the two electrons . see also ref . for a discussion of how such spin entanglement could be measured experimentally and possible quantum information applications . of eq . ( [ eq : unraveling ] ) as discussed in the text . times are measured in units of the inverse tunneling rate @xmath151 ; we assume @xmath152 for the spin - unselective and @xmath153 for the spin - selective scatterings . the black dashed line shows the sum of @xmath154 and @xmath155 which corresponds to the degree of spin entanglement . ] we finally list in table [ table : entangler ] the various contributions of the cascade decay to the density operator , i.e. , corresponding to the step before tunneling , after tunneling , after dephasing , and after disentanglement . the probabilities for the corresponding contributions are determined by our unraveling procedure and can be obtained as follows : initially , the system is in the charged - exciton state , i.e. , @xmath156 ; because of tunneling @xmath157 decays mono - exponentially ; correspondingly , the increase of the contribution after tunneling @xmath158 is proportional to the decrease of @xmath157 , which again diminishes because of dephasing , and so forth . within this scheme , we can easily obtain the time evolution of the various probabilities as shown in the last column of table [ table : entangler ] , with the initial conditions @xmath156 and @xmath159 for @xmath160 . fig . [ fig : entanglement ] sketches the various contributions for representative values of @xmath138 and @xmath149 . in conclusion , we have analyzed single scatterings in single semiconductor quantum dots for two representative examples . first , for the cascade decay of a biexciton it has been shown that an exciton fine - structure splitting results for appropriately chosen polarization filters in an oscillatory behavior of the two - photon correlations . in our second example we have investigated the tunneling decay of a charged exciton inside a field - effect structure into a photo current , and have shown that the spins of the electrons in the dot and reservoir become entangled . we have discussed that this entanglement is robust against dephasing and spin - unselective scatterings , and thus benefits from the long spin lifetimes in semiconductors . our theoretical approach has been based on a master - equation approach in lindblad form in conjunction with the quantum - jump approach for the description of photon measurements . we have seen that the description scheme in terms of lindblad operators , suitable for the environment couplings of our present concern , is superior over related quantum - transport descriptions based on the markov and adiabatic approximations . @xcite first , no assumption about the final states of a ` scattering ' has to be invoked , and it suffices to assume a sufficiently short - lived memory kernel of the reservoirs instead . secondly , lindblad operators associated to measurement , e.g. , photon detection , allow for the application of the quantum - jump approach and provide a means to describe single - system dynamics . finally , the distinction of out- and in - scatterings in the master - equation makes possible a flexible solution scheme through unraveling , which appears to be highly elegant in particular for the cascade processes discussed in this work . we expect our findings to be useful for the analysis of time - resolved single - system measurements and for the simulation of quantum - information applications . work supported in part by the _ fonds zur frderung der wissenschaftlichen forschung _ ( fwf ) under project no . p15752n08 . in this appendix we show how to evaluate contributions of the form @xmath161 . first , the polarization vector of a photon with wavevector @xmath43 and polarization @xmath44 can be expressed as @xcite @xmath162 the integral over the angular - dependent part can be evaluated to @xmath163 i.e. , only terms with @xmath164 give a nonzero contribution . finally , @xmath165 is the wigner - weisskopf decay rate for a dipole radiator embedded in a medium with refractive index @xmath136 . in this appendix we discuss the master equation of a generic two - level system . since the pauli matrices @xmath166 together with the unit matrix @xmath167 form a complete basis , we can expand the density operator as with the vector @xmath169 . from @xmath170 we find @xmath171 and @xmath172 implies that @xmath173 is real . analogously , a generic lindblad operator is of the form @xmath174 , where @xmath175 and @xmath176 can be decomposed into real and imaginary parts . the master equation corresponding to this lindblad operator is of the form then , for the lindblad operator @xmath179 of eq . ( [ eq : lindblad.disentangle ] ) for disentanglement we find @xmath180 and @xmath181 , which can be easily solved to arrive at the final result of eq . ( [ eq : disentangled ] ) . such selective biexciton population could be achieved similar to quantum - dot based single - photon experiments : @xcite first , a short laser pulse creates electron - hole pairs in continuum states in the vicinity of the quantum dot ; carrier capture results in a multi - exciton complex within the dot , which decays through successive emission of photons ; because of environment coupling , after each photon emission the system quickly relaxes to the electron - hole state of lowest energy . in what follows , it is sufficient to assume that during this cascade process the dot will eventually pass through state @xmath5 . such selective charged - exciton occupation can be achieved as follows . suppose that initially the spin direction of the doping electron is undetermined , i.e. , the corresponding density operator is a mixture @xmath182 . when at time @xmath183 the dot is subject to an unpolarized optical @xmath184-pulse @xcite the system will be excited in accordance to our assumption to state @xmath127 .
we employ the quantum - jump approach to study single scatterings in single semiconductor quantum dots . two prototypical situations are investigated . first , we analyze two - photon emissions from the cascade biexciton decay of a dot where the single - exciton states exhibit a fine - structure splitting . we show that this splitting results for appropriately chosen polarization filters in an oscillatory behavior of two - photon correlations , and carefully examine the proper theoretical description of the underlying scattering processes . secondly , we analyze the decay of a single - electron charged exciton in a quantum dot embedded in a field effect structure . we show how the quantum properties of the charged exciton are transferred through tunneling and relaxation to the spin entanglement between electrons in the dot and contact , and identify the pertinent disentanglement mechanisms .
in cryoelectron microscopy , unstained biological specimens are rapidly vitrified at cryogenic temperatures @xcite . consequently , artefacts due to heavy - metal staining , desiccation , and other sample preparation processes are no longer an issue . however , the frozen , hydrated biological specimen , consisting mostly of light elements , scatters electrons weakly . hence biological specimens generally are weak phase objects associated with low image contrast @xcite . in this setting , the resolution is limited by radiation damage by the probe electrons @xcite to approximately 5 - 10 nm in the case of _ single _ objects @xcite . this leaves much to be desired because 2 nm resolution would be needed to identify molecules in frozen vitrified slices of the cell in cryoelectron tomography @xcite , or 0.8 nm resolution would be required to observe the secondary structure of a single protein molecule . the reason why radiation damage limits the resolution is that the safe electron dose , which does not cause sizable damage to the specimen , is so small that the low - contrast image is dominated by shot noise . shot noise is a manifestation of the particle nature of the electron and hence is fundamental . several approaches to address the radiation damage problem are known . first , methods based on averaging , such as two - dimensional crystallography @xcite and single - particle analysis @xcite , represent an established branch of methodology in structural biology . in favorable cases , these methods essentially attained atomic resolution @xcite . however , in order to average out the noise , this approach requires at least thousands of copies of the molecule of interest without much structural variance and hence is not suited for soft or unique objects . second , the advent of in - focus phase contrast electron microscopy @xcite enabled researchers to see weak phase objects much clearer than hitherto possible . the reason is that it provides a well - behaving phase contrast transfer function ( ctf ) that does not fall to zero at low resolutions and does not oscillate at high resolutions . however , this method does not go beyond the standard quantum limit , as will be mentioned . third and finally , the use of low acceleration voltage down to , e. g. 20 kv reduces radiation damage . in particular , one may choose the electron energy below the onset of relevant knock - on damage thresholds , although the dominant damage mechanism in biological cryoelectron microscopy is considered to be ionization events due to radiolysis @xcite . there have recently been much effort on developing aberration - corrected low - voltage transmission electron microscopes ( tems ) and scanning tems ( stems ) around the globe @xcite . while this approach certainly makes sense , the inelastic mean free path is considerably shorter at lower electron energies , necessitating preparation of ultra - thin specimens @xcite . while there have been reports of impressive images of _ thin _ biological molecules @xcite and organic molecules @xcite since decades ago , hydrated macromolecules of biological interest can have the size more than 50 nm . this makes it necessary to study thicker specimens and hence , if we insist on studying biologically interesting frozen hydrated macromolecules or vitrified slices of the cell , the use of a sufficiently high acceleration voltage is necessary . overall , these considerations suggest that no current electron microscopy method produces , in a robust and widely applicable manner , images of single hydrated objects of biological interest at a resolution below 2 nm . furthermore , phase measurement in all electron microscopy methods developed thus far , including all mentioned above , is governed by the shot noise limit , or in other words , the standard quantum limit . in this case , the precision @xmath0 of the measurement of the small phase shift @xmath1 , associated with the biological specimen , scales with the number of electrons @xmath2 as @xmath3 . that the standard quantum limit can be beaten is well known in the field of quantum metrology @xcite . while the field has begun decades ago @xcite , a sizable fraction of recent activities in quantum metrology , mostly in the context of optics , revolves around the idea of employing entangled quantum states @xcite . in particular , measurement of a phase shift @xmath0 is a familiar objective in quantum metrology and it is relevant also to biological electron microscopy where weak phase objects are dealt with . a major question in quantum metrology is how the measurement precision @xmath0 varies with the amount of relevant resources @xmath2 . among others , query complexity emerged as one of the most useful resource count @xcite , which has been further elucidated and shown to be governed by the heisenberg limit @xmath4 @xcite . roughly , the number of queries equals the number of interactions with the entity to be measured . hence query complexity nicely captures the resource in biological electron microscopy , where the experimenter wants to get the most out of each query , which corresponds to passing of each electron in the specimen . it is worth noting here that , notwithstanding the recent theoretical @xcite and experimental @xcite reports of beating the heisenberg limit , as long as the number of query is taken as the resource count , the heisenberg limit does represent the fundamental limit @xcite . it should be noted that the heisenberg limit is by no means an easy target in real situations @xcite . it has increasingly been recognized that entanglement - assisted measurement is vulnerable to lossy processes and a number of recent studies address this problem @xcite . it is now established that quantum metrology , in lossy situations , offers merely a constant - factor improvement in the measurement precision , as opposed to the quadratic asymptotic improvement by a factor proportional to @xmath5 . nevertheless , a constant - factor improvement is all that we want in biological electron microscopy , or perhaps in any measurement for that matter . the real question is whether the value of the constant factor enables relevant specific resolutions , such as aforementioned 2 nm or 0.8 nm . this requires an analysis specific to biological cryoelectron microscopy , but in doing so , we may learn lessons of more general character . on a related note , a recent work @xcite proposes an efficient multi - pixel phase estimation method that takes advantage of quantum entanglement . while interesting , the scheme employs a delocalized probe state over all the pixels and hence it is unlikely to be robust against localizing lossy processes , which occur in biological electron microscopy . it should also be noted that , in general , beating the standard quantum limit does not necessarily require an entangled quantum state @xcite . repeated use of a probe particle would suffice . however , in the present case of entanglement - assisted biological electron microscopy , entangled quantum states are necessary for a somewhat mundane reason that the repeated use of an electron would result in too large a scattering angle to be handled by practical electron optics . the use of quantum advantage specifically in the context of biological electron microscopy has been discussed for some time now , albeit mostly from the theoretical perspective @xcite . in this paper we build on the recently proposed scheme that uses quantum entanglement between the probe electron and the cooper pair box ( cpb ) placed on an electron mirror @xcite . the scheme realizes what effectively amounts to multiparticle entanglement between the probe electrons , by way of sequential interactions between each electron and the cpb . a major objective of this paper is to investigate how and to what extent the effect of lossy processes may be mitigated in the proposed scheme . a distinct quantum approach @xcite to biological electron microscopy based on interaction - free measurement @xcite , also suggested in ref . @xcite , has been proposed . while interesting , we note that there is certain limit associated to this type of approach @xcite . this paper is organized as follows . we first review the entanglement - assisted electron microscopy scheme in sec . [ sec : review - ent - em ] . this is followed by sec . [ sec : coherent errors ] , in which we deal with errors that is not related to lossy processes . in sec . [ sec : effect - of - inelastic ] , we study how inelastic scattering processes make adverse effects to the scheme . we also show that an appropriate design of the electron probe geometry enables us to combat this type of effect . [ sec : conclusion ] concludes the paper . symbols @xmath6 respectively denotes the absolute value of the electron charge and the electron mass . a _ diffraction plane _ refers to any plane conjugate to the back focal plane of the objective lens . since aberration - corrected electron optics is becoming a commonplace , we generally neglect lens aberrations unless explicitly stated otherwise . the probe electron energy is assumed to be @xmath7 throughout the paper . here we briefly review the microscope proposed in ref . @xcite , in part because we also want to fix notations @xcite . the scheme takes advantage of superconducting quantum electronics that includes a single cpb @xcite , presumably in the circuit quantum electrodynamics configuration @xcite . the microscope is designed to measure the difference @xmath8 of phase shifts @xmath9 and @xmath10 that are respectively associated with two regions @xmath11 and @xmath12 on the biological specimen , which generally is a weak phase object . multitude of such a measurement form an image . on the face of it , the shapes of the two regions @xmath11 and @xmath12 may be chosen arbitrarily : for example , they may be adjacent two small pixels , or it may be that @xmath11 surrounds a smaller region @xmath12 . however , we will find in the present work that the latter choice is much better . the cpb is placed on the surface of an electron mirror , which in turn is inserted between the pulsed electron gun and the condenser lens so that the electron emitted from the electron gun first interacts with the cpb before going to the specimen ( see fig . [ fig : ent - asst - tem structure ] ) . we use only two states @xmath13 and @xmath14 of the cpb , that respectively denote quantum states of the cpb with zero or one excess cooper pair , where the subscript b stands for the word box. the main function of the electron mirror in this scheme is to modify the electron beam shape , depending on the charging state of the cpb : the electron mirror is designed in such a way that the beam trajectory is so sensitive to the cpb charge state that only a single cooper pair in the cpb is sufficient to direct the electron beam to the specimen region @xmath12 that would otherwise go to @xmath11 . the scattered electrons then go through more or less ordinary electron optics and are detected at a plane , which is not conjugate to the image plane , by an area detector with high quantum efficiency . alternatively , unlike the case of ref . @xcite , we will assume that the probe electrons are detected on a diffraction plane in the present work . in this case , it may well be possible to get rid of the objective lens ( ol ) and the projector lens system ( pls ) shown in fig . [ fig : ent - asst - tem structure ] altogether , resulting in an instrument resembling the scanning tem ( stem ) . we mention several engineering challenges involved . a more complete discussion of most of what follows has been presented elsewhere @xcite . there are mainly five aspects that need to be developed . first , while the electron mirror has long history , an electron mirror that works at temperatures generated by the dilution refrigerator , i.e. a few tens of millikelvin , must be developed . second , a pulsed electron gun @xcite , with @xmath15 energy spread and ps pulse width , must be developed . each pulse may or may not contain an electron for the scheme to work . we note that @xmath15 energy spread has already been realized in the field of high resolution electron energy loss spectroscopy ( hreels ) where very low energy ( @xmath16 ) electrons are dealt with @xcite . in a similar fashion , the part of the electron optical system shown in fig . [ fig : ent - asst - tem structure ] , consisting of the pulsed electron source , monochromator , electron beam separator and the electron mirror , controls electrons with such low kinetic energy . unlike the case of tem - eels , we do _ not _ need a several hundred kev high voltage source with the voltage stability comparable to the energy spread ( @xmath17 ) of the electron beam . third , the cooper pair box on the electron mirror must be operated in a controlled fashion and such operations must be synchronized with the pulsed electron gun . fourth , as noted earlier , the electron optical system must be sensitive enough so that a single excess cooper pair in the cooper pair box can be seen with the electron optics . fifth and finally , after interacting with the cooper pair box , the probe electron must be accelerated to several hundred kev before interacting with the biological specimen . in principle , this can be done by brute force , i.e. by electrically floating a large part of the electron microscope on either side of the accelerator acc shown in fig . [ fig : ent - asst - tem structure ] by several hundred kilovolts @xcite . a recent development in this connection , on the other hand , is the use of radio - frequency accelerator that has been applied to tems @xcite and it seems to have good compatibility with a pulsed electron beam . if this approach moves beyond the proof - of - principle stage , then it will be unnecessary to electrically float such a large part of the microscope as described above . structure of the proposed transmission electron microscope . electrons emitted form a pulsed electron source ( pes ) goes through a monochromator ( mc ) and are directed by an electron beam separator ( ebs ) towards a low temperature electron mirror ( em ) , on which a cooper pair box ( cpb ) is placed . all these processes happen with a very low kinetic energy of the electron . afterwards , the electron is accelerated in an accelerator ( acc ) , goes through a condenser lens ( cl ) , specimen ( sp ) , objective lens ( ol ) , projector lens system ( pls ) and finally detected by an area detector ( ad ) . the ol and pls may not be necessary in actual settings.,scaledwidth=45.0% ] we describe the _ single electron process _ that lies at the heart of our method . as shown later , mutiple single electron processes ( @xmath18 of them ) will comprise a _ k - electron process _ in our scheme . the single electron process involves a single imaging electron and proceeds as follows . suppose that the cpb is in a superposed charge state @xmath19 before the electron from the electron gun approaches the cpb . after the electron interacts with the cpb , it gets entangled with the cpb so that the whole quantum state of the cpb - electron system is@xmath20 where the electron states @xmath21 and @xmath22 respectively denotes electron waves incident on @xmath11 and @xmath12 regions on the specimen . upon transmission through the specimen , the electron wave in the state @xmath22 acquires a phase factor @xmath23 relative to the state @xmath21 , corresponding to the phase shift @xmath24 . hence the state becomes @xmath25 . let the electron state @xmath26 represent a diffracted wave to the @xmath27-th pixel of the area detector . we expand the electron states @xmath28 in terms of these states as@xmath29 using these , we obtain another expression of the state @xmath30 as@xmath31 upon detection of the electron at the @xmath27-th pixel of the area detector , the cpb is left in the state @xmath32 where @xmath33 is a real normalization factor . in reality , an electron wave is not a qubit . one way to make the above argument somewhat more precise is to expand the state @xmath22 as@xmath34@xmath35 where each state on the right hand side is spatially more localized on the specimen than the state @xmath22 is . the state becomes , after transmission through the specimen,@xmath36 in this case , the above phase shift value @xmath24 would represent something similar to , but not identical with , the average of @xmath37 . this way of representation would be more suitable when we deal with a specimen with a structure within the region of the electron beam . however , we will not pursue this representation . the point is also related to the _ similar intensity map condition _ @xcite , to which we now turn . to simplify the analysis , we assume that @xmath38 , which amounts to saying that the electron waves @xmath28 from the two small regions @xmath39 have similar intensity on the area detector . this assumption is what we call the similar intensity map condition . the assumption appears to be natural , but to what extent the similar intensity map condition is satisfied must be carefully evaluated , which will be done in sec . [ sec : coherent errors ] . once we accept @xmath38 , we may write @xmath40 , where @xmath41 represents a _ known _ phase shift associated with the optical path length difference between two electron trajectories , starting respectively at the specimen regions @xmath39 and ending at the @xmath27-th pixel of the area detector . the cpb state ( [ eq : post cpb state-1 ] ) may then be expressed as , up to the overall phase factor,@xmath42 since we know @xmath41 , by quantum information processing machinery developed for the cpb @xcite we may nullify it to obtain@xmath43 let us call this last step as the _ phase compensation step_. comparing equations ( [ eq : pre cpb state ] ) and ( [ eq : post cpb state-3 ] ) , we see that , assuming that the similar intensity map condition is satisfied , the net effect induced by the single electron process is multiplication of the phase factor @xmath23 to the cpb state @xmath14 relative to @xmath13 . the k - electron process in our scheme consists of , in the order of execution , the cpb initialization step , repeated single electron processes that is described above ( @xmath18 times ) , and the cpb readout step . first , the cpb is initialized to the state @xmath44 . second , the single electron process is repeated @xmath18 times , resulting in the state@xmath45 this state may be expressed as , up to the overall phase factor , @xmath46 where @xmath47 and @xmath48 . finally , in the cpb readout step we measure the cpb state with respect to the basis @xmath49 . probabilities to find the state in these basis states are respectively@xmath50 notice , by the way , that the @xmath41 correction step between the states ( [ eq : post cpb state-2 ] ) and ( [ eq : post cpb state-3 ] ) within the k - electron process may collectively be postponed until right before the cpb readout step . in this case , all @xmath18 phase correction values are recorded ; and their sum is used in the combined correction step . we show how the heisenberg limit can be approached . we write the electron dose for a measurement on a single spot on the specimen @xmath2 , which must be kept below a certain value in order not to damage the specimen significantly . hence we can repeat the k - electron process @xmath51 times ( that we assume to be an integer for simplicity ) . as is well known from the theory of binomial distribution , the random variable @xmath52 representing the number of finding the cpb state in the state @xmath53 has the expectation value @xmath54=np_{\leftarrow}$ ] and variance @xmath55=np_{\leftarrow}p_{\rightarrow}$ ] . to simplify the analysis , we assume that @xmath56 , although this may not necessarily be a good approximation in some of actual settings . then we obtain @xmath57 and @xmath55=n/4+o\left(nk^{2}\delta\varphi^{2}\right)\cong n/4 $ ] . introducing another random variable @xmath58 , we have@xmath59=\delta\varphi,\textrm { } var\left[y\right]=\frac{4}{k^{2}n^{2}}var\left[x\right]=\frac{1}{kn}.\ ] ] hence , the expected error in determining @xmath24 with the k - electron process is @xmath60 . note that this represents the standard quantum limit when @xmath61 and the heisenberg limit when @xmath62 . we remark that , despite being sequential , our scheme may also be regarded as one using something akin to the bosonic noon state @xcite . suppose that we had an unusually long electron microscope column between the cpb - incorporated electron mirror and the specimen . then all @xmath18 electrons would be flying simultaneously in the column before any of them colliding with the specimen . from the discussions above , we see that these electrons would be in the state@xmath63 where @xmath64 denotes the state in which all electrons are in the state @xmath21 , and likewise for @xmath65 . there are at least three sources of errors in the scheme reviewed in this section . first , to what extent the similar intensity map condition holds should be investigated , as remarked above . if this condition is significantly violated , then the absolute values of the coefficients @xmath66 in equation ( [ eq : pre cpb state ] ) deviate from the ideal @xmath67 , making the contrast of the final measurement weaker . second , when inelastic scattering happens , which is indeed @xmath68 times more likely to happen than the elastic counterpart in cryoelectron microscopy , we lose some coherence in the cpb state . in particular , if the inelastic event is _ localized _ , then the specimen would measure the probe electron state with respect to the localized basis states @xmath28 . in this case , the coherence would be completely lost . however , the actual situation is subtler and we will investigate how things should behave . third , we have not thoroughly analyzed the cpb - electron entangling interaction to estimate what and how much errors would be introduced when the electron bounces off the cpb in the electron mirror . we will consider the first two sources of errors in this paper , while leaving the last item for future analysis . what frequency of faulty measurement can we tolerate ? to get a rough idea , let us simplify the situation and consider a totally destructive event that projects the cpb state onto either @xmath13 or @xmath14 . suppose that such a destructive event occurs at random with the probability @xmath69 . the number of k - electron process is @xmath51 and any k - electron process fails if a destructive event occurs more than once during @xmath18 single electron processes . ( for simplicity , we assume that the experimenter , being somewhat lazy , does not abort the k - electron process upon such a destructive event . ) the success probability of a k - electron process is @xmath70 if @xmath69 is sufficiently small . the number of k - electron processes for the same dose @xmath2 is @xmath71 . retracing the argument in the previous subsection [ sub : a - brief - review ] , we have the expectation value @xmath72=n'p_{\leftarrow}$ ] , variance @xmath73=n'p_{\leftarrow}p_{\rightarrow}$ ] , @xmath74 , and eventually@xmath75=\delta\theta,\textrm { } var\left[y'\right]=\frac{1}{k^{2}n'}=\frac{1}{kne^{-kp_{d}}}.\ ] ] we want the variance @xmath76 $ ] smaller than the variance in the conventional case @xmath77 . to find @xmath78 that minimizes the ratio @xmath76/n^{-1}$ ] , we differentiate this expression and equate the result to zero . we find @xmath79 and this result remains valid when , e. g. , we minimize the ratio of standard deviations @xmath80/n^{-1}}$ ] instead of @xmath76/n^{-1}$ ] . when errors are not totally destructive , the error will still accumulate in the random - walk - like fashion and will eventually become totally destructive . in this case , we may redefine @xmath69 accordingly ( which we do not do in this paper ) so that , on average , it takes approximately @xmath81 single electron processes for the accumulated error to be fully destructive . finally , we remark on relativistic effects . since the tem we have in our mind may well operate at voltages as high as 300 kev , there will be relativistic effects . in the rest of the paper we use the so - called relativistically corrected schrodinger equation , which is a good approximation to the dirac equation in our case @xcite . the equation _ is _ the standard time - independent schrodinger equation , but the mass @xmath82 and the energy @xmath83 are interpreted respectively as @xmath84 and @xmath85 , where @xmath86 , @xmath87 , @xmath88 is the electron rest mass , and @xmath89 is the relativistic electron energy including the rest energy @xmath90 . in this section , we study the extent to which the similar intensity map condition @xmath38 is satisfied , for two cases of the probe electron beam geometry . first we study a simple case , in which the two regions @xmath11 and @xmath12 are adjacent two small regions of the same size of @xmath91 . we consider a @xmath92 thick specimen . we assume that the electrons are detected on a diffraction plane . then the problem is effectively about the diffraction pattern from a small spot and the question is how smooth the intensity on the detector is . let us focus on the region @xmath11 . let the @xmath93-axis be parallel to the direction of the incident wave propagation and let the position of the specimen be at @xmath94 . atoms in the specimen are labeled by an integer @xmath95 and are respectively located at @xmath96 . let @xmath97 be the scattering amplitude of the @xmath95-th atom corresponding to its elemental identity . we assume the electron waves to be monochromatic and write the wavenumber @xmath18 . we ignore multiple scatterings in the specimen . we want the incident beam to be gaussian with the waist size @xmath98 and assume that the diffraction - limited waist of the gaussian beam is on the specimen . without loss of generality , we compute the wave amplitude at a point p in the far field on the @xmath99-plane , where @xmath100 is large and @xmath101 coordinate is given by @xmath102 . far from the specimen , the transmitted gaussian wave has the known form in the paraxial approximation ( see appendix a ) @xmath103 that can be approximated as@xmath104 consider a single elastic scattering event . the amplitude of the scattered wave at the point p is found to be ( see appendix a ) @xmath105 for simplicity , let us first consider the contributions to the wave amplitude @xmath106 only from _ uniformly _ distributed carbon atoms , which we write @xmath107 . the scattering amplitude @xmath97 is that of carbon @xmath108 . let the number density of carbon atoms per unit area perpendicular to the optical axis be @xmath109 . the value of @xmath107 is@xmath110 @xmath111 simple addition will extend this result to the case of multiple chemical elements . comparing with equation ( [ eq : transmitted wave far field 2 ] ) , we see that the scattered wave @xmath107 does almost nothing more than shifting the phase of the transmitted wave , especially when the angular spread of @xmath112 is larger than the gaussian incident beam spread , which is about a few mrad in the present case . it is the random , as opposed to uniform , arrangement of atoms that we need to analyze to evaluate how well the similar intensity map condition is fulfilled . note that @xmath93-coordinates @xmath113 of atoms do not appear in equation ( [ eq : scattered wave 1 ] ) because of approximations discussed in appendix a. hence , what matters is only the projected atomic distribution of the specimen to a plane perpendicular to the optical axis . henceforth we assume that such a projected atomic distribution is completely random . this assumption obviously fails if , for example , the specimen is atomically thin because interatomic distance can not be too small in real specimens , and there may well be other objections to this assumption . however , we assume that this assumption will give us a good guide especially when the specimen is sufficiently thick . computation of equation ( [ eq : scattered wave 1 ] ) beyond the average case resembles the analysis of random walk . let the specimen plane be divided into many thin concentric rings centered at the optical axis with the radius @xmath114 and small width @xmath115 . label the rings with a natural number @xmath27 in the order of increasing @xmath116 . assume that , despite the small width @xmath115 , each ring still contains many atoms . again for simplicity , we consider only carbon atoms . let the number of carbon atoms in the @xmath27-th ring be @xmath117 . equation ( [ eq : scattered wave 1 ] ) is then rewritten as @xmath118 where @xmath119 is a set of integers consisting of the indices of carbon atoms that belong to the @xmath27-th ring . hence the cardinality of @xmath119 is @xmath117 . let @xmath120 be large enough so that all sort of phase angles in the range @xmath121 appear in the sum @xmath122 . in other words , we consider the region outside the peak of transmitted wave in the diffraction pattern . in this case , we may regard the sum as 2-dimensional random walk on the gaussian plane and we have @xmath123 , where @xmath124 is a random phase . more precisely , we accept that the angle @xmath125 behaves as a random variable with the uniform distribution because of the large @xmath120 and the supposedly random arrangement of the carbon atoms from location to location within the specimen . the phase factors @xmath126 are mutually independent random variables and @xmath107 is also a random variable in this view . clearly , the expectation value @xmath127 is zero because of the random phase , which is consistent with what equation ( [ eq : scattered wave 1c ] ) states . ( note that the expectation value here is with respect to the random arrangement of atoms and has nothing to do with the randomness in quantum measurement . ) however , the expectation value of the absolute square of the wavefunction is nonzero because multiplication of phase factors @xmath128 results in @xmath129 when @xmath130:@xmath131^{2}\sum_{j}n_{j , c}e^{-\frac{2r_{j}^{2}}{w_{0}^{2}}}=\left[\frac{f_{c}\left(\theta\right)}{r}\right]^{2}\sum_{j}n_{c}e^{-\frac{2r_{j}^{2}}{w_{0}^{2}}}2\pi r_{j}\delta r_{j}\ ] ] @xmath132^{2}\int_{0}^{\infty}n_{c}e^{-\frac{2r^{2}}{w_{0}^{2}}}2\pi rdr=\frac{\pi}{2}n_{c}w_{0}^{2}\left[\frac{f_{c}\left(\theta\right)}{r}\right]^{2}\ ] ] since we assume that the atomic arrangements are independent to each other among the chemical elements , the contributions from elements other than carbon can be accounted for by simple addition to the above expression . crucially , the variance of @xmath133 , that is @xmath134 , equals @xmath135 . this may be verified by noting that multiplication of four phase factors @xmath136 results in @xmath129 when either _ _ \{@xmath130 and @xmath137 } , or _ _ \{@xmath138 and @xmath139 } holds . ( we ignore the relatively rare case of @xmath140 . ) hence the diffraction pattern intensity fluctuates as much as its own average . ( the physical manifestation of it is known as the speckle pattern . ) this in turn means that the similar intensity map condition @xmath38 is _ not _ satisfied whenever an electron is detected outside the central peak described in equation ( [ eq : transmitted wave far field 2 ] ) . since typical angular spread associated with an elastic electron scattering @xmath141 is much larger than the spread of the incident beam @xmath142 ( see appendix a ) , essentially every elastic scattering process fails to satisfy the similar intensity map condition , and hence entails destruction of the cpb quantum state . this suggests that the ordinarily welcome high - angle elastic scattering processes do some harm here , because it provides unwanted high - resolution information that generates the uncontrollable speckle pattern . this is a rather curious feature of quantum measurement in this particular case , but the feature may be more generic . if so , we might state that we generally do not want to know more than needed in quantum measurement . to compute the frequency of elastic scattering , it suffices to know the total elastic scattering cross sections for each relevant chemical element@xcite , the number of atoms per unit area for these elements ( see appendix b ) , and the specimen thickness that we assume to be @xmath143 . it is found that @xmath7 electrons are elastically scattered with the probability @xmath144 . if inelastic processes _ were _ negligible , according to the argument in sec . [ sub : several - remarks ] , this means that we can use @xmath145 electrons on average in a k - electron process , which would results in a contrast enhancement by a rather modest factor @xmath146 . moreover , inelastic processes of course are not negligible and we will deal with them later . we now turn to the second case , where a circular region @xmath11 surrounds a smaller circular region @xmath12 . these two regions share the central point . this configuration appeared in a simulation study previously reported @xcite . to be specific , we consider the following situation : the incident beam has a relatively large divergence angle @xmath147 ( notice that we add primes to these variables in the present diverging beam case ) , which is larger than the characteristic angle @xmath141 of the scattering amplitude functions @xmath148 . for @xmath7 electrons , the diffraction - limited waist size @xmath149 of the gaussian beam with the divergence angle @xmath150 is @xmath151 , which appears to be feasible , considering there already is a @xmath152 probe @xcite . we set the distance @xmath153 between the incident beam waist and the _ central plane _ , which is at the midpoint between the entrance and exit surfaces of the @xmath143-thick specimen , to be @xmath154 or @xmath155 for the regions @xmath11 and @xmath12 , respectively . hence , on the entrance surface , the beam diameter is @xmath156 and @xmath157 for the regions @xmath11 and @xmath12 , respectively . the beam size increases to @xmath158 and @xmath156 respectively for the cases @xmath11 and @xmath12 at the exit surface of the specimen . this beam spread is large , but we must accept this because of plasmon scattering discussed in sec . [ sec : effect - of - inelastic ] . ( suitably designed data processing may solve this large - beam - spread problem . however , it goes beyond the scope of this paper . ) again , the electrons are detected on a diffraction plane . we define a parameter @xmath159 , which turns out to be small ( @xmath160 ) in our case ( appendix a ) . while for transmitted wave we can use equation ( [ eq : transmitted wave far field 2 ] ) with @xmath98 replaced with @xmath149 , the amplitude of the scattered wave in the present case is found to be ( see appendix a)@xmath161}e^{-\varepsilon\frac{k}{2\delta z}\left(x_{s}^{2}+y_{s}^{2}\right)}f_{s}\left(\theta-\frac{x_{s}}{\delta z}\right)\label{eq : scattered wave 2}\ ] ] where @xmath120 again is the scattering angle towards @xmath162 direction . to evaluate equation ( [ eq : scattered wave 2 ] ) , we again focus on carbon atoms and first assume that these atoms are uniformly distributed . then the above is@xmath163}e^{-\varepsilon\frac{k}{2\delta z}\left(x^{2}+y^{2}\right)}\ ] ] @xmath164 the integral is a convolution of two functions @xmath165 and @xmath166 . the former has two factors . the factor @xmath167 quickly goes to zero above @xmath168 and the other factor @xmath169 rapidly oscillates also above @xmath170 ( see appendix a ) . the latter function @xmath166 has a broader profile than the former when plotted against @xmath171 , by a factor @xmath172 . a crude but useful approximation to equation ( [ eq : uniform scattered wave 2 ] ) is to replace @xmath173 with @xmath174 because unless the argument of the function @xmath108 is close to zero , the factor @xmath175 oscillates anyway , making its contribution to the integral small . ( we neglected the fact that the width of @xmath108 is not so wide . this is why the approximation is crude . ) once we accept this , the integration in equation ( [ eq : uniform scattered wave 2 ] ) can be carried out , e.g. by noting that it is a convolution of two gaussian functions , to obtain@xmath176 or equivalently@xmath177 where @xmath178 is the half angle of the incident wave . comparing with equation ( [ eq : transmitted wave far field 2 ] ) , if we take the two @xmath179 factors as close to @xmath129 , then equation ( [ eq : uniform scattered wave 4 ] ) represents a phase shift of the incident wave by an angle @xmath180 , as long as this quantity is small . this finding is consistent with the result on weak phase object described in appendix a of ref . @xcite . again , in order to evaluate the validity of the similar intensity map condition , random atomic distributions must be considered instead of the uniform distribution . in this case , the scattered wave spreads with the half angle that roughly equals addition of @xmath181 and the characteristic angle of elastic scattering @xmath182 . ( in the above uniform atomic distribution case , the half angle does not exceed @xmath181 because beyond this angle , scattered waves from the atoms destructively interfere . ) let us consider two angular regions separately . first , if the angle @xmath120 is less than @xmath181 , the transmission wave @xmath183 is the largest component that goes into this angular region . the scattered waves from each atom , on the other hand , constructively interfere more or less , because of the small @xmath120 . the resultant @xmath184 , with @xmath185 phase shift , adds to @xmath183 . as noted above , @xmath186 is a phase - shifted ( by the angle @xmath180 ) version of the incident wave . if the atomic density @xmath109 varies due to randomness , then the phase shift also varies . however , this randomness represents exactly what we want to measure . the phase shift @xmath180 may not be small by itself , but the difference of two phase shifts @xmath24 between the regions @xmath11 and @xmath12 are small . indeed , at the @xmath187 scale , @xmath24 is expected to be @xmath188 for @xmath7 electrons @xcite and what this phase difference is expected to cause on the wave amplitude should be equal to or higher than the second - order correction , which roughly is @xmath189 . ( in order to exactly compute the second order correction , we must go beyond the first - order perturbation theory , which we do not do in the present work . ) hence in this region of @xmath190 , we do not expect problems regarding the similar intensity map condition . second , consider the outer region @xmath191 . here , we expect the speckle pattern and hence we assume that the scattered waves from each atom add with random phases . hence , the average intensity in this region is the sum of the intensities of each atom:@xmath192 @xmath193 @xmath194 or equivalently , @xmath195 where @xmath196 is the scattering amplitude normalized by the bohr radius @xmath197 that is provided in ref . @xcite and @xmath198 is the number of carbon atoms in the area @xmath199 . this expression may easily be extended to the multiple chemical element case by simple addition . we get the standard deviation of the amplitude @xmath200 , @xmath201 which provides essentially the expected amplitude , but this amplitude should fluctuate much , as in the focused incident beam case described in sec . [ sub : focussed - incident - beam ] . hence if we detect an electron in the region @xmath191 , then the measurement would be spoiled because the similar intensity map condition is not valid in this region . more precisely , at angles @xmath120 where @xmath202 , we expect that the similar intensity map condition is not satisfied . the condition is written down explicitly as@xmath203 where @xmath204 . let the smallest @xmath120 that satisfies equation ( [ eq : condition ] ) be @xmath205 . the function @xmath206 can be evaluated numerically using published data @xcite for elements h , c , n , o , s , which results in @xmath207 . finally , we compute the probability @xmath208 of undesirable elastic scattering that satisfies the condition ( [ eq : condition ] ) . this can be expressed as the ratio between the integrated scattered wave intensity into undesirable angular region and the integrated incident wave intensity.@xmath209 from equation ( [ eq : transmitted wave far field 2 ] ) , the denominator is computed as@xmath210 the numerator involving @xmath211 should be handled numerically . we obtain a much more favorable figure than the focused incident beam case@xmath212 the inverse of this is @xmath213 . this is more than sufficient because it would make @xmath214 for large values of @xmath24 . it should be noted that the above computation is little more than order estimation because we simply asserted that @xmath202 means complete spoiling of the measurement , and have not addressed the intermediate - level damaging of the cpb state etc . however , complete characterization of such processes would demand full computer simulations and is beyond the scope of the present paper . we begin by reviewing several known facts , many of them found in @xcite , about inelastic electron scattering , which have been established through decades of theoretical and experimental studies . inelastic scattering by definition involves excitation of electrons within the specimen , and is _ not _ characterized by energy loss _ per se_. ( in fact , the probe electrons do lose a small amount of energy upon elastic scattering , either by high - angle scattering or generation of phonons . ) conversely , elastic scattering processes are simply ones that are not inelastic scattering , and is not characterized by the ability to produce interference fringes . in fact , interference has been observed in experiments on inelastic scattering . inelastic scattering processes roughly divide into excitations of outer - shell electrons and the inner - shell counterpart . although inner shell excitation processes are generally important in eels because it carries elemental information , the associated cross sections are small and these are insignificant in the present context , as will be shown later . the outer shell excitations , on the other hand , are more complex because the states of valence electrons depends on the chemical environment of the atom under study . these excitations are typically collective in nature and in many cases , including the case of insulators and polymers , these are treated as plasmons with energy loss @xmath215 @xcite . because of the dependence on chemical bonding to neighboring atoms , inelastic scattering cross sections are difficult to predict with certainty . despite such uncertainty , it is generally accepted that energy loss at around @xmath215 does represent excitation of plasmons in vitrified biological specimens , while there are other relatively minor structures with smaller @xmath83 in the energy loss spectrum , which is thought to represent localized excitons . here are several additional facts about inelastic scattering . first , inelastic scattering experimentally has narrower scattering angle spread when compared to elastic scattering . as will be explained shortly , however , it is not easy to talk about a characteristic or average angle associated with it . second , inelastic scattering happens roughly twice as often as elastic scattering in organic materials . third , inelastic scattering comes from long - range coulombic interaction between the probe electron and a bound electron of an atom . hence inelastic scattering can happen even when the electron beam does not directly hit the atom @xcite . this is closely related to the concept of _ delocalization _ @xcite . bethe theory provides a double - differential inelastic scattering cross section with energy loss @xmath83 @xcite . it has the following dependence on the scattering angle @xmath120:@xmath216 where @xmath197 is the bohr radius , @xmath217 is the rydberg energy , @xmath88 is the electron rest mass , @xmath218 is velocity of the incident electrons , and @xmath219 is what is called the generalized oscillator strength ( gos ) . the gos is modified when a solid specimen is considered because of factors such as chemical bonding . however , what is modified is the energy dependence of the gos and the angular distribution of inelastic scattering is mostly robust @xcite . we evaluate equation ( [ eq : bethe formula ] ) at the typical plasmon energy loss @xmath220 . the angle @xmath221 is given as @xmath34 @xmath222 where @xmath82 is the relativistic electron mass . for @xmath7 electrons and @xmath220 , @xmath223 . let @xmath224 be the wavenumber of the incident electrons . it is tempting to identify @xmath225 with the delocalization length of inelastic scattering , which should give a limit of resolution when inelastically scattered electrons are used for imaging . however , the distribution @xmath226 has a long tail . consequently , identification of @xmath227 with the delocalization length has led to some confusion in the past because the apparent experimental resolution is much better than @xmath227 @xcite . equation ( [ eq : bethe formula ] ) allows us to compute the intensity pattern of inelastically scattered electrons at the diffraction plane . from this , one can infer the wavefunction @xmath228 , where @xmath116 is the distance from the atom that scatters the electron , of the inelastically scattered electron right after the scattering by assuming that @xmath228 is real . this is done by ignoring the @xmath120 dependence of the gos and fourier transform the square root of the diffraction intensity pattern to get the wavefunction back at the specimen plane . one obtains @xcite@xmath229 we first examine , at a relatively abstract level , how inelastic scattering affects entanglement - assisted electron microscopy . we saw in sec . [ sub : a - brief - review ] that the incident electron to the specimen is entangled with the cpb as equation ( [ eq : pre whole state ] ) indicates . we first simplify the situation and suppose that the electron beam excites a single atom in the specimen . in other words , we consider a localized excitation , such as excitation of an inner - shell electron . let the ground state and excited states of the atom be @xmath230 , where the subscript _a_ stands for the word atom. the initial state before interaction between the probe electron and the specimen is @xmath231 where we use the letter @xmath232 instead of @xmath233 ( see e. g. equation([eq : pre whole state ] ) ) to indicate that the atom is now included to the system under consideration . after collision between the probe electron and the specimen , probability amplitudes that correspond to the inelastic scattering develop , although these may be of relatively small amplitude . the wave function for the entire system is written as@xmath234 @xmath235 where @xmath236 represents zero - energy - loss component of the scattered electron wave and @xmath237 etc . are inelastically scattered electron waves . let these states be properly normalized , i. e. @xmath238 etc . then , for example , the coefficient @xmath239 is a probability amplitude for the probe electron going through the region @xmath11 to excite the atom to the second excited state . now , suppose that the state of the excited atom is found to be , for example , @xmath240 by a projective measurement on the atom . this corresponds to the situation where inelastic scattering actually happens . ( we of course do not measure the state of the atom experimentally , but we represent the whole complex processes after the inelastic scattering that involves thermalization , decoherence etc . by a single projective measurement where the measurement outcome is hidden to the experimenter . ) then the system consisting of the cpb and probe electron is left in the state @xmath241 we are in trouble here for the following reasons . the electron states @xmath237 are rather different from @xmath236 as shown in fig . if expressed as wavefunctions , for example , the state @xmath242 is a diverging gaussian beam as discussed in sec . [ sub : diverging - incident - beam ] . on the other hand , recall that the inelastically scattered electron wave generates the diffraction pattern consistent with equation ( [ eq : bethe formula ] ) and corresponds to the wavefunction ( [ eq : sqrt psf ] ) , _ if _ a plane incident wave is employed . because of the linearity of quantum mechanics , the wavefunction corresponding to the state @xmath243 equals the product of the gaussian incident wavefunction and the equation ( [ eq : sqrt psf ] ) . first , such a product may be vanishing if the overlap between the gaussian envelope and the function ( [ eq : sqrt psf ] ) is small . in this case , only one term of equation ( [ eq : delocalization principle3 ] ) survives and we get amplitude imbarance , or an amplitude error . second , the phase compensation step described in sec . [ sub : a - brief - review ] may fail because the optical path length difference is no longer the difference of distances between the detection pixel and the two regions @xmath11 and @xmath12 respectively . rather , this quantity now involves the _ modified _ two regions @xmath244 because the form of @xmath245 is localized at around the scattering atom and is different from @xmath246 . since we do not know the exact position of the inelastically scattering atom , we can not compensate for this modification when performing the phase compensation step . hence , we have problems in terms of both the amplitude and phase . is not shown because its amplitude is too small , due to the fact that the atom is outside the spatial extent of the state @xmath242.,scaledwidth=45.0% ] since the inner - shell excitations are problematic for the above reasons , we have used the sigmak3 and sigmal3 programs @xcite to evaluate the total k - shell and l - shell cross sections for @xmath7 electrons for the elements c , n , o , and s. ( hydrogen does not have an inner shell . ) we combined the result with data presented in appendix b , which can easily be translated to atomic number density - per - area data for a typical @xmath143-thick biological specimen . we find that the probability for a single incident electron to cause k - shell or l - shell ionization in such @xmath143-thick specimen is @xmath247 . this figure is small enough to perform entanglement assisted electron microscopy to our advantage . the most dominant outer - shell excitations are usually plasmons and here we focus on inelastic scattering by plasmons . let us first develop a simple but admittedly inaccurate model to grasp essential physics involved . the purpose of doing so is not to estimate quantities we want to know , but to make a connection between the localized case mentioned above to the case of delocalized scattering involving plasmons . we are interested in specimens consisting of biological molecules embedded in vitreous ice . clearly , it is not adequate to model such an insulating system by free electrons in a box , as done in studies of plasmons in metals . without further justifications , let us accept that essential physics of such an insulating system is captured by a set of @xmath2 1-dimensional harmonic oscillators , which are arranged on a line with a constant interval @xmath248 . ( the actual specimen is more like 2-dimensional , but we can oversimplify as long as the purpose is only to grasp essential physics . ) each harmonic oscillator models an atom with a single oscillating valence electron and the immobile rest . assume that all neighboring two electrons interact as if these are connected by a mechanical spring . this assumption is hard to justify , but for the sake of simplicity we proceed . then the model is similar to the 1-dimensional model for phonons . the hamiltonian is@xmath249 where @xmath250 and @xmath251 are the momentum and position of the electron at the @xmath95-th site , @xmath88 is the electron mass , @xmath252 and @xmath253 are spring constants of interaction between the @xmath95-th electron and the neighboring @xmath254-th electron ; and interaction between the @xmath95-th electron and its ion core , respectively . the last term in the curly bracket in equation ( [ eq : hamiltonian ] ) is the only difference from the phonon model . we apply the periodic boundary condition , i.e. , @xmath255 . let the plasmon coordinate and momentum operators be@xmath256 where @xmath257 , @xmath258 . ( we use notations @xmath259 that are used in other part of this paper , since there is no danger of confusion . ) following the standard procedure , we have @xmath260 where @xmath261\ ] ] though the model is crude , this reproduces the usual plasmon dispersion relation @xmath262 as @xmath263 . the standard method of quantization for phonons carries over to the case of plasmons , for which the creation operator is@xmath264\ ] ] now , consider the weak coupling limit @xmath265 . the creation operator for the individual bare harmonic oscillator at the @xmath95-th site is @xmath266\ ] ] where the bare oscillation frequency @xmath267 is given by @xmath268 . when @xmath269 , or in other words at the large wavelength limit , we have @xmath270 . in this case , combining some of the above relations , we obtain@xmath271e^{iksa}\cong\frac{1}{\sqrt{n}}\sum_{s}a_{s}^{\dagger}e^{iksa}\label{eq : creation operator relation}\ ] ] what equation ( [ eq : creation operator relation ] ) states is the following . _ an excitation of a plasmon is equivalent to the superposition of excitations of each scattering atom , with an additional phase factor that corresponds to the phase of the plasmon wave _ at least in the initial stage of time evolution . although our model is crude , we will henceforth assume that the preceding statement is robust so that it may be applied to real situations with reasonable degree of validity . from experiments we know that inelastic scattering is associated with a small scattering angle that corresponds to delocalization and long wave length plasmons . thus , unlike the case of inner - shell electron excitations , when one atom is excited in an inelastic scattering process generating a plasmon , the state must be superposed with other states , in each of which another atom , i.e. one of the surrounding atoms , is excited . in order to see how an excitation of a plasmon affects the measurement , we consider scattering of a plane wave . this suffices because , first , we can separately consider two terms of the whole wavefunction , containing respectively the cpb state @xmath13 and @xmath14 , and later add the results . second , all possible incident electron waves , including the diverging spherical wave , can be expanded as a sum of plane waves . we now consider all atoms that contribute their valence electron to form plasmons . for definiteness , let the number of these atoms be @xmath2 . we label them with an integer @xmath272 ( changing the meaning of @xmath2 somewhat ) and , for simplicity , we consider only the ground state @xmath273 and the first excited state @xmath274 for each of them . let the position of the @xmath95-th atom be @xmath275 . let the incident plane wave be @xmath21 . since there are many quantum entities , we will generally omit the tensor product symbol @xmath276 . the state before inelastic scattering is@xmath277 we confine ourselves to the case of single plasmon generation and we neglect terms with two or more atoms excited . a few definitions follow . let @xmath278 denote the electron state after interacting with the @xmath95-th atom , without exciting the atom s electron . ( one may view this state as a superposition of the transmitted wave and elastically scattered wave . ) we let the state @xmath279 represent the electron state after exciting the atom . then , after inelastic scattering , the state becomes@xmath280 @xmath281 @xmath282 where @xmath283 are probability amplitudes to reach the states @xmath284 . ( for simplicity we assume that @xmath283 are independent of @xmath95 . this may not be justified , but it will be found later that this will not affect our conclusion . ) for brevity , we define some of the state of the whole specimen atoms@xmath285 @xmath286 and also define an electron state @xmath287 . in terms of these , equation ( [ eq : many atoms excited ] ) is rewritten as@xmath288 the first term corresponds to elastic scattering , with the specimen left without excitation . now , according to the argument in the last paragraph , the state of a plasmon with wave vector @xmath289 is expressed as @xmath290 although the atoms are not regularly placed , by analogy with the fourier transform we write@xmath291 hence we have @xmath292 therefore , if we observe the plasmon in the state @xmath293 ( though this observation only models the decoherence process and we do not obtain the outcome of such an observation in practice ) , the electron wave is left in the state@xmath294 this is a superposition of electron waves inelastically scattered at each atoms , with a phase factor @xmath295 depending on the position on the specimen . if there are sufficiently many atoms , more - or - less uniformly distributed , then equation ( [ eq : scattered by a plasmon ] ) describes nothing but a tilted plane wave , whose wave vector is slightly changed by emitting a plasmon . this agrees with the standard picture of momentum conservation . at this point we make a few remarks . first , if the amplitude @xmath296 depends on @xmath95 ( for example , because @xmath296 depends on atomic number _ z _ ) , then we may divide the atoms into classes with the same @xmath296 values . as long as the number of atoms in each class is sufficiently large to retain the above fourier - transform - like property , the conclusion stands . conversely , one may simply neglect a class that comprises too few of atoms . second , one may argue that observing the plasmon with respect to the measurement basis @xmath297 is rather arbitrary and the use of other basis should be analyzed as well . however , there are no interactions between the electron and the specimen after scattering , and statistical measurement outcomes regarding the scattered electron should not depend on how the plasmon state is measured. ( of course , being entangled to each other , the measurement outcomes on the electron and the plasmon are _ correlated_. ) in addition , we will find the use of the basis @xmath297 highly convenient . third , as far as there are many atoms in the regions @xmath11 and @xmath12 , all these atoms contribute to the scattered wave amplitude and there will not be much variation of the wave amplitude within the respective regions . hence , unlike the case of inner - shell electron excitations , we do not expect to have much amplitude error . fourth , we may say that the effect of having plasmon scattering is equivalent to have a thin wedge prism right next to the specimen . every time inelastic scattering occurs , the rotation angle @xmath298 of the wedge prism around the optical axis is completely random . the angle between two planes of the wedge is also random but generally obeys the formula ( [ eq : bethe formula ] ) that specifies the distribution of inelastic scattering angles . when viewed in this way , it should be clear that no matter what incident beam we use , be it focused gaussian or diverging gaussian , the effect of plasmon generation is to tilt the exit wave from the specimen by an angle proportional to the size of the plasmon wave vector . the phase error due to plasmon excitations is more problematic than the amplitude error . the mechanism for the error is that the scattered electron wave is tilted by the aforementioned wedge prism and hence the phase compensation step ( see sec . [ sub : a - brief - review ] ) can not be done accurately . let us first roughly estimate the extent of such an error . the formula ( [ eq : bethe formula ] ) is in a sense misleading because the average scattering angle is much larger than the characteristic angle @xmath299 . in fact , the mean scattering angle does not even exist for the lorentzian distribution @xmath226 unless we introduce a cutoff angle @xmath205 , above which there is no scattering . the angle for the bethe - ridge @xmath300 is usually taken as a cutoff , which in our case is @xmath301 . let us compute the fraction @xmath219 of inelastically scattered electrons with the scattering angle less than @xmath302 . this is given by @xmath303 approximating @xmath304 , we get@xmath305}{\ln\left[1+\left(\theta_{c}/\theta_{e}\right)^{2}\right]}\label{eq : angular fraction}\ ] ] from this , one may compute , for example , that half of all inelastically scattered electrons have the scattering angle larger than @xmath306 . in other words , this is the median scattering angle and it is much larger than @xmath221 . a similar calculation yields the mean scattering angle @xmath307 . these numbers suggests that the probability distribution is quite unusual due to the long tail . we estimate the phase error due to failure in performing the phase compensation step . first , consider the case of focused incident beam ( sec . [ sub : focussed - incident - beam ] ) . as in appendix a , we take @xmath93-axis to be the optical axis . let the regions @xmath11 and @xmath12 be separated by a distance @xmath308 along @xmath101-axis . let the pixel , in which the scattered electron is detected , be @xmath34a . let the distance between the point a and the central points of the specimen regions @xmath11 and @xmath12 be @xmath309 and @xmath310 respectively . then , the optical path length difference is @xmath311 . the detection pixel is placed on the @xmath99 plane without loss of generality . the scattering angle is @xmath120 . simple geometrical consideration then yields @xmath312 . now , suppose that a plasmon is generated upon scattering , to the direction of @xmath101-axis . the path length difference is modified because of the presence of aforementioned wedge prism , to be @xmath313 where @xmath314 is the angle of the wedge prism , which actually is the inelastic scattering angle . the error in estimating the optical path length difference in the phase compensation step would therefore be @xmath315 , if we take mean angle error @xmath316 for plasmon scatterings as the angle of the wedge prism @xmath314 . the error , in terms of phase angle , is unacceptable @xmath317 , where @xmath318 is the wave number of @xmath7 electrons . in addition to the error due to the speckle pattern problem mentioned in sec . [ sub : focussed - incident - beam ] , this constitutes another reason why the focused - incident - beam configuration is not desirable . let us turn to the case of diverging incident beam , described in sec . [ sub : diverging - incident - beam ] . figure 3 shows that the gaussian electron beam focused at either @xmath154 or @xmath155 above the specimen central plane cp . the focal points corresponding to the regions @xmath11 and @xmath12 are labeled with the same symbol . the distance between these focal points is denoted as @xmath319 . let @xmath120 be the scattering angle and let @xmath314 be the deflection angle due to plasmon generation . the plasmon scattering occurs at cp in the figure , but exactly where the scattering occurs does not matter , as the reader will see shortly . we assume that the plasmon momentum vector is parallel to @xmath101-axis , which is the worst case . the four optical rays a , b , c and d go to the same direction under cp . the rays a and b represents optical paths taken by electrons that are not inelastically scattered , whereas the rays c and d shows optical paths associated with plasmon generation . in computing the optical path length difference , we always make the approximation @xmath304 for any angle . then , computation of the optical path length difference is a matter of simple exercise in geometry . the optical path length difference in the absence of inelastic scattering , between rays a and b is @xmath320 . on the other hand , the three thick portions of rays in fig . 3 contributes the path length difference between rays c and d , which turns out to be @xmath321 . thus , the change of the optical path length difference due to the plasmon scattering is @xmath322 . the magnitudes of @xmath314 is about the median inelastic scattering angle @xmath323 or the mean @xmath324 , depending on how one defines the typical scattering angle . the error in terms of phase , then , is @xmath325 or @xmath326 . the latter is not small and requires further investigation . and @xmath314 are much exaggerated.,scaledwidth=45.0% ] in order to resolve the question of how plasmon scattering affects the measurement , we resort to numerical simulation . we employ @xmath327 electrons in a k - electron process . we estimate the frequency of the plasmon scattering as follows . at near the end of sec . [ sub : focussed - incident - beam ] , it was found that the probability of elastic scattering for @xmath7 electrons is @xmath328 for a @xmath143 thick specimen . on the other hand , generally inelastic scattering is considered to occur twice as frequent as elastic scattering as mentioned earlier . hence we take a value of @xmath329 as inelastic scattering probability . we will ignore the failure due to high - angle elastic scattering ( probability @xmath330 as computed earlier ) and inner - core electron excitations ( probability @xmath247 ) , as these are sufficiently rare . indeed , when these probabilities are added we have @xmath331 . a simple consideration shows that the k - electron process fails with a probability @xmath332 . a @xmath333 waste of electron dose is acceptable . following ref . @xcite , we consider cryoelectron microscopy of the ribosome molecule . however , unlike ref . @xcite we consider @xmath7 electrons . otherwise the method of simulation is the same as ref . @xcite , except that we now consider inelastic scattering that occurs @xmath334 of times at random . when it is decided in our simulation program that inelastic scattering happened , the phase error is computed as follows . first , a uniform random number @xmath116 in the interval @xmath335 $ ] is generated . then we use equation ( [ eq : angular fraction ] ) by letting @xmath336 and compute @xmath302 . clearly , @xmath302 obeys the statistical distribution governing inelastic scattering angle . on the other hand , suppose that the plasmon wave vector is proportional to @xmath337 , where @xmath338 are unit vectors parallel to @xmath339 axes , respectively . geometrical consideration shows that the phase error in angle is @xmath340 , where @xmath341 is a random number drawn uniformly from the interval @xmath342 $ ] . since this error adds to the phase we want to measure during a k - electron process , equation ( [ eq : cpb reading prob ] ) is now modified as@xmath343 where @xmath344 represents the total error accumulated in the k - electron process . it is the sum of all individual error @xmath345 occurred during the k - electron process . more precisely , if @xmath346 inelastic scattering happened in a k - electron process , then there are @xmath346 independent random variables @xmath345 , that are summed up to produce @xmath344 . figure 4 shows the simulation result . each image consists of @xmath347 pixels and the size of the image is @xmath348 . the size of gaussian beams to the regions @xmath39 are respectively @xmath156 and @xmath157 at the entrance surface of the specimen . as mentioned in sec . [ sub : diverging - incident - beam ] , there is a significant beam spread in the specimen but we did _ not _ take the spread into account in the simulation , which has been performed as if there were no such spread @xcite . the electron dose is @xmath349 . perfect detector efficiency and the cpb readout efficiency is assumed . the raw image is smoothed with a gaussian filter with a standard deviation @xmath350 for better visibility . a simulated image of entanglement - assisted electron microscopy , which takes errors due to inelastic scattering into account , is shown in fig . 4 ( a ) . for the sake of comparison , a laplacian - filtered @xcite phase shift map of the ribosome , from a x - ray study @xcite , is shown in fig . 4 ( b ) ; the hypothetical case where inelastic scattering is absent , but otherwise identical situation , is shown in fig . 4 ( c ) ; and an image for the conventional case , where perfect in - focus phase contrast microscopy is employed with the same dose , also laplacian - filtered , is shown in fig . 4 ( d ) . we evaluated measurement errors in entanglement - assisted electron microscopy , due both to elastic and inelastic scattering processes at the specimen . despite remaining subtleties such as processing of data obtained by the significantly divergent electron beam , overall we have found no evidence that entanglement - assisted electron microscopy is not viable for some fundamental physical reasons related to electron scattering processes . there are several lessons learned along the way . first , despite the fact that lossy processes are often emphasized as the sole fundamental limit in the quantum metrology literature , elastic scattering processes present a limit just as fundamental in the present case . the reason is that the elastic scattering signal sometimes contains information that we do not need , because the information is about too high resolution features of the specimen that can not be fully obtained anyway . this leads to an error which can not be controlled nor corrected , but it has been computed to be sufficiently small . second , besides the inner - shell excitation processes associated with a sufficiently small cross sections , inelastic scattering processes involve generation of plasmons . the excitation of a plasmon results in bending of the probe electron wave , which in turn entails phase error because the phase compensation step of the entanglement - assisted electron microscopy can not be carried out precisely . we found that the magnitude of such an error is acceptable . rather remarkably , how the two probe electron states ( that are entangled to the two cpb states ) are geometrically arranged is found to be important in order to reduce this type of error . many issues are left for future investigations . for example , the most suitable energy of the primary electron beam , which presumably depends on the specimen thickness , has not been fully understood , although the present author feels that higher energy is better , unless knock - on damage comes in to play a significant role . furthermore , errors associated with the electron reflection process at the electron mirror equipped with the cpb , or errors associated with the cpb control processes , have largely been unexplored . i thank umpei miyamoto for his help with tex . this research was supported in part by the jsps `` kakenhi '' grant ( # 25390083 ) . here we derive equations ( [ eq : transmitted wave far field ] ) , ( [ eq : scattered wave 1 ] ) and ( [ eq : scattered wave 2 ] ) appeared in sec . [ sec : coherent errors ] and clarify what approximations are involved in the derivation . let us first deal with the case of sec . [ sub : focussed - incident - beam ] . the incident gaussian wave is a superposition of many plane waves . for a constituent plane incident wave @xmath351 , after elastic scatterings we have , in the far field @xmath352 where @xmath353 is the wave number vector after the scattering , whose primary purpose is to point at the point p , where we wish to compute the value of the wavefunction , and @xmath354 is the angle between the vectors @xmath355 and @xmath353 . we are dealing with elastic scattering and @xmath356 , at least to a good approximation . we ignore the imaginary part of @xmath97 , as is done in the born approximation though this can not be entirely correct because of the optical theorem . as usual , when @xmath357 appearing in the denominator in eq . ( [ eq : plane wave scattering-1 ] ) is much greater than all relevant @xmath358 , @xmath357 may be approximated to be @xmath359 . this is a representative distance between the whole specimen and the point p. let the @xmath360 component of the vector @xmath355 be respectively @xmath361 , which we write @xmath362 . we want the incident beam to be gaussian with the waist size @xmath98 and the above plane wave solutions are superposed as@xmath363\label{eq : gaussian wave scattering-1}\ ] ] the factor @xmath364 is introduced solely to simplify some of later expressions and keep them dimensionless . the diffraction - limited waist of the gaussian beam is on the specimen . at @xmath365 , the incident beam part is@xmath366 note that the @xmath93 component of the @xmath355-vector is @xmath367 in above equations and the limits of integrations in equation ( [ eq : incident beam-1 ] ) are approximated as positive and negative infinities . far from the specimen , the transmitted gaussian wave has the known form @xmath368\cong\frac{-ikw_{0}^{2}}{2z}\exp\left[-\frac{k^{2}w_{0}^{2}}{4z^{2}}\left(x^{2}+y^{2}\right)\right]e^{ikr}\label{eq : transmitted wave far field-1}\ ] ] this , when @xmath369 , is equation ( [ eq : transmitted wave far field ] ) of the main text . next , we consider the scattered part of the wave amplitude@xmath370\label{eq : scattered wave-1}\ ] ] note that , if the phase factor @xmath371 were identically @xmath129 , which it is not , the integration in the square brackets in equation ( [ eq : scattered wave-1 ] ) would represent , up to a constant multiplicative factor , something akin to , but not identical with , gaussian smoothing of @xmath372 . to evaluate the phase factor @xmath371 , we write it as @xmath373 . let us first consider the exit wave , and focus on a single pixel of the detector . since the pixel is on a diffraction plane , @xmath353 is independent of @xmath95 . in other words , it is independent of the place of the atom under consideration . without loss of generality , we assume that the pixel under consideration is on the @xmath99 plane . let the angle between @xmath93-axis and @xmath353 be @xmath120 . then the detector pixel is located at @xmath374 , where @xmath116 is much greater than any @xmath375 . writing @xmath96 , we have @xmath376 let us now turn to the incident wave and write @xmath377 . it is clear that the angle between @xmath355 and @xmath93-axis is @xmath378 , @xmath379 , and hence @xmath380 when these angles are small . putting them together , the phase factor @xmath373 equals@xmath381e^{ikr}\ ] ] @xmath382\exp\left[ik\left(-x_{s}\theta+z_{s}\frac{\theta^{2}}{2}\right)\right]e^{ikr}\label{eq : phase factor-1}\ ] ] this makes equation ( [ eq : scattered wave-1])@xmath383\label{eq : scattered wave 2 - 1}\ ] ] in order to further approximate , we check some numbers here . because of the waist size @xmath384 ( since we want a @xmath91 _ diameter _ ) , @xmath385 and @xmath386 are at most @xmath91 , while @xmath113 is up to the specimen thickness @xmath387 . from the theory of gaussian beams we know that the wave convergence half - angle is @xmath388 , where the wave number @xmath18 is @xmath389 for @xmath7 electrons . this makes @xmath390 ( and the focal depth , or the rayleigh range @xmath391 , is @xmath392 ) . noting that @xmath393 is at most of the order @xmath394 , the phase factor in the square bracket in equation ( [ eq : scattered wave 2 - 1 ] ) may be approximated as @xmath395 . on the other hand , the meaningful magnitude of @xmath120 should roughly be given by the `` characteristic angle of elastic scattering '' @xmath396 , where @xmath397 is the screening radius of the atom and is given by @xmath398 , where @xmath197 and @xmath399 are the bohr radius and the atomic number , respectively @xcite . for most relevant elements ( see appendix b ) @xmath400 and we have @xmath401 and @xmath402 and in the case of hydrogen this is even smaller . these numbers allow us to further approximate equation ( [ eq : scattered wave 2 - 1 ] ) to yield@xmath403\label{eq : scattered wave 3 - 1}\ ] ] this approximation should be understood as , for example , keeping only the phase shift @xmath404 associated with atoms with large @xmath385 values . ( we have the phase factor @xmath404 for atoms with smaller @xmath385 values also , but that may be less significant than the neglected @xmath405 . ) furthermore , since the presence of the factor @xmath406 implies that the integration , in terms of the angle @xmath354 , is over a narrow range of the order @xmath390 . this is fairly smaller than the spread of the function @xmath372 , that is @xmath407 , when plotted with respect to @xmath354 . these facts allow us to approximate equation ( [ eq : scattered wave 3 - 1 ] ) to obtain@xmath408\label{eq : scattered wave 4 - 1}\ ] ] the integral is the two - dimensional fourier transform of a gaussian function and we get @xmath409 this is equation ( [ eq : scattered wave 1 ] ) of the main text . we now turn to the second case described in sec . [ sub : diverging - incident - beam ] . the above discussion up to equation ( [ eq : scattered wave 2 - 1 ] ) holds also here . we keep the gaussian beam waist at @xmath365 , but the atoms in the specimen are no longer at @xmath410 but at near @xmath411 that is @xmath154 or @xmath155 , depending on whether the region under consideration is @xmath11 or @xmath12 . the @xmath93 coordinate of the @xmath95-th atom is now @xmath412 , where @xmath413 is at most @xmath387 . equation ( [ eq : scattered wave 2 - 1 ] ) is then rewritten as@xmath414 @xmath415\label{eq : scattered wave 6 - 1}\ ] ] here , we need to consider the range of @xmath120 up to @xmath416 , which is the sum of the incident beam divergence @xmath417 and the characteristic angle @xmath182 . the coordinates @xmath418 are at most @xmath419 and @xmath420 are at most @xmath421 . hence equation ( [ eq : scattered wave 6 - 1 ] ) is approximated as@xmath422\ ] ] @xmath423 @xmath424}f_{s}\left(\theta_{\boldsymbol{k}_{i},\boldsymbol{k}_{f}}\right)\right]\label{eq : scattered wave 7 - 1}\ ] ] though the approximation is not good . consider the integral in the square bracket . the phase factor @xmath425}$ ] has the associated phase angle up to @xmath426 . in our case @xmath427 and @xmath428 , thus one may say that the phase angle rotates fairly rapidly . hence the value of @xmath34@xmath372 matters only at around @xmath429 , where no first - order phase oscillation occurs . since @xmath430 and @xmath431 , we have @xmath432 to the first order in @xmath433 . hence , we further simplify equation ( [ eq : scattered wave 7 - 1 ] ) as@xmath434 @xmath424}\right]\label{eq : scattered wave 8 - 1}\ ] ] we must note , however , that this approximation is only barely justifiable because the phase factor @xmath425}$ ] is fairly constant in an angular range @xmath435 , where @xmath436 is the wavelength of @xmath7 electrons . taking @xmath437 , we have @xmath438 , which actually is comparable to the characteristic angle @xmath182 associated with the scattering amplitude @xmath97 . due to the lack of simple and good analytical alternatives we proceed . the integral is a convolution of two functions that equals@xmath439 it may be verified that @xmath440 and we define @xmath441 . ( the parameter @xmath442 has a physical meaning of the ratio between the waist size of the incident beam @xmath149 and the incident beam radius at the distance @xmath153 from the waist . alternatively , @xmath443 , where @xmath391 is the rayleigh range . ) hence , equation ( [ eq : scattered wave 8 - 1 ] ) approximately equals@xmath444}e^{-\varepsilon\frac{k}{2\delta z}\left(x_{s}^{2}+y_{s}^{2}\right)}f_{s}\left(\theta-\frac{x_{s}}{\delta z}\right)\label{eq : scattered wave 9 - 1}\ ] ] this is equation ( [ eq : scattered wave 2 ] ) in the main text . we consider a frozen , hydrated speimen that is typical in biological cryoelectron microscopy and compute the associated atomic number densities for the elements h , c , n , o , and s. the water content may vary considerably from specimen to specimen . however , for simplicity we take values of 76% water and 24% `` dry weight '' as typical composition , which has been reported as the composition of `` sediments obtained by predetermined dilution of pellets harvested by centrifugation '' of yeast suspension @xcite . the component which contributes to the dry weight may in general contain things other than protein . however , again for simplicity , we assume that the component consists solely of protein . then , given the known density of low density amorphous ( lda ) ice 0.94 g / cc @xcite , the standard value of protein density 1.35 g / cc ( but see also @xcite ) , and the reported amino acid frequencies in representative proteins @xcite , the representative atomic number densities for the elements h , c , n , o , and s may be computed . the results are shown in table 1 . at first glance , biological specimens may seem to be weak _ amplitude _ objects rather than weak phase objects because in a typical frozen hydrated biological specimen , the inelastic scattering cross section @xmath445 is about two times greater than the elastic counterpart @xmath446 @xcite . however , when phase contrast imaging is used , the phase shift @xmath447 of the electron wave , which directly translate to image contrast , is proportional to @xmath448 , whereas @xmath445 , representing intensity loss , contributes linearly to the contrast . the former contribution to the contrast is generally larger than the latter . u. kaiser , j. biskupek , j. c. meyer , j. leschner , l. lechner , h. rose , m. stoger - pollach , a. n. khlobystov , p. hartel , h. muller , m. haider , s. eyhusen , and g. benner , ultramicroscopy * 111 * , 1239 ( 2011 ) . for example , it is reasonable to estimate the inelastic mean free path of 20 kv electrons to be in the range of a few tens of nm , if we take the following three facts into account : 1 ) data presented in m. p. seah and w. a. dench , surf . interface anal . * 1 * , 2 ( 1979 ) show a trend at lower voltages . 2 ) the mean free path is 160 nm in vitreous ice ( b. feja and u. aebi , j. microsc . 193 , 15 ( 1999 ) and references therein ) . 3 ) the mean free path in frozen protein is somewhat shorter than the vitreous ice case , as described in chap 15 , ref . moreover , a short _ elastic _ mean free path would also be problematic , as multiple scatterings would make interpretation of images much more difficult . m. buttiker , phys . b * 36 * , 3548 ( 1987 ) ; p. lafarge , p. joyez , d. esteve , c. urbina , and m. h. devoret , nature ( london ) * 365 * , 422 ( 1993 ) ; y. nakamura , yu . a. peshkin , and j. s. tsai , nature ( london ) * 398 * , 786 ( 1999 ) . a. blais , r .- s . huang , a. wallraff , s. m. girvin , and r. j. schoelkopf , phys . a * 69 * , 062320 ( 2004 ) ; a. wallraff , d. i. schuster , a. blais , l. frunzio , j. majer , m. h. devoret , s. m. girvin , and r. j. schoelkopf , phys . lett . * 95 * , 060501 ( 2005 ) . v. a. lobastov , r. srinvasan , and a. h. zewail , proc . usa * 102 * , 7069 ( 2005 ) ; m. merano , s. collin , p. renucci , m. gatri , s. sonderegger , a. crottini , j. d. ganiere , and b. deveaud , rev . . instrum . * 76 * , 085108 ( 2005 ) . j. yang , k. kan , n. naruse , t. kondoh , y. yoshida , k. tanimura , t. takatomi , and j. urakawa , in proceedings of the 9th annual meeting of particle accelerator society of japan , 2012 ( unpublished ) p. 282 . the laplacian filter @xcite computes the difference of the following two data : the first data is the result of applying a gaussian filter with the standard deviation @xmath350 to the unprocessed data , while the second is the same except that the standard deviation is @xmath449 . it was done on figures ( b ) and ( d ) to extract high resolution components only , in order to make a comparison with figures 4 ( a ) and ( c ) . d. t. bowron , j. l. finney , a. hallbrucker , i. kohl , t. loerting , e. mayer , and a. k. soper , j. chem . phys . * 125 * , 194502 ( 2006 ) ; t. loerting , k. winkel , m. seidl , m. bauer , c. mitterdorfer , p. h. handle , c. g. salzmann , e. mayer , j. l. finney , and d. t. bowron , phys . chem . phys . * 13 * , 8783 ( 2011 ) .
the major resolution - limiting factor in cryoelectron microscopy of unstained biological specimens is radiation damage by the very electrons that are used to probe the specimen structure . to address this problem , an electron microscopy scheme that employs quantum entanglement to enable phase measurement precision beyond the standard quantum limit has recently been proposed [ phys . rev . a * 85 * , 043810 ] . here we identify and examine in detail measurement errors that will arise in the scheme . an emphasis is given to considerations concerning inelastic scattering events because in general schemes assisted with quantum entanglement are known to be highly vulnerable to lossy processes . we find that the amount of error due both to elastic and inelastic scattering processes are acceptable provided that the electron beam geometry is properly designed . + hiroshi okamoto + + electronic address : [email protected] + pacs : 67.81ee , 34.80.gs , 61.80.-x , 06.20.dk +
interplay between nonlinearity and periodicity is the focus of recent studies in different branches of nonlinear physics and applied mathematics . physical applications of nonlinear systems with periodic potentials range from nonlinear optics , in the dynamics of guided waves in inhomogeneous optical structures and photonic crystal lattices , to atomic physics , in the dynamics of bose einstein condensate droplets in periodic potentials , and from condensed matter , in josephson - junction ladders , to biophysics , in various models of the dna double strand . the paramount significance for these models is the possibility of spatial localization , that is emergence of nonlinear localized structures residing in the spectral band gaps of the periodic potentials . to describe this phenomenon , the primary equations of physics are typically simplified to the gross pitaevskii equation , which we shall study in our article in the space of two dimensions . more precisely , we consider the two - dimensional gross pitaevskii equation in the form @xmath0 where @xmath1 , @xmath2 , @xmath3 , and @xmath4 . time - periodic solutions of the gross pitaevskii equation are found from the solutions of the nonlinear elliptic problem @xmath5 where @xmath6 and @xmath7 arise in the substitution @xmath8 . it is known that localized solutions of the elliptic problem ( [ stationary ] ) with a periodic potential @xmath9 , called _ gap solitons _ , exist in _ every _ finite gap of the spectrum of the schrdinger operator @xmath10 and in the semi - infinite gap for @xmath11 @xcite . bifurcations of localized solutions from edges of the spectral bands were studied earlier in @xcite . coupled - mode equations were used by physicists for the analysis of existence , stability and dynamics of gap solitons @xcite . a justification of the one - dimensional coupled - mode equations in the context of the elliptic problem ( [ stationary ] ) with @xmath12 was carried out in @xcite in the limit of _ small - amplitude _ periodic potentials @xmath9 . ( a justification of time - dependent coupled - mode equations on finite - time intervals was done earlier in @xcite . ) in the limit of small - amplitude potentials , narrow gaps of the spectrum of the schrdinger operator @xmath13 bifurcate from resonant points of the spectrum of @xmath14 , while the bloch modes of @xmath15 bifurcate from the fourier modes of @xmath16 . in other words , photonic band gaps open generally for small - amplitude one - dimensional periodic potentials @xmath9 . _ small - amplitude _ gap solitons of the elliptic problem ( [ stationary ] ) in @xmath17 reside in the narrow gaps of @xmath15 for @xmath18 according to the approximation obtained from the coupled - mode equations @xcite . we refer to the opening of a spectral gap under a small change of the potential @xmath9 as to the bifurcation of the band gap . bifurcations of band gaps do not occur for small - amplitude multi - dimensional periodic potentials . this is caused by an overlap of the spectral bands of @xmath19 in the first brillouin zone if @xmath20 and @xmath21 @xcite . as a result , photonic band gaps in multi - dimensional potentials open only at some finite amplitudes of the periodic potential @xmath9 and the resonant eigenfunctions are given by the bloch modes of @xmath10 rather than by the fourier modes of @xmath16 . although the coupled - mode equations were also derived for multi - dimensional problems with small periodic potentials @xcite and the resonant fourier modes were used for the approximation of the full solution , the applicability of these coupled - mode equations remains an open issue for a rigorous analysis . bloch mode decomposition has been also used in one dimension for finite - amplitude periodic potentials to derive coupled - mode equations @xcite . the corresponding unperturbed one - dimensional potential , however , has to be of a special type to admit a finite number of open gaps , such that a new gap is opened under a small perturbation . in this paper we derive coupled - mode equations for wavepackets in narrow band gaps of a _ finite - amplitude _ periodic potential by using the fourier bloch decomposition and the rigorous analysis based on the implicit function theorem in the space of bounded continuous functions vanishing at infinity . the coupled - mode equations we derive here take the form of coupled nonlinear schrdinger ( nls ) equations . these equations differ from the first - order coupled - mode equations exploited earlier @xcite . similar coupled nls equations have been recently derived in @xcite near band edges of the well - separated spectral bands and in @xcite for tunnelling problems . unlike these works relying on numerical approximations , we justify the derivation of the coupled - mode equations and prove the persistence of localized solutions in the full nonlinear problem ( [ stationary ] ) . although details of our analysis are given only for the bifurcation of the first band gap in the spectrum of @xmath15 , a similar analysis can be developed for bifurcations of other band gaps and for bifurcations of the localized solutions near band edges of the well - separated spectral bands . our derivation is developed for the class of _ separable _ potentials @xmath22 , \quad \eta \in \mathbb{r},\ ] ] where the function @xmath23 is assumed to be real - valued , bounded , piecewise - continuous , and @xmath24-periodic on @xmath25 . to simplify the details of our analysis , we assume that @xmath26 on @xmath27 , such that the solution set of the elliptic problem ( [ stationary ] ) includes functions satisfying one of the following two _ reversibility _ constraints @xmath28 or @xmath29 where @xmath30 . our strategy is to show that there may exist a value @xmath31 , for which the first band gap opens due to the resonance of three lowest - order bloch modes for three spectral bands of the operator @xmath10 . the new small parameter @xmath32 is then used for the bifurcation theory of bloch modes which results in the _ algebraic _ coupled - mode equations for nonlinear interaction of the three resonant modes . the fourier bloch decomposition is used for the approximation of localized solutions and for the derivation of the _ differential _ coupled - mode equations with the second - order derivative terms . the main idea behind our technique is that a differential operator after the fourier bloch decomposition becomes a pseudo differential operator whose symbol is the multiple - valued dispersion relation . then only the relevant ( three ) branches of the resonant modes can be taken into account , which leads to a coupled - mode system . if the linearization operator of the coupled nls equations at the localized solutions is non - degenerate , the persistence of the localized symmetric solutions in the full nonlinear problem ( [ stationary ] ) is proved with the implicit function theorem . localized symmetric solutions of the coupled nls equations are approximated numerically and the convergence rate for the error of approximation is studied . finally , we study time - dependent localized solutions of the gross pitaevskii equation and the coupled - mode system and control smallness of the distance between the two solutions on a finite - time interval . the article is structured as follows . section [ s : strum_liou ] reviews elements of the sturm liouville theory for the separable potentials . section [ s : algeb_cme ] contains a derivation of the algebraic coupled - mode equations for three resonant bloch modes . section [ s : diff_cme ] gives details of the projection technique for the derivation of the differential coupled - mode equations for localized solutions . the persistence of localized reversible solutions under a non - degeneracy assumption on the linearization operator is proved in section [ s : reversible_gs ] . numerical approximations of the localized solutions , the associated linearization operators , and the convergence of the approximation error are obtained in section [ s : numerical ] . section [ s : t_dep_cme ] extends the results to the time - dependent case for finite time intervals . section [ s : generalizations ] discusses relevant generalizations . it is typically expected that the band gaps in the spectrum of the linear schrdinger operator with a two - dimensional periodic potential open at the extremal values of the bloch band surfaces @xcite . for a general potential , however , the extremal values may occur anywhere within the first irreducible brillouin zone @xmath33 in the quasi - momentum space @xmath34 @xcite . we will show here that the extremal values for a separable potential occur only at the vertex points @xmath35 , @xmath36 and @xmath37 on the boundary @xmath38 . figure [ f : b_ir ] shows the irreducible brillouin zone @xmath33 and the vertex points for a two - dimensional separable potential @xmath9 . ( 5.5,4.5)(-1.5,-0.5 ) ( 1.6,1.0)@xmath33 ( 0.65,3.2)@xmath39 ( 0.5,0.5)(3,0.5)(3,3)(0.5,0.5 ) ( -.3,.5)(3.9,.5 ) ( 0.5,-.3)(0.5,3.7 ) ( 3.9,.5)(4.1,.5 ) ( .5,3.7)(.5,3.9 ) ( -0.1,3.7)@xmath40 ( 4.1,0.1)@xmath41 ( 0.5,3)(3,3 ) ( 0.1,0)@xmath35 ( 3.2,0.7)@xmath36 ( 3.2,3.2)@xmath37 ( 0.1,2.85)@xmath42 ( 2.85,0.08)@xmath42 let us consider the spectral problem for the linear schrdinger operator associated with the separable periodic potential ( [ separable - potential ] ) : @xmath43 u(x_1,x_2 ) = \omega u(x_1,x_2).\ ] ] by using the separation of variables @xmath44 and the parametrization @xmath45 , we obtain the uncoupled eigenvalue problems @xmath46 the bloch modes and the band surfaces for the one - dimensional problems ( [ separated - ode ] ) are introduced according to the regular sturm liouville problem @xmath47 where @xmath48 is a quasi - momentum defined on the interval @xmath49 $ ] . the eigenfunctions of the sturm liouville problem ( [ e : bloch_eval_problem ] ) are periodic with respect to @xmath48 with the period one . by theorem 2.4.3 in @xcite , there exists a countable infinite set of eigenvalues @xmath50 for each @xmath51 , which can be ordered as @xmath52 by theorem 4.2.3 in @xcite , if @xmath23 is a bounded and @xmath24-periodic potential , eigenvalues @xmath50 have a uniform asymptotic distribution on @xmath51 , such that @xmath53 for some constants @xmath54 . let @xmath55 be the corresponding set of eigenfunctions of the sturm liouville problem ( [ e : bloch_eval_problem ] ) , such that @xmath56 . by theorem xiii.89 in @xcite , if @xmath23 is a bounded , piecewise - continuous and @xmath24-periodic potential , then the eigenvalue @xmath57 and the bloch function @xmath58 are analytic in @xmath59 and continuous at the points @xmath60 and @xmath61 . by theorem xiii.95 in @xcite , the eigenvalue @xmath57 is extended on a smooth riemann surface in the neighborhood of the points @xmath60 and @xmath62 if the @xmath63 spectral band is disjoint from the adjacent @xmath64 spectral bands by a non - empty band gap . by theorem xiii.90 in @xcite , if @xmath23 is a bounded , piecewise - continuous and @xmath24-periodic potential , then the spectrum of @xmath65 in @xmath66 is absolutely continuous and consists of the union of the intervals in the range of the functions @xmath57 on @xmath67 for @xmath68 , where @xmath69 . moreover , the extremal values of @xmath57 occur only at the points @xmath70 and @xmath61 . these points correspond to an alternating sequence of the maximum and minimum values of the eigenvalues @xmath71 according to the following formula @xmath72 for all @xmath73 . the bloch function @xmath58 is @xmath24-periodic if @xmath60 and @xmath74-antiperiodic if @xmath61 . let us denote the periodic and antiperiodic eigenfunctions and the corresponding eigenvalues by @xmath75 by theorems 2.3.1 and 3.1.2 in @xcite , these eigenvalues are ordered by @xmath76 and @xmath77 , while the corresponding eigenfunctions @xmath78 and @xmath79 have precisely @xmath80 zeros ( nodes ) on the interval @xmath81 . if @xmath26 , the eigenfunctions @xmath78 and @xmath79 are even for odd @xmath82 and odd for even @xmath82 . each set of the eigenfunctions @xmath83 and @xmath84 is orthogonal in @xmath85)$ ] . eigenvalues of the same one - dimensional operator @xmath86 for @xmath87-periodic eigenfunctions consist of the union of the eigenvalues @xmath88 and @xmath89 . since there are at most two eigenvalues of the second - order operator @xmath90 and by theorem 2.3.1 of @xcite , we obtain that @xmath91 in particular , the first band gap is always non - empty for a non - constant potential @xmath23 ( see theorem xiii.91(a ) in @xcite ) . due to this ordering , the lowest value of @xmath92 for the crossing of the bloch band surfaces associated with the two - dimensional separable potential @xmath93 occurs at @xmath94 . using these facts , we obtain the following results . extremal values of the bloch band surfaces @xmath95 for @xmath96 occur only at the vertex points of the boundary @xmath38 . because @xmath97 for a separable potential ( [ separable - potential ] ) , we have @xmath98^t$ ] , where @xmath57 are analytic on @xmath99 . if an extremal point occurs in the interior of the brillouin zone @xmath33 , then @xmath100 at @xmath101 . however , @xmath102 for any latexmath:[$0 < xiii.90 in @xcite . similarly , the extremal points can not occur in the interior of the boundary @xmath38 . therefore , the extremal values of @xmath104 occur only at the vertex points @xmath35 , @xmath36 and @xmath37 on @xmath38 . the derivative @xmath105 may be non - zero at @xmath60 and @xmath106 if the @xmath63 spectral band touches the adjacent @xmath64 spectral bands . this happens when the corresponding eigenvalue @xmath107 or @xmath108 is double degenerate with the equality sign in the ordering ( [ ordering ] ) . however , @xmath109 , @xmath110 and @xmath111 are always simple and , therefore , @xmath112 at least for the first three spectral bands @xmath92 at @xmath113 . assume that the resonant condition @xmath114 for the @xmath24-periodic eigenvalue @xmath109 and the @xmath24-antiperiodic eigenvalues @xmath115 of @xmath116 is satisfied for @xmath117 . then there are exactly three resonant bloch modes at @xmath118 and @xmath117 in the spectral problem ( [ lin_eval ] ) : @xmath119 the three resonant modes are orthogonal to each other with respect to the @xmath120-periodic inner product @xmath121 [ lemma - resonance ] there are exactly three resonant modes for _ any _ separable potential ( [ separable - potential ] ) because @xmath109 and @xmath110 are the smallest non - degenerate eigenvalues of the operator @xmath90 and the double degeneracy @xmath122 is due to the symmetry with respect to the interchange of the variables @xmath123 and @xmath124 . the eigenfunctions @xmath125 and @xmath126 are all orthogonal to each other in @xmath127)$ ] , which leads to the orthogonality of the three resonant modes . we note that the eigenfunctions @xmath128 and @xmath129 are not orthogonal on the interval @xmath130 $ ] since both of them are positive by theorem 3.1.2 in @xcite . however , a linear combination of @xmath128 and @xmath129 belongs to the class of @xmath87-periodic functions and the two eigenfunctions are orthogonal on the double - length interval @xmath131 $ ] . by theorem 6.10.5 in @xcite , there are finitely many gaps in the spectral problem ( [ lin_eval ] ) with a separable potential . ( in fact , there are finitely many gaps for general smooth periodic potentials in dimension 2 or higher @xcite . ) lemma [ lemma - resonance ] indicates a bifurcation when one of these gaps at the lowest value of @xmath104 may open . it is observed for many examples of the separable potential ( [ separable - potential ] ) that this bifurcation leads to the _ first _ band gap in the spectrum of @xmath132 $ ] if @xmath133 is increased from @xmath134 . [ example-1 ] let @xmath135 . numerical approximations of the eigenvalues of the sturm liouville problem ( [ e : bloch_eval_problem ] ) are computed with the use of the second - order central difference method . the eigenfunctions @xmath78 and @xmath79 are plotted in figure [ f : eigenfunctions ] for @xmath136 , while the dependence of the first eigenvalues @xmath107 and @xmath108 on @xmath133 is plotted in figure [ f : bands](a ) . figure [ f : bands](b ) shows the intersection of @xmath137 and @xmath138 at the bifurcation value @xmath139 , when @xmath140 , @xmath141 and @xmath142 , such that @xmath143 . the perturbation behavior shown in figure [ f : bands ] was studied analytically in @xcite . figure [ f:2d_band_diagram ] illustrates the band structure along @xmath38 of the full spectral problem ( [ lin_eval ] ) with @xmath144 clearly revealing the resonance at @xmath145 and the two resonant modes at the points @xmath36 and @xmath37 . the third resonant mode lies at @xmath39 because the symmetry @xmath146 implies @xmath147 . -periodic ( left ) and @xmath24-antiperiodic ( right ) eigenfunctions of @xmath148 with @xmath135.,title="fig:",height=226 ] -periodic ( left ) and @xmath24-antiperiodic ( right ) eigenfunctions of @xmath148 with @xmath135.,title="fig:",height=226 ] for @xmath24-periodic ( solid ) and @xmath24-antiperiodic ( dashed ) boundary conditions . ( b ) the lowest resonance occurs at @xmath118 and @xmath117 when @xmath149.,height=226 ] for the spectral problem ( [ lin_eval ] ) with @xmath135 and @xmath150.,height=226 ] in what follows , we consider the bifurcation of non - trivial solutions of the nonlinear elliptic problem ( [ stationary ] ) with the separable potential ( [ separable - potential ] ) in the lowest band gap described by lemma [ lemma - resonance ] . let @xmath151 , @xmath152 , @xmath153 and rewrite the nonlinear elliptic problem ( [ stationary ] ) in the form @xmath154 \phi(x ) - \epsilon \sigma |\phi(x)|^2 \phi(x),\ ] ] where @xmath155 and @xmath156 - \omega_0 $ ] . two classes of non - trivial solutions of the bifurcation problem ( [ bifurcation ] ) are considered for small @xmath157 : bounded @xmath120-periodic solutions ( section [ s : algeb_cme ] ) and bounded decaying solutions ( sections [ s : diff_cme][s : numerical ] ) . we consider here bounded periodic solutions @xmath158 of the bifurcation problem ( [ bifurcation ] ) for small @xmath157 . our results depend on the period of the periodic solutions @xmath158 . since the potential @xmath9 is separable , the @xmath24-periodic ( @xmath24-antiperiodic ) function @xmath158 can be represented by the series of bloch modes associated with the eigenvalue problem ( [ e : bloch_eval_problem ] ) on @xmath25 for @xmath60 ( @xmath159 ) , while the @xmath120-periodic functions @xmath158 can be represented by the series of both @xmath24-periodic and @xmath24-antiperiodic bloch modes . in what follows , we shall use symbols @xmath160 , @xmath161 and @xmath162 to denote the corresponding spaces of functions @xmath163 on the interval @xmath164 $ ] satisfying the boundary conditions @xmath165 . we shall also use symbol @xmath166 to denote the vector of elements of the sequence @xmath167 . let @xmath23 be a bounded and @xmath24-periodic function . let @xmath168 and @xmath169 be sets of eigenvalues and eigenfunctions of the sturm liouville problem ( [ e : bloch_eval_problem ] ) that depends on @xmath67 , such that @xmath170 where @xmath171 and @xmath172 is the kronecker symbol . for any fixed @xmath173 , the set of eigenfunctions is complete in @xmath160 , such that there exists a unique set of coefficients @xmath167 in the decomposition @xmath174 given by @xmath175 . [ proposition-1 ] the statement of the proposition follows by theorem xiii.88 of @xcite . [ corollary - decomposition-1 ] let @xmath176 be defined by the decomposition ( [ decomposition-1 ] ) , where @xmath177 belongs to the vector space @xmath178 with the norm @xmath179 if @xmath180 , then @xmath176 is a continuous function on @xmath25 and @xmath181 with @xmath182 . by the triangle inequality , we obtain @xmath183 such that @xmath184 for any @xmath180 , @xmath185 and @xmath67 . since @xmath23 is a bounded potential , the squared norm @xmath186 is equivalent to the integral @xmath187 where @xmath188 and @xmath189 is defined using the spectral family associated with the sturm liouville operator @xmath190 in @xmath160 . by the asymptotic distribution of eigenvalues ( [ asymptotic - distribution ] ) , we obtain that @xmath191 for some @xmath192 and any @xmath180 . if the potential @xmath193 is continued with respect to the parameter @xmath133 , the perturbation theory for an eigenvalue @xmath57 depends on whether @xmath57 is simple or multiple . the following lemma is a trivial statement of the perturbation theory , so we omit its proof . let @xmath23 be a bounded and @xmath24-periodic function of @xmath17 . fix @xmath67 , @xmath194 , and @xmath195 . if @xmath196 , then @xmath57 and @xmath58 depend analytically on @xmath133 near @xmath117 , such that @xmath197 and @xmath198 for some @xmath199 and sufficiently small @xmath200 . [ lemma - perturbation - theory ] due to the construction of the spectrum of the two - dimensional spectral problem ( [ lin_eval ] ) , the results of lemma [ lemma - perturbation - theory ] settle the question of the splitting of the three resonant bands near the point @xmath201 described in lemma [ lemma - resonance ] . indeed , each bloch band surface ( the graph of @xmath97 ) corresponding to the three resonant bands includes a different vertex point ( @xmath39 , @xmath36 , and @xmath37 ) at @xmath118 for different values of @xmath48 in the set @xmath202 . each eigenvalue of the spectral problem ( [ lin_eval ] ) is simple and isolated if @xmath48 is fixed , such that the continuation of each bloch band surface in @xmath133 near @xmath117 is determined by the expression ( [ leading - order ] ) and the separation of variables in @xmath95 . in other words , bifurcation of the lowest band gap that occurs at @xmath118 can not be studied if the function space is restricted to @xmath203 for a fixed value of @xmath204 . to detect the bifurcation , we shall work in the function space @xmath205\times [ -2\pi,2\pi])$ ] , where all three vertex points @xmath39 , @xmath36 and @xmath37 correspond to the same @xmath120-periodic boundary conditions on the function @xmath158 . according to lemma [ lemma - resonance ] , the three resonant bloch modes are orthogonal to each other in this space . to work with the @xmath120-periodic eigenfunctions , we shall introduce a number of notations . let @xmath206 and @xmath207 denote the sets of eigenvalues and the corresponding @xmath87-periodic eigenfunctions of the one - dimensional spectral problem ( [ e : bloch_eval_problem ] ) . these sets consist of the union of eigenvalues and eigenfunctions defined by ( [ eigenvalue - periodic ] ) . suppose that the eigenvalues are sorted in non - decreasing order , such that @xmath208 , @xmath209 , @xmath210 , @xmath211 , @xmath212 and so on , and that the eigenfunctions @xmath213 , etc . , are normalized such that @xmath214 , where @xmath215 . the statements of proposition [ proposition-1 ] and lemma [ corollary - decomposition-1 ] extend naturally to the new sets of eigenfunctions in space @xmath205)$ ] . to develop the nonlinear analysis of bifurcations of a non - trivial @xmath120-periodic solution @xmath158 on @xmath216 , we shall study first the nonlinear vector field @xmath217 acting on the decomposition @xmath218 , where @xmath166 belongs to the vector space @xmath178 . [ lemma - vector - field ] let @xmath23 be a bounded and @xmath24-periodic function . let @xmath219 and @xmath220 . if @xmath221 , then there exists a @xmath199 such that @xmath222 . we only need to prove that the space @xmath178 with @xmath223 forms a banach algebra in the sense @xmath224 for some @xmath199 , where @xmath225 and @xmath226 according to lemmas a.1 and a.2 of appendix a of @xcite , if @xmath227)$ ] and @xmath228 ) } = 1 $ ] for any @xmath195 , then there exists an @xmath82-independent constant @xmath199 , such that @xmath229 . by explicit computation , we obtain @xmath230 for some @xmath192 . if @xmath231 and @xmath232 , the bound is completed as follows @xmath233 for some @xmath234 . since @xmath235 , the parameter @xmath236 must satisfy @xmath221 . by using this construction , we prove the first main result of our analysis . let @xmath23 be a bounded , even and @xmath24-periodic function on @xmath27 . the nonlinear elliptic problem ( [ bifurcation ] ) has a continuous and @xmath87-periodic solution @xmath158 for sufficiently small @xmath157 if there exists a solution for @xmath237 of the algebraic coupled - mode equations @xmath238 + \epsilon r_3(a_1,a_2,a_3 ) , \end{array } \right.\ ] ] where parameters are given by @xmath239 and @xmath240 are analytic functions of @xmath157 near @xmath241 satisfying the bounds @xmath242 for some constants @xmath243 and sufficiently small @xmath244 . moreover , there exists an @xmath157-independent constant @xmath199 such that @xmath245 where @xmath246 are the modes defined by lemma [ lemma - resonance ] . [ theorem-1 ] @xmath247periodic solutions of the nonlinear elliptic problem ( [ bifurcation ] ) are expanded in the form @xmath248 let the vector @xmath249 with the elements of the sequence @xmath250 belong to the vector space @xmath251 equipped with the norm @xmath252 similarly to lemma [ corollary - decomposition-1 ] , it follows that the function @xmath158 is continuous if @xmath253 with @xmath180 . indeed , @xmath254 ) } \| v_{n_2 } \|_{c^0_b([-2\pi,2\pi ] ) } \\ & \leq & c \sum_{(n_1,n_2 ) \in \mathbb{n}^2 } |\phi_{n_1,n_2}| \| v_{n_1 } \|_{h^s_{\rm per}([-2\pi,2\pi ] ) } \| v_{n_2 } \|_{h^s_{\rm per}([-2\pi,2\pi ] ) } \\ & \leq & \tilde{c } \sum_{(n_1,n_2 ) \in \mathbb{n}^2 } ( 1 + n_1)^s ( 1 + n_2)^s |\phi_{n_1,n_2}|\end{aligned}\ ] ] for some @xmath255 and any @xmath180 . by construction , the solution @xmath158 is @xmath120-periodic in both coordinates of @xmath216 . the partial differential equation ( [ bifurcation ] ) is rewritten in the lattice form , which is diagonal with respect to the linear terms @xmath256 \phi_n = - \epsilon \sigma \sum_{(m , i , j ) \in \mathbb{n}^6 } m_{n , m , i , j } \phi_m \bar{\phi}_i \phi_j , \quad \forall n = ( n_1,n_2 ) \in \mathbb{n}^2,\ ] ] where @xmath257 depend on @xmath258 and @xmath259 since @xmath260 is a product of one - dimensional inner products and the weights in the norm in @xmath251 are separable , lemma [ lemma - vector - field ] applies and guarantees that the nonlinear vector field of the lattice equations ( [ bifurcation - lattice ] ) is closed in @xmath251 for @xmath221 . therefore , solutions of the lattice equations ( [ bifurcation - lattice ] ) can be considered in the space @xmath251 for any @xmath261 . by lemma [ lemma - resonance ] , the three resonant modes are isolated from all other @xmath120-periodic modes . by lemma [ lemma - perturbation - theory ] , the eigenvalues @xmath262 are then analytic in @xmath133 near @xmath117 . the three resonant modes correspond to the values of @xmath82 in the set @xmath263 . therefore , we decompose the solution @xmath249 of the lattice equations ( [ bifurcation - lattice ] ) into @xmath264 where @xmath265 are unit vectors on @xmath266 , @xmath267 is the kernel of the linearized system at the zero solution for @xmath241 , and @xmath268 lies in the orthogonal complement of the kernel such that @xmath269 . the linearized operator projected onto the orthogonal complement of the kernel is continuously invertible for sufficiently small @xmath157 . by the implicit function theorem in the space @xmath251 for any @xmath261 , there exists a unique map @xmath270 for sufficiently small @xmath157 . moreover , the map is locally analytic in @xmath157 near @xmath241 , such that @xmath271 . in addition , @xmath272 for any @xmath273 and @xmath274 is a homogeneous cubic polynomial of @xmath275 . let latexmath:[$\delta \equiv |a_1| + @xmath157-independent @xmath277 . then the map @xmath278 satisfies the bound latexmath:[\[\label{bound - remainder } \forall |\epsilon| < \epsilon_0 : \quad \| \vec{\mbox{\boldmath $ \psi$}}_{\epsilon}(a_1,a_2,a_3 ) \|_{l^1_s(\mathbb{n}^2 ) } \leq \epsilon c \left ( |a_1| + |a_2| + small , @xmath277 is finite , and the constant @xmath199 is independent of @xmath157 and @xmath280 . we can now consider the three equations of the system ( [ bifurcation - lattice ] ) for @xmath281 . after the taylor expansion of ( [ bifurcation - lattice ] ) in @xmath157 the formula ( [ leading - order ] ) yields the coefficients @xmath282 since @xmath283 . note that the @xmath24-inner product in ( [ leading - order ] ) is replaced by the @xmath120-inner product . although the bloch modes @xmath284 are now normalized over the @xmath120-long interval , the value of ( [ leading - order ] ) is unchanged . using the bound ( [ bound - remainder ] ) and the fact that the lattice equations are closed in @xmath251 , we derive the algebraic coupled - mode equations in theorem [ theorem-1 ] . the coefficients of these equations can be simplified due to the fact that @xmath23 is even on @xmath25 , such that the eigenfunctions @xmath128 and @xmath129 are even while @xmath285 is odd on @xmath25 . as a result , many coefficients of the coupled - mode system are identically zero , for instance @xmath286 , @xmath287 and so on . the residual terms @xmath288 are estimated from the map @xmath289 with the bound ( [ bound - remainder ] ) . the last estimate in the statement of theorem [ theorem-1 ] is obtained from ( [ bound - remainder ] ) and lemma [ corollary - decomposition-1 ] . we shall refer to the system ( [ algebraic - system - cme ] ) without remainder terms @xmath290 as to the truncated coupled - mode system . there exist five invariant reductions of the truncated coupled - mode system , namely ( i ) @xmath291 , ( ii ) @xmath292 , ( iii ) @xmath293 , ( iv ) @xmath294 , and ( v ) @xmath295 , which persist in the full lattice equations ( [ bifurcation - lattice ] ) . any of the five invariant reductions implies symmetry constraints on the function @xmath158 at the leading order of the decomposition ( [ decomposition-3 ] ) . for instance , if @xmath296 , then @xmath158 is @xmath24-antiperiodic with respect to both @xmath123 and @xmath124 ; if @xmath294 , then @xmath158 is @xmath24-antiperiodic in @xmath123 and @xmath87-periodic in @xmath124 , and so on . by the completeness results of proposition [ proposition-1 ] , all other terms of the decomposition ( [ decomposition-2 ] ) which violate the symmetry constraints on the solution @xmath158 can be set to be identically zero . by the implicit function theorem , the zero solution is unique near @xmath241 . therefore , the series ( [ decomposition-2 ] ) shrinks to fewer terms , the reduction persists for sufficiently small @xmath157 , and the proof of theorem [ theorem-1 ] applies . the invariant reduction @xmath297 of the truncated coupled - mode system may not satisfy the full lattice equations ( [ bifurcation - lattice ] ) since the solution @xmath158 with @xmath298 is still a @xmath87-periodic function of both @xmath123 and @xmath124 and the series ( [ decomposition-2 ] ) can not be shrunk to fewer terms for @xmath297 . [ remark - l2 ] the fourier bloch decomposition needed in theorem [ theorem-1 ] can be alternatively developed for @xmath166 in the vector space @xmath299 equipped with the squared norm @xmath300 is equivalent to the integral @xmath301 ) . by the asymptotic distribution ( [ asymptotic - distribution ] ) , @xmath302 is thus equivalent to @xmath303 . by sobolev s embedding theorem , @xmath304 for some @xmath199 and any @xmath305 therefore , the decomposition ( [ decomposition-1 ] ) produces a continuous function @xmath176 on @xmath25 if @xmath306 with @xmath305 . furthermore , using lemma 3.4 in @xcite , one can prove that the nonlinear term maps @xmath299 to @xmath299 for any @xmath180 , such that the arguments of the lyapunov schmidt reductions of theorem [ theorem-1 ] will work in the space @xmath299 for any @xmath180 . note that this approach would lift the upper bound on the index @xmath236 in theorem [ theorem-1 ] , where @xmath261 . [ example-2 ] for @xmath135 and @xmath307 , the parameters of the algebraic coupled - mode equations ( [ algebraic - system - cme ] ) are approximated numerically as follows : @xmath308 and @xmath309 the algebraic coupled - mode equations describe both linear and nonlinear corrections to the eigenvalues @xmath310 for fixed values of @xmath34 at the vertex points @xmath36 , @xmath39 and @xmath37 . in particular , these corrections show how the band edges of the three resonant bands split for @xmath311 and deform due to nonlinear interactions between resonant bloch modes . we skip further analysis of the algebraic coupled - mode equations and proceed with the decaying solutions of the nonlinear elliptic problem ( [ bifurcation ] ) described by the differential coupled - mode equations . we consider here bounded and decaying solutions @xmath158 of the bifurcation problem ( [ bifurcation ] ) for small @xmath157 . the decomposition of the decaying solution @xmath158 depends now on the completeness of the bloch modes of the one - dimensional sturm liouville problem ( [ e : bloch_eval_problem ] ) in @xmath66 , where all bloch modes for all @xmath51 must be incorporated in the fourier bloch decomposition . let @xmath23 be a bounded and @xmath24-periodic function . let @xmath168 and @xmath169 be sets of eigenvalues and eigenfunctions of the sturm liouville problem ( [ e : bloch_eval_problem ] ) on @xmath51 , such that @xmath312 where @xmath313 and @xmath314 is the dirac s delta function in the distribution sense . then , there exists a unitary transformation @xmath315 given by @xmath316 where @xmath317 is equipped with the squared norm @xmath318 [ proposition-2 ] the original proof of the proposition can be found in @xcite . orthogonality and completeness of the bloch wave functions @xmath319 on @xmath51 is summarized in theorem xiii.98 on p.304 in @xcite . existence of the unitary transformation ( [ direct - integral ] ) with the inverse transformation ( [ inverse - integral ] ) is summarized in theorems xiii.97 and xiii.98 on pp . 303304 in @xcite . [ corollary - decomposition-2 ] let @xmath176 be defined by the decomposition ( [ inverse - integral ] ) , where @xmath320 belongs to the vector space @xmath321 with the norm @xmath322 if @xmath180 , then @xmath176 is a continuous function on @xmath25 such that @xmath323 as @xmath324 . the proof is similar to that of lemma [ corollary - decomposition-1 ] since the asymptotic bound ( [ asymptotic - distribution ] ) is uniform on @xmath67 . therefore , @xmath325 for some @xmath255 and any @xmath180 . the decay of @xmath176 as @xmath324 follows from the riemann lebesgue lemma applied to the fourier bloch transform . indeed , the summation in @xmath195 of the integrals on @xmath67 can be written as an integral on @xmath326 in the form @xmath327 where @xmath328 for @xmath326 is related to the bloch functions @xmath58 for @xmath67 ( see @xcite for details ) . since @xmath329 , where @xmath330 is periodic in @xmath331 with period @xmath24 and is uniformly bounded in @xmath332 with respect to both @xmath331 and @xmath48 , the riemann lebesque lemma applies to the fourier bloch transform in the same manner as it applies to the standard fourier transform if @xmath333 . let @xmath23 be a bounded , piecewise - continuous and @xmath24-periodic function . let @xmath334 be the extremal value of @xmath57 for either @xmath335 or @xmath336 and assume that the adjacent bands are bounded away from the value @xmath334 . then , @xmath57 is an analytic function at the point @xmath337 , such that @xmath338 for some @xmath199 uniformly in @xmath67 . [ lemma - band - curvature ] by theorem xiii.95 in @xcite , the function @xmath57 for an isolated spectral band can be extended onto a smooth riemann surface which has no singularities other than square root branch points at @xmath60 and @xmath61 . therefore , the function @xmath57 is expanded in even powers of @xmath339 for @xmath335 or @xmath340 and @xmath341 . [ lemma - vector - field-2 ] let @xmath23 be a bounded and @xmath24-periodic function . let @xmath342 and @xmath343 . if @xmath221 , then there exists a @xmath199 such that @xmath344 . the proof is similar to that of lemma [ lemma - vector - field ] since the asymptotic bound ( [ bound - k ] ) is uniform on @xmath51 @xcite and the length of @xmath345 is finite . the banach algebra property ( [ banach - algebra ] ) holds in space @xmath321 for any @xmath221 . indeed , let @xmath346 where @xmath347 for all @xmath348 and @xmath349 . by an explicit computation , we obtain @xmath350 . the last sum is finite if @xmath351 and @xmath352 ( that is @xmath221 ) , similarly to the proof of lemma [ lemma - vector - field ] . we shall now define the coupled - mode system , which is a central element of our paper . let @xmath353 be functions of @xmath354 on @xmath216 which satisfy the differential coupled - mode equations @xmath355 \\ ( \omega - \beta_1 ) a_2 + \left ( \alpha_2 \partial_{y_1}^2 + \alpha_1 \partial_{y_2}^2 \right ) a_2 & = & \sigma \left [ \gamma_1 + \gamma_3 ( 2 |a_3|^2 a_2 + a_3 ^ 2 \bar{a}_2 ) \right ] \\ ( \omega - \beta_2 ) a_3 + \alpha_3 \left ( \partial_{y_1}^2 + \partial_{y_2}^2 \right ) a_3 & = & \sigma \left [ \gamma_4 |a_3|^2 a_3 + 2 \gamma_3 ( |a_1|^2 + |a_2|^2 ) a_3 + \gamma_3 ( a_1 ^ 2 + a_2 ^ 2 ) \bar{a}_3 \right ] , \end{array } \right.\ ] ] where parameters @xmath356 and @xmath357 are the same as in theorem [ theorem-1 ] , while @xmath358 since @xmath359 , @xmath360 , and @xmath361 are non - degenerate eigenvalues in the ordering ( [ ordering ] ) , the values of @xmath362 are non - zero according to lemma [ lemma - band - curvature ] . [ example-3 ] for @xmath135 and @xmath363 , the parameters of the differential coupled - mode equations are approximated numerically as follows : @xmath364 let @xmath365 on @xmath366 denote the fourier transforms of the functions @xmath353 on @xmath367 such that @xmath368 we shall use the standard norm in @xmath369 for the fourier transforms @xmath365 with any @xmath370 : @xmath371(p),\\ \left ( \omega - \beta_1 - \alpha_2 p_1 ^ 2 - \alpha_1 p_2 ^ 2 \right ) \hat{a}_2(p ) & = & \sigma \hat{n}_2[\hat{a}_1,\hat{a}_2,\hat{a}_3](p ) , \\ \left ( \omega - \beta_2 - \alpha_3 p_1 ^ 2 - \alpha_3 p_2 ^ 2 \right ) \hat{a}_3(p ) & = & \sigma \hat{n}_3[\hat{a}_1,\hat{a}_2,\hat{a}_3](p ) , \end{array } \right.\end{aligned}\ ] ] where @xmath372(p)$ ] denote the fourier transform of the cubic nonlinear terms of the differential coupled - mode equations . we now prove the second main result of our analysis . let @xmath23 be a bounded , piecewise - continuous , even and @xmath24-periodic function on @xmath25 . let @xmath373 . the nonlinear elliptic problem ( [ bifurcation ] ) has a continuous and decaying solution @xmath158 for sufficiently small @xmath374 if there exists a solution @xmath375 , compactly supported on the disk @xmath376 , which satisfies the extended coupled - mode equations in the integral form @xmath377 where @xmath378 , @xmath379 denote the cubic nonlinear terms @xmath380(p)$ ] truncated on @xmath381 , and @xmath382 are bounded by @xmath383 for some constants @xmath243 and sufficiently small @xmath244 . moreover , there exists an @xmath157-independent constant @xmath384 such that @xmath385 where @xmath386 are modes defined by lemma [ lemma - resonance ] and @xmath387 are defined by the fourier transform ( [ solution - integral ] ) . [ theorem-2 ] solutions of the nonlinear elliptic problem ( [ bifurcation ] ) are expanded in the form @xmath388 let @xmath389 denote the union of functions @xmath390 for all @xmath391 and @xmath392 and consider the vector space @xmath393 equipped with the norm @xmath394 one can show similarly to the proof of theorem [ theorem-1 ] that the statements of lemmas [ corollary - decomposition-2 ] and [ lemma - vector - field-2 ] extend in two dimensions to the vector space with the norm ( [ vector - space-2 - 2d ] ) . therefore , we convert the partial differential equation ( [ bifurcation ] ) to the integral form , which is diagonal with respect to the linear terms @xmath395 \hat{\phi}_n(k ) \\ \nonumber & = & - \epsilon \sigma \sum_{(m , i , j ) \in \mathbb{n}^6 } \int_{\mathbb{t}^6 } m_{n , m , i , j}(k , l,\kappa,\lambda ) \hat{\phi}_m(l ) \bar{\hat{\phi}}_i(\kappa ) \hat{\phi}_j(\lambda ) dl d \kappa d \lambda\end{aligned}\ ] ] for all @xmath396 and @xmath397 , where @xmath398 for all @xmath399 and @xmath400 . in view of lemmas [ corollary - decomposition-2 ] and [ lemma - vector - field-2 ] we consider solutions of the integral equations ( [ bifurcation - integral ] ) in the space @xmath393 for any @xmath401 . when @xmath117 , the multiplication operator @xmath402 vanishes at the points @xmath39 , @xmath36 , and @xmath37 , where the values of @xmath48 are given by @xmath403 with @xmath404 resp . therefore , we apply the method of lyapunov schmidt reductions @xcite and decompose the solution @xmath389 of the integral equations ( [ bifurcation - integral ] ) into @xmath405 where @xmath265 are unit vectors on @xmath266 , @xmath406 are disks of radius @xmath407 centered at the points @xmath48 of the set ( [ resonant - points ] ) , see figure [ f : k_sp_decomp ] . since @xmath408 , where @xmath409 is the first brillouin zone , we map @xmath410 on @xmath409 by periodic continuation and denote the resulting object by @xmath411 . in the representation ( [ decomposition-4 ] ) , @xmath412 is the characteristic function on @xmath408 ( @xmath413 if @xmath414 and @xmath415 if @xmath416 ) , and @xmath417 is zero identically on @xmath418 for the corresponding values of @xmath82 : @xmath419 ( 6,6)(0,0 ) ( 1,4.5)1 ( 4.5,4.5)1 ( 4.5,1)1 ( 0,1)(6,1 ) ( 1,0)(1,6 ) ( 4.5,4.5)(3.79,5.21 ) ( 1,4.5)(4.5,1)(4.5,4.5 ) ( 0.45,4.35)@xmath39 ( 1.75,5.25)@xmath420 ( 4.3,0.45)@xmath36 ( 5.25,1.75)@xmath421 ( 4.7,4.3)@xmath37 ( 5.25,5.25)@xmath422 ( 3.8,4.5)@xmath407 ( 5.8,0.6)@xmath41 ( 0.4,5.8)@xmath40 for any of the three resonant points of @xmath48 , it follows by lemma [ lemma - band - curvature ] that @xmath423 and @xmath424 for @xmath425 with the @xmath82 corrresponding to the resonance point @xmath48 . as a result , there exists a constant @xmath199 such that @xmath426 note that the minimal values of the multiplication operator occur for the three resonant bands with the indices @xmath82 in the set @xmath427 . for all other non - resonant bands , the multiplication operator is continuously invertible uniformly in @xmath157 for sufficiently small @xmath157 . if @xmath428 , the lower bound ( [ lower - bound ] ) is still larger than the perturbation terms of order @xmath157 . by the implicit function theorem in the space @xmath393 for any @xmath261 , there exists a unique map @xmath429 for sufficiently small @xmath374 . the right - hand side of the integral equation ( [ bifurcation - integral ] ) is a homogeneous cubic polynomial of @xmath417 and @xmath430 multiplied by @xmath157 . let @xmath431 for a fixed @xmath157-independent @xmath277 . the map @xmath432 satisfies the bound @xmath433 for sufficiently small @xmath434 , finite @xmath435 , and the @xmath436-independent constant @xmath199 . we shall now map the disks @xmath406 to the same disk @xmath437 in the stretched variable @xmath438 centered at the origin . to do so , we apply the scaling transformation @xmath439 where @xmath440 is the corresponding point of the set ( [ resonant - points ] ) . the new disk @xmath441 is now @xmath442 and it covers the entire plane @xmath443 in the limit @xmath444 if @xmath428 . note that the @xmath445-norm of @xmath446 on @xmath447 is invariant with respect to the transformation ( [ scaling - transformation ] ) in the new variable @xmath448 on @xmath449 , such that @xmath450 for any @xmath451 . when the integral equations ( [ bifurcation - integral ] ) are considered on the compact domains @xmath421 , @xmath420 , @xmath422 for the three resonant modes and the scaling transformation ( [ scaling - transformation ] ) is applied to map these domains into the domain @xmath441 , we obtain the extended coupled - mode system ( [ cme - integral - form - extended ] ) , where remainder terms occur due to three different sources . the first source comes from the component @xmath452 and it is estimated with the bound ( [ bound - remainder-2d ] ) . the remainder terms have the order of @xmath453 and are small if @xmath454 . the second source comes from the expansion of @xmath455 and @xmath456 in powers of @xmath157 and it is estimated with lemma [ lemma - perturbation - theory ] . the remainder terms have the order of @xmath457 . the third source comes from the expansion of @xmath455 and @xmath456 in powers of @xmath458 and it is estimated with lemma [ lemma - band - curvature ] . the remainder terms have the order of @xmath459 because of the following estimate : @xmath460 the remainder terms are small if @xmath461 . thus , the statement is proved if @xmath462 and the largest remainder terms have the order @xmath463 with @xmath464 . if @xmath465 , then @xmath466 and both remainder terms of the extended coupled - mode equations are of the same order in @xmath157 . the fastest convergence rate of the remainder terms is @xmath467 because @xmath468 . the differential coupled - mode system on @xmath353 has invariant reductions when one or two components of @xmath469 are identically zero . in particular , the reduction @xmath297 recovers coupled - mode equations near the band edge @xmath470 derived in @xcite , while the reduction @xmath291 recovers coupled - mode equations near the band edge @xmath471 in @xcite . however , these reductions do not generally persist in the extended coupled - mode equations near @xmath117 since the reductions do not imply any symmetry constraints on the function @xmath158 . the coupled - mode equations linearized at the zero solution describe the expansions of the function @xmath95 near the vertex points @xmath39 , @xmath36 , and @xmath37 in terms of the perturbation parameter @xmath472 and the small deviation of @xmath473 from the value @xmath474 in the set ( [ resonant - points ] ) @xmath475 we note that the values of @xmath476 are equal for the vertex points @xmath36 and @xmath39 , while the values of @xmath477 and @xmath478 have the same sign for all points @xmath36 , @xmath39 and @xmath37 . moreover , @xmath479 . a simple analysis of ( [ omega - expansion ] ) shows that the band gap between the three resonant bloch band surfaces exists if @xmath480 or @xmath481 examples [ example-2 ] and [ example-3 ] show that the first case occurs for the particular potential @xmath135 if @xmath482 , such that the band gap opens for @xmath374 in the interval @xmath483 , in a correspondence with a narrow band gap for @xmath482 in figure [ f : bands](b ) . the coupled - mode system in the integral form ( [ cme - integral - form ] ) is different from the extended coupled - mode system ( [ cme - integral - form - extended ] ) in two aspects . first , the convolution integrals are truncated in on the domain @xmath441 for @xmath484 , where @xmath462 . second , the remainder terms of order @xmath463 with @xmath485 are present in . the first source is small if solutions of the extended coupled - mode system are considered on @xmath443 in the space @xmath369 for some @xmath486 . the second source is handled with the implicit function theorem . we shall consider here a special class of decaying solutions of the coupled - mode system ( [ cme - integral - form ] ) on @xmath443 which lead to the corresponding solutions of the extended coupled - mode system ( [ cme - integral - form - extended ] ) on @xmath487 . [ definition - reversible ] a solution @xmath275 of the differential coupled - mode system on @xmath367 is called reversible if it satisfies one of the two constraints ( [ reversibility ] ) and ( [ reversibility-2 ] ) for each function @xmath488 , @xmath489 and @xmath490 . for notational simplicity , we write the differential coupled - mode equations in the form @xmath491 , where @xmath492 is a nonlinear operator on @xmath493 with the range in @xmath494 . the jacobian of the operator @xmath495 denoted by @xmath496 is a matrix differential operator , which is diagonal with respect to the unbounded differential part and full with respect to the local potential part . in many cases , the jacobian operator @xmath497 can be block - diagonalized and simplified but we can work with a general operator to prove the third main result of our analysis . [ theorem-3 ] let @xmath23 satisfy the same assumptions as in theorem [ theorem-2 ] . let @xmath498 belong to the interior of the band gap of the coupled - mode system @xmath491 . let @xmath275 on @xmath367 be a reversible solution of the differential coupled - mode system @xmath491 in the sense of definition [ definition - reversible ] such that their fourier transforms satisfy @xmath499 for all @xmath370 . assume that the jacobian operator @xmath496 has a three - dimensional kernel with the eigenvectors @xmath500 . then , there exists a solution @xmath375 of the extended coupled - mode system ( [ cme - integral - form - extended ] ) such that @xmath501 for some @xmath157-independent constant @xmath502 and sufficiently small @xmath244 . first , we consider system ( [ cme - integral - form - extended ] ) on the entire plane @xmath366 in the space @xmath369 for a fixed @xmath370 by extending the right - hand - side functions @xmath382 on @xmath503 with a compact support on @xmath441 . thanks to the assumption on the existence of a solution @xmath504 of @xmath505 and its fast decay in @xmath369 for all @xmath370 , we decompose the solution of the extended system @xmath506 on @xmath443 into @xmath507 , where @xmath508 solves the nonlinear problem in the form @xmath509,\ ] ] where @xmath510 is a linearized operator , @xmath511 is quadratic in @xmath512 , and @xmath513 maps an element @xmath514 to itself . the kernel of the jacobian operator @xmath515 is exactly three - dimensional , by the assumption , and it is generated by the two - dimensional group of translations in @xmath516 and a one - dimensional gauge invariance in @xmath517 . if @xmath498 belongs to the interior of the band gap of the coupled - mode system , the continuous spectrum of @xmath518 is bounded away from zero and the triple zero eigenvalue is isolated from other eigenvalues of the discrete spectrum of @xmath496 . the nonlinear elliptic problem ( [ stationary ] ) with a real - valued symmetric potential @xmath9 admits the gauge invariance @xmath519 for any @xmath520 and the reversibility symmetries ( [ reversibility ] ) and ( [ reversibility-2 ] ) . the extended coupled - mode system ( [ cme - integral - form - extended ] ) obtained from the integral system ( [ bifurcation - integral ] ) after the lyapunov schmidt decomposition ( [ decomposition-4 ] ) and the scaling transformation ( [ scaling - transformation ] ) inherits all the symmetries , such that after restriction of @xmath521 to functions satisfying ( [ reversibility ] ) or ( [ reversibility-2 ] ) and after setting uniquely the phase , e.g. by @xmath522 , system ( [ lin - abstract - form ] ) is formulated in the orthogonal complement of the kernel of @xmath510 . therefore , the linearized operator is continuously invertible for sufficiently small @xmath374 under this restriction . by the implicit function theorem , there exists a unique reversible solution in the form @xmath523 such that @xmath524 for some @xmath199 and any @xmath370 . this result implies the desired bound ( [ bound - final ] ) for @xmath525 as a solution of ( [ cme - integral - form - extended ] ) on @xmath366 . next , we consider the error between system ( [ cme - integral - form - extended ] ) on the disk @xmath487 and the same system on the plane @xmath443 . the error comes from the terms @xmath526 and @xmath527 , where @xmath528 , since @xmath529 for all @xmath486 and the cubic nonlinear terms @xmath530 of the coupled - mode system ( [ cme - integral - form ] ) map elements of @xmath369 to elements of @xmath369 . since @xmath531 and @xmath532 is a map from @xmath533 to @xmath534 for any @xmath370 , we obtain that @xmath535 for some @xmath536 . since @xmath537 , the error between system ( [ cme - integral - form - extended ] ) on the disk @xmath487 and the same system on the plane @xmath443 is of the same order as the right - hand - side terms of system ( [ cme - integral - form - extended ] ) , such that the desired bound ( [ bound - final ] ) holds on @xmath381 . [ c : reversibility ] the solution @xmath158 constructed in theorems [ theorem-2 ] and [ theorem-3 ] is a reversible localized solution of the bifurcation problem ( [ bifurcation ] ) satisfying one of the two constraints ( [ reversibility ] ) and ( [ reversibility-2 ] ) on @xmath216 . since the modes @xmath386 of lemma [ lemma - resonance ] and the solution @xmath387 of theorem [ theorem-3 ] satisfy one of the two reversibility constraints ( [ reversibility ] ) and ( [ reversibility-2 ] ) , the leading - order part of the representation ( [ decomposition-4 ] ) produces a solution @xmath158 which satisfies the same constraint . since the symmetry is also preserved in the integral equation ( [ bifurcation - integral ] ) , the map @xmath432 constructed in theorem [ theorem-2 ] inherits the symmetry and produces the remainder term in the decomposition ( [ decomposition-4 ] ) satisfying the same reversibility constraint . there are several classes of localized reversible solutions of the differential coupled - mode equations which satisfy the assumptions of theorem [ theorem-3 ] . these solutions are numerically approximated in section [ s : numerical ] . according to theorem [ theorem-3 ] , all these solutions persist as localized reversible solutions of the nonlinear elliptic problem ( [ stationary ] ) . [ c : conv_rate_final ] the solution @xmath158 constructed in theorems [ theorem-2 ] and [ theorem-3 ] satisfy the bound : @xmath538 where @xmath462 . the fastest convergence rate is @xmath467 , which is obtained at @xmath539 . consequently , the solution @xmath540 of the nonlinear elliptic problem is @xmath541 accurate to the approximation @xmath542 , with the fastest convergence rate being @xmath543 . using the triangle inequality and the bound , we obtain @xmath544 for some @xmath545 . by the hlder inequality we have @xmath546 with the last equality due to a compact support of @xmath547 on @xmath487 . the first term in the upper bound is @xmath548 by the bound and the second term is smaller than any power of @xmath157 if @xmath499 for all @xmath370 . we approximate here several classes of localized reversible solutions of the differential coupled - mode system numerically . we use the same potential as in the examples [ example-1][example-3 ] , i.e. @xmath549 so that the bifurcation takes place for @xmath550 . for selected representative cases we also verify the convergence rate of corollary [ c : conv_rate_final ] and the conditions of theorem [ theorem-3 ] on the jacobian operator , which guarantees persistence of these solutions in the full system . we limit our attention to the following classes of reversible solutions : * defocusing case @xmath551 : @xmath552 , @xmath553 , @xmath554 , @xmath555 + @xmath556 radially symmetric positive soliton + @xmath557 vortex of charge one * focusing case @xmath558 : @xmath559 * * @xmath560 , @xmath561 , @xmath562 , @xmath563 + @xmath556 ellipsoidal positive soliton + @xmath557 ellipsoidal vortex of charge one * * @xmath564 + symmetric real coupled soliton * * @xmath565 + @xmath566-phase delay coupled soliton * * @xmath567 + coupled vortex of charge one we will see that the localized solutions with @xmath568 bifurcate at @xmath569 from the upper edge @xmath570 of the band gap and the localized solutions with @xmath552 bifurcate at @xmath571 from the lower edge @xmath572 of the band gap . all solutions above satisfy the reversibility condition of definition [ definition - reversible ] . provided the assumptions of theorem [ theorem-3 ] are satisfied , corollary [ c : reversibility ] then guarantees that these solutions correspond to reversible solutions of the nonlinear elliptic problem . using the symmetries of @xmath573 , @xmath574 and @xmath575 , we can see that the function @xmath176 satisfies the following symmetry : * [ cols= " < , < " , ] note that ( b - iii ) and ( b - iv ) agree with the reversibility constraint ( [ reversibility-2 ] ) after multiplication by @xmath576 . the above solutions do not include any three - component gap solitons with all @xmath577 and @xmath578 nonzero . such solutions do not bifurcate from either upper or lower edge @xmath570 or @xmath572 , respectively , and it is thus hard to capture such solutions numerically . it is , nevertheless , possible that the three - component solitons still exist in the interior of the gap @xmath483 but we do not attempt here to locate these special solutions . one - component solutions correspond to classes ( a ) and ( b - i ) . the function @xmath579 for class ( a ) satisfies the ode @xmath580 where @xmath581 , @xmath582 for @xmath556 and @xmath583 , @xmath584 for @xmath557 . for @xmath585 the initial - value problem for the ode ( [ ode - radial ] ) is ill - posed but can be turned into a well - posed one via the transformation @xmath586 leading to @xmath587 with @xmath588 , such that @xmath589 as @xmath590 . similarly , the function @xmath579 for class ( b - i ) satisfies the ode @xmath591 we solve these equations numerically via a shooting method searching for @xmath579 vanishing at infinity as @xmath592 . figure [ f : cont_lines_1_comp ] shows the solution families in the frequency - amplitude space for both solutions with @xmath593 and @xmath594 . class ( a ) families bifurcate from the lower edge @xmath595 and class ( b - i ) families from the upper edge @xmath596 . examples of the gap solitons @xmath597 are in figure [ f : single_comp_townes ] and of the vortices ( @xmath557 ) in figure [ f : single_comp_vort ] . as expected , the class ( a ) solutions are radially symmetric and the class ( b ) solutions ellipsoidal . for the case of one - component solutions , the coupled - mode equations reduce to a scalar two - dimensional nls equation . conditions of theorem [ theorem-3 ] on the kernel of the jacobian operator are known to be satisfied for @xmath593 @xcite and , therefore , do not need to be checked numerically . figure [ f : eps_conv_1_comp ] presents the computed @xmath157-convergence of the error term @xmath598 described by corollary [ c : conv_rate_final ] for the solution classes ( a ) and ( b - i ) with @xmath556 . the convergence rate is seen around @xmath599 and @xmath600 , which is higher than the fastest convergence rate @xmath543 predicted by corollary [ c : conv_rate_final ] . the observed rate is close to @xmath601 , which is the rate predicted by a formal multiple scales asymptotic expansion of @xmath176 , for which @xmath602 is the leading order term . and @xmath603 correspond to the profiles in fig . [ f : single_comp_townes ] and [ f : single_comp_vort].,height=226 ] . ( a ) @xmath578 at point p in fig . [ f : cont_lines_1_comp ] ; ( b ) the corresponding leading - order term @xmath604 for @xmath605 ; ( c ) @xmath606 at point q in fig . [ f : cont_lines_1_comp ] ; ( d ) the corresponding leading - order term @xmath607 for @xmath608.,height=294 ] ) . ( a ) @xmath609 at point r in fig . [ f : cont_lines_1_comp ] ; ( b ) and ( c ) the modulus and phase , respectively , of the corresponding leading - order term @xmath604 for @xmath605 ; ( d ) @xmath610 at point s in fig . [ f : cont_lines_1_comp ] ; ( e ) and ( f ) the modulus and phase , respectively , of the corresponding leading - order term @xmath607 for @xmath611 . , height=294 ] -convergence of the error term for the gap soliton ( @xmath593 ) corresponding to ( a ) at @xmath612 and ( b - i ) at @xmath613.,height=151 ] the following algorithm is used to compute two - component solutions of classes ( b - ii ) , ( b - iii ) and ( b - iv ) . first , @xmath614 and @xmath615 are replaced by @xmath616 so that the first two equations of the coupled - mode system admit a reduction @xmath617 , where @xmath561 solves a one - component problem . the function @xmath579 is then computed as in section [ s:1-comp ] . next , homotopy in the coefficients @xmath614 and @xmath615 is employed to gradually deform the computed solution with @xmath618 into the one with the original values of @xmath614 and @xmath615 . in this homotopy continuation the full coupled - mode system needs to be solved numerically and this is done via newton s method on a central finite - difference discretization with zero dirichlet boundary conditions . figure [ f : cont_lines_2_comp ] shows the solution families of classes ( b - ii ) , ( b - iii ) and ( b - iv ) as curves in the frequency - amplitude space . profiles of examples of these solutions corresponding to the marked points in figure [ f : cont_lines_2_comp ] appear in figures [ f : two_comp_real ] , [ f : two_comp_pi_half_delay ] and [ f : two_comp_vort ] . note that the coupled vortex in figure [ f : two_comp_vort ] has been obtained via the above described homotopy continuation from a vortex of charge one with @xmath619 . as the phase plots of @xmath606 and @xmath620 show , the resulting coupled vortex is also of charge one . in order to verify the persistence conditions of theorem [ theorem-3 ] for the two - component solutions , the kernel of the jacobian operator @xmath621 has to be studied numerically . when rewritten in the real variables ( for real and imaginary part ) the first two equations of the coupled - mode system give rise to a jacobian operator @xmath621 . for the cases ( b - ii ) and ( b - iii ) , the jacobian operator can be block - diagonalized to the form @xmath622 such that @xmath623 . the elements of @xmath624 can , moreover , be easily reconstructed from the elements of @xmath625 and @xmath626 . this diagonalization reduces the expense of the eigenvalue solver . figure [ f : kernel_2_comp_real_delay ] shows the four smallest ( in modulus ) eigenvalues @xmath627 of the 4th order central finite - difference approximation of @xmath621 for examples of solutions ( b - ii ) and ( b - iii ) as functions of @xmath410 , where @xmath410 is the size of the computational box @xmath628\times [ -d , d ] \subset \mathbb{r}^2 $ ] with the step sizes @xmath629 and with zero dirichlet boundary conditions . clearly , the three smallest eigenvalues converge to zero as @xmath410 grows while the fourth eigenvalue converges to a nonzero value . therefore , the assumptions of theorem [ theorem-3 ] are verified for these solutions . we have also checked that the corresponding eigenvectors of the three zero eigenvalues are approximations of @xmath630 and @xmath631 . for the case ( b - iv ) the jacobian operator @xmath621 has to be treated in full because it can not be block - diagonalized for complex - valued solutions . figure [ f : kernel_2_comp_vort ] shows the four smallest eigenvalues of the discretized jacobian operator for an example of the solution ( b - iv ) . the fourth - order central finite - difference discretization was used once again with the step sizes @xmath632 . the fourth eigenvalue @xmath633 converges approximately to @xmath634 as @xmath410 grows , while the second and third eigenvalues @xmath635 converge to @xmath636 . we have checked that the limit of @xmath635 decreases as @xmath637 is decreased while the limit of @xmath633 remains practically unchanged . finally , figure [ f : eps_conv_2_comp ] verifies the @xmath157-convergence of the error term @xmath598 described by corollary [ c : conv_rate_final ] for the gap soliton of class ( b - ii ) at @xmath638 . the observed convergence rate is @xmath639 , which is , once again , close to @xmath601 expected from the formal asymptotic expansion . the @xmath157-convergence has not been checked for any complex - valued solutions of classes ( b - iii ) and ( b - iv ) due to four times larger memory requirements compared to the real case . the memory use grows rapidly as @xmath157 decreases since domains of the size @xmath640 need to be discretized to compute the solution @xmath176 of the elliptic problem and each period @xmath164 \times [ 0,2\pi]$ ] of the potential function @xmath9 requires about 100 grid points . correspond to the profiles in figs . [ f : two_comp_real ] and [ f : two_comp_pi_half_delay ] , while the point v at @xmath641 corresponds to the profiles in fig . [ f : two_comp_vort].,height=151 ] . ( a ) @xmath606 ; ( b ) @xmath620 ; ( c ) the leading - order term @xmath642 for @xmath611.,height=179 ] -phase delay gap soliton ( b - iii ) at point u in fig . [ f : cont_lines_2_comp ] . ( a ) @xmath606 ; ( b ) @xmath643 ; ( c ) and ( d ) modulus and phase , respectively , of the leading - order term @xmath642 for @xmath611.,height=151 ] . . ( a ) and ( b ) modulus and phase of @xmath606 ; ( c ) and ( d ) modulus and phase of @xmath620 ; ( e ) and ( f ) modulus and phase of the leading - order term @xmath642 for @xmath611.,height=226 ] for ( a ) the two - component real gap soliton ( b - ii ) and ( b ) the two - component @xmath566-phase delay gap soliton ( b - iii ) for @xmath644.,height=132 ] for the two - component vortex ( b - iv ) at @xmath645.,height=141 ] -convergence of the error term for the gap soliton ( b - ii ) with @xmath646.,height=151 ] we show here that dynamics of non - stationary localized solutions of the gross pitaevskii equation ( [ gp ] ) are close to the dynamics of the time - dependent coupled - mode equations for finite time intervals . these results are similar to the ones in @xcite and they give a rigorous basis for formal asymptotic results in @xcite . according to a formal asymptotic multiple scales expansion , solutions of the gross pitaevskii equation in the form @xmath647 where @xmath648 e^{i\omega_0 t},\end{aligned}\ ] ] are approximated by solutions of the time - dependent coupled - mode equations @xmath649 , \\ ( i\partial_t - \beta_1 ) a_2 + \left ( \alpha_2 \partial_{y_1}^2 + \alpha_1 \partial_{y_2}^2 \right ) a_2 & = & \sigma \left [ \gamma_1 + \gamma_3 ( 2 |a_3|^2 a_2 + a_3 ^ 2 \bar{a}_2 ) \right ] , \\ ( i \partial_t - \beta_2 ) a_3 + \alpha_3 \left ( \partial_{y_1}^2 + \partial_{y_2}^2 \right ) a_3 & = & \sigma \left [ \gamma_4 |a_3|^2 a_3 + 2 \gamma_3 ( |a_1|^2 + |a_2|^2 ) a_3 + \gamma_3 ( a_1 ^ 2 + a_2 ^ 2 ) \bar{a}_3 \right ] , \end{array } \right . \label{time - dependent - cme}\ ] ] where @xmath354 and @xmath650 . under the non - resonance condition @xmath651 where @xmath652 and the infinimum does not include direct resonances that give the nonlinear terms in system ( [ time - dependent - cme ] ) , the following theorem states the approximation result on the time - dependent solutions . let @xmath653 , and assume that the non - resonance condition holds . then for all @xmath654 and @xmath655 there exist @xmath656 and @xmath657 such that for all solutions @xmath658 , h^{s_*}(\mathbb{r}^2))$ ] of the time - dependent coupled - mode system with @xmath659 } \|{\bf a}(\cdot , t)\|_{h^{s_*}(\mathbb{r}^2 ) } \leq c_1,\ ] ] there exist solutions @xmath660 , h^s(\mathbb{r}^2))$ ] of the gross pitaevskii equation with @xmath661 } \|e ( \cdot , t ) - e_{ans}(\cdot , t)\|_{h^s(\mathbb{r}^2 ) } \leq c_2\epsilon^{3/2}.\ ] ] for all @xmath662 . the proof is very similar to the one given for a scalar nonlinear schrdinger equation in the one - dimensional case in @xcite . the non - resonance condition ( [ gs4 ] ) differs from ( * ? ? ? * eq.(4 ) ) by including the terms with @xmath663 . this is necessary to make the residual formally of order @xmath664 . due to the fact that we are working here on @xmath216 , we lose @xmath665 due to the scaling of the @xmath666-norm in contrast to @xcite , where we would only lose @xmath667 on @xmath25 with the same scaling . assuming that the non - resonance condition is satisfied , the @xmath668 terms in @xmath669 can be chosen in such a way that all terms up to order @xmath670 in the residual @xmath671 can be eliminated . by ( * lemma 3.3 ) bloch transform is an isomorphism between @xmath672 and @xmath673 . hence , similarly to ( * ? ? ? * lemma 3.4 ) the nonlinearity @xmath674 maps @xmath673 in bloch space , or @xmath672 in physical space , into itself for @xmath675 due to sobolev s embedding theorem in two dimensions . moreover , @xmath676 generates a uniformly bounded semigroup in @xmath673 , or respectively in @xmath672 , according to @xcite . hence , the error function @xmath677 satisfies an equation of the form @xmath678 using the variation of constant formula and gronwall s inequality gives the required @xmath679 bound for @xmath680 . for more details see ( * ? ? ? * section 4.2 ) . we have given a detailed justification of the three coupled - mode equations which describe gap solitons bifurcating in the first band gap of a two - dimensional separable periodic potential . these equations are generic ( structurally stable ) for the considered bifurcation but they can be modified for other relevant bifurcation problems . we review here several examples of other bifurcations , where the justification analysis is expected to be applicable in a similar manner . _ bifurcations from band edges . _ if the band edge separates the existing band gap of finite length and a single or double - degenerate spectral band , the coupled - mode equations persist without any modifications . a formal derivation is given in the recent paper @xcite , where a scalar nls equation is derived for a single band and two coupled nls equations are derived for a double band . note that the coupled nls equations for a double band may have terms destroying the reduction @xmath294 or @xmath681 in the truncated system of equations . this happens , for instance , at the band edge @xmath682 in the second band gap of a separable symmetric potential @xcite . since the analysis of separable potentials relies on the construction of one - dimensional bloch modes and one - dimensional spectral bands , the same analysis is valid for bifurcations from band edges in one - dimensional problems . the formal derivation of the nls equation for a band edge of a single spectral band in the one - dimensional gp equation was reported in @xcite . _ bifurcations in the higher - order band gaps . _ with larger values of the parameter @xmath133 in the separable potential ( [ separable - potential ] ) , more band gaps open , although the number of band gaps is always finite for finite @xmath133 , see theorem 6.10.5 in @xcite . since the bifurcation of new band gaps occurs due to the resonance of finitely many bloch modes in the separable potential , a system of finitely - many coupled - mode equations can be derived similarly in the higher - order band gaps . for instance , the bifurcation of the second band gap in the symmetric separable potential may occur for a resonance of either three modes when @xmath683 or four modes when @xmath684 or five modes when @xmath685 . _ bifurcations in asymmetric separable potentials . _ if the separable potential is asymmetric , such that @xmath686 where @xmath687 and @xmath688 are two different functions on @xmath17 , the degeneracy of spectral bands is broken and bifurcations of the band gaps occur with a smaller number of resonant bloch modes . our analysis remains valid but the spectrum of two different one - dimensional sturm liouville problems must be incorporated in the fourier bloch decomposition . in this case , two coupled - mode equations occur generally for the bifurcation of the first band gap in the anisotropic separable potential . _ bifurcations in finite - gap potentials . _ finite - gap potentials lead to the degeneracy of eigenvalues of the one - dimensional sturm liouville problem and non - zero derivatives of the functions @xmath57 at the extremal points @xmath60 and @xmath689 . adding a generic potential breaks the symmetry of the finite - gap potential and leads to the bifurcation of narrow band gaps in the space of one dimension . the corresponding coupled - mode equations must have first - order than second - order derivative operators . a formal derivation of such coupled - mode equations from the bloch mode decomposition was considered in @xcite . our analysis provides a rigorous justification of these coupled - mode equations . _ bifurcation in super - lattices . _ let the potential @xmath9 be represented in the form @xmath690 where @xmath23 and @xmath691 are the same as in the separable potential ( [ separable - potential ] ) and @xmath692 is a @xmath87-periodic potential . the perturbation term @xmath692 couples resonant bloch modes of the potential @xmath23 by linear terms and destroys reductions @xmath294 , @xmath681 and @xmath297 in the differential coupled - mode system . the system is still formulated in the form of the coupled nls equations with various ( linear and nonlinear ) coupling terms . _ bifurcations in three - dimensional periodic problems . _ the same analysis holds in three - dimensional separable potentials , since the spectral bands are still enumerated by a countable number of spectral bands of the one - dimensional potentials . bifurcations of new band gaps occur again due to a resonance of finitely many bloch modes of the one - dimensional potentials . _ bifurcations in non - separable periodic potentials . _ at the present time , the generalization of the analysis for non - separable periodic potentials meets a technical obstacle that the bound ( [ bound - k ] ) obtained in @xcite is needed to be extended to problems with two - dimensional periodic potentials . * acknowledgement . * the second author thanks a. sukhorukov for discussions at the early stage of the project . the work of t. dohnal and d. pelinovsky is supported by the humboldt research fellowship . the work of g. schneider is partially supported by the graduiertenkolleg 1294 `` analysis , simulation and design of nano - technological processes '' granted by the deutsche forschungsgemeinschaft ( dfg ) and the land baden - wrttemberg .
we address a two - dimensional nonlinear elliptic problem with a _ finite_-amplitude periodic potential . for a class of _ separable symmetric _ potentials , we study the bifurcation of the _ first band gap _ in the spectrum of the linear schrdinger operator and the relevant _ coupled - mode equations _ to describe this bifurcation . the coupled - mode equations are derived by the rigorous analysis based on the fourier bloch decomposition and the implicit function theorem in the space of bounded continuous functions vanishing at infinity . persistence of _ reversible localized _ solutions , called _ gap solitons _ , beyond the coupled - mode equations is proved under a non - degeneracy assumption on the kernel of the linearization operator . various branches of reversible localized solutions are classified numerically in the framework of the coupled - mode equations and convergence of the approximation error is verified . error estimates on the time - dependent solutions of the gross pitaevskii equation approximated by solutions of the coupled - mode equations are obtained for a finite - time interval .
in 4-dimensional general relativity , no - hair theorem @xcite guarantees that a neutral rotating astrophysical black hole is uniquely described by the kerr metric which only possesses two parameters , the mass @xmath4 and the rotational parameter @xmath5 . for the kerr black hole , the fundamental limit is the bound @xmath6 , and the central singularity is always behind the event horizon due to the weak cosmic censorship conjecture @xcite . however , the hypothesis that the astrophysical black - hole candidates are described by the kerr metrics still lacks the direct evidence , and the general relativity has been tested only for weak gravitational fields @xcite . in the regime of strong gravity , the general relativity could be broken down and astrophysical black holes might not be the kerr black holes as predicted by the no - hair theorem @xcite . several parametric deviations from the kerr metric have been suggested to study observational signatures in both the electromagnetic @xcite and gravitational - wave @xcite spectral that differ from the expected kerr signals . recently , johannsen and psaltis proposed a deformed kerr - like metric @xcite suitable for the strong field of the no - hair theorem , which describes a rotating black hole ( we named it the non - kerr black hole ) in an alternative theory of gravity beyond einstein s general relativity . the non - kerr black hole possesses the following novel features : there is no restriction on the value of the rotational parameter @xmath5 due to the existence of the deformation parameter @xmath0 . interestingly , for a positive parameter @xmath0 , the non - kerr black hole becomes more prolate than the kerr black hole and there are two disconnected horizons for high spin parameters , but there is no horizon when @xmath1 . therefore , for a negative parameter @xmath0 , the non - kerr black hole is more oblate than the kerr black hole , and the horizon always exists for an arbitrary @xmath5 and the topology of the horizon becomes toroidal @xcite . the non - kerr metric is an ideal spacetime to carry out strong - field tests of the no - hair theorem . therefore , a lot of effort has been dedicated to the study of the rotating non - kerr black hole recently @xcite . in ref . @xcite , we studied the properties of the thin accretion disk in the rotating non - kerr spacetime and found that the presence of the deformation parameter @xmath0 changes the inner border of the disk , energy flux , conversion efficiency , radiation temperature , spectral luminosity and spectral cut - off frequency of the thin accretion disk . moreover , for the rapidly rotating black hole , the effect of the deformation parameter @xmath0 on the physical quantities of the thin disk becomes more distinct for the prograde particles and more tiny for the retrograde ones . these significant features in the mass accretion process may provide a possibility to test gravity in the regime of the strong field in the astronomical observations . the power energy for a active galactic nuclei , x - ray binaries and quasars has always been concerned in the high energy astrophysics . several mechanisms ( i.e. the accretion disk model @xcite and blandford - znajek mechanism @xcite ) have been proposed to interpret how to extract energy from a black hole and the formation of the power jet . furthermore , the penrose process @xcite also provides an important method to extract energy from a black hole . the penrose process was also extended to the five - dimensional supergravity rotating black hole @xcite , higher dimensional black holes and black rings @xcite , and hoava - lifshitz gravity @xcite . in this paper , we will investigate in detail the ergosphere of the non - kerr black hole and how the deformation parameter @xmath0 of the non - kerr black hole affects the negative energy state and the efficiency of the energy extraction . the paper is organized as follows : in sec . ii , we review briefly the metric of the rotating non - kerr black hole proposed by johannsen and psaltis @xcite to test gravity in the regime of the strong field and then analyze the ergosphere structure . in sec . iii , we investigate the efficiency of the energy extraction by using the penrose process . iv is devoted to a brief summary . to test gravity in the regime of the strong field , johannsen and psaltis @xcite , starting from a deformed schwarzschild solution and applying the newman - janis transformation , constructed a deformed kerr - like metric which describes a stationary , axisymmetric , and asymptotically flat vacuum spacetime . in the standard boyer - lindquist coordinates , the metric can be expressed as @xcite @xmath7 with @xmath8,\end{aligned}\ ] ] where @xmath9 the constant @xmath0 is the deformation parameter . the quantity @xmath10 or @xmath11 corresponds to the cases in which the compact object is more prolate or oblate than the kerr black hole , respectively . as @xmath12 , the black hole is reduced to the usual kerr black hole in general relativity . the horizons of the black hole are described by the roots of the following equation @xcite @xmath13 clearly , the radii of the horizons depends on @xmath14 , which are different from that in the usual kerr case . for the case of @xmath10 , there exist two disconnected horizons for high spin parameters , but there is no horizon when @xmath1 . however , for @xmath11 the horizons never disappear for an arbitrary @xmath5 and the shape of the horizons becomes toroidal @xcite . the infinite redshift surface of a black hole is defined by the roots of @xmath15 . for the non - kerr black hole we note that @xmath15 gives @xmath16 obviously , for the case of @xmath17 , the outer infinite redshift surface is determined by @xmath18 . for @xmath11 , it seems that both of the positive root of @xmath19 and @xmath20 are the infinite redshift surfaces of the non - kerr black hole . however , the determinant of the metric @xmath21 vanishes if @xmath19 which shows that the surface defined by @xmath19 is an intrinsic singularity . the singularity is indicated by infinite curvature and can not been eliminated by coordinate transformation , and it s kretschmann scalar @xmath22 and @xmath23 becomes zero or infinity . in fig . [ htt ] , we depict the positive root of the equations @xmath19 , @xmath20 , and @xmath24 with different deformation parameter @xmath0 . the figure also shows that the surface defined by @xmath19 is the intrinsic singularity , because if we choose the surface defined by @xmath19 as the outer infinite redshift surface , we find that the metric of the non - kerr black hole does not satisfy @xmath25 , @xmath26 and @xmath27 in the ergosphere and the causality of the spacetime is violated . thus the surface defined by @xmath19 can not be the infinite redshift surface . of the rotating non - kerr black hole . the green , blue , and red lines correspond to the surfaces defined by @xmath19 , @xmath20 , and @xmath24 , respectively . the dashed line is the location of turning point of the surface defined by @xmath24 . here , we take @xmath28 . ] in fig . [ htt ] , there is an intersection point of the three surfaces and the value of the abscissa for the intersection point @xmath29 is @xmath30 . there also exists a turning point @xmath31 for the event horizon surface . the value of the abscissa for the turning point is located at the position @xmath32 , thus we get @xmath33 the maximum positive root of eq . ( [ soluti ] ) is @xmath34 using the eqs . ( [ soluti ] ) and ( [ solutione1 ] ) , we get the value of the abscissa for the turning point of the curve described by @xmath24 , which is @xmath35 the ergosphere is the region bounded by the event horizon @xmath36 and the outer stationary limit surface @xmath37 . a novel feature of the ergosphere of a black hole is that the timelike killing vector becomes spacelike crossing the infinite redshift surface . an observer moving along the timelike geodesics can not remain static but stationary in the ergosphere due to the frame - dragging effect " @xcite . according to the properties of the ergosphere , the deformation parameter @xmath0 should satisfy following relation @xmath38 in this range of @xmath0 , the non - kerr black hole has a event horizon given by largest root of @xmath24 , and a outer infinite redshift surface described by @xmath39 when @xmath40 or @xmath41 there is no the event horizon and the singularity becomes naked . how the deformation parameter @xmath0 affects the shape of the ergosphere is described by fig . [ ergosphere11 ] which shows that the ergosphere is sensitive to the deformation parameter @xmath0 . for @xmath42 ( such as @xmath43 ) , the non - kerr black hole becomes more prolate than the kerr black hole , and the ergosphere in the equatorial plane becomes thick as the parameter @xmath0 increases . it should be pointed out that , when @xmath0 exceed @xmath44 , the horizons become disconnected . for the case of @xmath45 , there exist the inner and outer horizons which coincide at the north and south poles , and the thickness of the ergosphere decreases when the deformation parameter @xmath0 takes a bigger negative value . for @xmath1 , @xmath0 can only take a negative value , and both the horizon and infinite redshift surface are not closed . a hole appears around the north and south poles in the range @xmath46 . thus a distant observer may see the central region of this compact object along the north or south pole . it is also interesting to note that the overspin compact object becomes more and more thin and looks like a disk @xcite as the deformation parameter @xmath0 increases . -plane of the ergosphere with the deformation parameter @xmath0 of the rotating non - kerr black hole . the red and the blue lines correspond to the infinite redshift surfaces and the horizons , respectively . here , we take @xmath28 . ] in this section , we will discuss the penrose process @xcite , by which we can extract a rotational energy from the non - kerr black hole . we focus our attention on how the deformation parameter @xmath0 of the non - kerr black hole affects the negative energy state and the efficiency of the energy extraction . let us now consider the trajectory of a test particle with the mass @xmath47 on the equatorial plane . with the help of the timelike killing vector @xmath48 and spacelike one @xmath49 , we have the following conserved quantities along a timelike geodesics on the equatorial plane @xmath50 where @xmath51 is the four - velocity defined by @xmath52 , @xmath53 is the proper time for the spacetime . in eq . ( [ eq : e ] ) or ( [ eq : l ] ) , the first equality is the basic definition of the energy or angular momentum @xcite , and the second equality describes the energy or angular momentum of the non - kerr black hole . moreover , we can introduce a new conserved parameter @xmath54 whose values are given by @xmath55 corresponding to the timelike , null and spacelike geodesics , respectively . from eqs . ( [ eq : e ] ) , ( [ eq : l ] ) and ( [ eq : phidot111 ] ) we can easily obtain the equation of motion @xmath56 with @xmath57 where @xmath58 from eq . ( [ hamjameqq ] ) , we can obtain the energy @xmath59 @xmath60 where we only choose @xmath61 to ensure that the 4-momentum of the particle is future directed . in the penrose process , the orbit of the particle with negative energy in the ergosphere is the key to extract energy from the non - kerr black hole . when a particle enters the ergosphere , the timelike killing vector becomes spacelike one , thus the energy of the particle @xmath62 becomes negative . the orbit of the particle with the negative energy @xmath59 must satisfy the conditions : @xmath63 , @xmath64 and @xmath65 , which can be achieved only if @xmath66 . allowed for the angular momentum @xmath67 and rest mass @xmath47 of the particles for the different deformation parameter @xmath0 at a given location near the event horizon inside the ergosphere.,title="fig:",width=302 ] allowed for the angular momentum @xmath67 and rest mass @xmath47 of the particles for the different deformation parameter @xmath0 at a given location near the event horizon inside the ergosphere.,title="fig:",width=302 ] in fig . [ negativeff ] , we describe the negative energy state @xmath59 for the different deformation parameter @xmath0 at a given location near the event horizon inside the ergosphere . we find that the negative energy @xmath59 increases as the deformation parameter @xmath0 increases for both the cases @xmath1 and @xmath42 . according to the penrose process , the mass of the black hole will change a quantity @xmath68 as a negative particle is injected into the central black hole . clearly @xmath69 can be made as large as we wished by increasing the mass @xmath47 of the injected particle . however , there is a lower limit on @xmath69 which could be added to the black hole corresponding to @xmath70 and @xmath71 @xcite . evaluating all of the required quantities at the horizon @xmath72 , we can get the lower limit @xmath73 from the eq . ( [ minenergy ] ) we can conclude that , in order to extract energy from the black hole , the angular momentum of the injected particle must satisfy @xmath74 , and the deformation parameter @xmath0 influences the value of @xmath75 . the efficiency of the energy extraction process is one of the most important questions in the energetics of the black hole . thus , it is interesting to study how the deformation parameter @xmath0 affects the efficiency of the penrose process @xcite for the non - kerr black hole . to calculate the maximum efficiency of the energy extraction , we take the radial velocity to be zero . let @xmath76 denotes the four - velocity of the @xmath77th particle of the locally nonrotating frame observer @xcite at a given radius @xmath78 , which can be expressed as @xmath79 with @xmath80 where @xmath81 is the angular velocity of the particle @xmath77 with respect to an asymptotic infinity observer . in the ergosphere , @xmath81 takes the value in the range of @xmath82 , where @xmath83 in the penrose process , an incident particle @xmath84 with the rest mass @xmath85 , i.e. @xmath86 , splits into the particle @xmath87 absorbed by the black hole and the particle @xmath88 escaping to infinity . according to the conservational laws of the energy and angular momentum , we get @xmath89 the efficiency of the penrose process is defined as @xmath90 the maximum efficiency @xmath91 can be obtained by the choice of @xmath92 and @xmath93 @xcite @xmath94 where @xmath95 and @xmath96 are constants to be determined . with the help of the eqs . ( [ four velocity1 ] ) , ( [ four velocity2 ] ) , ( [ kk3 ] ) and ( [ kk2 ] ) , we find @xmath97 when the incident particle @xmath84 splits at the horizon @xmath36 , we thus obtain the maximum efficiency @xmath98 of rotating non - kerr black hole . here we take @xmath28.,width=302 ] now , we would like to analyze the effects of @xmath0 on the efficiency of the rapidly rotating non - kerr black hole . we calculate the maximum efficiency with the numerical method and present the variation of the maximum efficiency in the energy extraction process with the deformation parameter @xmath0 , which takes the value in the range ( [ region2 ] ) to guarantee that the non - kerr black hole has the connected event horizon for a fixed @xmath5 in fig . [ effffffffffff ] . we also present the effects of the parameter @xmath0 on the maximum efficiency in table . [ table1 ] and [ table2 ] . .the maximum efficiency @xmath99 of energy extraction process in the non - kerr black hole depends on the parameter @xmath0 for @xmath6 . here we set m=1 . [ cols="^,^,^,^,^,^,^,^",options="header " , ] obviously , it is shown that the maximum efficiency of the penrose process can be enhanced as the parameter @xmath0 increases . it is interesting to note that , for @xmath1 , the non - kerr metric describes a superspinning black hole @xcite , the maximum efficiency can exceed @xmath2 , while it is only @xmath3 for the extremal kerr black hole . this result is reasonable , because in the sec . [ section2222 ] , we investigate the ergosphere and find that the maximum thickness of the ergosphere on the equatorial plane increases as the deformation parameter @xmath0 increases . especially , for @xmath100 and @xmath101 , the thickness of the ergosphere on the equatorial plane is much thicker than that of the extremal kerr black hole ( @xmath102 , @xmath12 ) in fig . [ ergosphere11 ] . if the values of @xmath5 and @xmath0 change gradually , by accretion or any other process , from @xmath42 to @xmath1 and from @xmath10 to @xmath11 we should expect a continuous change of the energy extraction intuitively . why is it not so ? from the eq . ( [ eq : efficiencymax ] ) , we find that the maximum efficiency ( @xmath103 ) is related to the event horizon @xmath36 of the non - kerr black hole . in fig . 5 , by taking @xmath104 and considering the values of @xmath5 changes gradually from @xmath102 to @xmath105 , we find the variation of the event horizon @xmath36 is not continuous . therefore , we can not obtain a continuous change of the energy extraction for the process by taking @xmath5 from @xmath42 to @xmath1 ( or from @xmath10 to @xmath11 ) . of a rotating non - kerr black hole with @xmath28 and @xmath104 .,width=302 ] in this paper , we present a detail analysis of the properties of the ergosphere in the rotating non - kerr black hole proposed recently by johannsen and psaltis @xcite to test gravity in the regime of the strong field in the future astronomical observations . we now summarize our results as follows : ( 1 ) we present the restricted conditions on the deformation parameter @xmath0 to guarantee that the non - kerr black hole has the connected horizons ( see eq . [ region2 ] and fig . [ htt ] ) . ( 2 ) we show that the ergosphere is sensitive to the deformation parameter @xmath0 ( see fig . [ ergosphere11 ] ) and the shape of the ergosphere becomes thick with increase of the deformation parameter @xmath0 . ( 3 ) we find that , comparing with the kerr black hole , the deformation parameter @xmath0 not only enlarges the negative energy @xmath59 ( see fig . [ negativeff ] ) but also enhances the maximum efficiency of the energy extraction process ( see tables [ table1]-[table2 ] and fig . [ effffffffffff ] ) . moreover , the maximum efficiency can exceed @xmath2 for the non - kerr compact objects with @xmath1 . the influence of the deformation parameter @xmath0 on the maximum efficiency presents a good theoretical opportunity to distinguish the non - kerr black hole from the kerr one and to test whether or not the current black - hole candidates are the black holes predicted by einstein s general relativity . however , we think such a test is not possible at present . this work was supported by the national natural science foundation of china under grant no . 11175065 , 10935013 ; the national basic research of china under grant no . 2010cb833004 ; the srfdp under grant no . 20114306110003 ; pcsirt , no . irt0964 ; the hunan provincial natural science foundation of china under grant no 11jj7001 ; hunan provincial innovation foundation for postgraduate under grant no cx2011b185 ; and construct program of the national key discipline . w. israel , phys . rev . * 164 * ( 1967 ) 1776 ; w. israel , commun . * 8 * ( 1968 ) 245 ; b. carter , phys . * 26 * ( 1971 ) 331 ; s. w. hawking , commun . * 25 * ( 1972 ) 152 ; d. c. robinson , phys . * 34 * ( 1975 ) 905 . s. a. hughes , arxiv:1002.2591 ; d. psaltis and t. johannsen , j. phys . conf . ser . * 283*,(2011 ) 2030 ; t. johannsen and d. psaltis , astrophys . j. * 716 * , ( 2010 ) 187 ; t. johannsen and d. psaltis , astrophys . j. * 718 * , ( 2010 ) 446 ; t. johannsen and d. psaltis , astrophys . j. * 726 * , ( 2011 ) 11 . f. d. ryan , phys . d * 52 * , ( 1995 ) 5707 ; f. d. ryan , phys . rev . d * 56 * , ( 1997)1845 ; f. d. ryan , phys . d * 56*,(1997 ) 7732 ; l. barack and c. cutler , phys . rev . d * 69*,(2004 ) 082005 ; j. brink , phys . d * 78 * , ( 2008)102001 ; c. li and g. lovelace , phys . d * 77*,(2008 ) 064022 ; t. a. apostolatos , g. lukes - gerakopoulos , and g. contopoulos , phys lett . * 103 * , ( 2009)111101 .
= 0.6 cm the properties of the ergosphere and energy extraction by penrose process in a rotating non - kerr black hole are investigated . it is shown that the ergosphere is sensitive to the deformation parameter @xmath0 and the shape of the ergosphere becomes thick with increase of the parameter @xmath0 . it is of interest to note that , comparing with the kerr black hole , the deformation parameter @xmath0 can enhance the maximum efficiency of the energy extraction process greatly . especially , for the case of @xmath1 , the non - kerr metric describes a superspinning compact object and the maximum efficiency can exceed @xmath2 , while it is only @xmath3 for the extremal kerr black hole . = 0.8 cm
the luminosity function ( lf ) , number of galaxies per unit volume per unit luminosity bin , is an important indicator of the distribution of galaxies over cosmological time . in particular , infrared ( ir ) lfs provide useful information regarding the amount of energy released by galaxy activities . we can obtain , however , different information depending on the wavelength at which the lf is measured . for example , nir luminosity function should be regarded as a proxy of stellar mass function of galaxies , while various emission features in the mir provide clues to detailed dust properties related to the star - forming activities . dust emission peaks in the far - ir ( fir ) wavelengths and fir luminosities give insight to total infrared luminosity ( @xmath11 ) which is directly connected to overall bulk of energy emitted by star - formation activities . the construction of the infrared luminosity functions has begun based on the data obtained by _ infrared astronomical satellite _ ( iras ) in the mid- and far - infrared , although the sample was restricted to bright galaxies because of relatively low sensitivity of the iras ( rowan - robinson , helou and walker 1987 ; saunders et al . 1990 ; rowan - robinson et al . subsequently , _ infrared space observatory _ ( iso ) data was used to determine the mid - infrared ( 12 @xmath4 m and 15@xmath4 m ) lfs with follow - up imaging and spectroscopic observations ( clements et al . 2001 ; xu et al . 2000 ; pozzi et al . far - infrared lf was also estimated ( e.g. , 90 @xmath4 m lf by serjeant et al . 2004 ) using the european large - area iso survey ( elais ) final ( optical - ir merged ) catalogue ( rowan - robinson et al . 2004 ) . based on the iras ( hacking et al . 1987 ; franceschini et al . 1988 ; fang et al . 1998 etc . ) and iso data ( elbaz et al . 1999 ; puget et al . 1999 ) , it has been revealed that dust - enshrouded star - forming galaxies undergo evolution in luminosity and density up to redshift around @xmath12 , and that their evolutionary rate may exceed those measured at any other wavelengths . the contribution of infrared luminous galaxies ( such as lirgs and ulirgs defined as @xmath13 and @xmath14 , respectively ) to the star formation ( sf ) history has been recognized and well established recently , implying the star forming activity was much more intense at higher redshifts ( @xmath2 ) than nearby universe . more accurate infrared luminosity functions ( irlfs ) have been derived for both local and distant galaxies observed from various fields based on the _ akari _ and _ spitzer _ data by many authors ( prez - gonzlez et al . 2005 ; babbedge et al . 2006 ; caputi et al . 2007 ; goto et al . 2010 ; rodighiero et al . 2010 ; goto et al . 2011 ; wu et al . 2011 ) . they have found a positive evolution in luminosity and density and increasing importance of the lirgs and ulirgs populations toward higher redshifts , indicating that the universe was much more active in the past . recent interests have moved to even higher redshift beyond the peak ( @xmath15 ) of the cosmic sfr density ( e.g. , prez - gonzlez et al . 2005 ; rodighiero et al . 2010 ; magnelli et al . 2011 ) to understand the early history of the star formation . however , we still have many questions about the local universe and uncertainties in the parameters that can relate local galaxy formation and evolution with the global history of the universe . especially , accurate measurements of the lfs in the local universe is important in order to determine the course of evolution over cosmic time . _ akari _ , an ir mission by jaxa / isas in japan with european and korean collaborations , has carried out all sky surveys at mid- and far - infrared ( murakami et al . _ akari _ also carried out a number of wide area surveys with better sensitivities than all - sky survey , including the north ecliptic pole ( nep ) surveys which used continuous nine spectral bands ranging from 2 to 24 @xmath4 m . the nep survey was composed of two programs : wide and deep surveys . the nep - deep survey covered about 0.4 deg@xmath1 field with higher sensitivity ( matsuhara et al . 2006 ; wada et al . 2008 ; takagi et al . 2012 ) , while the nep - wide survey covered much wider area of about 5.4 deg@xmath1 with lower sensitivity . the nep - wide survey is about 0.50.6 mag shallower than nep - deep survey in the nir and mir - s bands . the scientific purpose and overall plan of the nep survey projects were presented by matsuhara et al . ( 2006 ) , and detailed description on the nep - wide and nep - deep surveys including their ir source catalogues have been published by kim et al . ( 2012 ) and takagi et al ( 2012 ) , respectively . based on the data obtained by nep - deep survey , goto et al . ( 2010 , hereafter referred to as g10 ) have derived the mir ( 8 and 12 @xmath4 m ) luminosity functions for galaxies at redshift from 0.3 to 2 . they used @xmath3 4,000 ir sources and estimated photometric redshifts to determine the distances to galaxies based on sed model fitting . recently , they updated their lfs using newly obtained cfht data ( goto et al . 2015 ) . in this work , we use the nep - wide survey data which has more galaxy sample in the local universe , where g10 did not cover . the primary purpose of this work is to present more accurate luminosity functions for the redshift @xmath0 using only spectroscopic sample in the nep - wide field . therefore , our study complements to g10 in the following aspects . first , we use the nep - wide survey data covering the wider angular coverage ( 5.4 deg@xmath1 ) that allows us to minimize the cosmic variance . second , since the photometric depth of the nep - wide survey is shallower , we concentrate on the derivation of the luminosity function for the local universe within @xmath16 , whereas g10 measured the 8@xmath4 m and 12@xmath4 m lfs for the redshift range @xmath17 . finally , we take advantage of the spectroscopic redshifts ( spec - z ) for galaxies in the nep - wide field obtained by the follow - up spectroscopic surveys ( shim et al . 2013 ) with mmt / hectospec and wiyn / hydra . thus , our estimation of the distances to the galaxies would be accurate . this paper is organized as follows . in section 2 , we present ir photometric data from _ akari _ nep - wide survey and spectroscopic redshifts ( spec-@xmath2 ) data from optical survey with mmt / hectospec and wiyn / hydra used in this study . in section 3 , we describe our methodology to build the lfs . we present our results in section 4 , and conclusions and summary in section 5 . throughout this paper , we assume a @xmath18 cold dark matter ( @xmath18cdm ) cosmology with @xmath19 70 km s@xmath20 mpc@xmath20 , @xmath21 0.3 , @xmath22 0.7 . the nep - wide survey ( matsuhara et al . 2006 ; lee et al . 2009 ; kim et al . 2012 ) is one of the large area survey programs of _ akari _ space telescope ( murakami et al . 2007 ) . the imaging survey toward the north ecliptic pole ( nep ) was carried out on the circular area of about 5.4 deg@xmath1 with nine photometric bands of the infrared camera ( irc , onaka et al . 2007 ) , covering the spectral range of 2 24 @xmath4 m nearly continuously . the photometric bands are designated as @xmath23 , @xmath24 , and @xmath25 for the nir , @xmath26 , @xmath27 , and @xmath28 for the mir - s ( shorter part of the mir ) bands , and @xmath29 , @xmath30 , and @xmath31 for the mir - l ( longer part of the mir ) bands : the numbers indicate approximate effective wavelength in units of @xmath4 m . the detailed description of the data reduction methodologies , catalogue of point sources and their characteristics are presented by kim et al . ( 2012 ) . here we only make brief summary of the nep - wide data sets . all magnitudes of ir sources in this work are reported in units of ab , hereafter . the entire region surveyed by the nep - wide program centered at the nep ( @xmath32 , @xmath33 ) can be found in figure 1 ( left panel ) . background gray tiles in this figure represent 446 pointed observations carried out using irc instruments to cover the whole nep field . the 5-@xmath34 detection limits of nep - wide survey are approximately 21 magnitude ( mag ) in the nir bands , 19.5 19 mag in the mir - s bands , and 18.8 18.5 mag in the mir - l bands in ab mag , respectively . the nep - wide point source catalogue contains more than @xmath3 100,000 sources in the nir and up to @xmath3 18,000 sources in the mir bands . compared to the detection limits , 50% completeness levels are 0.5 0.6 mag brighter in the mir and @xmath3 1.2 mag brighter in the near - ir bands . the depth of the nir bands are affected by source confusion while those of the mir bands are less affected by confusion because of the lower source density . the number of sources in the mir bands typically ranges from a few thousands to about eighteen thousands . one of the advantages of using the nep - wide data is that we have many ancillary data sets such as high - quality optical data obtained with the megacam of 3.5 m cfht ( hwang et al . 2007 ) and the 4k @xmath35 4k snucam ( i m et al . 2010 ) of the 1.5 m telescope at maidanak observatory in uzbekistan ( jeon et al . in addition , near - ir @xmath36 , @xmath6 band data have been obtained ( jeon et al . 2014 ) using the florida multi - object imaging near - ir grism observational spectrometer ( flamingos ; elston et al . 2006 ) mounted at the kitt peak national observatory ( kpno ) 2.1 m telescope , and various radio data over a limited region ( kollgaard et al . 1994 , lacy et al . 1995 , sedgwick et al . 2009 , white et al . recently , the _ herschel space observatory _ ( pilbratt et al . , 2010 ) covered the entire _ akari _ nep - wide field with the spectral and photometric imaging receiver ( spire ; griffin et al . 2010 ) . for the ir sources detected by _ akari _ nep - wide survey , all the photometry results of irc bands as well as ancillary data were produced in the form of ( optical to mir ) band - merged catalogue covering @xmath5 band to 24 @xmath4 m band ( kim et al . 2012 ) . to construct a reliable source catalogue , all of the detected sources at one _ akari_/irc band were cross - matched with those in the other irc bands and all the ancillary data sets including optical ( cfht and maidanak ) and flamingos near - ir bands ( @xmath37 and @xmath38 ) . the catalog contains about 114,800 sources detected at least in one of the irc filter bands . no attempt to distinguish the sources into different types such as stars , galaxies has been made yet systematically for the entire sources in this nep - wide catalogue . however , we inspected sub - sample of nep sources based on various colour - colour diagrams with the stellarity parameters of optical counterparts , and assessed that about 70% of the mir sources ( with @xmath28 magnitude brighter than 18.5 mag ) are star - forming galaxies mostly at redshift range z @xmath39 1 ( see lee et al . 2007 ; lee et al . 2009 ; kim et al . 2012 ) . the active galactic nuclei ( agns ) and early - type galaxies comprise approximately 10% or more among the sample we inspected . the rest of them are most likely galactic stars . we expect that the composition of entire mir sources in the nep - wide catalogue should be similar . follow - up spectroscopic surveys over the entire _ akari _ nep - wide field were carried out with mmt / hectospec and wiyn / hydra ( shim et al . 2013 ) . the targets for spectroscopic follow - up observation were selected primarily based on mir fluxes at 11 @xmath4 m ( @xmath39 18.5 mag , or @xmath40 150 @xmath4jy ) and at 15 @xmath4 m ( @xmath39 17.9 mag , or @xmath41 250 @xmath4jy ) . an additional @xmath42-band magnitude cut was imposed to select optically bright objects ( @xmath43 for hydra , @xmath44 for hectospec observations ) that can yield the spectra with reasonable signal - to - noise ( s / n ) ratio . most of these flux - limited targets are considered to be various types of ir luminous star - forming galaxies . in addition to these primary targets , a smaller number of secondary targets were selected using their optical colours ( @xmath45 , @xmath46 , etc . ) and irc band colours ( @xmath47 , @xmath48 ) . these secondary criteria include high - redshift galaxy candidates based on dropout techniques ( @xmath49-dropout for @xmath50 3 and @xmath51 or @xmath52dropouts for z @xmath3 4 ) , active galactic nuclei ( agn ) candidates selected by the nir ( @xmath53 ) and mir ( @xmath54 ) colours reflecting power - law sed of an agn ( e.g. , lee et al . 2007 ) , radio sources ( white et al . 2010 ) , polycyclic aromatic hydrocarbon ( pah)-luminous galaxies ( ohyama et al.,2009 ; takagi et al . , 2010 ) , and so on . the supercluster member candidates in the nep field at z @xmath55 0.087 ( ko et al . 2012 ) were included in the targets for this spectroscopic observation . majority of the secondary targets are not used in the analysis for the lf derivation because they are cluster members or at @xmath17 . shim et al . ( 2013 ) presented spectra of 1796 sources ( primary targets : 1155 , secondary targets : 641 ) , among which spectroscopic redshifts are measured for 1645 objects . the success rate for redshifts identification are higher than 80% at r @xmath39 21 mag , though the quality of the spectra varies significantly depending on the targets in different fields . quality flag ranging from 1 to 4 was assigned to each source : 4 for a secure redshift , 3 for an acceptable and almost good redshift , 2 for a questionable , and 1 for unusable redshift . we did not use the sources having the flag 1 or 2 ( @xmath3 190 ) . the spectroscopic redshifts of @xmath3 90% of the sources are distributed over the redshift range of z @xmath39 1.0 ( 120 sources out of these are galactic stars ) , and the remaining 10% lie at z @xmath56 1.0 . due to the target selection seeking for high - redshift agns , most sources at z @xmath56 1.0 are type-1 agns . details of the line flux measurements and diagnostic line ratios can be found in shim et al . ( 2013 ) . the spatial distribution of the nep - wide sources observed by this spectroscopic survey is presented in the left panel of figure 1 along with the nep - wide survey area . the blue diamonds indicate the spectroscopic targets observed with mmt / hectospec , and the red triangles show the targets observed with wiyn / hydra . we also show the redshift distribution of this spectroscopic sample in the right panel of figure 1 . all the spectroscopic sources are shown in black colour . the blue line indicates the distribution of sample observed with mmt / hectospec , and the red one represents that observed with wiyn / hydra . also shown with a yellow curve is the shape of @xmath57 , where @xmath58 is the comoving volume within redshift @xmath2 . it is clear that our spectroscopic sample seriously underrepresents the galaxies beyond @xmath59 , but is consistent with constant spatial density below @xmath60 . we used the sample at @xmath0 , where the spectroscopic incompleteness is less severe . the redshift distribution of the sources with higher redshifts ( @xmath61 ) is shown in an inner small box . in figure 2 , we present the ir properties of the parent photometric sources and spectroscopic sample used in this work . using analysis of emission line fluxes , shim et al . ( 2013 ) identified the types of spectroscopic sources . they presented baldwin - phillips - terlevich ( bpt ) diagram for the sources using emission lines ratios such as [ o iii]/h@xmath62 and [ nii]/h@xmath63 , and classified them into star - forming galaxies , starburst - agn composites , and agns . the sources classified as ` galaxy ' are indicated by crosses ( blue colour ) while those classified as ` agn ' types are indicated by small boxes ( squares for type-1 and diamonds for type-2 ) , whose types are identified based on the analysis of emission lines ( shim et al . 2013 ) . in the upper left panel of figure 2 , we give the distribution of the optical ( @xmath42 ) versus mir ( @xmath28 ) band magnitudes of the nep - wide sources , together with the primary criteria from @xmath28 band and additional @xmath42-band magnitude cut for the spectroscopy . the background gray dots indicate parent photometric sources from the nep - wide catalogue . upper right panel shows the distribution of mir - band colour @xmath64 as a function of redshift , showing most of the galaxy sample gathered between -0.5 and 0.5 in this colour range with the redshift @xmath65 , compared to the wider spread of the agn types . in the lower left panel , the colour - colour diagram of the _ akari _ nir and mir bands shows that the sources classified as agns are clearly distinguished by the relatively red nir colour ( @xmath66 ) and narrow range of mir colour ( @xmath67 ) . the sources classified as ` galaxy ' are mostly distributed in the range of @xmath68 . the lower right panel shows the @xmath66 colour as a function of redshift . figure 2 shows that mir colours are efficient tool which can be used to demarcate between galaxies and agns . especially , the ` galaxies ' types ( which we use in this work ) are gathered at lower ( mostly @xmath65 ) redshifts and in a narrow range of nir colour ( @xmath69 ) , while having a wider distribution in the mir colour domains ( @xmath70 and @xmath64 ) suggesting the various kinds of spectral energy distributions ( seds ) in the mir wavelengths . also , there exist intriguing sources such as those classified as ` galaxy ' but having red nir colour ( e.g. , @xmath53 ) , which we should investigate further later when we classify entire sources in the nep - wide catalogue . in figure 3 , the number distributions of the entire photometric ( dashed ) and spectroscopic sample ( solid ) are shown in the bottom panels while the ratio of the spectroscopic sample is shown in the upper panels , as a function of observed mir bands ( @xmath27 , @xmath28 , and @xmath29 ) and optical @xmath42-band ( the rightmost panel ) magnitudes . the fraction of spectroscopic sample varies over the mir magnitudes , and shows broad peaks at around 16 ( 15.5 16.5 ) magnitudes ( ab ) . since the number of sources in optical band is much larger than those in the mir bands , the fraction at @xmath42-band is much lower than in the mir bands . in the lower right panel , a dotted line represents the distribution of r - band sources except for the star - like sources defined by the criteria based on the optical mag and nir colours ( jeon et al . 2010 ; kim et al . 2012 ; jeon et al . 2014 ) . for the spectroscopic sample , the variation from the darkest to brighter shades show relative amount of sample at the redshift bin @xmath71 , @xmath72 , @xmath73 , and so on . vertical dot - dashed lines indicate the magnitude cuts used for the spectroscopic target selection . there are many sources fainter than this criteria due to the deeper limit of the hectospec . the incompleteness is more significant at higher redshift bin . our limitation of the redshift range ( @xmath0 ) effectively minimize the incompleteness arising from the flux limited sample by the magnitude cuts . galaxy emission in the mir is the sum of the rayleigh - jeans tail of stellar emission , emission from heated dust , and a power - law component of agn , if it exists . since our interest is to investigate the star - forming activity and to construct their lfs , we have to know the type of ir sources and classify them into galaxies or agns . shim et al . ( 2013 ) identified the types of spectroscopic sources and classified them into star - forming galaxies , starburst - agn composites , and agns based on the line analysis . for this work , we adopted sample classified as ` galaxy ' having reliable spectroscopic redshift . after excluding agns and unreliable spectroscopic redshift , we have 1090 galaxies . by limiting the redshift range of galaxies to @xmath0 to construct the local luminosity functions , we are left with about @xmath3 600 galaxies . to estimate mir luminosities for our sample with the spectroscopic-@xmath2 , we carried out sed fit analysis . we utilized the publicly available code _ ( ilbert et al . 2006 ) . the fitting to sed models was done over all the available photometric band data with the spectroscopic redshift ( by an option zfix@xmath55yes ) . we selected an option for the calzetti extinction law ( calzetti , 2001 ; fischera et al . 2003 ) in the software ( [email protected] ) . details of the software are given in ilbert et al . ( 2006 ) . for each observed sed , _ le phare _ carried out single fit from the @xmath5 to the mid - ir band , using various templates defined throughout this wavelength range ( polletta et al . 2007 ; ilbert et al . we searched for a best - fit model among all types and chose a template giving a minimum @xmath74 . in figure 4 , we present some of our local ( @xmath0 ) galaxy sample ( e.g. , id@xmath55 73741 , 53756 , 93654 , 28848 , 40162 , 34840 , 38402 , 4552 , 13077 , 70827 , and 61376 ) fitted to various templates . the small squares indicate available photometric data points from @xmath5 , @xmath75 , @xmath76 , @xmath77 , @xmath78 ( cfht : diamonds ) , @xmath79 , @xmath42 , @xmath80 ( maidanak : triangles ) , @xmath37 , @xmath38 ( kpno : squares ) , @xmath23 , @xmath24 , @xmath25 ( _ akari _ nir bands : asterisks ) , @xmath26 , @xmath27 , @xmath28 ( mir - s : filled circle ) , and @xmath29 , @xmath30 , @xmath31 ( mir - l : x - symbols ) . the first row shows galaxies that are fitted to ellipticals ( ell5 and ell13 ) , the second row shows sample fitted to spiral galaxies ( sb , sc , and sd types ) , and the third and bottom rows show those fitted to various starburst galaxies , ulirgs , ( e.g. , m82 , n6090 , and n6240 type ) and agn ( e.g. , seyfert type 2 ) . some sample fits better to agn template which gave a smaller @xmath74 than that of a best - fit galaxy template , although we used objects classified as various types of ` galaxy ' based on the optical line analysis . in this case , both template showed similar acceptable fit over the entire data points ( see the last sample presented in figure 4 ) . considering rather complex mir features and larger photometric uncertainties in the longest part of mir bands , slight confusion between galaxies and agns for some sample seems inevitable . because the seds from both types are very similar in the mir , the sources fitted better to agn types do not make significant difference in our resulting mir lfs . the k - correction ( oke @xmath81 sandage , 1968 ) , which transforms the measured flux at a given wavelength to the unredshifted flux at that band , is also computed in this procedure . we checked the amount of k - corrections for the mir bands where we derive the lfs , assuming simple box function centered at 8 and 12 @xmath4 m having full - width of 2 @xmath4 m and using l15 band response function . k - correction , depending on the sed of galaxy , its redshift ( @xmath2 ) , and the filter response ( @xmath82 ) , can be computed by @xmath83 figure 5 shows the k - corrections at 8 @xmath4 m , 12 @xmath4 m , and 15 @xmath4 m bands for a few representative sed models from the polleta et al . these figures show substantial differences in the k - correction behavior for different types of galaxies because the seds in mir portion vary significantly depending on galaxy types . for example , the 8 @xmath4 m bands show sensitive dependencies , because there are pah features at 6.2 @xmath4 m , 7.7 @xmath4 m , and 8.6 @xmath4 m and some of them can be included or excluded in this band depending on the redshift . in particular , 8.6 @xmath4 m feature is located near the boundary of the 8 @xmath4 m band , and thus can be excluded for the galaxies at low but non - zero redshift . from z @xmath3 0.11 , the 7.7 @xmath4 m pah feature begins to move out of this band . the k - correction is small but not negligible in the mir bands for nearby galaxies . in the case of early - type galaxies , k - correction decreases monotonically with redshifts since the flux decreases with wavelength near these bands . our sample used for this work is located at @xmath0 and mostly appears to be normal spiral / sf galaxies ( strongly dominated by late - types ) , explainable with m51 type ( cyan dot - dashed lines in figure 5 ) , while actively star forming populations such as arp 220 ( or lirg / ulirgs ) or agn types appear to take a very small faction . luminosity function @xmath84 is the number density of galaxies having luminosity in the range @xmath85 and residing in a specific volume . therefore , we should take a volume limited sample and count all the galaxies in a given luminosity bin and divide it by the surveyed volume . however , it is not as simple since the sample is usually limited by flux rather than volume . it means that the survey volume depends on the brightness or luminosity of the objects in the observation . in order to account for this simple fact , @xmath7 method ( as described by schmidt , 1968 ) is widely used to obtain the lfs of flux limited sample . this method is known to be relatively insensitive to the incompleteness of the observations , and there is no parametric dependence or assumed model . one can calculate luminosity ( @xmath86 ) in the rest frame or absolute magnitude ( @xmath87 ) based on a redshift and observed sed ( observed magnitudes @xmath88 at each band ) of each source , by following formula , @xmath89 where @xmath90 is the luminosity distance ( in units of mpc ) and k is @xmath91-correction . using this equation , we can compute @xmath92 , which is the maximum redshift at which a source could be observed with the detection limits of the survey ( at which @xmath93 detection limit ) . when converting the observed flux ( or magnitude ) in observed wavelength to luminosity of the rest - frame wavelength , we need to apply @xmath91-correction , which is calculated using the best - fit template for each source ( sec . a comoving volume @xmath94 associated with a source is the maximum volume corresponding to the maximum redshift within which the source could be still detected , where @xmath95 is the lower limit of the redshift bin . if @xmath92 is larger than the upper limit of the interested redshift bin ( i.e. , @xmath96 ) , we take smaller one , i.e. , @xmath92 = min ( @xmath97 of the z - bin , @xmath92 of a source ) . we estimated @xmath92 and @xmath98 at each band where we derive lfs . we collected the sources having same luminosity ranges ( i.e. , finite bin size , @xmath99 ) . after assigning @xmath98 to each source , the lf(@xmath100 ) can be obtained by @xmath101 where @xmath84 is the number of objects per mpc@xmath102 in the rest frame luminosity range @xmath103 ) , @xmath104 is correction factor to compensate the selection bias or incompleteness for @xmath105th galaxy , which will be described in detail in the following section 3.2 . since our sample is not a complete set and has photometric uncertainties , we should take these facts into account in deriving the lfs . first , the number of galaxies with spectroscopic redshifts are much smaller than that of parent photometric sources in the nep - wide field , as shown in figure 3 . we compared the number distribution of the spectroscopic sample with that of entire s11 sources from the nep - wide catalogue as a function of magnitude to obtain the selection function for spectroscopic sample because the targets for the spectroscopic observations are selected based mainly on the s11 magnitude . in figure 6 , the orange solid line indicates all the s11 sources from the _ akari _ nep - wide survey . the gray lines show the number distribution of the sources from spectroscopic observation : from top to the bottom , all the spectroscopic sample , sample with reliable redshifts , and those classified as galaxy . the solid brown line shows the sources classified as galaxies , which are used for the estimation of lfs . the uncertainty associated with the correction for the number of the spec - z sample can be explained by the poisson statistics in the number of the parent photometric and spectroscopic sample . it is normally a few % level at @xmath106 mag ( ab ) , but reaches around @xmath107 at @xmath108 mag due to the smaller number of sample . this correction uncertainty appears to affect error - bar sizes of lf by several % on average ( @xmath107 at most ) . second , the parent photometric data becomes partially incomplete near the detection limit . at the mir - s ( s7 s11 ) bands , the detection limit ranges 19.5 19 mag and the 50% completeness levels are @xmath3 0.6 mag brighter than the detection limits . the magnitude difference between the 90 @xmath109 and 10 @xmath109 completeness is about 0.7 0.8 mag , indicating that the completeness begins to drop quite rapidly from around 90 @xmath109 level ( 18.3 17.9 mag ) . we should also be careful at the bright end because the low sampling rate could derive unnatural selection function . in figure 6 , the green dotted line with the right - side vertical axis indicates the completeness level at s11 band obtained by simulation of artificial source injection and re - extraction tests as described by kim et al . a vertical dot - dashed gray line roughly corresponds to the primary selection criteria for the spectroscopic observation ( 50@xmath109 completeness limit of the nep - wide survey ) . we applied the correction for this photometric incompleteness based on the completeness estimation of the @xmath28 data that was obtained by kim et al . ( 2012 ) in our derivation of the lfs . there is also redshift bias as shown in the right - hand panel of figure 1 . we compared the number density of spectroscopic sources in the comoving volume corresponding to each redshift bin in order to estimate the redshift incompleteness . assuming constant number density , we estimated the correction for incompleteness as a function of @xmath2 . since our derivation of lfs is limited to @xmath0 where the source distribution is roughly consistent with the constant spatial density , we regard the redshift bias is not significant . rest frame luminosity is directly derived from sed - fit results , therefore , error estimation for luminosity function ( lf ) depends on the sed - fitting errors arising from redshift , @xmath110-correction error , and photometric uncertainties . since we used the redshifts measured by spectroscopic observation , we can ignore uncertainties from the redshifts ( which is very small ) . thus , our lf errors can be determined based mostly on the photometric error . in order to estimate the errors of the lfs , we carried out monte carlo simulation using random number generator . we generated more than 10,000 simulated catalogues by putting random errors into the measured flux of each source . in the simulated catalogues , those errors are realized by assuming a normal gaussian distribution centered at the observed flux . standard deviation of this gaussian distribution for simulated errors is determined based on photometry measurement error . this means we generated @xmath3 10,000 input catalogues for le phare runs doing sed - fit and obtained result sets , which finally results in slightly different 10,000 lfs . we determined final luminosity function and its uncertainties based on weighted average and the dispersion of the simulated results . it is known that 8 @xmath4 m luminosity of galaxy is well correlated with the total ir luminosity ( babbedge et al . 2006 ; caputi et al . 2007 ; bavouzet et al . 2008 ; goto et al . 2010 ; galametz et al . 2013 ) because the rest - frame 8@xmath4 m fluxes are dominated by prominent pah features which are sensitive to star formation activity . in this section , we present 8 @xmath4 m luminosity function of local galaxies in the nep - wide field . to estimate 8@xmath4 m fluxes at observed frame , we used a simple boxy filter centered at 8@xmath4 m with a 2@xmath4 m bandwidth across the filter ( constant transmission over the bandwidth ) and convolved the filter response with best - fit template for each galaxy . the observed flux and the rest - frame luminosity are related by following equation @xmath111 where @xmath112 is the luminosity distance for a given redshift based on the cosmological parameters assumed in this work , @xmath113 the flux density ( erg @xmath114 hz@xmath20 ) , and @xmath115 and @xmath116 are the observed and rest - frame frequencies , respectively , related by @xmath117 . m luminosity functions ( lfs ) for local sf galaxies . the filled circle ( magenta colour ) represents the lf of local ( @xmath0 ) galaxy sample in the nep - wide survey field ( this work ) . shaded area indicates the lf of rodighiero et al . ( 2010 ) using vvds - swire data . meshed region shows the results from babbedge et al . ( 2006 ) based on the swire data . triangles are from huang et al . ( 2007 , noao deep ) , and ` x ' symbols from the g10 . [ fig 8 ] ] in figure 7 , we show the distribution of the sources used to derive 8 @xmath4 m luminosity function . the top panel ( figure 7a ) shows the number of sources versus redshift . the width of the redshift bin is 0.1 , as indicated by vertical gray broken lines . from all the spectroscopic sample to those classified as galaxies , the amount of sample is shown by solid gray line ( all spec - z sample ) , broken lines ( except for stars and reliable spec - z ) , and the brown solid line ( galaxies ) . we give the number of galaxies on histogram bar chart for each redshift bin . the bottom left panel ( figure 7b ) shows the 8 @xmath4 m luminosity distribution of sample as a function of redshift . the squares represent all the spectroscopic sample , but for @xmath118 , the sample in each redshift bin over the detection limit is over - plotted with diamonds , triangles , and squares , respectively . the dark red dots represent those classified as agns , and a dashed curve indicates the luminosity corresponding to the 8 @xmath4 m flux limit as a function of redshifts . the estimation of the flux limit depends on sed model , but in the range @xmath0 , different models do not make significant changes ( @xmath3 a few % ) . we estimated the 8 @xmath4 m flux limit based on a simple flat sed . for the nep - wide survey , the detection limits in the mid - ir bands of irc increases almost linearly as a function of a filter s central wavelength . therefore we derived detection limit at 8 @xmath4 m through the interpolation . our sample is brighter than the detection limit of s11 band , but it is not always in 8 @xmath4 m band . since 8 @xmath4 m flux and luminosities are estimated from the sed - fit results , some sample can have 8 @xmath4 m luminosities lower than the detection limit curve . in the bottom right panel ( figure 7c ) , the number distribution of these sources are shown as a function of 8@xmath4 m luminosity for each redshift bin using the same colour as used in the left panel . we constructed the 8 @xmath4 m luminosity function for star - forming galaxies ( at @xmath119 ) using the sample presented in the figure 7c . in the redshift bin of @xmath120 , a vertical clustering ( figure 7b ) at around z @xmath3 0.09 is due to the presence of a supercluster at z @xmath55 0.087 in the nep - wide survey data . we did not include it ( @xmath3 90 sample at @xmath121 , see ko et al . 2012 ) for the estimation of luminosity function because its local over - density could affect the luminosity function . we present the 8@xmath4 m luminosity function for local sf galaxies ( sfg ) in the nep - wide field in figure 8 . while a large fraction of the sample occupy @xmath122 @xmath123 luminosity range ( figure 7c ) , there are too small number of sources fainter than @xmath124 . at around @xmath125 , the 8 @xmath4 m detection limits according to the redshift exceeds @xmath124 indicating many local sample fainter than @xmath126 could not be detected in this survey . therefore , it is hard to produce a statistically meaningful @xmath84 for this luminosity range . we compared our lf with previous results from various literatures . in fig 8 , shaded area indicates the lf from rodighiero et al . ( 2010 ) , which is based on the data from _ surveys on the vimos vlt deep survey ( vvds - swire ) and goods fields . triangles show the lf of local ( @xmath0 ) galaxies in the noao deep field in botes field ( huang et al . , 2007 ) . meshed area indicates the lf of local ( @xmath127 ) galaxies in the swire field from babbege et al . these different works are summarized in table 1 . rodighiero et al . ( 2007 ) used a library of template seds based on polletta et al . ( 2007 ) and franceschini et al . babbedge et al . ( 2006 ) used the sed models from rowan - robinson et al . ( 2005 ) and photometric - redshifts . their sfg / agn separation is based on sed - fit results . huang et al . ( 2007 ) used models from lu & hur ( 2000 ) with spectroscopic redshifts . a part of works from g10 presented here shows a lf for a higher redshift range ( @xmath128 ) , which was based on the nep - deep survey and photometric redshifts ( @xmath3 500 sample ) . we fit our 8 @xmath4 m lf using a double -power law ( marshall 1987 ; babbedge et al . 2006 ; goto et al . 2010 ) as @xmath129 { { dl } \over { l^ { * } } } .\ ] ] the best fit parameters for the normalization factor ( @xmath130 ) , and the characteristic luminosity ( @xmath131 ) are @xmath132 ( mpc@xmath133dex@xmath134 and @xmath135 , respectively . we obtained the faint - end slope @xmath8 , and the bright - end slope @xmath9 , which are comparable to those of the adjacent redshift bin ( @xmath136 ) from g10 . our local lf spans wide luminosity ranges and is clearly shifted toward the lower density at a given luminosity in comparison with the lf from g10 , indicating luminosity evolution . our lf shows a good agreement with that of rodighiero et al . ( 2010 ) but , in general , is consistent with other works . the difference between lfs also exist ( e.g. , some deviation of huang s ) . at the faintest luminosity ranges , rather larger photometric errors ( with smaller number statistics if any ) can count for the fluctuation . at higher luminosity ranges , it is more difficult to track down the origin of the differences in detail . but the differences seem to originate mostly from the different incompleteness in each sample and corrections for them . we regard that different number of sources contributing to each luminosity bin and agn fraction to be excluded are also entangled and seem to affect the disagreement , even though it s hard to compare how the individual works deal with correction factors . m luminosity distribution of local galaxies in the nep - wide field as a function of redshift ( left ) and number of sources in each luminosity bin ( right ) . bottom panel shows the 12 @xmath4 m luminosity function derived from the upper panels . in the bottom panel , filled circles represent this work for local ( @xmath0 ) galaxies in the nep - wide survey field . triangle represent the local lf derived by prez - gonzlez et al . shaded area and diamonds indicate the lf for @xmath137 and @xmath138 from rodighiero et al . ( 2010 ) using vvds - swire data . ` x ' symbols indicate lf for galaxies at @xmath139 from the g10 . [ fig 9],title="fig : " ] m luminosity distribution of local galaxies in the nep - wide field as a function of redshift ( left ) and number of sources in each luminosity bin ( right ) . bottom panel shows the 12 @xmath4 m luminosity function derived from the upper panels . in the bottom panel , filled circles represent this work for local ( @xmath0 ) galaxies in the nep - wide survey field . triangle represent the local lf derived by prez - gonzlez et al . shaded area and diamonds indicate the lf for @xmath137 and @xmath138 from rodighiero et al . ( 2010 ) using vvds - swire data . ` x ' symbols indicate lf for galaxies at @xmath139 from the g10 . [ fig 9],title="fig : " ] we construct 12 @xmath4 m luminosity function ( lf ) of local galaxies similar way . 12@xmath4 m luminosity has also been studied well based on the iras and iso observations ( spinoglio et al . 1989 ; rush et al . 1993 ) . it is also known to have good correlation with total ir luminosity , and recognized as one of the indicator of galaxy sf activity . ( spinoglio et al . 1995 ; prez - gonzlez et al . 2005 ) . the continuous wavelength coverage of our photometric data ( from the near - ir ) all the way through 24 @xmath4 m band is very useful / helpful to obtain accurate 12 @xmath4 m luminosity based on good sed fit with spectroscopic redshift . the detection limit at 12@xmath4 m band corresponds to about 18.8 mag and 50% completeness level is around 18.0 mag ( ab ) . due to the slightly different limit and luminosity distribution from those of the 8 @xmath4 m , the number of sample selected in each 12 @xmath4 m luminosity bin differs a few @xmath109 from that of 8 @xmath4 m luminosity function ( lf ) , as shown in the upper panels of figure 9 . in the lower panel of figure 9 , we present our 12 @xmath4 m luminosity functions ( for @xmath0 ) along with other results . the filled circles ( magenta ) represent the lf of local ( @xmath0 ) galaxies in the nep - wide field ( this work ) . the shaded area ( in light blue ) and diamond ( green ) symbols represent the lfs from rodighiero et al . ( 2010 ) based on the vvds - swire and goods data , for the redshift @xmath137 and @xmath138 , respectively . the filled triangle indicate the lf determined by prez - gonzlez et al . ( 2005 ) for @xmath137 . we also compared with the 12 @xmath4 m lf derived by g10 for galaxies at @xmath139 from the nep - deep data . the redshift ranges from these various works are partly overlapped with each other , but the filled circles , triangles and shade area represent the lfs for the local universe , while the diamond and cross symbols correspond to the lfs of higher redshift ranges . our result ( filled circles ) for local galaxies agrees well with the rodighiero s local lf , except for a deviated point at the luminosity bin @xmath140 , seemingly due to the different redshift ranges . for the higher - z lfs , the redshift range @xmath138 ( diamonds ) of rodighiero et al . ( 2010 ) is slightly higher than that from g10 ( @xmath139 ) , but their lf is systematically lower than that of goto et al . both works are based on the photometric redshifts , thereby seem to contain the inevitable uncertainties . besides , due to the samplings by different redshift bins , probably leading the different correction schemes for each incompleteness / bias might have caused the deviation from each other . however , the measurements of lfs for both fields ( the nep and the vvds - swire field ) consistently suggest the evolution of 12@xmath4 m lf . m luminosity distribution of local sf galaxies in the nep - wide field as a function of redshift and number of sources in each luminosity bin . bottom panel shows the 15 @xmath4 m luminosity function derived from the upper panels . in the bottom panel , the circles ( magenta colour ) represent the lf of our ( @xmath0 ) sample in the nep field . triangles indicate the local lf from pozzi et al . ( 2004 ) , and shaded area indicates the local lf from rodighiero et al . meshed region represents the results given by wu et al . also local lf from le floch ( 2005 ) was presented by crosses . [ fig 9],title="fig : " ] m luminosity distribution of local sf galaxies in the nep - wide field as a function of redshift and number of sources in each luminosity bin . bottom panel shows the 15 @xmath4 m luminosity function derived from the upper panels . in the bottom panel , the circles ( magenta colour ) represent the lf of our ( @xmath0 ) sample in the nep field . triangles indicate the local lf from pozzi et al . ( 2004 ) , and shaded area indicates the local lf from rodighiero et al . ( 2010 ) . meshed region represents the results given by wu et al . also local lf from le floch ( 2005 ) was presented by crosses . [ fig 9],title="fig : " ] we also derive 15@xmath4 m luminosity function of local ( @xmath141 ) galaxies in the nep - wide field and compare with various other results . we used the same galaxy sample as used for the 8 @xmath4 m and 12 @xmath4 m lf , but slight different number of sample contributes to each luminosity bin ( see figure 10 ) . while the observed magnitudes ( or fluxes ) at 8 @xmath4 m and 12 @xmath4 m bands derived from the assumed simple boxy filters with fitted sed templates , the observed 15 @xmath4 m fluxes used in this section were based on the measurement by l15 filter of irc . the 15@xmath4 m lf was needed for interpretation of 15@xmath4 m source properties ( xu , 2000 ; gruppioni et al . , 2005 ) since mir observations by isocam , its significance as an indicator to estimate the star formation activity has also been recognized . since xu et al . ( 2000 ) derived local ( @xmath142 ) luminosity function based on isocam 15@xmath4 m observation on the iras 60@xmath4 m galaxies ( @xmath3100 sample ) in the nep region , various determination of 15@xmath4 m lf has been carried out . we present the 15 @xmath4 m luminosity distribution of our sample and the resulting lf in figure 10 in comparison with several previous works . here , the filled circles ( magenta ) represent our results ( @xmath0 ) . pozzi et al . ( 2004 ) determined the local ( @xmath143 and @xmath138 ) 15@xmath4 m luminosity function using about 150 sample from the european large area iso survey ( elais ) southern ( s1 : 2@xmath144@xmath35 2@xmath144 , s2 : @xmath145 ) field , based on the sed analysis mainly on m51 and m81 types as reference templates . they did not include active galaxies like arp 220 ( @xmath146 ) considering that those types are rare and tend to appear at higher - z ( e.g. , @xmath147 ) . this aspect seems consistent with our local spectroscopic sample but we used many types of normal galaxy templates in the sed fit analysis . ( 2011 ) determined the lf for the redshift range of @xmath0 based on the galaxies ( @xmath3 230 ) from the 5 milijansky unbiased extragalactic survey ( 5muses ) , with the redshifts measured by infrared spectrograph ( irs ) . their lf is also presented by meshed area in figure 10 . there is no 15@xmath4 m lf measurement for the higher redshift from the nep - deep data . but the lfs from le floch et al . ( 2005 ) and rodighiero et al . ( 2010 ) indicated by cross and diamond represent higher - z ( @xmath148 ) lfs . here , le floch et al ( 2005 ) used _ spitzer _ mips 24@xmath4 m selection at cdf - s field ( 0.6 deg@xmath1 ) , carrying out analysis about 2600 sample based on both spectroscopic and photometric data from literatures and combo-17 . these various works are also summarized in table 1 . [ cols="<,<,^,^,>,<,<,^ , < , < , < " , ] [ tab : comparison_lf_works ] we presented mid - infrared luminosity functions for the galaxies in the local universe . we used the sample from the nep - wide survey data observed by akari space telescope , which covered the near- to mid - ir ( 2 @xmath3 25 @xmath4 m ) wavelengths continuously . to obtain more accurate luminosity functions , we took advantage of the spectroscopic redshift information obtained by optical follow - up surveys carried out with the mmt / hectospec and wiyn / hydra as well as various ancillary data sets covering from optical u@xmath149 to nir @xmath38 band . therefore the sed - fit analysis to determine the rest frame luminosities at mir bands does not have serious source of uncertainties . for better statistics and completeness , we limited this study to the lower redshifts ( @xmath0 ) range . our sample appears to be composed of various types of local sf galaxies including a small fraction of ellipticals and normal galaxies with some of these having similar seds to those of active galaxies . for the purpose of this work to derive the mir luminosity functions at the 8 , 12 , and 15 @xmath4 m bands , we employed the @xmath7 method , which is widely accepted because of its merits as described in sec . 3.1 . the lfs appear to be consistent with other previous studies , in general , within error bars although the differences among various determinations seem to originate from different method for source classification / selection and different photometric data and sed models , different redshift measurements and redshift ranges of sample , different correction methods for incompleteness or selection effect , different fields / areal coverage , and so on . errors also arises from those components . but , compared to the other works , the advantage of this work is the redshift information determined by spectroscopic observation and source classification based on line analyses as well as extensive wavelength coverage from optical u@xmath150 to mir 25 @xmath4 m including the _ akari _ s continuous filter coverage in the mir wavelengths , which allow us to obtain good sed - fit without large uncertainties such as those from photometric redshifts or type decision by colours or sed fit . also , since our sample is distributed over the large area of 5.4 @xmath151 , our results are not susceptible to the statistical uncertainty arising from the observation on a small part of the sky . while akari data do not drastically change earlier results , we can give more accurate luminosity function for the luminosity ranges 10@xmath152 10@xmath153 l@xmath154 . although it is not easy to say about evolution based on this work alone , valuable comparisons are possible with the higher - z lfs from the nep - deep survey which is complementary to nep - wide or various lfs of similar redshifts . the purpose of this work was achieved but exact measurements for more fainter luminosity bins and faint end slope seem to be pending . we expect to extend this study to the fir wavelengths range and to higher redshifts using much larger number of sample than those used in this work , if we can measure photometric redshifts for the sources in the nep - wide catalogue . we would like to thank the referee for the careful reading and constructive comments for this paper . this work is based on observations with _ akari _ , a jaxa project with the participation of esa , universities and companies in japan , korea , the uk , and so on . this work was supported by the grant 2012r1a4a1028713 from national research foundation of korea ( nrfk ) . mi acknowledges the support from the national research foundation of korea grant , no . 2008 - 0060544 . sjk thanks dani chao and the institute of astronomy , national tsing hua university , for their hospitality during his visit . prez - gonzlez p. g. et al . , 2005 , apj , 630 , 82 pilbratt g. l. et al . , 2010 , a&a , 518 , l1 polletta m. et al . , 2007 , apj , 663 , 81 pozzi f. et al . , 2004 , apj , 609 , 122 puget j .- et al . , 1996 , a&a , 308 , l5 puget j .- et al . , 1999 , a&a , 345 , 29 rodighiero g. et al . , 2010 , a&a , 515 , a8 rowan - robinson m. et al . , 1987 , mnras , 227 , 589 rowan - robinson m. et al . , 2004 , mnras , 351 , 1290 rowan - robinson m. et al . , 1997 , mnras , 289 , 490 rush b. , spinoglio l. , 1993 , apjs , 89 , 1 saunders w. et al . , 1990 , mnras , 242 , 318 schmidt m. , 1968 , apj , 151 , 393 sergeant s. et al . , 2004 , mnras , 335 , 813 sedgwick c. et al . , 2009 , aspc , 418 , 519 shim h. et al . , 2013 , apjs , 207 , 37 spinoglio l. , malkan , m. a. , 1989 , apj , 342 , 83 spinoglio l. et al . , 1995 , apj , 453 , 616
we present mid - infrared ( mir ) luminosity functions ( lfs ) of local ( @xmath0 ) star - forming ( sf ) galaxies in the _ akari _ s nep - wide survey field . in order to derive more accurate luminosity function , we used spectroscopic sample only . based on the nep - wide point source catalogue containing a large number of infrared ( ir ) sources distributed over the wide ( 5.4 deg@xmath1 ) field , we incorporated the spectroscopic redshift ( @xmath2 ) data for @xmath3 1790 selected targets obtained by optical follow - up surveys with mmt / hectospec and wiyn / hydra . the _ akari _ s continuous 2 24@xmath4 m wavelength coverage as well as photometric data from optical @xmath5 band to near - infrared @xmath6band with the spectroscopic redshifts for our sample galaxies enable us to derive accurate spectral energy distributions ( seds ) in the mid - infrared . we carried out sed - fit analysis and employed @xmath7 method to derive the mir ( e.g. , 8 @xmath4 m , 12 @xmath4 m , and 15 @xmath4 m rest - frame ) luminosity functions . we fit our 8 @xmath4 m lfs to the double power - law with the power index of @xmath8 and @xmath9 at the break luminosity @xmath10 . we made extensive comparisons with various mir lfs from several literatures . our results for local galaxies from the nep region are generally consistent with other works for different fields over wide luminosity ranges . the comparisons with the results from the nep - deep data as well as other lfs imply the luminosity evolution from higher redshifts towards the present epoch . [ firstpage ] galaxies : evolution galaxies : luminosity function infrared : galaxies
since the discovery of the cosmological distance scale to gamma ray bursts in 1997 ( djorgovski et al . 1997 ) several grb redshifts have been spectroscopically measuredjcg / grbgen.html for a continuously updated table . ] . the small sample of bursts , now comprising 42 events , allowed in the latest years the investigation of the grb rest frame properties . several correlations have been discovered between grbs spectral properties : 1 . the spectral lag - luminosity correlation @xmath4@xmath5 ( norris , marani & bonnel 2000 ; norris 2002 ) 2 . the variability - luminosity correlation @xmath6@xmath7 ( fenimore & ramirez ruiz 2000 ; reichart et al . 3 . the peak energy - isotropic energy correlation @xmath0@xmath8 ( amati et al . 2002 ; lloyd & ramirez - ruiz 2002 ) . we will call it the amati correlation . 4 . the peak energy - isotropic peak luminosity correlation @xmath0@xmath9 ( yonetoku et al . 2004 ) . 5 . the peak energy - collimation corrected energy correlation @xmath0-@xmath10 ( ghirlanda , ghisellini , lazzati 2004 , ggl04 hereafter ) . we will call it the ghirlanda correlation . all these correlations can be represented by powerlaws : their slopes are summarized in tab . 1 . note that , for the last three correlations , the peak energy @xmath11 of the @xmath12 spectrum is obtained by fitting the grb spectrum _ time integrated over the burst duration . _ despite several attempts to explain these correlations ( see e.g. eichler d. & lenvison a. 2004 ; liang e. , dai z. & wu x.f . 2004 ) , there is no universally accepted interpretation yet . one of the major drawback of these correlations is that they were found for a small sample of bursts , mainly due to the limited number of reliable redshift measurements . however , one test which can be performed even without knowing the redshift is to check if grbs with known fluence and @xmath11 are consistent with say the amati correlation , by considering all possible redshifts . this kind of consistency check has been performed very recently by two groups which tested the amati and the ghirlanda correlation . nakar & piran ( 2004 , hereafter np04 ) tested the amati correlation in the observed plane @xmath13@xmath14 ( where @xmath14 is the burst observed fluence ) using a sample of bright batse bursts ( from band et al . 1993 ; jimenez , band & piran 2000 ) . they concluded that @xmath1540% of the batse long ( bright ) bursts are inconsistent with this relation . band & preece ( 2005 , hereafter bp05 ) extended this work with a larger sample of batse grbs ( 760 bursts from mallozzi et al . 1998 ) and tested both the amati and the ghirlanda correlation . they claim that at least 88% of their bursts are inconsistent with the amati correlation and 1.6% with the ghirlanda correlation . they also claim that these conclusions are robust , since they are model independent . an important point to notice is what is meant by amati correlation . originally , amati et al . ( 2002 ) found @xmath16 with a small scatter , but using only 9 bursts . later , ggl04 ( see their tab . 1 and tab . 2 ) collected 23 grbs ( including the above 9 ) with redshift measurements and published peak energy , finding a slightly different slope and , more importantly , a larger scatter around the correlation , whose significance was nevertheless improved ( due to the increased number of points ) . in the same paper , it was demonstrated that for all bursts with a good measurement of the jet opening angle ( through the detection of a break in the light curve of the afterglow ) , the scatter of the amati correlation was entirely due to the different opening angles of the jet : bursts with the same collimation corrected energy @xmath10 and the same intrinsic @xmath0 have different @xmath1 because they have different jet opening angles . the scatter of the amati correlation , far from being due , i.e. to ill measured quantities , has a physical origin and is revealing of the distribution of the opening angles of the jets . bearing this in mind , in this letter we test if @xmath0 correlates with @xmath1 even if we do expect a large scatter around the best fitting line ( if any ) . to this aim we use a large sample of bursts with pseudo redshifts " which have been estimated through the lag luminosity relation ( sec . 2 ) . having found that indeed the @xmath0-@xmath1 correlation exists ( sec . 3 ) , we follow ggl04 and interpret the scatter as due to the distribution of jet opening angles , finding it ( sec . 4 ) . finally 5 ) , we discuss why np04 and bp05 , who also use a large sample of batse bursts , arrive to different conclusions . during the completion of this work , we received a manuscript from bosnjak et al . by studying the same problem with an independent and complementary approach , they arrive at the same conclusions presented in this letter . in the following we adopt a standard @xmath17cdm cosmology with @xmath18 , @xmath19 and @xmath20 . . slopes of the correlations between rest frame grb quantities ( for instance : @xmath21 ) . ref : ( 1 ) norris et al . 2000 ; ( 2 ) this work ( sec.3 ) ( 3 ) fenimore & ramirez ruiz 2000 , reichart et al . 2001 ; ( 4 ) yonetoku et al . 2004 ; ( 5 ) amati et al . 2002 ; ( 6 ) ghirlanda , ghisellini & lazzati 2004 . [ cols="<,<,<,<,<",options="header " , ] through a large sample of 442 batse grbs , with known spectral peak energy ( y04 ) and pseudo redshift ( bnb04 ) , we have computed the rest frame peak energy @xmath0 and the k - corrected equivalent isotropic energy @xmath1 . we have found that these two quantities are correlated : the power - law fit to this correlation gives a slope slightly flatter and an @xmath1 a little smaller than the slope and the @xmath1 found with the 27 grbs with spectroscopic redshifts , respectively . the scatter of the 442 data points around the fitting line is equal to the scatter of the 27 grbs with known redshift , the existence of outliers being marginal . we have further interpreted the scatter of these data points as due to the distribution of the jet opening angle of grbs . by requiring that they obey the ghirlanda correlation , we could estimate this distribution , which is a log - normal with an average angle of about 6.5@xmath22 . our main conclusion is that the existence of an @xmath0-@xmath23 ( amati ) correlation is confirmed , its scatter is probably rather well known , but its definitive estimate needs a wider sample of grbs with spectroscopic measured redshifts . we thank z. bosnjak , a. celotti , g. barbiellini & f. longo for sending us their manuscript before submission . the referee j. bonnell is thanked for his useful comments . we thank the italian miur for funding through cofin grant 2003020775_002 . amati , l. et al . 2002 , a&a , 390 , 81 band , d.l . , et al . 1993 , apj , 413 , 281 band , d.l . & preece , r.d . , 2005 , subm . to apj ( astro ph/0501559 ) ( bp05 ) band , d.l . , norris , j.p . & bonnell , j.t . , 2004 , apj , 613 , 484 ( bnb04 ) djorgovski , s.g . 1997 , gcn , 298 eichler d. & lenvinson d. 2004 , apj , 614 , l13 fenimore , e.e . & ramirez ruiz , e. , 2000 , preprint , astro ph/0004176 firmani , c. , ghisellini , g. , ghirlanda , g. & avila reese , v. , 2005 mnras ( letters ) , in press ( astro ph/0501395 ) ghirlanda , g. , ghisellini , g. & lazzati , d. , 2004 , apj , 616 , 331 ( ggl04 ) ghirlanda , g. , ghisellini , g. , firmani , f. , celotti , a. , bosniak , z. , 2005 , mnras , in press . jimenez , r. , band , d.l . & piran , t. , 2001 , apj , 561 , 171 liang e. , dai z. & wu x. f. , 2004 , apj , 606 , l29 lloyd ronning , n.m . & petrosian , v. , 2002 , apj , 565 , 182 . lloyd ronning , n.m . & ramirez - ruiz e. , 2002 , apj , 576 , 101 . mallozzi , r.s . , et al . , 1998 , in gamma ray bursts , 4th huntsville symposium . proceedings 428 , eds . c. meegan , r. preece and t. koshut ( aip : woodbury , ny ) , 273 nakar , e. & piran , t. , 2004 , preprint , astro ph/0412232 ( np04 ) norris , j.p . , marani , g.f . & bonnell , j.t . , 2000 , apj , 534 , 248 norris , j.p . , 2002 , apj , 579 , 386 reichart , d.e . , lamb , d.q . , fenimore , e.e . , ramirez ruiz , e. , cline , t.l . , & hurley , k. , 2001 , apj , 552 , 57 yonetoku , d. et al . 2004 , apj , 609 , 935 ( y04 )
we probe the existence of the @xmath0@xmath1 correlation in long grbs using a sample of 442 batse bursts with known @xmath0 and with redshift estimated through the lag luminosity correlation . this sample confirms that the rest frame peak energy is correlated with the isotropic equivalent energy . the distribution of the scatter of the points around the best fitting line is similar to that obtained with the 27 bursts with spectroscopic redshifts . we interpret the scatter in the @xmath2 plane as due to the opening angle distribution of grb jets . by assuming that the collimation corrected energy correlates with @xmath0 we can derive the observed distribution of the jet opening angles , which turns out to be log normal with a peak value of @xmath3 . [ firstpage ] cosmology : observations distance scale gamma rays : bursts
the study of hyperon production in hadron induced multi particle exit channels like @xmath4 involves several aspects . the nucleon hyperon interaction can be extracted by analyzing the @xmath5 subsystem in the appropriate kinematical region . closely related to that is the issue of the reaction mechanisms of the hyperon production which have to be clarified for an unambiguous interpretation of the data . if the hyperon production is due to the excitation and a subsequent decay of intermediate nucleon resonances than information about the structure of the relevant resonances can be extracted . the excitation function near threshold shows a clear deviation from a pure s wave phase space distribution , and the proton hyperon final state interaction ( fsi ) has to be included to describe the data @xcite . in the channel the @xmath6 fsi seems to be negligible and the pure s wave phase space calculations follow the data points reasonably well . the cross section ratio @xmath7 below excess energies of @xmath8 20 mev is about 28 @xcite in contrast to the value of about 2.5 determined for excess energies higher than @xmath9 mev @xcite ( see figure [ all_models ] ) . this value is in good agreement with the @xmath10 isospin relation . the question arises if this drastic increase of the cross section ratio near threshold is a mere effect of the @xmath11 fsi or wether it is partly due to the reaction mechanisms in the @xmath5 channels . to explain the increase of the @xmath12 cross section ratio in the near threshold region , different scenarios were proposed . calculations of the strangeness production based on incoherent @xmath13 and @xmath14 exchange have been performed in ref . the @xmath15 and @xmath16 scattering amplitudes for pion and kaon exchange , respectively , were taken from the existing data in the higher energy region @xcite . incoherent @xmath17 exchange model _ describes the cross section over the whole energy range , however in the near threshold region the channel is overestimated , and thus the predicted @xmath12 ratio is too low for @xmath18 20 mev ( see curve ( 1 ) in figure [ all_models ] ) . a better description of the strong rise of the ratio towards lower @xmath1-values is achieved by the _ resonance model _ ( curve ( 2 ) in figure [ all_models ] ) @xcite . in this model the nonresonant direct contributions like @xmath13 or @xmath14 exchange were not included , but the @xmath13 , @xmath19 and @xmath20 meson exchange with the excitation of the intermediate baryonic resonances @xmath21(1650 ) , @xmath21(1710 ) , @xmath21(1720 ) and @xmath22(1920 ) are taken into account . in this _ resonance model _ the near threshold region of the @xmath12 cross section ratio seems to be better reproduced than the higher energy values . it should be stressed that in these calculations the parameters were fixed on the basis of higher energy data , before the near threshold @xmath3 and @xmath23 data were available . other calculations by shyam @xcite ( based on the _ effective lagrangian model _ ) for the strangeness production include meson exchange ( @xmath24 and @xmath25 ) together with the excitation of resonances . the coupling constant was determined from data of the @xmath26 , @xmath27 and @xmath28 reaction channels . the coherent sum of resonant states and meson exchange processes describe the experimental data for the and channels very well . the _ effective lagrangian model _ is depicted by the curve ( 3 ) in figure [ all_models ] . the jlich theory group has performed calculations including @xmath13 and @xmath14 exchange @xcite . in their approach the interaction between the hyperons ( @xmath3 , @xmath29 ) and the nucleon is described by a microscopic ( @xmath3n-@xmath29n ) coupled channel model @xcite with a coherent superposition of the production amplitudes . the @xmath3 production is dominated by the @xmath14 exchange and therefore the contribution due to an interference between @xmath13 and @xmath14 exchange is negligible in this hyperon channel . on the other hand the @xmath13 and @xmath14 exchanges give a comparable contribution to the cross section in the case of @xmath2 production . an interference between @xmath13 and @xmath14 exchange amplitudes act very differently on the two channels . within the _ jlich meson exchange model _ the large cross section ratio can be described by a destructive interference of the @xmath13 and @xmath14 exchange amplitudes only . for excess energies above 20 mev the model is not valid any more but qualitatively the cross section ratio given by the model stays at a nearly constant level . although the various descriptions of the cross section ratio differ even in the dominant reaction mechanism , all reproduce more or less the trend of an increase of the @xmath12 cross section ratio in the threshold region ( see figure [ all_models ] ) . the present data are not sufficient to unambiguously identify the dominant reaction mechanism . to clarify this point further data are needed . especially the other isospin channels should allow information about the production mechanisms to be extracted . recently , after the installation of a neutron detector in addition to the @xmath3 and @xmath2 production the reaction channel became accessible at the detection system . the measurement of the @xmath30 hyperon production via this reaction was performed at two beam momenta , and @xmath31 = 2.74 gev / c , corresponding to excess energies of 13 mev and 60 mev , respectively . is an internal magnetic spectrometer at the cooler synchroton and storage ring cosy @xcite in jlich . the interaction between a proton in the beam and a proton from the @xmath32 cluster target @xcite may lead to the production of the @xmath0 hyperon in the reaction . the charged reaction products are separated from the circulating beam in the magnetic field of one of the regular cosy dipoles @xcite . the @xmath0 hyperon is identified via the missing mass technique by detecting the @xmath33 and the neutron . the momentum vector ( @xmath34 ) of the @xmath33 meson can be determined by tracking the @xmath33 trajectory reconstructed in the drift chambers ( dc1 and dc2 in figure [ tracks ] ) through the known magnetic field back to the target point . the momentum vector together with the velocity ( @xmath35 ) measurement in the two scintillators s8 and s1 are used to identify the kaon via its invariant mass @xmath36 . assuming that a hit in the neutron detector is from a neutron , the four momentum vector of the neutron is calculated by the measured velocity , the direction of the neutron ( given by the first module hit ) and the known mass . the background from charged particles incident on the neutron detector is discriminated by veto scintillators . in figure [ mm1-fig ] the experimental distributions of the squared missing mass ( @xmath37 ) of the system for the two beam momenta are shown . for the higher momentum , an enhancement around the squared @xmath38 mass is clearly seen on a large background ( figure [ mm1-fig]b ) , but for the lower beam momentum ( figure [ mm1-fig]a ) a @xmath38 peak is not directly visible . in order to determine the number of @xmath0 events in the higher energy data set , the background has been parametrized by a polynomial function which is added to the expected missing mass distribution of the @xmath39 system for the reaction obtained from simulation studies . in figure [ mmh_wielo - reac]a the experimental missing mass spectrum of the system is compared with the fitted polynomial function . the expected distribution from monte - carlo studies with @xmath40 is depicted in the figure as well . figure [ mmh_wielo - reac]b shows the result of the subtraction of the polynomial from the experimental missing mass distribution together with the monte - carlo distribution . in order to understand the background distribution , 22 reaction channels ( mostly multi - pion reactions but also @xmath41 ) were simulated and their contributions to the missing mass distribution were determined . these studies showed that the reactions , and are the dominant background channels in the @xmath38 region . the monte carlo code includes the realistic geometry and physics processes like energy loss and straggling which occasionally cause the misidentification of the particle type . all background channels result in a rather smooth distribution of the missing mass spectrum as can be inferred from calculations from monte - carlo studies and by comparing the two experimental distributions ( see figure [ mm1-fig ] ) . for the lower energy data set a @xmath38 peak is not obviously visible via the missing mass distribution . therefore , a simple polynomial background parametrization can not be used . to determine the number of @xmath38 events it was assumed that the background shape for this data set is the same as that at the higher energy . this assumption is justified since there are no new open channels for the higher energy . at the cosy-11 experiment the shape of the missing mass distribution is mainly determined by the acceptance of the detection system and is dependent on the excess energy of an individual event . from the analysis of @xmath19 and @xmath42 production studies at cosy-11 it was verified that the background shape resulting here mainly from multi pion production is in very good agreement with the expectations from monte carlo studies taking into account the detector characteristics and is comparable at different beam momenta . in addition , monte carlo data of the reaction channels which contribute dominantly to the background in the @xmath43 production were compared in view of the background shape by adjusting the kinematical limits . within error bars their shapes were identical . therefore it is justified to assume that the background shape is the same for both beam momenta . for a detailed discussion on the background shape at cosy-11 we refer to @xcite . the background shape from the experimental missing mass distribution for the higher energy data set was shifted such that the kinematical limits for both spectra were the same . afterwards , it was added to the missing mass distribution of the reaction which was determined from monte - carlo studies at the lower energy ( see figure [ mml_fit - reac ] ) . the relative magnitudes were determined by the @xmath44 minimisation of the functional forms to the data sample at lower energy . for the lower energy data set , no clear enhancement near the kaon mass in the invariant mass distribution is visible . therefore the assumption was made , that the kaon peak in the experimental distribution has the same position and width as in the simulated distribution . as a cross check , event samples in different regions of experimental invariant mass but still within the expected range for kaons were taken and the missing mass distributions were determined . the assumed background shape was subtracted from the experimental distribution and the enhancement around the mass of the @xmath0 was interpreted as a signal from the @xmath0 hyperon ( see figure [ mml_fit - reac ] ) . this procedure was repeated for five different invariant mass regions and the number of events under the remaining peak was determined . the results are shown in figure [ inv_scan ] . the vertical error bars correspond to the statistical error of the number of events . the horizontal error bars indicate the widths of the appropriate invariant mass intervals . in figure [ inv_scan ] curve ( 1 ) shows the expected invariant mass distribution of kaons from monte - carlo calculations . curve ( 2 ) is a gauss fit to the experimentally determined number of @xmath0 events in the different invariant mass regions . this can be interpreted as an experimental distribution of kaons . the assumed experimental kaon invariant mass distribution is consistent with the distribution expected from monte - carlo studies . for the higher beam momentum a kaon peak is clearly visible in the invariant mass distribution and can be fit by a gauss function . in a systematical study of the result at the higher beam momentum three different selections on the experimental invariant mass distribution were applied . events from the following regions : @xmath45 , @xmath46 and @xmath47 centered around the nominal reconstructed mass ( @xmath48 ) were taken and the corresponding missing mass distributions were generated . @xmath49 is the experimental resolution of this observable . next the number of @xmath0 events for each of these distributions was determined . the results are listed in the middle column of table [ number_higher ] and in the last column the number of the @xmath0 events corresponding to the full gaussian distribution . .[number_higher ] the number of @xmath0 events as a function of the width of the invariant mass range . only statistical errors are given . [ cols="^,^,^ " , ] among the models described in the introduction only two give predictions for the reaction , namely the _ jlich meson exchange model _ @xcite and the _ resonance model _ @xcite . calculations of the @xmath0 production within the _ jlich meson exchange model _ predict a total cross section of at @xmath1 = 13 mev for the destructive interference ( which was necessary to describe the high @xmath50 cross section ratio at threshold ) . this is about a factor of 20 below the experimental value of 4.56 @xmath51b given in table [ results - tab - sys ] . a constructive interference would result in a cross section even a factor of 53 too low . for the _ resonance model _ the predictions for the channel for the near threshold region deviate even more from the data . in figure [ res - sib ] the model predictions and the available data for the ( a ) , ( b ) and ( c ) channels are shown . the filled data points presented by triangles , dots and squares in the near threshold region were measured by the collaboration @xcite . the data point in the channel indicated by the arrow was determined in parallel to this work by selecting the @xmath52 exit channel which was included in the events sample at 2.74 gev / c . the high energy data for the given reactions were taken from @xcite and @xcite . the model calculations for each channel are given by the solid lines @xcite . the data point for the channel at is underestimated in the total cross section calculated using the resonance model @xcite by about a factor of 500 and for by about a factor of 50 . for the channel , this model calculation describes the existing data set and in the case of the channel the underestimation of the cross section in the near threshold region is about a factor of 16 being 30 times smaller than for the @xmath0 production . at high excess energies , the @xmath0 data points are by factor 3 4 below the model calculations . previous hyperon production studies conclude that final state interactions ( fsi ) plays an important role in the near threshold @xmath3 production @xcite . in the _ resonance model _ the fsi is not included @xcite and therefore the deviation of the model calculations from the data points in the near threshold region is expected if a strong fsi is present . this effect is clearly seen for the and barely observed for the reaction channel . investigations of hyperon production in @xcite indicate that the data can not be described by a pure 3-body phase space ( ps ) dependent cross section expressed as @xcite : @xmath53 where @xmath14 is a normalization factor and @xmath1 the excess energy . therefore a modification is needed which takes into account the _ proton - hyperon _ fsi . in order to describe the near threshold region , the parametrisation of the excitation function including the fsi proposed by fldt - wilkin @xcite was used . it is expressed by : @xmath54 where @xmath55 and @xmath56 are parameters related to the fsi strength . in figure [ fw - all ] the cross sections for different production channels for the @xmath3 , @xmath2 and @xmath0 hyperons are compared to predictions based on the 3body phase space ( ps , dotted line ) and the 3body phase space calculations modified by the @xmath6 fsi ( ps + fsi , solid line ) , following equation [ fw - eq ] with @xmath57 and @xmath55 as free parameters . these parameters are related to the scattering length _ a _ and the effective range _ r _ of the @xmath6 potential @xcite . for the data the resulting @xmath56 and @xmath55 parameters have similar values as for the channel . it seems that for the @xmath0 production via the reaction a rather strong @xmath58 fsi is present , however , within the error bars also the curve obtained without fsi describes the two data points . therefore an unambiguous conclusion about the @xmath58 fsi requires more data to disentangle the reaction mechanisms and especially the role of nuclear resonances . the total cross section of the reaction was determined by means of the detection system for excess energies of @xmath1 = 13 and 60 mev . the values established are more than an order of magnitude larger than the expectations of currently available model predictions . the unexpected large total @xmath0 production cross section is consistent with an observation by tan @xcite who concluded that when assuming charge symmetry in @xmath59 and @xmath60 scattering , the contribution from the @xmath2 diagram is less than one seventh of the contribution from the @xmath0 channel . further , recently it was suggested that for the @xmath61 production @xcite a strong enhancement of the reaction amplitude towards threshold might be due to the presence of a crypto exotic baryon with hidden strangeness . although this observation is not directly applicable to other isospin channels , ( in the @xmath62 system no corresponding structure was observed ) it might be a hint of an exotic mechanism . in any case , present theoretical predictions of the cross sections strongly underestimate the experimental data . the comparison of the excitation function expected based upon phase space distribution including @xmath63 fsi to the data results in fsi parameters comparable to the @xmath64 system . this may indicate a strong @xmath58 interaction but due to the large systematic uncertainties the data are also consistent with a pure phase space distribution without @xmath58 fsi . further studies of the @xmath0 production are necessary to clarify the picture . on the experimental side additional data points should be added for which an improved event selectivity is favorable to reduce the large uncertainties introduced by the background subtraction . a @xmath65 detection system for neutral and charged particles which will be soon available with wasa at cosy could be used @xcite . on the theoretical side an improved model has to be developed which consistently reproduces the hyperon cross section data near threshold . this work has been supported by the european community - access to research infrastructure action of the improving human potential programme , by the ffe grants ( 41266606 and 41266654 ) from the research center jlich , by the daad exchange programme ( ppp - polen ) , and by the polish state committee for scientific research ( grant no . pb1060/p03/2004/26 ) . during the evaluation process of the article we have been made aware of the predictions of the excitation function of the total cross section for the reaction which is closer to the data in comparison with the models discussed , yet still underpredict the determined total cross sections by more than an order of magnitude @xcite . j. t. balewski et al . , phys . lett . * b420 * ( 1998 ) 211 . s. sewerin et al . , phys . * 83 * ( 1999 ) 682 . p. kowina et al . j. * a22 * ( 2004 ) 293 . g. fldt and c. wilkin , z. phys . * a357 * ( 1997 ) 241 . a. baldini et al . , _ total cross - section for reactions of high - energy particles _ , springer , berlin , 1988 . a. sibirtsev et al . a646 * ( 1999 ) 427 . a. sibirtsev et al . nucl - th/0004022 * v2 * ( 2000 ) . r. shyam et al . , phys . rev . * c63 * ( 2001 ) 022202 . n. tsushima et al . , phys . rev . * c59 * ( 1999 ) 369 . r. shyam , hep - ph/0406297 ( 2004 ) . a. gasparian et al . , phys . lett . * b480 * ( 2000 ) 273 . a. gasparian et al . , nucl . phys . * a684 * ( 2001 ) 397 . b. holzenkamp et al . a500 * ( 1989 ) 485 . r. maier , nucl . instr . & meth . * a390 * ( 1997 ) 1 . h. dombrowski et al . , nucl . instr . & meth . * a386 * ( 1997 ) 228 . s. brauksiepe et al . , nucl . instr . & meth . * a376 * ( 1996 ) 397 . p. moskal et al . , j. phys . * g 32 * ( 2006 ) 629 . d. hesselbath , phd . thesis , universitt bonn , 2000 . m. fritsch , phd . thesis , universitt erlangen , 2002 . v. flamino et al . , _ compilation of cross sections _ , cern - hera 84 - 01 1984 a. sibirtsev , ikp , fz jlich , germany , priv . j. balewski et al . , eur . j. * a2 * ( 1998 ) 99 . r. bilger et al . , phys . lett . * b420 * ( 1998 ) 217 . e. byckling and k. kajantie , _ particle kinematics _ , john wiley & sons ltd . , london 1973 . tai ho tan , phys . 23 ( 1969 ) 395 . a. sibirtsev et al . , eur . phys . j.*a27 * ( 2006 ) 263 . adam et al . , proposal for the wasa at cosy experiment , nucl - ex/0411038 ( 2004 ) . r. shyam , phys . rev . * c 73 * ( 2006 ) 035211 .
@xmath0 hyperon production was measured at the spectrometer via the reaction at excess energies of @xmath1 = 13 mev and @xmath1 = 60 mev . these measurements continue systematic hyperon production studies via the / @xmath2 reactions where a strong decrease of the cross section ratio near threshold was observed . in order to verify models developed for the description of the @xmath3 and @xmath2 production we have performed the measurement on the @xmath0 hyperon and found unexpectedly that the total cross section is by more than one order of magnitude larger than predicted by all models investigated . after the reconstruction of the kaon and neutron four momenta , the @xmath0 is identified via the missing mass technique . details of the method and the measurement will be given and discussed in view of theoretical models .
( a)the top view of a coo@xmath2 plane . ( b)the side views of the crystal structures of a parent na@xmath14coo@xmath2 , a na - deintercalated na@xmath15coo@xmath2 and a @xmath16layer hydrated na@xmath15coo@xmath1h@xmath2o . open arrows indicate soft chemical treatment.,width=415 ] na@xmath14coo@xmath2 is a high - performance thermoelectric power material @xcite . the crystal structure is constructed from the stacking of a co triangular lattice in the layered edge - sharing coo@xmath2 octahedrons and of a na layer . the discovery of superconductivity through soft chemical treatment of sodium de - intercalation and hydration has renewed our interests in spin and charge frustration effects on the coo@xmath2 triangular lattice @xcite . [ fig : crystal](a ) and [ fig : crystal](b ) illustrate the crystal structure and the soft chemical treatment , respectively . the double - layer hydrated cobalt oxides na@xmath0coo@xmath1h@xmath2o ( @xmath30.35 , @xmath41.3 ) are triangular - lattice superconductors with the optimal @xmath17 5 k @xcite . except the charge ordered insulator na@xmath0coo@xmath2 with @xmath3 1/2 , na@xmath0coo@xmath2 and the hydrated compounds are itinerant electronic systems . for @xmath30.7 , @xmath18type spin fluctuations ( in - plane ferromagnetic and out - of - plane antiferromagnetic spin fluctuations ) @xcite , the motion of na ions @xcite and co charge disproportionation @xcite have been reported . for anhydrated na@xmath0coo@xmath2 , the na concentration changes the carrier doping level of a coo@xmath2 plane , which develops the electronic state @xcite . for double - layer hydrated na@xmath0coo@xmath1h@xmath2o , the na concentration is not a unique parameter of @xmath5 @xcite . since no one has ever observed superconductivity of the anhydrated compounds , the carrier doping level is not a unique parameter of @xmath5 . the chemical diversity of the site ordering of na ions @xcite and the charge compensation by oxonium ions h@xmath8o@xmath9 @xcite may realize a delicate electronic system on the triangular lattice . the effect of the intercalated water molecules on the electronic state of the coo@xmath2 plane is poorly understood . the followings have been debated so far , whether the normal - state spin fluctuations are ferromagnetic , antiferromagnetic or the other type , whether or not the superconductivity occurs in the vicinity of the magnetic instability , whether the charge fluctuations play a key role in the superconductivity , how different the triangular - lattice superconductivity is from the square - lattice one , what the pairing symmetry is . in this paper , we present the highlights of our nmr studies and findings for the double - layer hydrated compounds ; the various @xmath5 samples synthesized in controllable way , @xmath7h nmr evidence for the existence of the h@xmath8o@xmath9 oxonium ions , @xmath5-sensitive @xmath10na nuclear spin - lattice relaxation , two superconducting phases classified by @xmath11co nuclear quadrupole resonance ( nqr ) frequency , a magnetic phase located between the two superconducting phases , in - plane two components in @xmath11co nmr knight shift and nuclear spin - lattice relaxation , non - korringa behavior above @xmath5 , and power law behaviors in nuclear spin - lattice relaxation rates . polycrystalline samples of the starting compound na@xmath14coo@xmath2 were synthesized by conventional solid - state reaction methods . the powdered samples of na@xmath14coo@xmath2 were immersed in br@xmath2/ch@xmath8cn solution to deintercalate na@xmath9 ions and then in distilled water to intercalate h@xmath2o molecules @xcite or in br@xmath2/h@xmath2o solution to make ion - exhange reaction of na and h@xmath2o @xcite . after the powders of na@xmath15coo@xmath1h@xmath2o were separated from the solution by filtration , they were exposed in various humidity air . for the synthesized samples , we observed various duration ( keeping time in the humidity - controlled chamber in a daily basis ) effects on @xmath5 @xcite . @xmath11co ( spin @xmath19 = 7/2 ) , @xmath10na ( spin @xmath19 = 3/2 ) and @xmath7h ( spin @xmath19 = 1/2 ) nmr experiments were performed by a pulsed nmr spectrometer for the powder samples . the nuclear spin - lattice relaxation times were measured by inversion recovery techniques . magnetic phase diagram of @xmath5 and @xmath20 plotted against @xmath11co nqr frequency @xmath21 at 10 k. we reproduced the data of y1 from @xcite , h1 from @xcite , h2 from @xcite , and h3 from @xcite . the solid and dashed curves are guides for eyes . we call two superconducting phases characterized by @xmath22 12 . 5 mhz and by @xmath23 12 . 6 mhz _ @xmath24 phase _ and _ @xmath25 phase _ , respectively . the samples denoted by blh ( @xmath26layer hydrates ) are referred to the nmr experiments . , width=377 ] zero - field nqr is sensitive to local crystal structure and charge distribution around the nuclear site . although no distinct changes in x - ray diffraction patterns have been observed for the various double - layer hydrated compounds , systematic changes in the peak values of @xmath11co nqr spectra were observed @xcite . thus , we present the results of @xmath11co nqr spectrum measurments to characterize the samples and to classify the electronic states . for an electric field gradient tensor @xmath27 ( @xmath28 = xx , yy , and zz , principal axis directions ) with an asymmetry parameter @xmath29[@xmath30(@xmath31-@xmath32)/@xmath33 @xmath34 1 , three transition lines of @xmath11co ( spin @xmath19 = 7/2 ) nqr should be observed as @xmath35 ( @xmath36 = @xmath373/2 @xmath381/2 ) , @xmath39 ( @xmath36 = @xmath375/2 @xmath383/2 ) , and @xmath40 ( @xmath36 = @xmath377/2 @xmath385/2 ) . for @xmath41 1 , one easily finds @xmath40 @xmath42 1.5@xmath39 @xmath42 3@xmath35 . in fig . [ fig : phasediagram ] , superconducting transition temperatures @xmath5 and magnetic transition temperatures @xmath20 are plotted against @xmath11co nqr frequency @xmath21 at 10 k @xcite . the existence of two superconducting phases were confirmed . we call the two superconducting phases at @xmath22 12 . 5 mhz and at @xmath23 12 . @xmath24 phase _ and _ @xmath25 phase _ , respectively . in the magnetic phase diagram of fig . [ fig : phasediagram ] , the optimal samples of @xmath5 = 4.5 @xmath6 4.8 k in the @xmath24 @xmath43 are located in 12.1 mhz @xmath34 @xmath21 @xmath44 12.4 mhz . the magnetic ordering phase is located between two superconducting phases . thus , the magnetic instability and the superconductivity occur closely to each other . it should be emphasized that for the samples with 12.5 mhz @xmath45 12 . 65 mhz , a possibility of charge - density - wave ( cdw ) ordering is excluded by the temperature dependence of both @xmath11co nqr spectra of @xmath39 and @xmath40 @xcite . the cdw ordering at a transition temperature @xmath20 must result in the broad nqr line widths of @xmath46 @xmath42 1.5@xmath47 @xmath42 3@xmath48 . however , @xmath49 was observed below @xmath50 @xcite . thus , an internal magnetic field due to magnetic ordering is concluded . here , since no divergence behavior at @xmath20 was observed in nuclear spin - lattice relaxation rates , this magnetic ordering is unconventional @xcite . no wipeout effect below @xmath20 excludes the existence of slow modes due to spin glass transition @xcite . it should be noted that the experimental magnetic phase diagram against @xmath40 in fig . [ fig : phasediagram ] is similar to the theoretical one against the thickness of a coo@xmath2 layer @xcite . in the theoretical phase diagram , two superconducting phases are separated through a magnetic ordering phase and result from two competing fermi surfaces of @xmath51 cylinders and @xmath52 hole pockets . two superconducting phases should have different pairing symmetry @xcite . we present the nmr studies of the @xmath24 phase below . @xmath7h proton nmr spectrum at 200 mhz for a double - layered hydrated sample at room temperature . the dashed curve is a numerical simulation of @xmath7h nmr powder pattern for h@xmath8o@xmath9 illustrated in the right figure.,width=302 ] in the double - layer hydrated superconductors , not only the concentration of na@xmath9 ions @xcite but also the h@xmath8o@xmath9 ions @xcite play key roles in the occurrence of superconductivity . the existence of the oxonium ions h@xmath8o@xmath9 was evidenced by the observation of a bending mode and stretching modes of h@xmath8o@xmath9 ions using ramann spectroscopy @xcite . nmr is also a powerful technique to detect such a local structure . [ fig : proton ] shows a high resolution fourier - transformed @xmath7h nmr spectrum at 200 mhz for a double - layered hydrated sample in the @xmath24 phase at room temperature . we observed one peak and two shoulders in the @xmath7h nmr spectrum . for a frozen water molecule , a pake doublet of two peaks should be observed @xcite . in fig . [ fig : proton ] , a numerical simulation for an oxonium ion h@xmath8o@xmath9 with triangular coordination of three protons is illustrated by a dashed curve @xcite . similarity between the multiple structure in the @xmath7h nmr spectrum and the numerical simulation in fig . [ fig : proton ] indicates the existence of the oxonium ion h@xmath8o@xmath9 in the double - layer hydrated compound . it is a future work to estimate how much h@xmath8o@xmath9 ions and h@xmath2o molecules are involved in each sample . since na ions are located between coo@xmath2 layers , it is expected that na nuclear spins can probe interlayer correlations . ( a)fourier - transformed @xmath10na nmr frequency spectra of na@xmath14coo@xmath2 , double - layer hydrated na@xmath15coo@xmath1h@xmath2o ( blh1 and blh2 denoted in fig . [ fig : phasediagram ] ) , and a dehydrated na@xmath15coo@xmath2 . the dashed line is @xmath54 = 84.2875 mhz ( @xmath54=@xmath55 , @xmath56 = 11.262 mhz / t and @xmath57 7.484 t ) . ( b)temperature dependence of @xmath10na nuclear spin - lattice relaxation rates 1/@xmath58 for double - layer hydrated na@xmath15coo@xmath1h@xmath2o ( blh1 , blh2 , and blhs2 denoted in fig . [ fig : phasediagram]).,width=453 ] log - log plots of @xmath10na nuclear spin - lattice rate 1/@xmath58 against temperature for na@xmath14coo@xmath2 @xcite , double - layer hydrated na@xmath15coo@xmath1h@xmath2o ( blh1 , blh2 , and blhs2 denoted in fig . [ fig : phasediagram ] ) . the solid lines are fits by power laws of @xmath59 . the right figure illustrates a snapshot of @xmath53-type spin fluctuations . , width=453 ] fig . [ fig : nanmr](a ) shows the central transition lines ( @xmath36 = @xmath371/2 @xmath601/2 ) of the fourier transformed @xmath10na ( nuclear spin @xmath19 = 3/2 ) nmr spectra for na@xmath14coo@xmath2 , a double - layer hydrated non - superconducting blh1 ( @xmath61 1.8 k ) and a superconducting blh2 ( @xmath62 4.5 k ) , both of which are denoted in fig . [ fig : phasediagram ] , and a dehydrated na@xmath15coo@xmath2 @xcite . the dehydrated na@xmath15coo@xmath2 was obtained by heating a double - layer hydrate at about 250 @xmath63c . in na@xmath14coo@xmath2 , the na nmr lines are affected by large knight shift and electric quadrupole shift . for the double - layer hydrates blh1 and blh2 , however , the @xmath10na nmr lines show small knight shifts . for the na@xmath15coo@xmath2 dehydrated from na@xmath15coo@xmath1h@xmath2o , the @xmath10na nmr lines show large knight shift once again . the recovery of knight shift by dehydration indicates that not the na deficiency but the intercalated water molecule blocks the co - to - na hyperfine field . [ fig : nanmr](b ) shows the temperature dependence of the @xmath10na nuclear spin - lattice relaxation rate 1/@xmath58 for blh1 , blh2 and blhs2 ( @xmath62 3 k ) @xcite . 1/@xmath58 largely decreases with duration from blh1 to blh2 and then blhs2 . the local field fluctuations probed by 1/@xmath58 @xcite are sensitive to the duration effect . fig . [ fig : loglognat1 ] shows the log - log plots of 1/@xmath58 for the parent na@xmath14coo@xmath2 which are reproduced from @xcite and for the double - layer hydrated blh1 , blh2 and blhs2 @xcite . at low temperatures in the normal states above @xmath5 , 1/@xmath58 s show the power - law behaviors ( solid lines ) of @xmath59 with @xmath64 0.7 and 0.6 for blh1 and blh2 , respectively . these non - korringa behaviors indicate that the samples are not conventional fermi liquid systems . the magnitude of 1/@xmath58 significantly decreases from the non - hydrated to the double - layer hydrated samples . the temperature dependence of 1/@xmath58 in blh1 is nearly the same as that in the parent na@xmath14coo@xmath2 @xcite . both systems show the power law of @xmath65 in 1/@xmath58 at @xmath66 20 k and the rapid increase at @xmath67 200 k. the electron spin dynamics probed at the na site in the double - layer hydrates is similar to that in na@xmath14coo@xmath2 . the @xmath53-type spin fluctuations , i.e. intra - plane ferromagnetic and inter - plane antiferromagnetic fluctuations , are observed in na@xmath14coo@xmath2 by inelastic neutron scattering experiments @xcite . the intercalated water molecules may block the inter - plane antiferromagnetic couplings . nearly ferromagnetic intra - plane spin fluctuations may persist in the double - layer hydrates . @xmath11co nqr and nmr for an optimal sample ( blh1 m in fig . [ fig : phasediagram ] ) of @xmath5 = 4.8 k. ( a)@xmath11co nuclear quadrupole resonance frequency @xmath68 and the asymmetry parameter @xmath29 . ( b)the central transition lines ( @xmath36 = 1/2 @xmath691/2 ) of @xmath11co nmr spectra at @xmath70 plane and a numerical simulation ( dotted curve ) of two dimensional powder pattern of the central transition including anisotropic knight shift of @xmath71 . ( c)@xmath11co knight shifts @xmath72 and @xmath73 at @xmath74 and @xmath75 . ( d)@xmath11co nuclear spin - lattice relaxation rates divided by temperature 1/@xmath76 at @xmath74 and @xmath75.,width=453 ] from @xmath11co nqr and nmr experiments for an optimal sample of @xmath5 = 4.8 k ( blh1 m denoted in fig . [ fig : phasediagram ] ) , we obtained the significant results that the electric field gradient tensor @xmath77 at the @xmath11co nuclear site is asymmetric ( @xmath78 0.2 ) , that the in - plane @xmath11co knight shift has two components ( inplane - anisotropic @xmath79 or magnetic disproportionation ) , and that the @xmath11co nuclear spin - lattice relaxation rate 1/@xmath13 is different at the two signals @xcite . [ fig : conmr](a ) shows the temperature dependence of @xmath11co nuclear quadrupole resonance frequency @xmath68 and the asymmetry parameter @xmath29 @xcite . the temperature dependence of @xmath68 in fig . [ fig : conmr](a ) is widely observed in many materials . the finite asymmetry parameter @xmath78 0.2 suggests an unique underlying electronic wave funtion and/or charge distribution at the co site . [ fig : conmr](b ) shows the temperature dependence of the central transition ( @xmath36 = @xmath371/2 @xmath601/2 ) of @xmath11co nmr spectra at @xmath80(=7.4847 t ) @xmath81-plane @xcite . for the @xmath11co nmr experiments , we prepared the powder sample na@xmath15coo@xmath1h@xmath2o oriented by a magnetic field of @xmath57 7.5 t in fluorinert fc70 ( melting point of 248 k ) . stycast 1266 was the worst for the hydrated samples . hexane was much better than stycast 1266 and actually applied to nmr experiments @xcite but hard to handle . fluorinert was the best . the @xmath12-planes of powder grains are aligned to the external magnetic field . thus , the observed nmr spectrum should be a two dimensional powder pattern . the dotted curve in fig . [ fig : conmr](b ) is a numerical simulation of two dimensional powder pattern including anisotropic knight shift of @xmath71 . here we define the knight shift @xmath82 for the line @xmath83 and @xmath84 for the line @xmath85 of the central transition ( @xmath36 = @xmath371/2 @xmath601/2 ) . the full - swept @xmath11co nmr spectra and their field dependences were well reproduced from two dimensional powder pattern due to quadrupole shifts with second order perturbation and anisotropic knight shifts @xcite . in principle , the asymmetric quadrupole shift with @xmath78 0.2 yields two peaks in the central transition but could not reproduce the actual width between the two resonance peaks ( @xmath74 and @xmath75 ) in fig . [ fig : conmr](b ) . from the magnetic field dependence of the two peaks , two signals have nearly the same quadruple shifts @xcite . since twice magnetic field yields twice split width between the two resonance peaks in fig . [ fig : conmr](b ) , we should introduce at least two components of knight shift @xcite . since imperfect orientation of powder grains easily suppresses the singularity at @xmath83 , then the disagreement of the intensity ratio of @xmath74 to @xmath75 in the nmr spectrum may not be significant . however , if the intensity difference of @xmath74 and @xmath75 is intrinsic , one must introduce two magnetic co sites with nearly the same quadruple shifts . the charge disproportionation was observed in the parent na@xmath14coo@xmath2 @xcite . then , magnetic disproportionation may occur in the double - layer hydrates in that case . after the second order correction of quadrupole shifts based on @xmath68 and @xmath29 in fig . [ fig : conmr](a ) , we estimated the respective knight shifts @xmath72 and @xmath73 at the lines @xmath74 and @xmath75 in fig . [ fig : conmr](b ) @xcite . fig . [ fig : conmr](c ) shows the temperature dependences of the knight shifts @xmath72 and @xmath73 @xcite . the knight shift @xmath72 shows a curie - weiss - type upturn below 50 k , being similar to a low temperature bulk magnetic susceptibility @xcite . thus , a curie - weiss behavior in the magnetic susceptibility below 50 k is intrinsic . [ fig : conmr](d ) shows the temperature dependence of 1/@xmath86 at @xmath74 and @xmath75 @xcite . the difference in 1/@xmath86 indicates inplane ( xy)-anisotropy or disproportionation in the co electron spin fluctuations . ( a)zero - field @xmath11co nuclear spin - lattice relaxation rates 1/(@xmath13)@xmath87 for blh1 and blh sample of @xmath5 = 4.6 k @xcite and the sample of @xmath5 = 4.7 k @xcite . ( b)magnetic field dependece of the superconducting @xmath11co knight shift @xmath73 for the optimal sample of @xmath5 = 4.7 k reproduced from @xcite . the horizontal lines indicate the orbital shifts @xmath88 = 2.9 , 2.95 , 3.0 , and 3.1 @xmath89 reported so far . the arrows indicate the respective spin shifts @xmath90 . , width=491 ] fig . [ fig : cosct1](a ) shows zero - field @xmath11co nuclear spin - lattice relaxation rates 1/(@xmath13)@xmath87 for blh1 , the sample of @xmath5 = 4.6 k @xcite and the sample of @xmath5 = 4.7 k @xcite . no korringa behavior in 1/(@xmath13)@xmath87 just above @xmath5 and no coherence peak just below @xmath5 suggest non - fermi liquid and unconventional superconductivity . the power law behavior of 1/(@xmath13)@xmath87 below @xmath5 indicates the existence of line nodes on a superconducting gap parameter . in a triangular lattice , such a line - nodal order parameter is expected for @xmath91-wave , @xmath92-wave and @xmath93-wave pairing @xcite and extended @xmath94-wave inter - band pairing @xcite . [ fig : cosct1](b ) shows the magnetic field dependence of the superconducting @xmath11co knight shift @xmath73 for the optimal sample with @xmath5 = 4.7 k @xcite . the horizontal bars indicate the reported orbital shifts @xmath88 = 2.7 ( out of the figure scale ) @xcite , 2.9 @xcite , 2.95 @xcite , 3.0 @xcite , and 3.1 @xmath89 @xcite . the knight shift @xmath95 is expressed by the sum of the spin shift @xmath90 and the van vleck orbital shift @xmath88 . then , the arrows indicate the respective spin shifts @xmath90 . if @xmath88 = 2.9 @xmath89 , the finite @xmath90 indicates a spin - triplet pairing . if @xmath88 = 3.1 @xmath89 , the diminished @xmath90 indicates a spin - singlet pairing . thus , the choice of @xmath88 changes the conclusion for the issue whether the spin singlet or triplet is realized . the existing data on the impurity ga- and ir - substituion effects @xcite support non-@xmath94-wave pairing . although any impurity substitution effects on @xmath5 as anderson localization effect @xcite could not serve as the tests of identification of any pairing symmetry , weak impurity potential scattering for the line - nodal pairing could account for the observed weak suppression of @xmath5 . double - layer hydrated cobalt oxides na@xmath0coo@xmath1h@xmath2o were found to have two superconducting phases labeled by @xmath11co nqr frequency @xmath21 . non - superconducting magnetic phase is located between two phases in the @xmath21 classification . the peculiar electronic states , magnetic disproportionation or inplane ( xy ) anisotropy of @xmath11co local spin susceptibility , was observed for the @xmath24 @xmath43 . although the superconducting gap parameter with line nodes seems to be established , there does not seem to be a robust answer to a question which is realized , a spin singlet or triplet state . we thank y. yanase , m. mochizuki , m. ogata , j. l. gavilano , j. haase , t. imai , and m. takigawa for their fruitful discussions . this study was supported by a grant - in - aid for science research on priority area , `` invention of anomalous quantum materials '' from the ministry of education , science , sports and culture of japan ( grant no . 16076210 ) .
we present our nmr studies of double - layer hydrated cobalt oxides na@xmath0coo@xmath1h@xmath2o ( @xmath30.35 , @xmath41.3 ) with various @xmath5 = 0 @xmath6 4.8 k and magnetic transition temperatures . high - resolution @xmath7h nmr spectrum served as an evidence for the existence of h@xmath8o@xmath9 oxonium ions . @xmath10na nuclear spin - lattice relaxation rates served to detect local field fluctuations sensitive to @xmath5 . @xmath11co nuclear quadrupole resonance ( nqr ) spectra served to classify the various @xmath5 samples . from the classification by @xmath11co nqr frequency , the double - layer hydrated compounds were found to have two superconducting phases closely located to a magnetic phase . in the normal state and at a magnetic field in the @xmath12-plane , two @xmath11co nmr signals with different knight shifts and different @xmath11co nuclear spin - lattice relaxation times @xmath13 were observed . the two @xmath11co nmr signals suggest magnetic disproportionation of two co sites or inplane ( xy ) anisotropy of a single co site . non korringa behavior and power law behavior in zero - field nqr 1/@xmath13 above and below @xmath5 suggest non - fermi liquid and unconventional superconductivity .
the recent announcement of the discovery @xcite of the so called penta - quark states has rekindled interest in the possibility of exotic hadron states . in the last decade , a large number of exotic mesonic states known as the x , y , z states @xcite have been observed by different experimental collaborations , belle , babar , besiii , cdf , cleo , lhcb etc . long ago , a proposal was made to identify the @xmath0 baryon as a possible molecular state of a colourless baryon and meson @xcite . recently this proposal was confirmed through lattice calculations @xcite . while the existence of such exotic states has been theoretically studied over decades @xcite using models of quark confinement with hyperfine interaction , we explore the possibility of identifying states which differ in their underlying dynamics due to different interaction scales responsible for forming composite states such as mesons . this is a model independent analysis which depends entirely on the information already contained in the experimental data on the spectra of composite states . the method itself is not new . it has been in vogue in the study of semiconductors with impurities ( which may have closely spaced electronic spectra ) or in the analysis of spectra of deformed nuclear states for a long time . in principle when the number of states are large , the high temperature behaviour of the specific heat directly yields the information about the relevant degrees of freedom in the spectrum- this is the well known dulong - petit law . this has been effectively put to use in the analysis of the light baryon spectrum by bhaduri and dey @xcite where they have shown that even with truncated spectrum of light quark baryon states , the degrees of freedom of the system may be inferred by comparing the models with the experimental spectrum through the so called schottky peak . however , such an analysis can not be applied to the meson spectrum to infer the degrees of freedom as may be inferred from an analysis of the meson spectrum . on the other hand , when a truncated spectrum is available with no saturation possible , the specific heat at low temperature displays a schottky peak ( or peaks ) which is an indication of the relevant scale ( scales ) in the system . therefore the method we use here , consists of analyzing the heat capacity @xmath1 of a spectrum of states which may contain finitely many states that may not lead to saturation of the specific heat . the existence of schottky peak ( peaks ) is taken as an indication of the presence of an interaction scale ( scales ) and analyzed further . this provides a possible model independent diagnostic of the presence of unusual or exotic states . the only input used is the experimentally measured spectrum of states . in sec.ii , we illustrate the method with a set of assumed ideal spectra closely following the illustrations adopted from ref . @xcite . though the method has been widely used in other fields , it may not be familiar to the practitioners in particle physics phenomenology . in sec.iii we first discuss the charmonium spectrum to illustrate the applicability of the method to the measured spectrum of states using the lists provided in pdg @xcite . charmonium spectrum provides a template for the analysis of other states . further , we discuss other cases including the bottomonium and open - charm / bottom mesons and show that interesting facts emerge by a simple application of the idea of the schottky anomaly . we conclude with a mention of caveats and how the method may be fruitfully used as and when more data becomes available particularly for the conjectured exotic bottomonium states . in order to illustrate the method , consider the simplest case of a two level system with an energy gap given by @xmath2 . the energy gap is an indication of the scale in the hamiltonian . the canonical partition function of the system is simply given by @xmath3 where @xmath4 is the inverse temperature . hereafter we set the boltzmann constant @xmath5 and the temperature is given in energy units . for the purpose of illustration here we have not assumed any occupancy factors . the temperature is introduced here simply as a mathematical parameter to define the partition function of the system and no assumption is made regarding the system being in a heat bath in equilibrium . it is a parameter that is used to calculate the specific heat or more precisely energy fluctuations . the specific heat of the system at constant volume may be defined as @xmath6 = \beta^2[\langle e^2\rangle-\langle e\rangle^2 ] . \label{cvdef}\ ] ] in systems with constant density , we may replace @xmath1 by @xmath1 . substituting for the partition function of the two level system given in eq.([idealz ] ) we have @xmath7 when @xmath1 is plotted against @xmath8 the schottky peak appears at a value @xmath9 with an exponential tail at higher values of @xmath10 . the location of the peak is a function of the energy gap in the system . in general the schottky peak occurs in systems with few energy levels where gaps are a result of a single scale parameter . if however , there is more than one independent scale parameter responsible for the energy levels , then peaks appear whenever the temperature is sufficient to cross the gap signaling a change in the entropy of the system . at high temperatures when all the levels are equally possible , a plateau appears signaling very little change in the entropy . to further expand on this theme , we next consider an ideal spectrum , namely the spectrum of a three - dimensional harmonic oscillator . the partition function of the system is given by @xmath11 where the oscillator parameter @xmath12 defines the scale in the problem and @xmath13 is the degeneracy of the level . in fig . [ fig1 ] , we show the effect of truncation on the specific heat when plotted as a function of @xmath14 with @xmath15 ( say ) . as we include more and more orbitals , the specific heat tends to reach the required saturation , while the peak persists in the truncated spectrum . as we shall see later , in any realistic hadronic spectrum , especially in the heavy quark sector , we do not need more than 2 to 3 orbitals to count the observed spectrum of states . for these cases the location of the schottky peak is close to @xmath16 . in order to simulate a realistic spectra where there is a possibility of more than one energy scale in operation , we combine the spectra of two such systems which differ in @xmath17 significantly . in fig . [ fig2 ] we show the individual spectra separately as well as the spectra for the combined single set . the effect of combining is to normalize the specific heat with a single partition function given by , @xmath18 this is somewhat artificial but nevertheless we use this for the purpose of illustrating the effect of the presence of multiple scales in the spectra when the states are combined in a listing in the absence of any dynamical information . as can be seen from fig . [ fig2 ] the amplitude and location of the two separate schottky peaks depend on truncation apart from the oscillator frequencies . when combined , the peak corresponding to the lower frequency remains unchanged while the one corresponding to higher frequency is sensitive to truncation . this is simply due to the domination of the lower scale in the partition function . + we exploit this sensitivity to different scales in the problem , and apply it in the analysis of experimental spectra of sets of mesons in the next section . different scales may arise from different terms in the same hamiltonian , for example the confinement scale could be very different from the splitting of states with different spins , namely the hyperfine splitting . alternatively the data set could contain states which may arise from different underlying dynamics and hence have different scales . often it may so happen that the data set is missing a certain number of states which also introduces an artificial scale ( or energy gap ) in the analysis . nevertheless , the analysis of schottky peaks in the experimental spectra may reveal some hidden scales on the average . we must however caution that this analysis will not reveal any detailed dynamics ( or the system hamiltonian ) underlying the formation of the bound states . we consider an analysis , based on the template given in the previous section , of the experimental data on mesons . we divide the data sets according to their flavours and start from the simplest cases , where there is no complication arising from the isospin degeneracies . in this analysis we only use the identification given in the particle data group tables without any further theoretical bias . we focus on the heavy quark sector . we first choose to analyze the spectra of the @xmath19 ( charmonium ) states as identified in the pdg @xcite , listed in table [ tab1 ] . .charmonium masses given in mev along with their total angular momentum @xmath20 . other quantum numbers are not needed for this analysis . [ cols="^,^,^,^,^,^",options="header " , ] nevertheless some features are already visible . for example when all the states are plotted together as shown in fig . [ bmes ] , we discern the confinement as well as the hf peak ( at a reduced temperature ) , similar to the open charm scenario . separation into various @xmath20 and @xmath21 lists can not be carried out here since there are only two states each in the @xmath22 , @xmath23 ; @xmath24 , @xmath23 and @xmath24 , @xmath25 sectors . it is not possible to plot the @xmath22 , @xmath23 states against @xmath26 due to different flavour contents . for example , 5366.77 mev state is a @xmath27 ( b and s quark ) state whereas the 6275.6 mev state is a @xmath28 ( b and c quark ) state . for completeness , we may however plot the @xmath24 , @xmath23 and @xmath24 , @xmath25 states each of which has only two states . we have showed the plot in fig . [ bmesj1 ] where again the peaks coincide indicating the confinement scale . more detailed analysis must wait for more data . we have presented a model independent analysis of the data on the meson spectra using the schottky anomaly . given a spectrum of states , the specific heat ( or equivalently energy fluctuation ) when plotted as a function of temperature displays peaks , known as schottky peaks , corresponding to the different scales present in the interaction hamiltonian which gives rise to such states . the peaks are well pronounced especially if the spectrum is truncated with very few orbitals as in the case of meson spectra . the corresponding temperature at which the schottky peaks occur may be converted into energy scales relevant to the spectra on hand . this method is well known in other areas of physics . in this analysis temperature is simply treated as a parameter which is used to extract the scale and should not be confused with the thermodynamic temperature . the information so obtained is nothing new since it is already contained in the spectra , but it provides a new way of looking at the data and analyzing the same . this is especially useful if a given set of states , in the absence of any other information , contains states which may have their origin in different types of interaction hamiltonians , for example the presence of exotics in the quarkonium states . unlike the models of quarkonium states , no dynamical information may be extracted . but the intuition from the potential models or lattice calculations may be used to gain more insight . * the simplest to analyze are the quarkonium states , especially the charmonium states . using this as a template we show how the two main scales corresponding to flavour independent confinement interaction and the hf interaction may be seen from the representation of the data through schottky peaks . typically the average confinement scale ( corresponding to the radial potential in a model ) results in a peak around @xmath29 mev . if we are to interpret this information in terms of the actual scale of confinement we need to invoke a model . if there are only two to three orbitals involved ( as is the case in most of the data ) then a ball park estimate of the confinement scale is given approximately by @xmath30 . it varies very slowly with truncation of the orbitals as seen in the ideal case . the schottky peak corresponding to hf interaction results in a peak around @xmath31 mev which results in an average splitting of about 90 mev as given by potential models . + an analysis of the so called exotic states in the charmonium sector seems to indicate that they are indeed unusual states due to the absence of the usual confinement peak , but exhibit a peak at a lower @xmath26 , about @xmath32 mev , corresponding to a lower @xmath33exotic confinement scale at @xmath34 mev . * our conclusions in the case of bottomonium states are similar to the charmonium states . the hf interaction is much weaker here as it should be if one uses the intuition from potential models . these states also exhibit the confinement peak at around @xmath35 mev . interestingly , the so called exotics again display a peak at about @xmath32 mev , similar to the exotics in the charmonium sector . * the analysis of open charm and open bottom states is more complicated but nevertheless we do get some insight into the scales involved . finally without going into details , we add a few comments on the light quark sector . the light quark sector is the most complicated since here the intuition from the potential models are not as clean as in the case of the heavy quark sector . furthermore , the hf splitting is comparable to the confinement scale . this results in a single broad peak which may be due to a combination of confinement and hf splitting . both in non - strange and strange quark pseudo scalar bound states the confinement scale shifts to a much larger value due to the complication arising from the masses of pion and k - meson . however , when @xmath36 states are analyzed they display the confinement peak as witnessed in all other sectors . this buttresses the well known problem with pions , that they are too light to be simple bound states of quark and an antiquark @xcite . this is also true , to a lesser extent , with the pseudo scalar k - mesons around 495 mev . while we have analyzed the data on meson spectra here , we may analyze the baryon spectra also from this perspective . in some ways this later analysis is likely to provide complimentary information further simplified by the fact that only quarks are involved unless the exotics are considered . this analysis is underway and will be published later @xcite . we thank r. k. bhaduri for many suggestions and careful reading of the manuscript . the ideas outlined in the paper originated during the discussions in the hep journal club . we thank all the members of jc for their helpful comments , especially srihari gopalakrishna , d indumathi and rahul sinha . 78ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevlett.115.072001 [ * * , ( ) ] , for a detailed review , see link:\doibase 10.1007/s11467 - 014 - 0449 - 6 [ * * , ( ) ] , for more detailed experimental references for all the exotic states , see @xcite and the references therein . link:\doibase 10.1103/physrev.153.1617 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0003-4916(60)90001-4 [ * * , ( ) ] link:\doibase 10.1103/physrevd.5.610 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.114.132002 [ * * , ( ) ] link:\doibase 10.1142/s0217751x15300021 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.44.1369 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.71.1411 [ * * , ( ) ] , in http://inspirehep.net/record/1245049/files/arxiv:1307.7425.pdf [ ] ( ) link:\doibase 10.1140/epjc / s10052 - 010 - 1534 - 9 [ * * , ( ) ] , link:\doibase 10.1146/annurev.nucl.58.110707.171145 [ * * , ( ) ] , link:\doibase 10.1016/0370 - 2693(83)91337 - 0 [ * * , ( ) ] link:\doibase 10.1016/0375 - 9474(88)90518 - 0 [ * * , ( ) ] link:\doibase 10.1088/1674 - 1137/38/9/090001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.262001 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevd.71.071103 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevlett.93.072001 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevlett.93.162002 [ * * , ( ) ] , ; link:\doibase 10.1007/jhep04(2013)154 [ * * , ( ) ] , link:\doibase 10.1140/epjc / s10052 - 012 - 1972 - 7 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevlett.98.082001 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.99.182004 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevlett.102.242002 [ * * , ( ) ] , ; link:\doibase 10.1016/j.physletb.2014.05.055 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.100.202001 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.95.142001 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevd.74.091104 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.99.142002 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevlett.98.212001 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.101.172001 [ * * , ( ) ] , first reference in @xcite . link:\doibase 10.1103/physrevlett.110.252001 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevlett.110.252002 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.111.242001 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.78.072004 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevd.85.052003 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.90.112009 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.100.142001 [ * * , ( ) ] , ; link:\doibase 10.1103/physrevlett.112.222002 [ * * , ( ) ] , link:\doibase 10.1103/physrevc.69.055202 [ * * , ( ) ] , link:\doibase 10.1016/j.physletb.2004.11.003 [ * * , ( ) ] , link:\doibase 10.1016/j.physletb.2004.03.033 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.69.114012 [ * * , ( ) ] , link:\doibase 10.1016/j.physletb.2003.11.005 [ * * , ( ) ] , link:\doibase 10.1007/jhep04(2013)154 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.103.162001 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.100.112001 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.108.122001 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.88.052016 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.114.062002 [ * * , ( ) ] , under preparation .
we show that the hadronic heat capacity " calculated as a function of temperature may be used to infer the possible presence of different scales underlying the dynamical structure of hadronic resonances using the phenomenon of schottky anomaly . we first demonstrate this possibility with well known meson spectrum in various channels and comment on the possibility of using this method as a diagnostic to distinguish the exotic states .
surface fit using two - variable orthogonal polynomials has reasonable advantages.@xmath0}$ ] one among many is not necessary to solve a normal equation which is probably ill - conditioned . a second one usually mentioned in literatures is that , the coefficients of orthogonal polynomials in the fitting expression do not depend on those of preceding polynomials . hence , fit with two - variable orthogonal polynomials has been widely used in the field such as image processing . while applying it to magnetization data , another advantage that might be worth mentioning but rarely mentioned is rooted in physics . when pressure in the sample chamber does not change , magnetization of the sample , @xmath1 , as one of the thermodynamic functions , can be completely determined as a function of two independent parameters , magnetic field @xmath2 and temperature @xmath3 . application of this method to magnetization data is not new . for example , a two - step category has been employed to fit the magnetization data of gadolinium , and the fitting expression is subsequently used to estimate the corresponding magnetic entropy change.@xmath4}$ ] the main features of our method are as follow . firstly , in recursively generating orthogonal polynomials with classical or modified gram - schmidt schemes , orthogonality of the polynomials will progressively deteriorate . so an iterating scheme is selected in order to preserve the orthogonality . ( generally speaking , as has been pointed out before , one more orthogonalization process is sufficient to significantly improve the orthogonality . ) secondly , a regularization method is introduced . in previous fittings to magnetization data , the overfitting problem has rarely been taken into account . the reason for this probably originates in that not very - high - order polynomials are included in the final fitting expression . however , our numerical results suggest that fluctuations indeed appear between experimentally recorded data , and such fluctuations become even severer near to the boundaries . so , we are convinced that overfitting occurs and has to be addressed . that s why the regularization scheme is introduced to relieve the probable overfitting . although the specific formulations are different , the present idea of regularization is similar to that used in reference,@xmath5}$ ] which uses gridding and interpolation to obtain the normal equation to be solved . thirdly , a cross - validation scheme is employed to select out the proper regularization parameter . the main idea is to divide the experimentally recorded data into three groups , namely , the training , cross - validation , and test groups . among them , the training group includes the most elements , and the other two have less data . for each of different regularization parameters , use the training group to generate orthogonal polynomials and determine the unknowns in the fitting expression by minimizing training error ; apply the cross - validation group to selecting out the optimal regularization parameter that corresponds to the minimum of cross - validation error ; and assess the applicability of the determined expression by computing test error . for orthogonal polynomials fitting , it s found that the common cross - validation method can not efficiently select out the proper regularization parameter . to overcome this problem , the concept of overfitting degree is introduced and used to monitor fitting performance . fourthly , two methods to implement sampling are provided . in principle , a good fitting expression should not be affected by the sample method to obtain the training group . however , our results show an apparent difference in the cross - validation performance , which is associated with the sampling with respect to magnetic field or temperature . the reason for the different performance is attributed to the ( dimensionless and size - normalized ) averaged increment of recorded magnetic fields unequal to that of recorded temperatures within the experimental measurement range . fifthly , an extended - precision version of the algorithm is used . on completely determining the fitting expression , the magnetization at particular magnetic field and temperature calculated from the orthogonal polynomials , should not be different from the value estimated from the linearly independent functions , which are used to generate those orthogonal polynomials . however , in performing a double - precision version of the algorithm , we noticed in practice that the values obtained by the two methods are actually different . so , during recursively computing the values of linearly independent functions and subsequently generating orthogonal polynomials , the error accumulation is significant . that s why we implement the algorithm at an extended - precision level in spite of more computational time . sixthly , an equally - spaced distribution of both @xmath6 and @xmath7 is not a prerequisite to fit the magnetization data @xmath8 . in principle , our method is also suitable for magnetization data randomly distributed over the @xmath2-@xmath3 plane . finally , after obtaining the fitting expression , it s ready to estimate the magnetocaloric quantities at arbitrary magnetic field and temperature within the measuring range , and comprehensively investigate magnetic - phase - transition properties including the order of transition , critical exponents , anomalous specific heat , and so forth . the rest of this paper is organized as follows : next section provides the general formulations and key algorithms used in this work ; section 3 applies the algorithm to magnetization data ; and conclusions are put in section 4 . following formulations are quite general and not restricted to magnetization data . so the experimental data are denoted by as @xmath9 , @xmath10 and @xmath11 instead of @xmath6 , @xmath12 and @xmath13 . for experimental data @xmath14 with @xmath15 , two - variable orthogonal polynomials @xmath16 are defined as satisfy @xmath17 where , @xmath18 is the kronecker @xmath19-symbol . by using the gram - schmidt orthogonalization process ( classical or modified scheme ) to the linearly independent functions @xmath20 with integer numbers @xmath21 . @xmath16 can be recursively generated as follows @xmath22 in above expression , @xmath23 is the largest index of the generated orthogonal polynomials . it is smaller than or equal to the number of created linear independent function @xmath24 . if we assign @xmath25 , coefficients @xmath26 and @xmath27 @xmath28 are determined as @xmath29 @xmath30 note that the orthogonal polynomials defined above are not normalized . the normalized version can be conveniently obtained through dividing coefficients @xmath27 @xmath31 by @xmath32-norm @xmath33^{1/2}$ ] as @xmath34^{-1 } , \label{e3}\ ] ] @xmath35 after the orthogonal polynomials are obtained , we can expand the fitting expression @xmath36 with @xmath37 as @xmath38 where @xmath39 denotes the maximum index of orthogonal polynomials used in the fitting expression . by minimizing fitting error @xmath40 ^ 2 \label{sigma1}\ ] ] the coefficient of normalized orthogonal polynomials in the fitting expression is determined as @xmath41 using above orthogonalization processes , it s found that orthogonality becomes poorer and poorer . although it s better than the the classical ( cgs ) scheme , performance of the modified gram - schmidt ( mgs ) scheme unavoidably becomes poor with increasing the largest index of orthogonal polynomials , @xmath23 . since the orthogonality is closely related to fitting precision , we use the following iterating scheme ( igs ) to improve the orthogonality . * step 1 * recursively compute the values of linearly independent functions @xmath42 ( @xmath43 ; @xmath44 ) . ( assume that the first @xmath45 polynomials @xmath46 with @xmath47 have been orthonormalized and assigned to @xmath48 with @xmath47 . estimate the value of the @xmath49-th orthonormalized polynomial @xmath50 and assign it to @xmath51 ; and save coefficients @xmath52 with @xmath53 and @xmath54 . ) * step 2 * re - orthogonalize @xmath51 and update @xmath55 . ( 2-a ) compute the modification coefficient @xmath56 , @xmath57 ; ( 2-b ) re - orthogonalize @xmath58 ; ( 2-c ) update coefficients @xmath59 , @xmath60 ; ( 2-d ) judge whether the orthogonality criterion is satisfied . if true then continue ; else go back to ( 2-a ) . * step 3 * normalize @xmath51 and update @xmath55 . ( 3-a ) compute the 2-norm @xmath61 ; ( 3-b ) normalize @xmath51 as @xmath62 ; ( 3-c ) update coefficients @xmath63 ; ( 3-d ) update coefficients @xmath64 , @xmath60 . * step 4 * compute coefficients of @xmath51 in the fitting expression @xmath65 . * step 5 * judge whether fitting precision matches the criterion . if true then @xmath66 and break out the loop ; else continue . * step 6 * subtract the projection of @xmath48 from @xmath51 and update @xmath55 . ( 6-a ) compute the coefficients of subsequent orthogonal polynomials @xmath67 , @xmath68 ; ( 6-b ) subtract the projection of @xmath48 from @xmath51 @xmath69 , @xmath68 . * step 7 * update @xmath45 and go back to * step 2*. if the largest index @xmath23 is not a big number , then the changing tendency of experimental data can not be properly reflected . now fitting error is large and underfitting happens . for decreasing fitting error , more orthogonal polynomials are successively generated and added to the fitting expression until the wished precision is achieved . ( the more orthogonal polynomials , the higher fitting precision . ) however , too many polynomials will lead to strong local fluctuations in the fitted surface , and overfitting happens . the reason for this is that higher - order polynomials generally imply more inflection points . the degree of overfitting can be controlled by regularization like adding so - called penalty functions into the error expression , in order to strengthen the stiffness of fitted surfaces . in contrast to the method of using penalty functions , we implement the regularization by adding a laplace term to the error expression . in essence , the laplace method aims at suppressing the changing rate of curve slope . after adding a laplace term @xmath70 with regularization parameter @xmath71 , the error expression ( [ sigma1 ] ) is rewritten as @xmath72 ^ 2 + \lambda \left [ \nabla^{2 } f(x_{i},y_{i } ) \right]^2.\label{sigma2}\ ] ] it s easy to see that the laplace term in ( [ sigma2 ] ) affects only coefficients @xmath73 . by minimizing ( [ sigma2 ] ) , it is obtained that @xmath74 where , @xmath75 with @xmath76 . we next examine whether the laplace method above really leads to regularization . firstly , when @xmath77 , equation([bt2 ] ) reduces to the non - regularized case ( [ bt1 ] ) . secondly , if @xmath78 , the laplace term has no contribution to @xmath73 since @xmath79 ( corresponding to linear fitting ) . thirdly , the laplace term starts to play its role when @xmath80 . if @xmath71 is large enough so that equations ( [ a1 ] ) and ( [ a2 ] ) are satisfied , @xmath81 then @xmath73 reduces to @xmath82 namely , @xmath83 if we assume that @xmath84 when @xmath85 , then @xmath86 , which implies that @xmath87 since @xmath88 . hence , the laplace term introduced above makes @xmath73 rapidly decay with increasing @xmath89 , so that overfitting is avoided and regularization is realized . now , we invoke a cross - validation process to select out a proper regularization parameter @xmath71 . dividing the whole data into three groups , one of which includes much more data , labelled `` training group '' , and the other two has fewer data , labelled cross - validation group " and test group " , respectively . for each fixed value of @xmath71 , the training group is used to determine coefficient @xmath90 by minimizing @xmath91 in ( [ sigma2 ] ) , and the corresponding training error is computed as @xmath92 ^ 2 . \label{err_tr}\ ] ] subsequently , the cross - validation group selects out the value of @xmath71 that minimizes the cross - validation error @xmath93 ^ 2 . \label{err_cv}\ ] ] finally , the test group assesses the applicability of the determined fitting expression by calculating the test error @xmath94 ^ 2.\ ] ] in contrast to the ordinary scheme used in the field such as machine learning , the model used here has two parameters that require determining , namely , the number of orthogonal polynomials @xmath23 and the regularization parameter @xmath71 . by setting certain routine - terminating criterion , the optimal choice of @xmath23 can be determined by minimizing the training error @xmath95 . a useful criterion can be defined by assessing the changing tendency of fitting error . for example , on increasing @xmath23 from @xmath96 , if the fitting error is not apparently decreased , it is reasonable to consider @xmath96 to be the optimal value of @xmath23 . another task is to determine parameter @xmath71 . practically , we find that both @xmath95 and @xmath97 decrease with enlarging s at fixed @xmath71 or with increasing @xmath71 at fixed @xmath23 . thus , the optimal value of @xmath71 can not be identified to the one that minimizes @xmath97 . this motivates us to construct a new quantity to characterize the degree of overfitting ( and also underfitting ) . typically , when @xmath97 is approximately equal to @xmath95 , underfitting happens ; if @xmath97 is much larger than @xmath95 , overfitting occurs . we can define overfitting degree @xmath98 as @xmath99 it is identified as underfitting if @xmath100 , and overfitting when @xmath101 . it s noticed that the calculated value of fitting error depends on the measurement unit used for experimentally recorded data @xmath102 @xmath103 where @xmath104 $ ] , @xmath105 $ ] , and @xmath106 $ ] . for comparison purposes , we do size normalization as follow @xmath107 where , @xmath108 $ ] . hence , the fitting error is calculated from the after - transformed data . with an inverse transformation , the physical quantities are obtained in the measurement unit . in dividing the experimental data into three groups , a uniform sampling algorithm is executed in order to optimize fitting performance . original experimental data are firstly sorted in terms of a sample parameter , @xmath109 or @xmath110 . then the uniform sample is executed on a pro - rata basis , which is regulated by the sampling factor , and the sorted data are put into the training , cross - validation , and test groups according to the sampling factor . numerical results show that difference appears between different sampling methods , which is attributed to the different data density along @xmath109 axis with that along @xmath110 axis . after coefficients @xmath27 and @xmath90 being determined , the functional value at arbitrary location @xmath112 within the measuring range can be estimated in a similar way to that used in the fitting routine . another method is to estimate from linearly independent functions as @xmath113 where @xmath114 is the coefficient of the @xmath89-th linearly independent functions in the fitting expression , and can be readily calculated from @xmath115 and @xmath73 . the latter scheme is recommended for lower computation cost . practically , in implementing a double - precision version of the algorithm , it s found that the fitting value calculated from linearly independent functions significantly differs from that computed from orthogonal polynomials , when the power exponent of the fitting expression is very high . after an extended - precision algorithm is applied , the difference decreases . if the extended - precision operation is also used to recursively generate linearly independent functions , the difference is not longer obvious . these facts suggest that round - off error is rapidly accumulated while recursively computing the linearly independent functions . hence , it is required to carefully consider the influence of error accumulations in fitting using two - variable orthogonal polynomials . since the fitting expression is essentially the linear combination linearly independent functions @xmath116 , utilizing recursive properties of the corresponding partial derivative and integral , it s quite convenient to compute the physical quantities of interest . followings are typical recursive algorithms at given @xmath117 and @xmath118 used in this work . @xmath120 @xmath122 @xmath124 the formula given above are quite general . for application to magnetization data , it s only needed to replace @xmath109 , @xmath110 , @xmath125 and their corresponding size normalizations with @xmath2 , @xmath3 , @xmath1 and @xmath126 , @xmath127 , @xmath128 . here , we fit the magnetization data of polycrystalline samples la@xmath129sr@xmath130mn@xmath131o@xmath132 obtained with physical property measurement system ( ppms ) of quantum design company . more details can be found in reference . @xmath133}$ ] the total number of data used is @xmath134 . with one half of the data ( @xmath135 ) uses as the training group , the fitting results are shown in fig . 1 . note that satisfactory fitting precision can be reached using two - variable orthogonal polynomials . however , one can not assess the degree of overfitting from the fitting precision , which suggests the necessity to introduce cross - validation . the whole data are uniformly sampled according to temperature and the sampling factor is set to @xmath136 , namely , two thirds of the data being put into the training group , half of the rest one third into the cross - validation group and half into the test group . by setting regularization parameter @xmath77 , we discuss the effect of the number of orthogonal polynomials ( @xmath23 ) on fitting performance . the calculated results are shown in table 1 . for reference , the overfitting degree corresponding to the test error is defined in a similar way with that in ( [ overfit1 ] ) @xmath137 it noted that both @xmath98 and @xmath138 reflect fitting performance , since the data in the cross - validation and test groups can be interchanged . with increasing @xmath23 , fitting error decreases and overfitting degree increases , which suggests that the overfitting degree can be used to monitor fitting performance , although it can not always select out the optimal regularization parameter . fitting errors and overfitting degrees without regularization ( @xmath139 ) . for comparison , the training error with @xmath140 ( i.e. , only @xmath141 is used ) is @xmath142 . [ cols="^,^,^,^,^,^",options="header " , ] table 3 summarizes the optimal fitting precision and corresponding overfitting degrees with different regularization parameter @xmath71 . the sampling factor is the same to preceding tables . it s noted that overfitting begins to occur at @xmath143 . since the cross - validation and test group can interchange data , one needs to compare @xmath98 with @xmath138 in order to select out the proper @xmath71 . here we discuss aspects that have not been mentioned above . first of all , the algorithm in this paper can be further optimized so that the computing efficiency is increased and memory decreased . for example , in the re - orthogonalization step of iterating orthogonalization , the projection of the just orthogonalized polynomial is subtracted from all those subsequent polynomials which are not orthogonalized yet ; this definitely increases the the computing cost since redundant polynomials are generated to ensure the fitting precision . secondly , if only physical quantities that are at experimental - recorded magnetic fields and temperatures are concerned , the computing results for the linearly independent functions and orthogonal polynomials used to fit can be saved to compute these physical quantities . thirdly , after the fitting expression obtained , uniform gridding and interpolation of the data can be readily achieved . thus , besides the iterative method used in this work , operations like numerical derivatives and integrals can be easily executed . fourthly , when the coefficients of linear independent functions in orthogonal polynomials and those of the orthogonal polynomials in the fitting expression are computed , we have not considered the effect of measuring errors . however , the measuring error is always there . what s more , the measuring error of experimental data at extremely weak field is usually bigger . because we have taken into accounted all experimental data with equal weight , the measuring error under weak field will do harm to global fitting performance . to solve this problem , one method is to abandon the data under weak field ; another method is to introduce a local weight factor that depends on magnetic field and temperature so that the influence of experimental data is confined within a nearby area . fifthly , it s noted that in above fitting , rotational symmetry of magnetic systems has not been considered . because of this symmetry , the terms involving even - number powers of magnetic field should not appear in the analytic expression of magnetization . while talking about the properties closely associated with this symmetry , it needs eliminating such terms in the set of linearly independent functions . finally , since the algorithm considers only the dependence on magnetic field and temperature , the effect of other processes such as rotation of crystal grains and the change of sample volume are not adequate disclosed , that may be the reason for larger fitting error at weak fields . another issue not considered here is the influence of demagnetization factor whose value in polycrystalline materials is hard computed . to conclude , by adding the laplace term into the fitting error expression , the regularization method and corresponding cross - validation scheme are introduced to two - variable orthogonal polynomials fitting . after applying to magnetization data , it s found that the regularization scheme does play its role through rapidly suppressing the coefficients of higher - order terms in the fitting expression and therefore effectively relieving the overfitting problem . with the aid of the concept of overfitting degree , it s also shown that the cross validation scheme can be used to select out the proper regularization parameter . the influences of sampling parameter and sampling factor are also analysed . thus it offers the quite reliable base for the following investigations of the magnetic - entropy - change and phase - transition properties of magnetic functional materials . i would thank wu hong - ye for many helps in software usage and constructive discussions in developing the method in this work . and my thanks also give to wu ke - han and zhou min for providing experimental data before their paper published . leon s j , bjrck and gander w 2012 _ numer . linear algebra appl . _ * 20 * 492 - 532 li z t , wu p f , tao y q and mao d k 1999 _ acta phys . sin . _ * 48(s ) * s126 ( in chinese ) derrico j r 2006 _ understanding gridfit_. available : + http://www.mathworks.com/matlabcentral/fileexchange/8998-surface-fitting-using-gridfit . wu k h , wan s l , xu b , liu s b , zhao j j and lu y , to be published .
an obstacle encountered in applying orthogonal - polynomials fitting is how to select out the proper fitting expression . by adding a laplace term to the error expression and introducing the concept of overfitting degree , a regularization and corresponding cross validation scheme is proposed for two - variable polynomials fitting . while the fortran implementation of above scheme is applied to magnetization data , a satisfactory fitting precision is reached , and overfitting problem can be quantitatively assessed , which therefore offers the quite reliable base for future comprehensive investigations of magnetocaloric and phase - transition properties of magnetic functional materials .
let @xmath0 be some function from the positive integers @xmath1 into @xmath2 , and let @xmath3 be a graph and @xmath4 some vertex of it . then we define _ reinforced random walk _ on @xmath3 starting at @xmath4 with reinforcement function @xmath0 as follows . it is a walk on @xmath3 i.e. a random sequence of vertices @xmath5 with @xmath6 a neighbour of @xmath7 ; and the transition probabilities depend only on the past . specifically , for an edge @xmath8 and time @xmath9 we define the number of traversals @xmath10 using @xmath11 where @xmath12 is an unordered edge , so @xmath13 counts traversals of @xmath8 in both directions . now apply @xmath0 and normalize to get probabilities . totally , the definition is @xmath14 where @xmath15 is the neighbourhood relation in our graph . this finishes the definition of the process . this is well defined assuming all degrees of all vertices are finite , and we assume this from now on on all our graphs . there are also versions where reinforcement is applied to vertices or to directed edges , but for simplicity we concentrate for now on edge reinforced walks . it comes as no surprise that different @xmath0 give different processes , sometimes dramatically so . to make things more concrete let us discuss a few examples . 1 . once - reinforced random walk is the process defined by @xmath16 in other words , edges which have already been traversed are given double weight when compared to unknown edges , but the number of traversals does not matter . 2 . linearly reinforced random walk is the process defined by @xmath17 . in other words , whenever you traverse an edge , you increase its weight by @xmath18 . 3 . strongly reinforced random walk , where @xmath0 grows faster than linear . for example , one may take @xmath19 . perhaps surprisingly , the order of difficulty seems to be decreasing . once - reinforced random walk is the least understood , while the superlinear case is the simplest . let us start with the superlinear case . it turns out that in this case the walk gets stuck on a single edge , going back and forth endlessly . to understand why , let us prove the following weaker claim : let @xmath3 be an arbitrary graph , @xmath4 a vertex of @xmath3 and @xmath8 an edge going out of @xmath20 . examine reinforced random walk on @xmath3 starting from @xmath4 with reinforcement function @xmath21 . then with positive probability the process goes back and forth over @xmath8 endlessly i.e. for all @xmath9 , @xmath22 and @xmath23 is the other vertex of @xmath8 . denote the other vertex of @xmath8 by @xmath24 . let @xmath25 be the event that @xmath22 and @xmath26 for all @xmath27 and @xmath28 , respectively . let us calculate the probability of @xmath29 given @xmath25 . the numbers of traversals are @xmath30 and @xmath31 for any edge @xmath32 . hence the probability is @xmath33 for some constant @xmath34 that depends only on the degrees of @xmath4 and @xmath24 . therefore the probability of @xmath25 is @xmath35 for some positive @xmath36 . this proves the claim . there is no zero - one law for the behaviour of processes like reinforced random walk , so one can not easily conclude from the theorem above that some event of interest happens with probability 1 . nevertheless , this has been proved with almost complete generality . let @xmath3 be a graph and @xmath0 a function satisfying @xmath37 then reinforced random walk on @xmath3 with reinforcement function @xmath0 eventually gets stuck on a single edge with probability @xmath18 , under either of the following conditions 1 . @xmath0 is increasing , or 2 . @xmath3 contains no cycle of odd order . see limic @xcite and limic and tarrs @xcite . it is conjectured that neither condition is necessary , but clearly , the result is quite complete as is . let us now consider once - reinforced random walk , but let us add a parameter @xmath38 and examine the reinforcement function @xmath39 an exciting conjecture due to beffara & sidoravicius is that , say for the graph @xmath40 , @xmath41 , there is a _ phase transition _ in @xmath38 . for @xmath38 very small the walk is essentially simple random walk . for @xmath38 very large the walk looks like a man trapped in a balloon : it constantly jumps against the boundaries of the balloon , occasionally inflating it , but the balloon keeps its round shape , as it inflates . the `` balloon '' , of course , is the set of visited sites . in particular for @xmath41 there should be a phase transition between recurrence and transience . let us stress that @xmath38 does not depend on @xmath9 . it is not difficult to establish results if one first fixes @xmath9 and then makes @xmath38 very small or very large . in this case for @xmath42 one can directly check that the process is identical to simple random walk ( with high probability ) , while for @xmath43 the process gets stuck on its first edge . but the conjecture is that for a fixed @xmath38 there is a marked difference in behaviour as @xmath44 . some simulation results can be seen in figure [ fig : once ] . = 0.48 once - reinforced random walk , scaledwidth=100.0% ] @xmath45 once - reinforced random walk , scaledwidth=100.0% ] @xmath46 once - reinforced random walk , scaledwidth=100.0% ] @xmath47 once - reinforced random walk , scaledwidth=100.0% ] @xmath48 once - reinforced random walk , scaledwidth=100.0% ] @xmath49 once - reinforced random walk , scaledwidth=100.0% ] @xmath50 each simulation was done on a @xmath51 grid and stopped when the process reached the boundaries . the 2-dimensional case was chosen since it is easiest to depict , but one should probably caution that in fact it is not clear what is the true behaviour in 2 dimensions , even on a conjectural level . colour depicts the number of reinforced edge . this means that the second edge which is reinforced is given colour @xmath52 , not the second edge which is traversed ( colour change rate is not identical in the 6 images ) . the fractal - like colour patterns seen in @xmath47 are indicative of the process leaving holes all around and then going back to fill them at a macroscopically later time . presumably @xmath53 is quite close to the point of criticality where the processes of roughening and smoothing balance exactly . the case of large @xmath38 in this conjecture is related to a series of similar models . internal diffusion - limited aggregation is a process where the walker , upon reaching a new vertex , is transported to the starting point @xmath4 . for this model it is known since the early 90s that the set of visited vertices is spherical @xcite , with impressive improvements in the precision of the results achieved lately @xcite . a variation where , when the walker , upon reaching a new vertex is transported to a random vertex in the existing cluster also gives a ball @xcite . getting closer to once - reinforced random walk we may examine _ excited to the center_. in this model the walker , upon getting to a new vertex gets a small drift towards the center . less is known about this problem , basically only that the process is recurrent in any dimension , regardless of the strength of the drift [ unpublished , 2006 ] . once - reinforced has all the difficulties of excited to the center and more : there is no preferred direction and there is a phase transition . hence it seems quite challenging at this point . existing rigorous results on the model are on a tree @xcite , and it is not difficult to solve the one - dimensional case , but neither case exhibits a phase transition , so for now this phenomenon is beyond reach . let us close this comparison by returning to the simulation results above . as can be seen , the shapes for @xmath48 or @xmath54 are not that round . this is unlike internal diffusion - limited aggregation where even quite small simulations are very close to a ball . a similar phenomenon can be seen in simulations of excited to the center done by yadin . this is simply indicative that the boundary smoothing mechanism is far less efficient in these models . we now move to the main topic of this talk , linearly reinforced random walk , lrrw for short . naively one might conjecture that , since @xmath55 diverges so slowly , maybe the behaviour of lrrw is not that different from the behaviour of strongly reinforced random walk . let us start with a simple example that shows that this is not at all the case . the example will be a graph with @xmath56 vertices connected in a line , and @xmath4 being the middle vertex . the reinforcement function will be @xmath57 . with this weight , the process is equivalent to the classic plya urn . recall that in the plya urn model one repeatedly takes out a ball , and then returns two balls of the same colour as the ball you took out . looking at the weights of the lrrw at time @xmath58 ( i.e. the values of @xmath59 ) , and dividing them by @xmath52 , one gets a plya urn . let us recall plya s theorem @xcite=10.4 cm = 2.2 cm 8 0pt [ thm : polya]at time @xmath9 ( i.e. when the urn contains @xmath60 balls ) , the probability that it contains @xmath61 black balls is exactly @xmath62 , for any value of @xmath61 between @xmath18 and @xmath9 . assume inductively that the claim has been proved for @xmath63 . now , to get @xmath61 black balls at time @xmath9 you need one of the following : 1 . have @xmath64 black balls at time @xmath63 and take a black ball out of the bin ; or 2 . have @xmath61 black balls at time @xmath63 and take a white ball out of the bin . the probability of ( 1 ) is @xmath65 where the term @xmath66 is our inductive assumption , and the term @xmath67 is the probability to pick a black ball when you have @xmath64 black balls out of a total of @xmath9 balls . this formula also holds for @xmath68 , though for a different reason : here the probability is zero since @xmath69 black balls are not allowed , and the formula indeed gives @xmath69 . a similar calculation shows that the probability of ( 2 ) is @xmath70 summing ( 1 ) and ( 2 ) gives @xmath71 and the induction is complete . thus we see that the picture is very different from that of strongly reinforced random walk . looking at strongly reinforced random walk at some very large time would give that with probability very close to @xmath18 , the walk visited one of the edges a very small number of times , and the other edge all the rest ( from symmetry each one gets probability close to @xmath72 ) . here this is not the case , typically each edge is visited a number of times proportional to @xmath9 . on the other hand , this is also very different from what we expect for , say , once - reinforced random walk . it is not difficult to analyze this explicitly and see that once - reinforced would have the number of visits to each edge at time @xmath9 being @xmath73 , with the random errors being of order @xmath74 . so we see that the linear case is quite special . 0 the analogy to plya s urn works in a number of other cases . examine the graph @xmath75 . examine a single vertex . since there are no cycles , you know that when you leave your vertex through some edge @xmath8 , you are bound to return through @xmath8 , if you return at all . the other vertices can only effect you through preventing return , not in any other way . thus you can couple lrrw on @xmath75 to a sequence of i.i.d . plya urns . the same holds for any tree , except that you get multicoloured plya urns . but these are also well understood . using these ideas , pemantle @xcite was able to analyze lrrw on regular trees . he showed that there is a phase transition in the initial weight @xmath38 [ thm : pmntl]let @xmath3 be a regular tree of degree @xmath76 , and let @xmath77 be the reinforcement function . then there exists an @xmath78 such that for @xmath79 lrrw is ( strongly ) recurrent , while for @xmath80 it is transient . here strong recurrence means that the expected time to return to your starting point is finite . this is significantly stronger than usual recurrence : infinite graphs ( like @xmath75 or @xmath81 ) are recurrent but not strongly so ( for @xmath75 this is the well - known `` gambler s ruin '' theorem ) . in a way it means that the process behaves like random walk on a _ finite _ graph again , compare to the case of once - reinforced random walk on a tree which is transient for any value of the parameter @xcite . theorem [ thm : pmntl ] shows that importance of the parameter @xmath38 , and from now on we will simply say `` lrrw with @xmath38 '' rather than the cumbersome `` lrrw with reinforcement function @xmath77 '' . note that the parameter @xmath38 has the opposite effect compared to once - reinforced random walk : there small @xmath38 gave random - walk - like behaviour , while here it is the large @xmath38 that give this . another case where the theory of urns is applicable is _ directed _ reinforced random walk . one modifies the definition of @xmath13 to count directed traversals i.e. @xmath82 and then again define @xmath83 by taking @xmath84 and normalizing it . here it is not important what graph one looks at , each vertex can be isolated and treated as a multicoloured urn independent of all others . nevertheless , this urn argument only reduces the model to a kind of _ random walk in random environment _ , a notorious problem in its own right . this was tackled on the graphs @xmath40 by enriquez and sabot @xcite . interestingly , there is no phase transition when @xmath85 , the walk is transient for any choice of @xmath38 . we will return to plya s urn later , but for now let us go back to general graphs . another way to understand why lrrw is special is that it has a property called _ partial exchangeability_. this means that for a given path , its probability depends on how many times each edge is visited , but not on the order in which these visits happen . figure [ fig : partial - exchangeability ] shows two such paths in a triangle . both paths have length 4 , and both cross one edge twice and the other two edges once , but are different . the numbers in the figure are the probability of each traversal for lrrw with @xmath86 . the reader can verify that the numbers are not the same , but their product is @xmath87 indeed the numerators on the left are a permutation of the numerators on the right ; and ditto for the denominators . the general proof is the same : lrrw is partially exchangeable . we may assume without loss of generality that the reinforcement function is @xmath88 for some @xmath89 . let @xmath90 be a path in the graph @xmath3 . then the probability @xmath91 that @xmath92 is the beginning of the lrrw path is @xmath93 where @xmath94 is of course the analog of @xmath13 above for the path @xmath90 i.e. @xmath95 . now , if @xmath96 for some @xmath97 , then the numerators contain the terms @xmath38 , @xmath98 at the @xmath97 places where @xmath8 is an edge of @xmath90 . this is not specific to linear reinforcement . now assume that the vertex @xmath99 appears @xmath100 times in the path @xmath101 ( note that we do not count @xmath102 ) . then the denominators contain @xmath103 , @xmath104 at the @xmath100 places where @xmath105 . for @xmath106 you get a similar picture , except the denominators contain @xmath107 . in other words it does not matter at which order did @xmath90 traverse the edges coming out of @xmath108 , since each such traversal increases the total weight of the edges surrounding @xmath108 by @xmath18 ( and another @xmath18 is added when returning to the vertex ) . this is the point which is special to linear reinforcement . hence , if you know @xmath109 for all @xmath8 you can reconstruct both numerators and denominators up to a permutation ( this requires the observation that you can identify the last vertex of @xmath90 , since it is the only one with odd total number of appearances , or @xmath4 if no vertex has this property ) . thus you can calculate @xmath91 from @xmath94 , and the process is partially exchangeable . partial exchangeability is interesting because of the following theorem of diaconis & freedman @xcite let @xmath110 be a random sequence of elements in some state space , which is recurrent . then @xmath20 is partially exchangeable if and only if it is a mixture of markov chains . here recurrent means that the sequence returns to its starting vertex with probability @xmath18 . this has to be specified as for general random sequences the various definitions of recurrence are not identical . in fact , the counterexamples in @xcite which show that recurrence is necessary consist of two states and a process that gets stuck in the second state . a mixture of markov chains means that there is some measure @xmath111 on the space of markov chains such that the probability of every event @xmath112 is given by @xmath113 where @xmath114 is a markov chain and @xmath115 is the probability of @xmath112 under the markov chain @xmath114 . the same concept is known under the name of random walk in random environment . again , you first pick an environment @xmath114 randomly , then perform random walk on it . we already mentioned random walk in random environment when we discussed the directed case above . but there each vertex was independent , while in the general case this is typically not the case . we will not prove the theorem , but just make some remarks on its proof . recurrence is used to break the infinite process into blocks , from one visit of @xmath4 to the next . these blocks then turn out to be _ fully _ exchangeable , i.e. any permutation has the same probability . this allows to apply de finetti s theorem which states that such a process is a mixture of i.i.d . processes . de finetti s theorem itself is not difficult either , it is an application of krein - milman . the point to note here is that the proof is soft and abstract , without any explicit calculations . let us demonstrate the utility of this fact by applying it to plya s urn , i.e. to lrrw on a line of length @xmath56 with @xmath116 . there is no question as to recurrence , since our walker returns to its starting point every 2 steps without failure . our random environment can only relate to the central vertex , so asking `` what is the random environment ? '' is equivalent to asking `` what is the probability to go left from the central vertex ? '' . in other words , there is only one parameter . let us denote it by @xmath117 . [ thm : polyaunif]the parameter @xmath117 defined above is uniform in @xmath118 $ ] . fix @xmath117 and examine the markov chain on a line of length @xmath56 which moves from the central vertex to the left vertex with probability @xmath117 and to the right vertex with probability @xmath119 . denote by @xmath120 the number of times the left edge was crossed by time @xmath9 . by the law of large numbers we know that @xmath121 the random walk in random environment tells us that for the measure @xmath111 that we seek , @xmath122 this holds for every measurable @xmath112 , but let us restrict our attention to intervals . by the explicit solution of plya s urn ( theorem [ thm : polya ] above ) , the left hand side is @xmath123 where @xmath124 is the lebesgue measure of @xmath112 . taking @xmath44 ( which can be done inside the integral as @xmath111 is a probability measure and the integrand is bounded ) gives @xmath125 since this holds for all intervals @xmath112 , we get that @xmath111 is uniform , as needed . in other words , the theorem of diaconis and freedman allows us to get from the `` static '' description of the plya urn we had before , a `` dynamic '' description which is fuller . we now know that plya s urn is equivalent , _ as a process _ , to choosing some number @xmath126 $ ] uniformly and then picking each time a black ball with probability @xmath117 and a white ball with probability @xmath119 . in general graphs it proved to be a significant challenge to find an explicit formula for the environment , and this is what we turn to next . let us consider for a moment finite graphs . even a triangle would prove to be highly nontrivial ! lrrw is recurrent on any finite graph this is not difficult to show , but we will skip it here . the theorem of diaconis & freedman promises us that _ some _ random environment exists such that lrrw is equivalent to random walk in the environment . but what is it ? it turns out there is a formula for it , known fondly as the `` magic formula '' . the history of its discovery is complicated and the reader would be best served by reading about it in detail in @xcite . it turns out that the environment is always _ reversible _ ( a fact which does not follow from diaconis & freedman ) . this means that there is a ( random ) weight function @xmath127 where @xmath112 is the set of edges of @xmath3 such that the probability to go from @xmath20 to @xmath24 is @xmath128 it is common to denote the denominator by @xmath129 and we shall do so below . reversibility is perhaps the most important property a markov chain may or may not possess . most notably a reversible markov chain gives rise to a _ self - adjoint _ operator ( on @xmath130 with the weight being @xmath114 ) which is amenably to spectral analysis . now , @xmath114 is not unique since multiplying @xmath114 by a constant gives rise to exactly the same markov chain . so we need to normalize and we normalize so that @xmath131 . for completeness , let us state the magic formula for non - constant initial weights , i.e. assume for every edge @xmath8 there is some @xmath132 and the transition probabilities at time @xmath9 are @xmath133 , normalized . the magic formula is then : let @xmath134 be a finite graph . examine lrrw on @xmath3 , starting from @xmath4 with initial weights @xmath132 , and let @xmath111 be the corresponding random environment . then the density of @xmath111 at a given @xmath114 is given by @xmath135 where the sum inside the square root is a sum over all spanning trees of @xmath3 ; where @xmath136 is a normalization constant ; and where @xmath137 . the most problematic term is the sum over trees , as it introduces long - range dependencies . by the matrix - tree theorem this sum can be written as a determinant , but that does not simplify the formula significantly . as an exercise , let us analyse plya s urn again : [ second proof of theorem [ thm : polyaunif ] ] we use the magic formula . all @xmath138 , @xmath139 and for @xmath99 ( i.e. one of the side vertices ) , @xmath140 . since the normalization is @xmath141 , we have @xmath142 . denote the weight of the left edge by @xmath117 , so for one of the side vertices @xmath143 while for the other it is @xmath119 . all in all this gives @xmath144 the expression inside the square root is so simple because a line graph has exactly one spanning tree ( the graph itself ) , so the sum is only over one term . everything cancels and we get that the density is @xmath145 , which of course means that @xmath146 , proving the theorem . throughout the last decade the magic formula stood in the center of research on lrrw . even the case of the ladder was not so easy , and was done in @xcite . the next result achieved was a tightness result that allowed to show that lrrw is always equivalent to a random walk in random environment , even in the transient case ( which is not covered by diaconis & freedman ) @xcite . the most impressive result , also due to merkl & rolles @xcite , was the proof that a certain two - dimensional graph is recurrent . a conjecture that goes back to the 80s @xcite is that lrrw on @xmath81 is recurrent , at least for @xmath38 sufficiently small . merkl & rolles got quite close , but their proofs required to replace each edge by a line of length 130 or more . around the same period a surprising connection to certain quantum models appeared , a topic we will discuss next . the hyperbolic sigma model has its roots in physics . as some of the readers ( and , frankly , the author ) may not be familiar with the necessary background , we will be short and vague . hopefully the description will still be useful to some . our starting point is supersymmetry . supersymmetry started its life as a branch of particle physics that postulated that each of the known particles has a partner ( a `` superpartner '' ) which is similar in most aspects , but the superpartner of a fermion is a boson and vice versa . the existence of these particles leads to cancellations which would solve several theoretical problems with the current model . these superpartners were never found in particle accelerators , but the mathematics of these cancellations turned out to be very fruitful . one potential application of supersymmetry to problems which have nothing to do with bosons or fermions is the wegner - efetov approach to random matrices and to certain disordered quantum systems @xcite . the titular hyperbolic sigma model captures some of the mathematical difficulties of this approach , and , conjecturally , most of the phenomenology . it has been studied by disertori , spencer & zirnbauer @xcite . the model , roughly , is as follows . we are looking at functions @xmath147 from the vertices of our graph ( say @xmath40 ) into hyperbolic space . the action ( or energy ) @xmath148 is a sum over the edges of some interaction _ which respects the symmetries of the hyperbolic space_. then each @xmath147 is given probability density proportional to @xmath149 where @xmath150 is a parameter which is analogous to the initial weights @xmath38 in lrrw . the approach taken in @xcite to tackle the hyperbolic sigma model is to make a change of variable in the hyperbolic plane ( change to the so - called horospherical coordinates ) and then integrate some of the variables . the result was a sigma model with values in @xmath151 but with a much more complicated interaction term , which included a determinant very similar to the one appearing in the magic formula . this caused significant curiosity but for a few years it was not clear whether the similarity is coincidental or not . this mystery was solved in november 2011 . it was sabot and tarrs who managed to make the connection between lrrw and the hyperbolic sigma model rigorous @xcite . they investigated a variation on lrrw called the vertex - reinforced jump process , defined in continuous time . on the one hand the vertex - reinforced jump process has a random walk in random environment representation with the density of each environment _ identical _ to the formulas that appear in the hyperbolic sigma model . this allowed them to harness the results of @xcite to show that the vertex - reinforced jump process is recurrent for small @xmath38 in any graph ( the jump process has a parameter which acts analogously to @xmath38 for lrrw ) . on the other hand , they showed that lrrw is given by a vertex - reinforced jump process with random weights . thus , after some modification of the techniques of @xcite to work with different , unbounded @xmath150 in different edges , they were able to show : for every graph @xmath3 there exists some @xmath152 such that lrrw is recurrent for all @xmath153 . simultaneously , with angel and crawford , we arrived at the same result @xcite . our approach was completely different and did not use the magic formula at all . the basic idea is very simple . [ basic idea of the proof]examine some edge @xmath154 and assume you have traversed @xmath8 for the first time from @xmath20 to @xmath24 . let @xmath0 be the edge through which you arrived into @xmath20 . if @xmath38 is extremely small , then upon arriving to @xmath20 you see one edge , @xmath0 , with weight @xmath155 , and all the others with weight @xmath38 . hence you are very likely to return via @xmath0 . this will repeat on the second , third etc . visit to @xmath20 . at each time , because @xmath0 has much larger weight than @xmath8 ( its weight is even increasing , but we will not use that ) , it will choose to exit @xmath20 through @xmath0 multiple times until finally exiting through @xmath8 . denote the number of times it exited through @xmath0 before the first exit through @xmath8 by @xmath13 . we now recall that the process is also a random walk in random environment . the environment might be complicated , but we have earned a bit of information about it . we know that the walker exited the vertex @xmath20 @xmath13 times through @xmath0 before the first time it exited through @xmath8 . this must mean that the weight of @xmath0 is approximately @xmath13 times larger than that of @xmath8 ! in other words , we see that the weight of edges decays exponentially in the `` first entrance distance '' , i.e. if we define @xmath156 as the edge through which we entered the other vertex of @xmath0 , @xmath157 as the edge through which we entered the other vertex of @xmath156 etc . , we finally reach @xmath158 . then the argument says that the weight of @xmath8 should decay exponentially in @xmath159 . comparing to the approach of disertori - spencer - sabot - tarrs , our approach is much softer , and has almost no explicit calculations . we rely on the existence of a random walk in random environment representation , i.e. on the theorem of diaconis & freedman , but that too , as explained above , is a soft result . trying to formalize this leads to some minor technical difficulties . the first is that , to get exponential decay , different edge ratios need to be independent . however , different edges are not independent , because the `` first entrance path '' is a global property which introduces dependencies between all quantities . to tackle this difficulty we use a somewhat brutish approach and simply count over all such paths , i.e. write @xmath160 this allows to just fix two edges @xmath8 and @xmath0 sharing a vertex @xmath20 , and define a variable as in the basic idea . these variables ( call them @xmath161 ) are independent and have expectation @xmath162 . thus we argue , very roughly , as follows @xmath163 the first equality is because under the event that @xmath90 really is the first entrance path , @xmath164 is really the product of the @xmath161 along @xmath90 . the inequality that follows comes from just throwing this event away . this seems very wasteful , but look at the final result : we show that the expectation is exponentially small , and not only that , we can make the exponent base as small as we want by reducing @xmath38 . so losing a fixed exponential term by counting over the @xmath90 is something that our approach can sustain . the third step in ( [ eq : wr ] ) is simply independence , and the fourth is the estimate that came from the `` basic idea '' sketch above . another technical difficulty that might be worth mentioning is that , even though the weights typically decrease along the first entrance path , sometimes they increase . once you incorporate this phenomenon into the basic sketch , you see that the @xmath161 , despite being typically @xmath165 , in fact have infinite expectation . we solve this by taking everything to some small fixed power ( @xmath166 worked ) . thus ( [ eq : wr ] ) as written is false , and what we actually prove is @xmath167 returning to our favourite example , plya s urn , we see that the phenomenon is real : the ratio between the weights of the two edges has no first moment . so some power must appear in the formulation of the result , though @xmath166 is not optimal . with these two problems solved , we were able to show the exponential decay of weights . let us state this formally : let @xmath3 be a graph with all degrees bounded by @xmath168 . then there exists @xmath169 such that for all @xmath170 , @xmath171 the next step to understand the phase transition is to understand the transient regime . using our techniques we were able to show the existence of a transient regime on any non - amenable graph . but getting to @xmath40 seems to require a new idea . next one would like to show that the phase transition is unique , namely that there is some @xmath78 such where the process moves from recurrence to transience , and not , say , that the process switches from recurrence to transience and back a few times before finally becoming transient for high enough @xmath38 . in many models proving this kind of fact requires monotonicity , but there are also other approaches , so e.g. @xcite . there are also many other models we would like to know similar results . we already covered other reinforced walks , but the connection to the hyperbolic sigma models suggests many more questions . are there other sigma models where one can get an interacting representation for random walk on the generated field ? all - in - all this is an exciting and fast - moving field , with a lot more to understand . margherita disertori and thomas spencer , _ anderson localization for a supersymmetric sigma model . phys . 300:3 ( 2010 ) , 659671 . available at : http://www.springerlink.com/content/m6583504661746lj/ [ ] , http://arxiv.org/abs/0910.3325 [ ] margherita disertori , thomas spencer and martin r. zirnbauer , _ quasi - diffusion in a 3d supersymmetric hyperbolic sigma model . phys . 300:2 ( 2010 ) , 435486 . available at : http://www.springerlink.com/content/61xj2m4860u40844/ [ ] rick durrett , harry kesten and vlada limic , _ once edge - reinforced random walk on a tree_. probab . theory related fields 122:4 ( 2002 ) , 567592 . available at : http://www.springerlink.com/content/xw5b9462m2kwtf6j/ [ ] nathanal enriquez and christophe sabot , _ edge oriented reinforced random walks and rwre_. c. r. math . paris 335:11 ( 2002 ) , 941946 . available from : http://www.sciencedirect.com/science/article/pii/s1631073x02025803 [ ] , http://www.proba.jussieu.fr/pageperso/enriquez/pub/cras.pdf [ ] david jerison , lionel levine and scott sheffield , _ logarithmic fluctuations for internal dla_. j. amer . 25:1 ( 2012 ) , 271301 . available at : http://dx.doi.org/10.1090/s0894-0347-2011-00716-9 [ ] , http://arxiv.org/abs/1010.2483 [ ] vlada limic and pierre tarrs , _ what is the difference between a square and a triangle ? _ in and out of equilibrium . 2 , 481495 , progr . probab . , 60 , birkhuser , basel , 2008 . available at : http://arxiv.org/abs/0712.0958 [ ] franz merkl , aniko ry and silke w. w. rolles , _ the `` magic formula '' for linearly edge - reinforced random walks_. statistica neerlandica 62:3 ( 2008 ) , 345363 . available at : http://onlinelibrary.wiley.com/doi/10.1111/j.1467-9574.2008.00402.x/abstract [ ] , http://www-m5.ma.tum.de/foswiki/pub/m5/allgemeines/silkerollespublications/magic-snversion.pdf [ ] franz merkl and silke w. w. rolles , _ recurrence of edge - reinforced random walk on a two - dimensional graph_. ann . 37:5 ( 2009 ) , 16791714 . available at : http://projecteuclid.org/euclid.aop/1253539854 [ ] , http://arxiv.org/abs/math/0703027 [ ] george plya , _ sur quelques points de la thorie des probabilits _ [ french : on some points of the theory of probability ] . henri poincar , 1:2 ( 1930 ) , 117161 . available at : http://www.numdam.org/numdam-bin/fitem?id=aihp_1930__1_2_117_0 [ ] christophe sabot , _ random walks in random dirichlet environment are transient in dimension @xmath172_. probab . theory related fields 151:1 - 2 ( 2011 ) , 297317 . http://www.springerlink.com/content/24480204un6v607x/ [ ] , http://arxiv.org/abs/0811.4285 [ ] christophe sabot and pierre tarrs , _ edge - reinforced random walk , vertex - reinforced jump process and the supersymmetric hyperbolic sigma model . _ preprint ( 2012 ) . available at : http://arxiv.org/abs/1111.3991 [ ] thomas spencer and martin r. zirnbauer , _ spontaneous symmetry breaking of a hyperbolic sigma model in three dimensions_. commun . phys . 252:1 - 3 ( 2004 ) , 167187 . available at : http://www.springerlink.com/content/06tcd263nyxkxyxu/ [ ] , http://arxiv.org/abs/math-ph/0410032 [ ]
we show that linearly reinforced random walk has a recurrent phase in every graph . the proof does not use the magic formula . primary 60k35 ; secondary 60k37 , 60g09 , 60j10 . linearly reinforced random walk , random walk in random environment , partial exchangeability , plya urn .
one of the first few detected exoplanets was a planetary companion to 55 cnc @xcite . at the time , eight years of doppler measurements from lick observatory revealed a 14.6-day wobble in 55 cnc as it was gravitationally perturbed by a jupiter - mass planet . superimposed on this 14.6-day doppler periodicity was an additional trend showing clear curvature and indicating that 55 cancri was host to a second orbiting body , likely of planetary mass . additional doppler measurements through 2002 uncovered the full doppler cycle with a period of @xmath514 yr , caused by a planet with a minimum mass @xmath0= 4 m@xmath6orbiting @xmath55.5 au from 55 cnc @xcite . this was the first giant planet found with an orbital radius similar to the giant planets in our solar system . a third doppler periodicity of 44.3 days was also apparent in those data , indicating a third jupiter - mass planet in the system , this one orbiting 0.25 au from the star @xcite . this was the second planetary system found to have three planets , the first being that around @xmath7 andromedae @xcite . by combining these velocities from lick observatory with over 100 precise doppler measurements obtained in one year at the hobby - eberly telescope , along with measurements from the `` elodie '' spectrometer at haute provence , @xcite identified a fourth planet having a small minimum mass of @xmath0= 14 m@xmath8with an orbital period of 2.8 d. this planet was one of the first three neptune - mass planets discovered , along with the planets orbiting gj 436 @xcite and @xmath9 arae @xcite . the detection of this neptune - mass planet made 55 cancri the first system known to contain four planets . @xcite also used the fine guidance sensor on the hubble space telescope to carry out astrometry of 55 cnc , and estimated the inclination of the orbital plane of the outer planet to be @xmath10 deg . this four - planet system left a large , dynamically empty gap between 0.25 and 6 au . numerical simulations suggested that hypothetical planets in this gap would be dynamically stable , including the interesting possibility of terrestrial mass planets in the habitable zone between 0.5 and @xmath52 au @xcite . in 2004 , we noticed modest peaks in the periodogram at 260 and 470 d , indicating possible planets at those periods that motivated our continued intense doppler observations . @xcite carried out an independent analysis of the combined published doppler measurements and identified a 260 d periodicity , implying a new planet with a minimum mass of 1.8 m@xmath11= 31 m@xmath1 . it is not uncommon for modest peaks in the periodogram to fluctuate in their confidence level with the addition of new data , especially for cases where the radial velocity amplitudes are comparable to the precision of doppler measurements , so we intensified our observations of this star . here , we add 115 additional radial velocity measurements from lick observatory and 70 radial velocity measurements from keck observatory to our previous data set and find that the false alarm probability for a 260-d signal has strengthened and is present in data sets from both observatories independently . a stellar companion orbits 55 cancri as well . it is a 13th magnitude m dwarf located roughly 1000 au away , certainly bound to 55 cancri a as the radial velocities are nearly the same . the occurrence , dynamics , and final properties of planetary systems may well be affected by such stellar companions , as indicated in observational studies by @xcite and @xcite . thus 55 cancri offers a test of the effects of binary companions on the architecture of complex planetary systems . spitzer space telescope results for 55 cnc by @xcite show that the observed 24 and 70 micron flux densities are comparable to the predicted brightness of the stellar photosphere , indicating no infrared excess above the errors . the corresponding upper limit to the fractional infrared luminosity is 8@xmath12 , or about 80 zodis . a detectable scattered light disk was also ruled out by its non - detection in hst nicmos data by @xcite . here we provide doppler measurements from both lick and keck observatories that significantly augment the 2002 set from lick observatory alone . these measurements span a longer time baseline and contain higher doppler precision with the addition of new keck velocities , offering a chance to reassess all of the planets around 55 cancri . 55 cnc ( = hd 75732=@xmath13 cnc a = hr3522=hip 43587 ) has an apparent brightness @xmath14 and hipparcos parallax of @xmath15 mas @xcite , implying a distance of @xmath16 parsecs and absolute visual magnitude @xmath17 . using spectra from the california & carnegie planet search , @xcite derived @xmath18k , @xmath19 , @xmath20 , and [ fe / h]@xmath21 . indeed , 55 cnc is so metal - rich as to be in the fifth metallicity percentile of stars within 25 pc @xcite . using a bolometric correction that accounts for the high metallicity @xcite , we calculate a stellar luminosity of 0.6 l@xmath22 . the effective temperature , spectroscopic surface gravity and intrinsic luminosity are all consistent with a spectral classification of this star as k0/g8v . the star is chromospherically inactive , with a mt . wilson @xmath23-value of 0.22 ( averaged during the past seven years of our measurements ) implying @xmath24 , indicating a modest age of 2 - 8 gyr where 2 gyr is a strong lower limit on age . the rotation period , calibrated to this chromospheric activity , is estimated to be 39 days @xcite . however , for metal - rich stars , chromospheric emission at the h & k lines remain poorly calibrated as a diagnostic of rotation and age . a more complete discussion of the chromospheric activity and implied stellar properties are given by @xcite . the mass of 55 cnc is best determined by associating its measured effective temperature , luminosity , and metallicity with models of stellar interiors . using the well known `` yale models , '' @xcite , @xcite found a stellar mass of @xmath25 m@xmath22 . @xcite have also derived modified stellar evolutionary models using the yale stellar evolution code to match the observed spectroscopic parameters from @xcite . they derive a stellar mass for 55 cnc of @xmath26 m@xmath27with the uncertainty corresponding to the 99.7% credibility intervals of the bayesian posterior probability distributions . here we simply adopt the average of these two estimates , giving @xmath28 m@xmath27for the mass of 55 cnc . we note that the adopted uncertainty in stellar mass implies fractional errors in the derived planetary masses of 8 percent in addition to errors in the orbital parameters . we have obtained 636 observations of the g8 main - sequence star 55 cnc over the past 18 years . we generally obtain two or three consecutive observations and bin them to increase the velocity precision and accuracy . here , we present 250 binned velocity measurements made at lick observatory from 1989 - 2007 , and 70 binned velocity measurements made at the keck observatory from 20022007 . the lick spectra were obtained using both the 3-meter telescope and the coud auxiliary telescope , which both feed the hamilton optical echelle spectrometer @xcite . a detailed description of the setup of the hamilton spectrometer , its calibrating iodine absorption cell , and the method of extracting doppler measurements for 55 cnc are given in @xcite and in @xcite . the keck spectra were obtained with the hires spectrometer @xcite , and a description of that setup and doppler measurements are given in @xcite . at both telescopes , we place a cylindrical , pyrex cell filled with molecular iodine gas in the light path of the telescope , just before the spectrometer slit , to superimpose sharp absorption lines of known wavelength on the stellar spectrum . the iodine lines provide calibration of both wavelength and the spectrometer psf ( butler et al 1996 ) . while twenty percent of the starlight is absorbed by iodine , the cell s inclusion is worthwhile because the dense iodine absorption lines provide a permanent record of the wavelength scale and behavior of the spectrometer at the instant of each observation , producing long - term doppler precision free of systematic errors to the level of 1 . the velocity measurements are listed in table 1 ( the full table is available in the electronic version of this paper only ) and shown in figure [ fig1 ] with different symbols for measurements made at lick and keck . we carried out a preliminary five - keplerian fit to the combined velocities from both telescopes , allowing one extra parameter to be the difference in the velocity zero - point from the two spectrometers , found to be @xmath29 . once established , we applied the + 28.8 correction to the keck data , putting the two spectrometers measurements on the same velocity scale before showing them in figure [ fig1 ] and listing them in table 1 . the first 14 doppler measurements made between 1989 and november 1994 typically have uncertainties of 810 , worse than most of the subsequent observations due to the unrepaired optics of the hamilton spectrometer . observations made at lick since december 1994 have typical uncertainties of 35 . at keck , between 1999 and 2004 , the typical doppler uncertainty is 3 . in august 2004 the optics and ccd detector for hires were upgraded , reducing the doppler errors . in november 2004 we began making three consecutive observations ( and sometimes five ) of 55 cnc to average over stellar p - mode oscillations that can add 1 velocity noise to g8v main sequence stars @xcite . the resulting doppler precision at keck since august 2004 has been 1.01.5 . the doppler measurements of 55 cnc were fit with a series of keplerian models , each model having an increasing number of planets , beginning with the two well established periods of 14.65 d and 14.7 yr @xcite . we polished all models with a marquardt minimization of @xmath2to establish the best - fit model . the weights assigned to each doppler measurement are the inverse square of each measurement s uncertainty , which are approximated as the quadrature sum of the internal uncertainty in the doppler measurement and the `` jitter '' that stems from photospheric motions and instrumental errors @xcite . experience with similar g8 main - sequence stars suggests that the combined astrophysical and instrumental jitter is 3 at lick and 1.5 at keck , both values being uncertain by 50 percent . the jitter prediction is complicated by the high metallicity of 55 cnc , [ fe / h ] = + 0.3 . the radiative transfer of h & k in 55 cnc will be different from that in solar metallicity stars , because of higher line and continuous opacities , rendering the calibration of emission with stellar age , rotation , and jitter even more uncertain . however , the estimated rotation period of 39 d from periodicities in the h & k emission and the star s low rotational @xmath30of 2.5 confirm that the star has , at most , a modest level of magnetic activity , indicating correspondingly modest jitter with an upper limit of 4 . for this analysis , we adopt a jitter value of 1.5 and 3.0 for keck and lick , respectively . after fitting a model with the two well established planets , we assessed the statistical significance of any periodicities remaining in the residuals to motivate addition of another planet to the model , as described in detail below . we determine false alarm probabilities for peaks in the periodogram attributed to any additional planets by testing the null hypothesis that the current velocity residuals are merely incoherent noise . in such tests , the velocity residuals to our best - fit model are scrambled and their periodograms computed to assess the fraction of trials with scrambled residuals that have a stronger peaks . this fap assessment makes few assumptions about the width or shape of the distribution of noise . our initial model consisted of the sum of two keplerian orbits ( no gravitational interactions ) for the two planets having secure orbital periods of 14.65 d and @xmath514.7 yr , both strongly supported by all of our past doppler analyses of this star @xcite . a two - planet fit yields periods of 14.65 d and 14.7 yr and eccentricities of 0.002 and 0.06 for the two planets , respectively . the residuals have an rms of 11.28 and @xmath31of 3.42 . an accurate model would have an rms on the order of the errors in the data plus `` jitter '' ( @xmath55 ) and @xmath31near 1 . these large values of rms and @xmath31indicate that the model is inadequate . the periodogram of the residuals ( figure [ fig2 ] ) exhibits a tall peak at a period of 44.3 d and power of 55 , clearly significant above the noise . this period corresponds to the orbit of the planet suspected in @xcite . this 44.3 d period is most likely caused by a third planet as the only other explanation would be rotational variation from surface inhomogeneities . such rotational explanations are ruled out both by the shorter stellar rotational period , 42.7 d , found in the photometry as shown in 6 , and by the large velocity amplitude of 10.6 , which is never seen in such chromospherically quiet stars . furthermore , photospheric features generally only survive for a few rotation periods of the star . it seems unlikely that surface inhomogeneities would persist for more than a decade and maintain rotational phase coherence . a levenberg - marquardt minimization was used to find the best - fit orbital parameters for a three - planet keplerian model with periods near 14.65 d , 44.34 d , and 14.7 yr . the best fit yielded residuals with an rms scatter of 8.62 and @xmath31= 2.50 . this result represents an improved fit to the two - planet model , but is still clearly inadequate , not surprising as the model did not include a periodicity near 2.8 d as found by @xcite . indeed , the periodogram of the residuals to the three - planet fit , shown in figure [ fig3 ] , reveals two additional strong peaks near 2.8 d and 260 d. we proceeded to test a 4-planet model by including a fourth planet with a period near 2.8 d @xcite . the best - fit 4-planet model gave periods of 2.81 d , 14.65184 d , 44.32 d , 14.4 yr , all with eccentricities less than 0.3 . the residuals have rms of 7.87 and @xmath31= 2.12 , both representing a significant improvement over the 3-planet model . ( in computing both the rms and @xmath32 , the denominator was appropriately diminished by the greater number of free parameters , i.e. , five per planet . ) thus both the periodogram in figure [ fig3 ] and the superior fit with four planets offer support for the existence of the planet with 2.81 d , corresponding to the planet with @xmath33 = 2.808 @xmath34 0.002 d in @xcite . however this 4-planet model remains inadequate for two reasons . the residuals reveal a poor fit with @xmath31= 2.12 and an rms of 7.87 , larger than explainable by the doppler errors and jitter . also , a periodogram of the residuals reveals a peak at a period of 260.1 d , as shown in figure [ fig4 ] , and some additional smaller peaks . the periodogram peak near 260 d ( figure 4 ) in the residuals to the 4-planet model could be spurious , caused by fluctuations arising from photon - limited doppler errors in the spectra or by aliases in the window function of the sampling times . the ccd detector at lick observatory has been upgraded four times in the past eighteen years , which could produce discontinuities of 12 in their zero points and even create an alias . such abrupt , one - time instrumental changes should not produce periodicities . nonetheless , to check for such effects , the four - planet keplerian model was fit separately to the lick and keck velocities . the 250 velocities from lick against only 70 from keck cause the periodogram in figure 4 to be heavily weighted toward the lick measurements . the prominent period at 260 d certainly reflects the lick velocities more than those from keck , leaving open the question of independent confirmation of the 260 d period in the keck data . we fit a 4-planet model to the 70 keck velocities alone . the keck velocities offer higher precision ( @xmath51.5 ) than those from lick but carry the disadvantage of a duration of only 5 1/2 years . the four - planet fit to the keck velocities alone yielded residuals with rms = 4.3 and @xmath31= 2.59 . the periodogram of the residuals is shown in figure [ fig5 ] , and it reveals a peak at a period of 266 d with a power of 7.7 . there is no significant power at any other periods . the power in the 266 d peak is higher than all peaks for periods between 1 - 3000 d. importantly , the peak at 266 d is roughly twice as high as the noise peaks . although this peak is not overwhelming by itself , the independent occurrence of a periodicity near 265 d in the keck velocities along with the similar period found in the lick velocities , supports the reality of that period and argues against systematic errors as the cause . one might be concerned that the keck velocities yielded such a modest peak at @xmath5266 d ( figure [ fig5 ] ) as compared to the relatively strong peak in the lick data ( figure [ fig4 ] ) . we addressed this concern by augmenting the keck velocities with artificial velocities corresponding to a planet having a period of 260 d in a keplerian orbit that causes a semiamplitude of @xmath35 . the idea is that if the power in the periodogram doubles , then the modest peak in figure 5 is probably reasonable . we performed a four - planet fit and computed the periodogram of the residuals . a peak at @xmath36 d was seen with a power of 13 , roughly double the power of the peak that emerged from the original velocities . thus , the 266 d peak in the periodogram from the original keck velocities ( figure 5 ) constitutes a confirmation of the @xmath5265 d planet seen in figure 4 having that period and amplitude . of course , the lick velocities alone also exhibit the 260 d peak indepedently . we also checked to see if the 260 d signal might be an alias of the possible 470 d peak seen in figure 4 . we fit the combined lick and keck velocities with a five - planet model having a fifth planet with a period near 470 d instead of near 260 d. the periodogram of the residuals to this five - planet model still has a strong peak with period near 263 d , with a power of 19 . apparently the period at 260 d does not vanish by including a 470 d period in the model and thus is not an alias of it . we assessed the probability that the 260 d signal was caused by chance fluctuations in the velocities by performing a conservative false alarm probability test . we fit the combined velocities with only a four - planet model and tested the null hypothesis that no periodicity near 260 d actually exists in the residuals , implying that the peak is due merely to noise . we scrambled the residuals to the four - planet fit but kept the times of observation the same , and recomputed the periodogram for each of 500 trials . we recorded the power of the tallest peak in the periodogram from each of 500 trials . the histogram of those peak powers is shown in figure [ fig6 ] . the typical peaks from the scrambled residuals have powers of 713 , with the tallest being 16 . in contrast , the periodogram of the original residuals had a peak height of 31.5 , shown both in [ fig4 ] and as the vertical dashed line in [ fig6 ] . thus , the null hypothesis ( that the residuals have no coherence ) is unlikely and the associated false alarm probability of the peak at 260 d is less than 0.002 , indicating that the periodicity is real . the analysis above strongly supports the existence of a planet with a period of 260 d. the period of 260 d does not correspond to any known time scale of stellar interiors or atmospheres , nor to the rotation period of the star which is 42.7 d ( see below ) . thus , a plausible interpretation is a planet with a period near @xmath37 d , making it the fifth planet in the 55 cnc system . we constructed a keplerian model that included a fifth planet having a period near 260 d. a best - fit model to the combined lick and keck velocities was found easily , yielding five periods of 14.65162 d , 44.344 d , 5218 d , 2.817 d , and 260.0 d ( see table 2 ) . the residuals have rms = 6.74 and @xmath31= 1.67 ( including the jitter in the expected variance ) , and a periodogram of them is shown in figure 7 . the values of the rms and @xmath31are 15% and 20% lower , respectively , than the corresponding diagnostics of the four - planet model . table 3 gives the rms and @xmath31for all multi - planet models considered in this paper , showing the significant improvement with each additional planet . this major improvement in the quality of the fit of 320 measurements , coming from a fifth planet with its five additional free parameters , indicates that the new model has considerable merit . the five - planet model containing the 260 d planet is clearly superior to the four - planet model . the period agrees with that found by @xcite from a periodogram analysis of our earlier , published velocities from lick observatory . as this model contains a proposed planet with @xmath37 d , we present in table 2 all of the orbital parameters for all five planets self - consistently computed with a levenberg - marquardt least squares algorithm . considerable trial and error with various starting guesses for the 26 free parameters was carried out to ensure that the least squares search began near the deepest minimum . the @xmath2fit was virtually unchanged for orbital eccentricites between 0.0 to 0.4 for the 260 d planet . this is not surprising since the amplitude of the planet is comparable to the single measurement precision for most of our data . although the orbit is consistent with circular , we adopted an intermediate eccentricity of @xmath38 to indicate the indistinguishable range of eccentricity . the best - fit parameters for the 260 d planet are @xmath39 = 0.2 @xmath34 0.2 , @xmath40 = 4.879 @xmath34 0.6 , implying @xmath0= 0.144 @xmath34 0.04 m@xmath3 . the innermost planet has @xmath33 = 2.81705 @xmath34 0.0001 d , @xmath39 = 0.07 @xmath34 0.06 , @xmath40 = 5.07 @xmath34 0.53 , and @xmath0= 0.034 m@xmath6= 10.8 m@xmath1 . in comparison , @xcite found the inner planet to have a period , @xmath33 = 2.808 @xmath34 0.002 d , @xmath39 = 0.174 @xmath34 0.127 , @xmath40 = 6.67 , and @xmath0= 0.045 m@xmath6= 14.2 m@xmath1 . there is no question that the planet with @xmath33=2.817 d is the planet previously identified as having a period of 2.808 d @xcite . the new minimum mass @xmath0= 10.8 m@xmath8is lower than the 14.2 m@xmath8previously reported in mcarthur et al .. these differences are not surprising as some of the excess velocity variation previously left to be absorbed by the four known planets is now accounted for by the fifth planet . the phase - folded velocities for the 260-day planet are shown in figure 8 after subtracting the sum of the computed velocities of the other four planets from the measured velocities . the orbital eccentricity has been fixed to 0.2 . the resulting residual velocities are plotted versus orbital phase and shown in figure 8 . the residuals reveal the 260 d period that had been detected in the periodogram and the keplerian model is overplotted . the scatter has an rms of 6.74 . the error bars shown in figure [ fig8 ] include the quadrature sum of the internal errors ( typically 2 for keck and 4 for lick ) and the `` jitter '' ( 1.5 for keck and 3 for lick ) . thus , the scatter of 6.74 is only slightly larger than the known internal errors and expected jitter ( astrophysical and instrumental ) . several explanations for the modest 6.74 scatter in the residuals are possible . perhaps we are underestimating our internal errors . perhaps the jitter for this metal - rich star is somewhat higher than the average for g8 main - sequence stars . or perhaps there are other planets that cause a sufficiently low signal that they are not apparent in the periodograms but nonetheless add a few of `` noise '' to the velocities . we assessed the detectability of a hypothetical 6th planet by adding the velocities that would be induced by it to the observed velocities . we fit a 5-planet model to these augmented velocities , allowing all 26 parameters to float . we searched the periodogram of the residuals for peaks that loom above those arising in the 5-planet fit of the actual velocities ( figure 7 ) . such peaks would have been identified as a candidate 6th planet . we considered orbital periods from 300 d - 4000 d and determined the minimum @xmath0that produced a peak 50% above any of the peaks in the actual periodogram ( i.e. above a power of 15 ) . the minimum detectable mass of a hypothetical 6th planet is a sensitive function of its period and phase as those parameters determine how easily the signal can be absorbed in the 5-planet model , avoiding detection . neighboring periods differing by a mere few percent can produce periodogram peaks differing by a factor of two simply due to differing commensurability with the 5 existing planets . a 6th planet in a mean motion resonance is particularly capable of avoiding detection in the face of the five existing planets . such fine structure aside , the simulations can be characterized as follows . for orbital periods of 300 - 850 d , a 6th planet with msini below 50 m@xmath8would have eluded detection as the periodogram peaks would not have loomed even 50% above the noise . for periods 850 d - 1500 d , a 6th planet could avoid detection by having @xmath0below 100 m@xmath1 . for periods 1750 - 4000 d , planets below 250 m@xmath8would elude detection . thus such planets could exist around 55 cnc and yet have avoided detection by our current 18 years of doppler measurements . indeed several such planets could exist in the large gap between periods of 260 d and 13 yr and probably maintain dynamical stability . the models in this paper are based on the approximation that the planetary orbits are keplerian ellipses . in actuality , the radial velocity variation of the parent star over nearly two decades of observation is also affected by the mutual gravitational perturbations between the planets . as a concrete example , one can interpret the 5-planet fit in table 2 as describing a set of osculating orbital elements at the epoch jd 2447578.730 of the first radial velocity observation . by making a choice of epoch , one creates a unique initial condition for a six - body integration of newton s equations of motion . when this integration is carried out , one finds radial velocity deviations of @xmath41 25 in comparing keplerian and newtonian models at epochs near jd 2454000 . these deviations arise primarily from the orbital precessions of planets b , c , and d that occur in the newtonian model that are absent from the keplerian model . because the orbits are nearly circular , a keplerian 5-planet fit can , however , compensate for nearly all of the precession through small adjustments to the orbital periods . it is likely that one can obtain an improved chi - square by adopting a self - consistent n - body model for the stellar reflex velocity ( e.g. rivera et al . in addition to lowering the rms of the fit , a definitive model of this type allows for the correct characterization of the possible 3:1 resonant relationship between planets b and c , and can therefore give important clues to the formation history of the system . adopting the keplerian fit in table 2 as an intial guess , we used levenberg - marquardt minimization to obtain a self - consistent 5-planet dynamical fit to the radial velocity data sets . the resulting orbital parameters of our dynamical fit are all quite similar to their corresponding values in the keplerian model , and are listed in table 4 . our dynamical fit has @xmath42=2.012 ( without including any jitter ) , and rms=7.712 . a more computationally expensive search should be able to find orbital parameters that provide a slight improvement to these values . we leave such an analysis to future work . in order to assess the dynamical stability of our five - planet model , we adopt the self - consistent orbital elements in table 4 and integrate the system forward for one million years from epoch jd 2447578.730 . we used a bulirsch - stoer integrator ( press et al . the system remained stable throughout a one million year integration . the evolution of the five planetary eccentricities during a representative @xmath43 year interval are shown in figure [ fig10 ] . as is true throughout the @xmath44 year integration , the eccentricity variations experienced by all five planets are quite modest , and the system appears likely to be dynamically stable for long periods . it is interesting to note that during the course of the numerical integration , the 3:1 resonant arguments for planets b and c are all circulating . this indicates that planets `` b '' and `` c '' do not currently participate in a low - order mean motion resonance , despite the near commensurability of their orbital periods . we have computed the eccentricity variations that result when the system is modeled using a secular perturbation theory that includes terms up to second order in eccentricity ( see , e.g. , murray & dermott 1999 ) , and which includes the leading - order effects of general relativity as outlined by adams & laughlin ( 2006 ) . the results are quite similar to those in figure [ fig10 ] . this indicates that the bulk of the planet - planet interactions in the system can be accounted for with a simple second - order theory , thus improving the likelihood that the configuration of planets can remain dynamically stable for the lifetime of the star . we have used the t8 0.8 m automatic photometric telescope ( apt ) at fairborn observatory to obtain high - precision photometry of 55 cnc during 11 observing seasons between 1996 november and 2007 april . the t8 apt is one of several automated telescopes at fairborn dedicated to observing long - term , low - amplitude brightness variations in solar - type stars associated with stellar magnetic cycles as well as to measuring short - term , low - amplitude variations caused by rotational modulation in the visibility of surface magnetic features @xcite . apt photometry of planetary candidate stars helps to establish whether observed radial velocity variations are caused by stellar activity or planetary - reflex motion , and direct measures of stellar rotation periods provide good age estimates of the planetary systems ( e.g. , * ? ? ? @xcite and @xcite have published several examples of periodic radial velocity variations in solar - type stars caused by photospheric spots and plages . the apt observations are also useful to search for transits of the planetary companions . the rare transiting systems allow direct determination of basic planetary parameters such as mass , radius , and mean density and so provide observational constraints on models of planetary composition and internal structure ( e.g. , * ? ? ? bright transiting systems enable detailed follow - up studies of exoplanet atmospheres ( e.g. , * ? ? ? finally , monitoring a planetary host star s long - term luminosity variations provides a measure of the star s climate forcing ability on its system of planets ( e.g. , * ? ? ? . the t8 apt is equipped with a two - channel precision photometer employing two emi 9124qb bi - alkali photomultiplier tubes to make simultaneous measurements in the strmgren @xmath45 and @xmath46 pass bands . the apt measures the difference in brightness between a program star and one or more nearby constant comparison star(s ) ; the primary comparison star used for 55 cnc is hd 76572 ( @xmath47 , @xmath48 , f6 iv - v ) . the strmgren @xmath45 and @xmath46 differential magnitudes are corrected for differential extinction with nightly extinction coefficients and transformed to the strmgren system with yearly mean transformation coefficients . finally , the @xmath49 and @xmath50 observations are combined into a single @xmath51 pass band to increase the photometric precision . the external precision of a single differential magnitude is typically around 0.0015 mag for the t8 apt , as determined from pairs of constant stars . further details on the telescope , photometer , observing procedures , and data reduction techniques can be found in @xcite . the complete 11-yr set of differential magnitudes computed with the primary comparison star is plotted in the top panel of figure 11 . intercomparison of the primary comparison star with two secondary comparisons ( hd 77190 , @xmath52 , @xmath53 , a8vn ; hd 79929 , @xmath54 , @xmath55 , f6v ) revealed that the annual means of the primary comparison vary over a range of 0.003 mag from year to year . rather than switch to one of the more stable secondary comparison stars , we have instead normalized the 11 seasons with the primary comparison so they all have the same annual mean . this was done because the secondary comparison stars have been used only for the past seven years . the normalization removes any long - term variation in the primary comparison star as well as in 55 cnc , but this improves the sensitivity of our transit search described below for orbital periods under one year . after normalization , outliers exceeding three standard deviations were removed . the final data set in the top panel of figure 11 contains 1349 nightly observations ; the standard deviation of an individual observation from the normalized mean is 0.0017 mag . the 0.0017 mag standard deviation of the full data set is only slightly greater than the nominal measurement precision of 0.0015 mag but suggests that low - amplitude , short - term intrinsic variability might be present at times in 55 cnc . ( long - term variability has been removed by the normalization ) . we searched each annual set of measurements for evidence of coherent , low - amplitude variability that might be the result of rotational modulation in the visibility of starspots . the middle panel of figure 11 , which shows photometry from a portion of the 9th observing season , exhibits the clearest example of coherent variability in the data set . two cycles of brightness variation are visible with an amplitude of approximately 0.006 mag . we interpret this as evidence for a small starspot region ( covering less than 1% of the star s visible surface ) that has survived for two rotation cycles of the star . a power spectrum of the observations in the middle panel is computed with the method of @xcite and shown in the bottom panel of figure 11 . this gives a period of 42.7 @xmath34 2.5 days , which we interpret to be the stellar rotation period . this confirms the rotation period of 55 cnc reported by @xcite , who used rotational modulation of the ca ii h & k flux measured by the hk project at mount wilson observatory @xcite . we searched the photometric data for evidence of transits of the four inner planets ; the results are summarized in table 4 and plotted in figure 12 . we first computed the semi - amplitudes of the light curves ( column 3 ) with least - squares sine fits of the complete data set phased to the four shortest radial velocity periods . the resulting amplitudes are all extremely small and consistent with zero . these very tight limits on photometric variability on the radial velocity periods clearly support planetary - reflex motion as the cause of the radial velocity variations . while our measured 42.7 day rotation period is consistent with the 44.35 day radial velocity period because of the relatively large uncertainty of 2.5 days in the rotation period , the absence of any photometric variability on the more accurate 44.35 day radial velocity period is strong support for the existence of 55 cnc c. in figure 12 , we have plotted light curves of the photometric data phased with the orbital periods of the inner four planetary companions . zero phase in each panel represents the predicted phase of mid transit for each of the companions . only phases from 0.94 to 0.06 are plotted to improve visibility of any possible transits . the solid curve in each panel approximates the predicted transit light curve , assuming a planetary orbital inclination of 90(central transits ) . the out - of - transit light level corresponds to the mean ( normalized ) brightness of the observations . the transit durations are calculated from the orbital elements , while the transit depths are derived from the estimated stellar radii and the planetary radii computed with the models of @xcite . the horizontal error bar below each predicted transit curve represents the approximate @xmath56 uncertainty in the time of mid transit , based on monte carlo simulations and the uncertainties in the orbital elements . the vertical error bar represents the typical @xmath57 mag uncertainty of a single observation . column 4 of table 5 lists the geometric probability of transits for each of the five companions , computed from equation 1 of @xcite and assuming random orbital inclinations . the predicted transit depths for each planet determined as described above are given in column 5 , and the `` observed transit depths '' are recorded in column 6 . the observed depths are computed as the difference in the mean light levels between observations that fall inside and outside of the transit windows plotted in figure 12 ; a positive depth indicates a brightness drop in the transit window . unfortunately , we see no evidence for transits of the inner , low - mass companion 55 cnc e ; the mean of the 51 observations within its predicted transit window agree with the out - of - transit observations within [email protected] mag , a result consistent with the absence of 0.00065 mag transits but still allowing a small possibility for their existence . since the uncertainty in the time of mid transit is rather large compared to the transit duration , we searched for shallow transits over the full range of phases and over orbital periods between 2.702.90 days with null results . the secure detection of transits of such a small body would require reducing the uncertainty of the in - transit brightness mean by a factor of about two , which would require a factor of four more observations . this could be accomplished with the apt over the next observing season or two by concentrating brightness measurements around the times of predicted transits . we are forced to leave our non - detection of transits of 55 cnc e as an uncertain result , as indicated by the colon in column 7 of table 5 . table 5 and figure 12 demonstrate that transits with the expected depths of 55 cnc b and 55 cnc c do not occur . the observed transit depths are both consistent with zero . figure 12 shows that we have no photometric observations during the predicted time of transit of 55 cnc f. however , given the uncertainty in the transit timing and the density of observations within the uncertainty range , we conclude that transits of planet f probably do not occur . we have insufficient radial velocities to predict accurate transit times of the outermost planet 55 cnc d , so we can say nothing about their occurrence . we note that our non - detection of transits is consistent with the likely inclinations of the planetary orbits as discussed in 7 ( below ) . finally , we comment on the long - term variability of the host star 55 cnc . although our normalization of the light curve has removed any such variation from the present analysis , an examination of the light curves computed with the secondary comparison stars mentioned above show that 55 cnc clearly exhibits year - to - year variations in mean brightness with an amplitude of 0.001 mag over a time scale of several years or more ( henry et al.,in preparation ) . thus , long- and short - term brightness variations in 55 cnc are very similar to irradiance variations in our sun ( .e.g . planet formation in the protoplanetary disk around 55 cnc was apparently extraordinarily efficient , yielding at least five planets . our extensive radial velocity data set , with its 18-year baseline , gives no indication that additional jupiter - mass companions exist beyond 6 au , although saturn - mass or smaller planets would easily go undetected . if , as a thought experiment , we grind up the currently known planets , we may infer the properties of the protoplanetary disk around 55 cnc . we first assume an edge - on , co - planar geometry with @xmath58 , and _ in situ _ formation of the planets directly from the disk gas , e.g. @xcite . in this case , the masses of planets total @xmath59 and together imply a lower limit of 410 g @xmath60 for the average surface density of the disk interior to 6 au . for a gas - to - dust ratio of 100 , this implies an average surface density in solids of 4 g @xmath60 . if the radial surface density profile of 55 cancri s protostellar disk declined as @xmath61 , this implies a solid surface density of 1.4 g @xmath60 at 5 au , which is approximately half the value of the minimum - mass solar nebula at jupiter s current position . in all likelihood , however , the surface density of solids in 55 cancri s protostellar disk was higher than in the solar nebula . if we assume that the planets formed via the core accretion mechanism , as described , for example , by @xcite we estimate that they contain at least 150 m@xmath8of heavy elements . here we include the high metallicity of the host star , 55 cnc , with its [ fe / h]=+0.3 as representative of the planet s interior . reconstituting this mass of solids to recover 55 cancri s metallicity implies an original protostellar disk mass of @xmath62 within 6 au . assuming that the nascent disk extended to 30 au with a @xmath63 surface density profile , the total mass would have been @xmath64 , and the surface mass density in solids at 5 au would have been 7 g @xmath60 . adopting the reported orbital inclination of 53 deg for the outer planet @xcite and assuming the orbits to be co - planar augments all masses by 1/@xmath65 = 1.25 . the resulting simplistic minimum mass protoplanetary nebula has 510 g @xmath60 for the average surface mass density of the combined gas and dust within 6 au . adopting a nominal gas to dust ratio of 100 yields a dust surface mass density of 5 g @xmath60 . but again considering the likely enrichment of solids within giant planets , and the associated h and he , yields an original mass within 6 au of at least 0.031 m@xmath22 . extending this disk to 30 au gives a total mass of 0.075 m@xmath22 . the estimated surface mass density of solids in the disk at 5 au would have been 8.7 g @xmath60 . for expected equilibrium disk temperatures , this minimum mass disk is below the threshold required for the development of non - axisymmetric gravitational instabilities @xcite , but likely high enough to support the formation of planets via core accretion @xcite . in the context of this disk - profile scenario , the core accretion theory suggests that additional objects with masses ranging from neptune to saturn mass should be present beyond the frontier marked by the orbit of planet d , i.e. beyond 6 au . if the outer planet migrated during or after its formation , the estimated disk properties computed here would be affected . our eighteen year campaign of doppler measurements of 55 cnc at the lick and keck observatories has gradually revealed additional superimposed wobbles , each best interpreted as due to another orbiting planet . the previously identified four planets revealed a large gap between 0.24 and 5.8 au raising questions about unseen planets there and the planet formation history in the protoplanetary disk . the velocities presented here reveal a fifth periodicity with @xmath37 d , consistent with keplerian motion for which the most reasonable interpretation is another orbiting planet . the five - planet model suggests the new planet has a minimum mass of 45 m@xmath8 in a nearly circular orbit with a semimajor axis of @xmath66au . thus , 55 cnc is the first quintuple - planet system known . this fifth planet apparently resides in the previously identified gap between 0.24 - 5.8 au , and it remains between 0.73 au ( periastron ) and 0.84 au ( apastron ) , preventing orbit crossings with both the next inner planet , `` c '' , whose apastron is at 0.26 au and the outer planet , `` d '' , whose periastron is at 5.5 au , ensuring dynamical stability that is demonstrated numerically by n - body simulations . as the star s luminosity is @xmath67 l@xmath27(from its effective temperature and radius ) , this fifth planet resides within the classical habitable zone . with a minimum mass of 45 m@xmath1 , we speculate that it contains a substantial amount of hydrogen and helium , not unlike saturn ( @xmath68 m@xmath1 ) in the solar system . the four previously published planets around 55 cancri now have revised orbital parameters and masses because the fifth planet had been contaminating the doppler signal but was not taken into account . the orbital semimajor axes and masses of all five planets ( moving outward from the central star ) are @xmath69 au and @xmath0= 10.8 m@xmath1 ; 0.115 au and 0.824 m@xmath3 ; 0.24 au and 0.169 m@xmath3 ; 0.781 au and 0.144 m@xmath3 ; and 5.77 au and 3.83 m@xmath3 . all quoted minimum masses are uncertain by @xmath55% due to the uncertain mass of the host star 55 cnc . the planets in this system all have nearly circular orbits , with the caveat that the orbital eccentricity for the 260 d planet is poorly constrained by radial velocity data . the inclination of the orbital plane of the outer planet has been estimated from the apparent astrometric motion of the star , as measured with the fine guidance sensor of the hubble space telescope @xcite . the derived orbital inclination is @xmath70 deg ( 37 deg from edge - on ) for that outer planet , implying that its actual mass is 4.9 m@xmath3 . assuming coplanarity for the other four planets , their actual masses ( proceeding outward ) are @xmath71 m@xmath8(nearly one uranus mass ) , @xmath72 m@xmath3 , @xmath73 m@xmath6= 66.7 m@xmath1 , and @xmath74 m@xmath6= 57 m@xmath1 . normally , co - planarity should not be a foregone conclusion , as the eccentricities of many exoplanets imply a dynamically perturbative history . but for the 55 cnc system with its five planets so vulnerable to instabilities , such an active past seems unlikely . any great perturbations would have ejected the smaller planets . thus the planetary orbits in 55 cnc are likely to reside in a flattened plane , analogous to the ecliptic , coinciding with the original protoplanetary disk out of which the planets presumably formed . if 55 cnc did have a quiescent past , the star s spin axis should be nearly coincident with the normal of the system s orbital plane . the inclination of the spin axis can be determined from the spectroscopically measured rotational @xmath30= 2.46 and the photometrically determined spin period of 42.7 d for the star , along with its radius of 0.93 r@xmath22 . the 42.7 d spin period implies an equatorial velocity of only 1.24 , which is lower than @xmath75 , an impossibility . either the measured @xmath30is too high by a factor of two ( quite possible given the many line broadening sources ) or the spin period is much shorter ( not likely , given the low chromospheric activity ) . as is common , the inclination of the spin axis of a nearby star , as attempted here , carries uncertainties so large as to be of little value . nonetheless the crude interpretation would be that the star is not viewed pole - on . indeed , if the orbits were viewed nearly pole - on the implied planetary masses would be so large as to render the system dynamically unstable . we conclude that the orbital plane of 55 cnc is not being viewed nearly pole - on , consistent with the astrometric value of @xmath76=53 deg . as the 55 cnc system has more planets than any previously discovered system , its overall structure , including its planet mass distribution , its density of orbits , and its orbital eccentricities , offers direct constraints about its protoplanetary disk and subsequent planetary dynamics . the 55 cnc system may initially be sketched as having one massive planet , likely a hydrogen - helium gas giant , in a nearly circular orbit at 5.8 au . inward are four less massive planets , the innermost being roughly uranus mass , the next outward having roughly jupiter mass and likely gaseous , and the next two having somewhat sub - saturn masses , also probably gaseous . the outer planet has an orbital angular momentum quickly shown to be at least 8.2@xmath7710@xmath78 kg m@xmath79 s@xmath80 , certainly 100 times greater than the star s _ spin _ angular momentum , 6.1@xmath7710@xmath81 kg m@xmath79 s@xmath80 ( from its 42.7 d spin period ) . thus , 55 cnc contains a dominant outer planet at 5.8 au of roughly 5 m@xmath6and four lower mass planets , all five in nearly circular orbits ( though two orbits appear to be somewhat more eccentric than found for the more massive planets in our solar system ) . the large eccentricities found in the majority of exoplanets @xcite are not seen in 55 cnc . the five orbits in 55 cnc are probably nearly co - planar , as discussed above , lest some planet masses be too large to allow stability . thus , 55 cnc system has some basic structural attributes found in our solar system : nearly coplanar , circular orbits , with a dominant gas giant between 5 - 6 au . this similarity suggests that such solar system architectures are not extremely rare . the formation of multi - planet systems with outer , dominant planets may occasionally form such that they persist for billions of years without disruptive gravitational perturbations that cause large eccentricities and eject planets . because nested , coplanar , circular orbits could hardly be obtained unless they began that way , the 55 cnc system , along with the solar system , supports the hypothesis that planets form in viscous protoplanetary disks , as has long been predicted by standard planet formation theory , e.g. , @xcite . one puzzle is whether the 55 cnc planets suffered significant migration . the 44.35 d and 14.65 d have a period ratio of 3.027:1.000 , thus leaving open the possibility of a mean motion resonance identified previously @xcite . as planets have no reason to form with integer period ratios , any resonance suggests that some differential migration occurred , allowing the two planets to capture each other . however the 3:1 mean motion resonance was found to be absent in the current n - body model as none of the relevant resonant arguments are librating . however , if migration occurred , we wonder what prevented the outer planet , and indeed all of the planets , from migrating inward . perhaps the disk dissipated just as this last crop of five planets formed , as suggested in some migrational models @xcite . indeed , the proximity of the three inner planets to the host star , especially the jupiter - mass planet at 0.115 au ( @xmath82 d ) , suggests that they migrated inward to their present locations , assuming they did not form in situ . if so , protoplanetary disk material likely orbited outside 0.24 au , exerting an inward torque on those planets and carrying away orbital angular momentum in the system . during the migration period , the implied disk material would have had a mass comparable to ( or exceeding ) the jovian - mass planets , from which the most recently identified planet at 0.78 au could have formed . it is interesting that the third and fourth planets ( at 0.24 and 0.78 au ) have small minimum masses , under 0.2 m@xmath3 , but are surrounded by much larger giant planets with minimum masses of 0.824 and 3.8 m@xmath3 . one wonders why the acquisition of material was apparently so different among these four planets . one also wonders why this particular star ended up with five planets while 90 percent of stars on doppler surveys do not have any detected giant planets . a statistical analysis of the planet detectability and observational incompleteness has been carried out by cumming 2007 . perhaps the high metallicity of 55 cnc ( [ fe / h ] = + 0.30 ) played a role in the efficient planet formation . fischer & valenti ( 2005 ) find a correlation not only between stellar metallicity and the occurrence of planets , but also between high metallicity and multi - planet systems . but we doubt that a mere factor of two enhancement in heavy elements could account entirely for the five planets in this system . some stochasticity in planet formation and subsequent stability must play a role . ) by 0.47 arcseconds making it a good target for next - generation adaptive optics systems . the space interferometry mission , `` sim planetquest '' , operating in narrow angle mode with astrometric precision of 1 @xmath9as could measure the astrometric wobble caused by all four outer planets , providing definitive masses and orbital inclinations for them . a spaceborn coronagraph or a spaceborn interferometer might be capable of imaging the outer planet and taking spectra of it . nasa and esa have a wonderful opportunity to fund such an imaging telescope , thereby detecting and spectroscopically assessing a mature extrasolar planet . moreover , as 55 cancri is metal - rich , the planets may also be abundant in heavy elements , offering an opportunity to study rich atmospheric chemistry , clouds , and weather , if spectra could be obtained . transits , if any occur , would provide planet radii offering information about potential rocky cores . this rich planetary system portends a fruitful future for the doppler technique of studying exoplanets . it shows that extending the time baseline of doppler measurements can reveal multiple planets , the existence or absence of which provides information about the formation , structure , and evolution of planetary systems . we gratefully acknowledge the efforts and dedication of the lick and keck observatory staff . we thank karl stapelfeldt for helpful comments . we thank the anonymous referee for comments that improved the manuscript . we appreciate support by nasa grant nag5 - 75005 and by nsf grant ast-0307493 ( to ssv ) ; support by nsf grant ast-9988087 , by nasa grant nag5 - 12182 and travel support from the carnegie institution of washington ( to rpb ) . gwh acknowledges support from nasa grant ncc5 - 511 and nsf grant hrd-9706268 . we are also grateful for support by sun microsystems . we thank the nasa and uc telescope assignment committees for allocations of telescope time toward the planet search around m dwarfs . this research has made use of the simbad database , operated at cds , strasbourg , france . the authors wish to extend special thanks to those of hawaiian ancestry on whose sacred mountain of mauna kea we are privileged to be guests . without their generous hospitality , the keck observations presented herein would not have been possible . , s. l. , donahue , r. a. , soon , w. , & henry , g. w. 1998 , in cool stars , stellar systems , and the sun , r. a. donahue and j. a. bookbinder , eds . , volume 154 of _ astronomical society of the pacific conference series _ , pp . 153+ , g. , beichman , c. a. , trilling , d. e. , rieke , g. h. , holmes , e. k. , lawler , s. m. , stapelfeldt , k. r. , werner , m. w. , gautier , t. n. , blaylock , m. , gordon , k. d. , stansberry , j. a. , & su , k. y. l. 2006 , , 636 , 10981113 , c. , anklin , m. , crommelynck , d. , finsterle , w. , & willson , r. c. 1998 , in new eyes to see inside the sun and stars , f .- deubner , j. christensen - dalsgaard , and d. kurtz , eds . , volume 185 of _ iau symposium _ , pp . 89+ , n. c. , bouchy , f. , mayor , m. , pepe , f. , queloz , d. , udry , s. , lovis , c. , bazot , m. , benz , w. , bertaux , j .- l . , lo curto , g. , delfosse , x. , mordasini , c. , naef , d. , sivan , j .- , & vauclair , s. 2004 , , 426 , l19l23 , b. , fischer , d. a. , henry , g. w. , laughlin , g. , butler , r. p. , marcy , g. w. , vogt , s. s. , bodenheimer , p. , ida , s. , toyota , e. , wolf , a. , valenti , j. a. , boyd , l. j. , johnson , j. a. , wright , j. t. , ammons , m. , robinson , s. , strader , j. , mccarthy , c. , tah , k. l. , & minniti , d. 2005 , , 633 , 465473 , s. s. , allen , s. l. , bigelow , b. c. , bresee , l. , brown , b. , cantrall , t. , conrad , a. , couture , m. , delaney , c. , epps , h. w. , hilyard , d. , hilyard , d. f. , horn , e. , jern , n. , kanto , d. , keane , m. j. , kibrick , r. i. , lewis , j. w. , osborne , j. , pardeilhan , g. h. , pfister , t. , ricketts , t. , robinson , l. b. , stover , r. j. , tucker , d. , ward , j. , & wei , m. z. 1994 , in proc . spie instrumentation in astronomy viii , david l. crawford ; eric r. craine ; eds . , volume 2198 , p. 362 362+ rrrrrrrrrr [ candid ] planet e & 2.81705 & 2449999.83643 & 0.07 & 248.9 & 5.07 & 0.034 & 0.038 + @xmath34 & 0.0001 & 0.0001 & 0.06 & 38 & 0.53 & 0.0036 & 1.0@xmath12 + planet b & 14.65162 & 2450002.94749 & 0.014 & 131.94 & 71.32 & 0.824 & 0.115 + @xmath34 & 0.0007 & 1.2 & 0.008 & 30 & 0.41 & 0.007 & 1.1@xmath12 + planet c & 44.3446 & 2449989.3385 & 0.086 & 77.9 & 10.18 & 0.169 & 0.240 + @xmath34 & 0.007 & 3.3 & 0.052 & 29 & 0.43 & 0.008 & 4.5@xmath84 + planet f & 260.00 & 2450080.9108 & 0.2(f ) & 181.1 & 4.879 & 0.144 & 0.781 + @xmath34 & 1.1 & 1.1 & 0.2 & 60 . & 0.6 & 0.04 & 0.007 + planet d & 5218 & 2452500.6 & 0.025 & 181.3 & 46.85 & 3.835 & 5.77 + @xmath34 & 230 & 230 & 0.03 & 32 & 1.8 & 0.08 & 0.11 + rlllllllll [ selfconfit ] 55 cancri b & 14.651262 & 7572.0307 & 0.0159 & 164.001 & 71.84 & 0.8358 & 0.115 + 55 cancri c & 44.378710 & 7547.5250 & 0.0530 & 57.405 & 10.06 & 0.1691 & 0.241 + 55 cancri d & 5371.8207 & 6862.3081 & 0.0633 & 162.658 & 47.20 & 3.9231 & 5.901 + 55 cancri e & 2.796744 & 7578.2159 & 0.2637 & 156.500 & 3.73 & 0.0241 & 0.038 + 55 cancri f & 260.6694 & 7488.0149 & 0.0002 & 205.566 & 4.75 & 0.1444 & 0.785 + ccccccc e & 2.79565 & 0.00004 @xmath34 0.00006 & 9.7 & @xmath850.00065 & @[email protected] & no ? + b & 14.65165 & 0.00006 @xmath34 0.00006 & 4.1 & @xmath850.0143 & @[email protected] & no + c & 44.3401 & 0.00008 @xmath34 0.00006 & 2.0 & @xmath850.0086 & @[email protected] & no + f & 260.81 & 0.00008 @xmath34 0.00006 & 0.8 & @xmath850.0090 & & no : + d & 5223 & & 0.1 & @xmath850.0155 & & ?
we report 18 years of doppler shift measurements of a nearby star , 55 cancri , that exhibit strong evidence for five orbiting planets . the four previously reported planets are strongly confirmed here . a fifth planet is presented , with an apparent orbital period of 260 days , placing it 0.78 au from the star in the large empty zone between two other planets . the velocity wobble amplitude of 4.9 implies a minimum planet mass @xmath0= 45.7 m@xmath1 . the orbital eccentricity is consistent with a circular orbit , but modest eccentricity solutions give similar @xmath2fits . all five planets reside in low eccentricity orbits , four having eccentricities under 0.1 . the outermost planet orbits 5.8 au from the star and has a minimum mass , @xmath0= 3.8 m@xmath3 , making it more massive than the inner four planets combined . its orbital distance is the largest for an exoplanet with a well defined orbit . the innermost planet has a semi - major axis of only 0.038 au and has a minimum mass , @xmath4 , of only 10.8 m@xmath1 , one of the lowest mass exoplanets known . the five known planets within 6 au define a _ minimum mass protoplanetary nebula _ to compare with the classical minimum mass solar nebula . numerical n - body simulations show this system of five planets to be dynamically stable and show that the planets with periods of 14.65 and 44.3 d are not in a mean - motion resonance . millimagnitude photometry during 11 years reveals no brightness variations at any of the radial velocity periods , providing support for their interpretation as planetary .
a charged particle moving in two dimensions under the influence of a spatially varying perpendicular magnetic field can exhibit snake state trajectories . these are open two - dimensional orbits perpendicular to the direction of the magnetic field gradient . snake trajectories occur in both the classical and quantum limits of this problem and are of fundamental interest with potential applications for electron transport in multidomain ferromagnets , two - dimensional electron gases @xcite , and in nanomaterials like graphene @xcite . in the quantum limit the snake states can be interpreted as the protected modes that occur at domain walls that separate topologically mismatched gapped ground states . this picture suggests that snake trajectories can arise even in a _ uniform _ magnetic field if the particle density is suitably modulated laterally , e.g. by electrostatic gating patterned to form interfaces between distinct quantum hall ground states . indeed exactly this possibility has been explored theoretically @xcite and examined experimentally for graphene in a uniform perpendicular magnetic field via measurements of the hall conductance and of fabry - perot like oscillations in the inter - edge conductance across graphene _ pn _ junctions @xcite . graphene is an excellent candidate for this application because it can be electrostatically switched from _ n _ to _ p _ carrier types and studied in the ballistic transport regime @xcite . the converse problem of snake trajectories for a uniform carrier density in a spatially varying magnetic field is even more technically challenging and it has not been examined experimentally ( for a theoretical discussion see ref . ) . on the other hand , a variant of this latter problem _ is _ routinely encountered in present - day experimental environments . in single - layer graphene subject to elastic lattice strains , the low - energy electronic structure is described by a dirac hamiltonian containing a strain - induced gauge field that mimics the effects of a perpendicular ( albeit valley antisymmetric ) magnetic field . for generic smoothly varying strain fields the presence of nodal lines that separate regions of positive " and negative " pseudo - magnetic field in a single valley is a nearly unavoidable consequence of the symmetry of this strain coupling . the antisymmetric @xmath0 " and antisymmetric doping ( @xmath1 ) " problems break both time reversal symmetry @xmath2 ( due to the presence of a magnetic field ) and particle - hole symmetry @xmath3 ( they require a nonzero carrier density ) . nonetheless they retain different composite symmetry operations that combine these discrete symmetries with twofold rotation about the layer normal @xmath4 : the former problem is symmetric under @xmath5 and the latter under @xmath6 . this difference manifests itself in many of the spectral properties presented in section i. in fact , this distinction persists even into the classical limit , and reflects the different underlying dynamics of these two problems . in the antisymmetric-@xmath0 problem a snake state trajectory arises from the compensation of the circulation of cyclotron orbits in regions where the magnetic field is reversed . in the antisymmetric-@xmath1 it arises from one - sided skipping orbits due to an electric field at the interface of a @xmath7 junction . it is perhaps surprising that these problems exhibit the _ same _ pattern of boundary and interface currents . this can be understood as a consequence of confinement of these boundary modes at the interface between topologically mismatched gapped ground states on either side of the interface . it is therefore of interest to understand precisely how these different problems are related in the bulk . in this paper we observe that these two situations are in fact gauge equivalent representations of the same problem . however our demonstration of this equivalence requires that we extend both problems in a nambu - doubled formulation , explicitly restoring particle hole symmetry about a nonzero chemical potential . in this doubled representation we find that the problems are interconverted by local gauge transformations exploiting the particle and hole degrees of freedom in the nambu basis . among the insights provided by this approach , we observe that the interfacial degrees of freedom common to the two problems ( the snake states ) are mapped to a model for the bogoliubov quasiparticles in a one - dimensional superconductor along the tangent line that supports a p - wave " pairing field . this paper is organized as follows . in section i we begin by comparing the spectra for antisymmetrically doped graphene ( a _ pn _ junction ) in uniform field with the spectrum for uniform doping in an antisymmetric field . in section ii we present a family of local gauge transformations that allow one to map one problem onto the other . in section iii we implement this procedure for the case of the graphene _ pn _ junction and analyze the structure of the uniformly doped antisymmetric field problem to which it maps . in section iv we present a topological analysis of the ground state manifolds in these models , concluding that they are the same and are members of the altland - zirnbauer chiral symmetry class c , i.e. they are indexed by even integer - valued winding numbers . in section v we use a variational approach to determine the spatial structure of the interface modes . we consider snake state solutions in two limits of a tight - binding theory for electrons on a honeycomb lattice . the hamiltonian is @xmath8 where @xmath9 is the peierls phase accumulated in a ( possibly nonconstant ) perpendicular magnetic field @xmath10 . we adopt a coordinate system where the scalar potential @xmath11 and the vector potential @xmath12 are spatially varying in the @xmath13 direction and constant along @xmath14 . the chemical potential @xmath15 is set so that the left and right sectors are simultaneously gapped . the calculations are carried out for zigzag " interfaces , where a domain wall at @xmath16 is tangent to a primitive translation vector along @xmath14 . we examine two limiting domain wall geometries . in the first we assume that the system is uniformly doped , say @xmath17-type on both sides of an interface where @xmath18 changes sign . in the second we consider the complementary case where the @xmath0 field is uniform ( or at least a symmetric function of @xmath13 with no sign changes ) , and instead the external scalar potential @xmath19 with zero mean changes its sign on an interface defining a @xmath7 junction . we show that the edge - state solutions for these two limits are the same despite the different microscopic dynamics . this manifests a topological equivalence of their bulk ground states . indeed we find that these can be mapped into each other by gauge transformations that mix the particle and hole degrees of freedom when the problem is rewritten in a nambu particle - hole basis . this leads to the possibility of inventing architectures that simulate unusual ballistic transport effects like andreev reflection even in the absence of a physical superconducting condensate . in the following , we will first discuss the two limits separately and then analyze their gauge equivalence . we assume that the system is uniformly slightly doped @xmath17-type and has an antisymmetric magnetic field profile @xmath20 . the spectrum for this problem is displayed in fig . [ fig1 ] . the vertical red lines denote the projections of the bulk @xmath21 and @xmath22 dirac points , i.e. in the absence of a field these are the interface - projected locations of the bulk gap closures . for an antisymmetric @xmath18 the vector potential in landau gauge is an even function with @xmath23 . consequently the system supports normalizable ( near ) zero - energy states that are one - sided " in momentum space as shown in fig . there are two types of momentum - space anisotropy evident in these spectra : ( a ) the @xmath0-induced zero modes occur only for @xmath24 in both valleys ( b ) the @xmath25 spectrum near @xmath22 supports an additional pair of zero modes due to the undercoordinated atoms at the zigzag edges ( the total orbital degeneracy of the @xmath25 spectrum is actually four in this region ) . the additional low - energy modes bridge the @xmath22 and @xmath21 points where they smoothly evolve into the field - induced zero modes in the opposite valley . in either case the transition from @xmath25 to @xmath26 marks a crossover where the zero - energy degrees of freedom hybridize to produce a pair of particle - hole symmetric propagating modes that are confined to the domain wall . for @xmath17 type doping ( as illustrated ) this pair of interfacial modes copropagate along @xmath27 . physically the pair of domain wall modes combine cyclotron orbital states of opposite circulation to confine their motion near the interface . the dispersion of the outer - edge modes near the right - hand @xmath22 valley is particularly instructive . note that this band is nearly flat for small @xmath25 but it becomes strongly dispersive with positive group velocity for sufficiently large negative @xmath28 . this occurs via hybridization of the sublattice - polarized edge degree of freedom with the landau zero mode _ on the opposite sublattice _ when their guiding centers are forced to the outer edges of the ribbon . note that the antisymmetry of @xmath18 requires that the guiding centers are forced to opposite outer edges at the same value of the crystal momentum @xmath29 . these dispersive outer edge modes constitute a return path for the topological current induced in the domain wall . these features can be identified in the spatial distribution of the probability ( or charge ) densities plotted in figs . [ fig2 ] and [ fig3 ] . to summarize , for constant @xmath1 and antisymmetric @xmath0 we observe ( a ) four interface / edge modes at the fermi energy . ( b ) a pair of co - propagating modes at the domain wall which combine cyclotron motions of opposite circulation . ( c ) outer edge modes that hybridize the zeroth landau level with the zigzag surface state . ( d ) valley asymmetry : domain wall modes occur in both valleys but there is support for the outer edge modes only in a single valley . reversal of the direction of @xmath0 everywhere will select the other valley . numerically calculated spectrum for graphene with uniform doping in an antisymmetric magnetic field profile . we plot @xmath30 as a function of @xmath31 , where @xmath32 is the inter - atomic distance in graphene , and assume @xmath17-type doping ( i.e. , @xmath33 , and a uniform scalar potential @xmath1 that shifts the system away from neutrality ) . the spectrum shows four dispersing features at the fermi energy : two with positive " velocity on the outer edges , and a pair of modes with negative " velocity in the domain wall . the flat band that extends from the @xmath21 to @xmath22 points is the surface state for a zigzag edge . the horizontal dashed lines show the dispersionless bulk landau levels for a dirac system . ] numerically calculated probability ( or charge ) density for one of the interface zero - energy modes shown in fig . the mode is associated with valley @xmath21 for uniform doping in an antisymmetric magnetic field profile , and @xmath34 ( solid blue ) and @xmath35 ( red dashed ) are the two spinor components . for the other valley , @xmath22 the density of the @xmath32 and @xmath36 components of the wave function are interchanged . the @xmath13-axis is measured in units of the interatomic distance @xmath32 in graphene . ] numerically calculated probability ( or charge ) density for the zero - energy modes shown in fig . the two peaks in the center correspond to the interface modes of fig . the peaks on the left and right side correspond to the edge modes . the solid blue ( red dashed ) curve represent the @xmath34 ( @xmath35 ) spinor component . inset : schematic of the current pattern due to opposite cyclotron motions in the two halves of the setup . ] we now consider the opposite limit that occurs with uniform magnetic field and an antisymmetric bias @xmath37 . this creates a graphene @xmath7 junction in a uniform field which is the situation studied in two recent experiments @xcite . the spectrum calculated for this configuration is displayed in fig . [ fig4 ] where we plot @xmath30 as a function of @xmath29 . here the system is @xmath38-doped for @xmath39 and @xmath17-doped for @xmath40 . again one finds four dispersing modes at the fermi energy : two with negative velocity at the domain wall , and two with positive velocity confined on the outer edges . despite this similarity , the mechanism producing the edge state structure is quite different . we note that the @xmath41 signature of the landau quantization of the dirac spectrum is observed for @xmath42 in the @xmath21 valley but for @xmath43 in the @xmath22 valley , i.e. the landau quantized spectra are both one - sided in momentum space , but with opposite senses in the two valleys . in the forbidden regions @xmath25 and @xmath44 the spectrum collapses to a pair of nearly degenerate orbital doublets that connect the two valleys . this degeneracy is exact at @xmath45 : the energy jump that is produced by the transition @xmath26 to @xmath25 ( and vice versa for @xmath46 ) is the quantized energy spacing between the zeroth and first landau levels , so all the levels for @xmath47 are twofold degenerate . here the dispersion of the confined interfacial modes can be understood as a response to the lateral electric field produced in the @xmath7 junction . as @xmath13 crosses zero the scalar potential @xmath19 switches its sign and the internal electric field @xmath48 is nonzero . thus a state with drift velocity @xmath49 sees no deflection and can propagate freely . this can be contrasted with the guiding - center mechanism that liberates these modes in the former antisymmetric-@xmath0 problem where @xmath50 and one requires the compensation of the circulation in orbits in reversed @xmath0 fields to produce freely propagating interfacial snake states . interestingly , the appearance of dispersive edge modes on the _ outer _ boundaries follows exactly the same recipe as for the antisymmetric @xmath0 problem . the guiding center of the landau zero mode which is sublattice polarized , is forced to the outer edge of the ribbon where it hybridizes with the zigzag surface state on a complementary sublattice to form the one - way dispersive excitation . however , because the @xmath0 field is constant in this problem , the guiding centers are forced to the outer edges of the ribbon at opposite momenta @xmath51 . the entire spectrum of fig . [ fig4 ] is then invariant under the combined transformation @xmath52 and @xmath53 . the probability ( or charge ) densities associated with the zero modes is plotted in figs . [ fig5 ] and [ fig6 ] . we note that the charge density shows a sublattice polarization , favoring the sublattice found in landau zero mode . this sublattice polarization is captured in the variational approach presented in section v. to summarize the main results from this model for the graphene @xmath7 junction in a uniform @xmath0 : ( a ) four interface / edge modes at the fermi energy . ( b ) co - propagating modes in the domain wall determined by their drift velocity specified by @xmath0 and the potential gradient in the wall . ( c ) conventional outer edge modes that hybridize a landau zero mode with the surface state . spectra for the @xmath7 junction in a uniform field . the plot gives @xmath30 as a function of @xmath29 for a geometry where @xmath0 is constant but @xmath37 . the system supports four edge and interface modes : two positive velocity modes on the outer edges and two negative velocity modes at the domain wall . the flat feature is a zigzag edge state that morphs into the zeroth landau level . the spectra are one - sided , and show graphene character near a shifted neutrality point of one sign of @xmath54 in one valley and the opposite sign in the other . at @xmath45 the spectra are twofold degenerate . ] numerically calculated probability ( or charge ) density for one of the interface zero - energy modes shown in fig . the mode is associated with valley @xmath21 for a graphene _ pn _ junction with the potential profile @xmath37 in a homogeneous perpendicular magnetic field , @xmath34 and @xmath35 are the two spinor components . for the other valley , @xmath22 the density of the @xmath32 and @xmath36 components of the wave function are interchanged . ] numerically calculated probability ( or charge ) density for the zero - energy modes shown in fig . the two peaks in the center correspond to the interface modes of fig . the peaks on the left and right side correspond to the edge modes . the solid blue ( red dashed ) curve represent the @xmath34 ( @xmath35 ) spinor component . inset : schematic of the current pattern showing the two co - propagating modes at the interface between the two half of the setup . the electric field @xmath55 corresponds to the potential gradient at the @xmath7 junction . ] it is striking that despite the different microscopic origins of the domain wall solutions and the different structure of the full spectra apparent in figs . [ fig1 ] and [ fig4 ] , the basic pattern of the edge state currents is the same . this is evidence of the topological character of these modes . by negating either @xmath0 or @xmath1 at the interface we reverse the sign of the chern number in the first fundamental gap between landau levels and therefore we require the same pattern of boundary currents . this suggests that the ground states of these two systems can be adiabatically mapped into each other . this conclusion is surprising since the momentum - space structures of their spectra examined in the previous section are evidently controlled by the underlying dynamics which are quite different and in fact incompatible for the two states . in this section we show that their ground states can nonetheless be mapped into each other using a particle - hole extension of the original formulations of both problems . the required mapping is a rotation in the particle and hole degrees of freedom expressed in a nambu basis . local gauge transformations in this basis interconvert the two problems at the expense of introducing a fictitious pairing field within the domain wall . in this section we develop a family of such mappings and discuss the consequences of the induced pseudo - pairing field in section iii . the problems of section i.a and i.b are distinguished by the coupling of external potentials to bilinear terms in the fermion operators @xmath56 and @xmath57 . for example , the peierls phase in eq . ( [ tb_hamiltonian ] ) is coupled to nearest - neighbor bilinear terms in the form @xmath58 while the scalar potential that defines the local doping is coupled through the site density operator @xmath59 sign reversal of the magnetic field direction negates the exponentiated phase in eq . ( [ peierls ] ) while a reversal of the scalar potential flips the sign of the coupling to the net charge operator eq . ( [ charge ] ) . ignoring the physical spin of the electrons , these reversals are introduced by the particle - hole transformation @xmath60 whereby @xmath61 for our application it is useful to collect these operators in two - component spinors that resolve the two degrees of freedom at each bloch wavevector @xmath62 @xmath63 and write the original problem in a doubled nambu four - component basis @xmath64 @xmath65 when summed over @xmath62 this theory gives a doubled redundant " description of the original problem . using eqs . ( [ reversal ] ) a global rotation in particle - hole space can globally choose " the signs of @xmath0 and @xmath1 . for example , consider a hamiltonian @xmath66 $ ] parameterized by the fields @xmath19 and @xmath18 . then define @xmath67 pauli matrices @xmath68 acting on the two sublattice degrees of freedom and @xmath69 acting on the particle and hole degrees of freedom in the nambu representation . a global operator of the form @xmath70 where @xmath71 and @xmath72 , has the property of formally flipping the signs of @xmath1 and @xmath0 everywhere in the manner @xmath73 { \cal s}(\theta ) = { \cal h}[-v(x),-b(x)]\:.\end{aligned}\ ] ] we will discuss the behavior of @xmath74 first and then return to the interpretation of the remaining phase degree of freedom @xmath75 . in an analogous manner , if we promote @xmath76 to a local gauge degree of freedom we can introduce a gauge transformation @xmath77 that _ locally _ defines the signs of @xmath1 and @xmath0 . specifically , we can use this to interconvert the domain wall configurations of sections i.a and i.b . to keep track of the signs of @xmath1 and @xmath0 in the left and right spaces we use a shorthand notation @xmath78 $ ] where @xmath79 and @xmath80 specify the asymptotic signs of the potential and magnetic field strength . in this notation , @xmath81 $ ] denotes a situation with @xmath82 on the left and @xmath83 on the right , all immersed in a uniform positive field @xmath84 . we now introduce a local gauge transformation @xmath85 where @xmath86 , @xmath87 and @xmath88 . this has the effect of implementing a one - sided particle - hole transformation , where the local gauge transformation evolves smoothly through the interface . we retain the original problem for @xmath89 but swap particle and hole amplitudes for @xmath90 to invert the signs of @xmath1 and @xmath0 . this transformation is unitary and performs the mapping @xmath91 { \cal s}^\dag = { \cal h}[+,+;-,+]\ : , \label{mapping}\end{aligned}\ ] ] thereby swapping the representation of a @xmath7 junction in a uniform field with a system with uniform doping in an antisymmetric @xmath0 field . in the doubled space the ground states can be identified implying that the zero - mode structure is unchanged . in fig . [ fig7 ] we overlay the spectra calculated for the two problems in the nambu representation , illustrating this correspondence . bogoliubov spectrum for the topological domain wall in graphene corresponding to both figs . [ fig1 ] and [ fig4 ] . the spectrum is the particle - hole doubled version of the antisymmetric @xmath0 spectrum , which is plotted only in the particle channel in fig . the blue lines correspond to the antisymmetric magnetic field spectrum as shown in fig . [ fig1 ] ; the orange lines correspond to the spectrum of a @xmath7 junction in a uniform magnetic field as shown in fig . the horizontal dashed green lines show the landau - level spectrum for a dirac particle in a uniform magnetic field for a chemical potential lying symmetrically between the zeroth and the first landau level . ] an interesting consequence of this mapping is the structure of the spectrum inside the domain wall . the control parameter @xmath92 varies smoothly through the wall and has the value @xmath88 exactly at its center . the transformation in eq . ( [ local_gauge_transf ] ) when @xmath93 is @xmath94 as an example , for graphene we explicitly show the complex structure of its sublattice off - diagonal terms : @xmath95 inserting this in eq . ( [ h_nambu ] ) , in the domain wall the transformed hamiltonian has the form @xmath96 & -iv & { \rm im}[\gamma ] \\ { \rm re}[\gamma ] & 0 & { \rm im}[\gamma ] & iv \\ iv & { \rm im}[\gamma ] & 0 & -{\rm re}[\gamma ] \\ { \rm im}[\gamma ] & -iv & -{\rm re}[\gamma ] & 0 \\ \end{array } \right)\:.\end{aligned}\ ] ] thus , there is a pairing amplitude @xmath97 $ ] defined by the local potential @xmath1 and the _ imaginary part _ of the hopping amplitude @xmath98 $ ] . it is a matrix pairing operator acting on the sublattice degrees of freedom . it is symmetric and thereby represents an effective p - wave pairing . the appearance of an interfacial pair amplitude in the transformed problem can be interpreted as follows . suppose we introduced the local gauge transformation in the absence of a domain wall . then in the particle language , electrons would pass through the putative interface undeflected . but in the transformed language this means that an incident electron is converted into a hole with unit probability . this is an andreev process that requires a pairing field in the bogoliubov hamiltonian . note that a relation of this form indicates a formal mapping between the problem of klein tunneling @xcite and andreev reflection at a boundary in the dirac theory . a closely related observation was made by beenakker and colleagues a few years ago @xcite who noticed that the reflection amplitude for electrons incident on a symmetric graphene @xmath7 junction from the @xmath38-doped side has a precise analogy with a problem where the particles were actually being andreev reflected by a contact with a superconductor . this work did not present the problem in a nambu basis and did not access the physics that we discuss below . we now look at the symmetries of the original and transformed problem in the domain wall . the original problem can be expressed , using @xmath99 $ ] , @xmath98 $ ] , and suppressing the @xmath100 , as @xmath101 while after the transformation we have @xmath102 in both cases the @xmath1 " term commutes with the @xmath103 and @xmath104 terms while the @xmath103 and @xmath104 terms anticommute . for the @xmath7 junction problem , @xmath105 in the interface . for the antisymmetric @xmath0 problem @xmath1 is constant and can be absorbed in the chemical potential . then because of the anticommutation rules for the @xmath106 terms in eq . ( [ h_domain ] ) , zero - energy solutions can only occur when @xmath103 and @xmath104 simultaneously vanish , as they do precisely at the ( projected ) dirac point . in the transformed language this means that a zero - energy solution in the bogoliubov spectrum for the pseudo - superconductor will similarly require that @xmath107 which reveals a momentum - space linear node in the pairing field , consistent with its p - wave symmetry . the bogoliubov spectrum is invariant under global @xmath108 gauge transformations of its matrix - valued pair field @xmath109 in the interface @xmath110 this gauge degree of freedom can be identified with the continuous family of possible transformations in eq . ( [ s_theta ] ) that interchange the particle and hole degrees of freedom in the nambu representation . thus any global @xmath108 gauge transformation at the interface ( @xmath16 ) can be absorbed in a redefinition of the global phase angle @xmath75 that defines the rotation that is used to switch the particle and hole subspaces . note that this choice is invisible in the two bounding states but it does appear in the theory of the interface . however , a variation of the phase @xmath111 along the interface is a gauge choice and it is not associated with a physical charge current along its tangent line . to see this we observe that if @xmath75 is promoted to a local ( i.e. @xmath14-dependent ) @xmath108 degree of freedom it also can be eliminated by a local @xmath14-dependent gauge transformation back to a number conserving representation at the expense of introducing a connection for operators that transport electrons ( holes ) along the @xmath14 direction . here one can verify that for generic hamiltonians in the form @xmath112 the generalization of eq . ( [ alpha_pi_over4 ] ) @xmath113 satisfies @xmath114 so that any variation @xmath111 in the interfacial pair field disappears completely and leaves no residual signature in the number - conserving blocks of the back - transformed hamiltonian . the interfacial problem presents the more interesting case of a chiral theory where the hamiltonian instead has the structure @xmath115 . here the connection introduces terms in the hamiltonian @xmath116 \partial_y \theta /2\:.\end{aligned}\ ] ] the first term commutes with the number conserving blocks and leads to a momentum boost of the spectrum @xmath117 . the remaining terms anticommute with the number conserving blocks so their effects appear only at higher order @xmath118 and are unimportant for smoothly varying @xmath111 . the momentum boost has no effect on the boundary currents which depend on the number of zero - energy intersections of the chiral branch of the hamiltonian . the boost can be interpreted as producing a valley polarization within the occupied manifold , however lacking a sharp definition of this polarization ( the valleys are connected in the full lattice theory ) , we expect that any variation @xmath111 has no effects on physically measurable quantities . particle - hole doubling of the domain wall problem promotes its hamiltonian to a bogoliubov de gennes ( bdg ) symmetry class . in the original number - conserving representation , this system is gapped breaking time - reversal symmetry ( @xmath119 ) , particle - hole symmetry and chiral symmetry ( @xmath120 ) . it is described by the altland - zirnbauer @xcite unitary symmetry class a which supports topologically nontrivial ground states in two dimensions which are indexed by an integer - valued invariant @xmath121 . the count of the interface states in sections i.a and i.b manifests the mismatch @xmath122 for the two bounding gapped states across the domain wall . the nambu - doubled version of this problem explicitly restores particle - hole symmetry @xmath3 so that the extended problem can be described by either altland - zirnbauer symmetry classes d or c , depending on whether @xmath123 equals @xmath124 ( class d ) or @xmath125 ( class c ) . for our application this distinction is important since in the former case the ground state becomes topologically trivial and does not support any symmetry protected modes at its edges or the interfaces . in the latter case ( c ) the ground state has a @xmath126 topological classification requiring an _ even _ number of interfacial / edge modes in the spectrum . in either case odd _ number of topologically protected edge modes is excluded . combining the two conditions @xmath127 we have @xmath128 , so that @xmath129 and the extended problem is a member of class c. we have verified this by an explicit calculation of the winding number for the ground state . we carry this out in the number conserving representation where the topological invariants from the particle and hole sectors can be summed . the chern number ( @xmath130 ) for the ground state can be calculated by choosing a chemical potential lying in the first fundamental gap between landau levels and summing up the chern numbers for all the individual bands below the chemical potential . the chern number of the @xmath38th bloch band is @xmath131 where the berry connection @xmath132 for @xmath133 . here @xmath134 represents the normalized wavefunction of the @xmath38th bloch band . we compute the chern number by discretizing the brillouin zone ( bz ) and summing up the berry curvature defined on each of the discretized plaquettes@xcite . after discretization , the chern number for the @xmath38th band is @xmath135 where @xmath136 are within the first bz , @xmath137 and @xmath138 the column vectors of the matrix @xmath139 are given by the bloch eigenstates @xmath134 , where we include all the bloch bands below the chemical potential . the bloch eigenstates @xmath134 are obtained by numerically diagonalizing the tight - binding hamiltonian for the graphene lattice with nearest - neighbor hopping , where the phase of the hopping amplitude is determined via the peierls substitution . assuming that system is of @xmath38-type and the magnetic field is uniform , adding up the chern numbers for the particle sector @xmath140 and the hole sector @xmath140 , we find that in the ground state @xmath141 . similarly , we find that @xmath142 for a @xmath17-type system in a uniform magnetic field . this mismatch of chern numbers suggests that there should be four topologically protected edge states at the interface defining the @xmath7 junction . however , since we have artificially doubled the spectrum by going to the nambu basis , the genuine number of edge states at the interface is two . the following thought experiment provides an alternative route to demonstrating the equivalence of the @xmath143 = [ + , -,+,+]$ ] and @xmath144 $ ] interfaces . the two situations are distinguished by the substitution of @xmath145 $ ] by @xmath146 $ ] in the right hand space . this replacement simultaneously negates the field direction _ and _ the chemical potential and therefore it leaves the chern number of the gapped ground state unchanged . this substitution can also be regarded as resulting from the gauge transformation in eq . ( [ mapping ] ) imposed on a uniform @xmath147 state and implemented on a line _ displaced from the physical interface_. thus this second wall is topologically trivial and does not support any protected interfacial modes . reducing the separation between the original ( physical ) @xmath148 $ ] interface and the second ( fictitious ) interface to zero generates the @xmath144 $ ] problem without changing the pattern of domain wall currents . although the full energy spectrum of the graphene honeycomb lattice can be calculated using a tight - binding approximation , additional insight can be obtained from the continuum approximation . in this case a taylor expansion centered at two non - equivalent dirac points @xmath149 and @xmath150 produces the dirac - like hamiltonians @xmath151 where @xmath152 acts on the @xmath21 and @xmath22 valley indices and @xmath153 is the fermi velocity . together these hamiltonians can be combined to reproduce a @xmath154 two - valley hamiltonian in terms of the dirac gamma matrices@xcite . due to the similarities of the two hamiltonians , here we will only consider the @xmath149 point where the hamiltonian with an external potential @xmath19 becomes @xcite @xmath155 where @xmath156 are pauli matrices in the sublattice basis , @xmath157 is the momentum operator , @xmath158 is the identity matrix , and @xmath19 is an external or bias potential associated with an applied electric field . below we consider the two cases of antisymmetric electric and symmetric magnetic fields and vice versa . in either case the vector potential is written in landau gauge with @xmath159 and the dirac hamiltonian associated with the @xmath21 valley and normalized for @xmath160 has the form @xmath161 the solution for the dirac equation @xmath162 can be written in the bloch form @xmath163 indexed by the conserved wave vector @xmath29 . after this substitution the dirac equation is rewritten as a pair of coupled second - order equations : @xmath164 \hat{\mathbb{i } } - \partial_x a_y \sigma_z \right ) \psi \nonumber\\ = \left ( ( \epsilon - v)^2 \hat{\mathbb{i } } + i \partial_x v \sigma_x \right ) \psi\end{aligned}\ ] ] which defines a complex matrix - valued potential @xmath165 in the case of an antisymmetric @xmath0 field with uniform doping , @xmath166 and these equations are decoupled in the sublattice basis so they can be solved separately . for both sublattices we have a particle described by a massive schrdinger - like equation in a double parabolic potential . in the case when @xmath167 the potentials are shown in the upper panel of fig . the minima in these potentials occur at the values @xmath168 , offset in energy by the pseudo - zeeman @xmath169 term in eq . ( [ matrix_potential ] ) . the @xmath32 and @xmath36 sublattice potentials are exchanged by the reflection @xmath170 . equation ( [ valley_hamiltonian ] ) reveals that these two amplitudes are not independent , but are coupled through the @xmath171 term in the linearized hamiltonian . the form of the effective potential @xmath172 suggests a useful variational basis for studying the states bound to the domain wall , @xmath173 which are two overlapping gaussians each of width @xmath174 and separated by @xmath175 ; @xmath176 is a normalization factor . in this variational space the off - diagonal coupling describes a tunneling between states on opposite sides of the interface and with opposite sublattice polarizations . evaluating the matrix element one finds @xmath177 which is optimized for @xmath178 in a gauge where a minimum in the effective potential occurs at @xmath16 when @xmath179 . this demonstrates that the localized variational states are most effectively tunnel - coupled at a finite range from the domain wall . the charge density in the lowest energy state is therefore sublattice polarized on opposite sides of the interface as seen in the charge density plotted in fig . the tunneling hamiltonian in this variational basis is proportional to @xmath180 . the energy - optimized state is a coherent combination of these trial basis states forming an eigenfunction of the velocity operator @xmath181 with negative eigenvalue : this is a domain wall state propagating along the @xmath182 direction . the same physics occurs in the @xmath22 valley . in either case the variational solution describes a situation where an incident electron bound in a cyclotron orbit is transmitted through the domain wall where the field is reversed producing a counter circulation of the orbit and liberating the average motion along the tangent line of the wall . upper panel : potential of the interface created by an antisymmetric magnetic field @xmath183 . for both sublattices we have a pseudo - zeeman shifted double parabolic potential . lower panel : potential of the interface created by an antisymmetric electric bias field @xmath184 . it consists of one parabola , which has a jump at the interface related to the value of the electric field in the @xmath7 junction , e.g. as @xmath185 . ] in the case of a homogeneous magnetic field and antisymmetric doping , when @xmath186 , the @xmath32 and @xmath36 sublattice solutions are coupled by the @xmath187 term in the potential eq . ( [ matrix_potential ] ) . for the specific example considered here , when @xmath188 this coupling appears as boundary conditions at the interface , @xmath189 where @xmath190 is the jump of the electrostatic potential at the interface . the potentials in shown in the lower panel of fig . [ fig8 ] again suggest a variational basis . here the dominant low - energy degree of freedom is @xmath32-sublattice polarized in the lowest landau level on the right side ( low potential side ) of the interface . it is coupled to an evanescent mode on the right which is @xmath36-sublattice polarized penetrating a barrier from the pseudo - zeeman field , and to evanescent modes on the left penetrating the electrostatic barrier on both sublattices . this physics is captured in the variational basis @xmath191 where @xmath192 and @xmath193 are decay lengths for penetrating the pseudo - zeeman barrier and the electrostatic barrier , respectively . the matching condition can be written in matrix form @xmath194 a nontrivial solution occurs when the determinant of the matching matrix is zero , giving @xmath195 where @xmath196 . the interfacial mode is a vector in the nullspace of eq . ( [ matching ] ) @xmath197 here , @xmath198 so the group velocity is directed along the @xmath182 direction . this velocity is evidently proportional to @xmath190 , consistent with its identification with the classical drift velocity @xmath199 expected in the classical theory of an edge / interface state . note also the sublattice asymmetry in this boundary solution : the interfacial mode is not in a velocity eigenstate as was found in the antisymmetric @xmath0 problem and has its dominant amplitude on the sublattice found in the zero - energy mode ( lowest landau level ) in each valley . this is clearly seen in the charge densities calculated for a version of this interface plotted in fig . the evanescent form of the trial functions in eq . ( [ var_basis ] ) reflects the different physics operative in the electrostatic barrier . here an electron bound in a cyclotron orbit is backscattered by an electric field at the interface , weakly penetrating both an electrostatic barrier and a zeeman barrier due to the sublattice polarization . this behavior is illustrated in fig . [ fig9 ] . the two different types of domain walls support two different kinds of semiclassical trajectories . solid line : skipping orbits typical for the uniform magnetic field case at a @xmath200 domain wall . dashed lines : snake orbits typical for a @xmath201 domain wall . ] graphene interfaces that support electrostatic barriers ( at a @xmath7 junction ) or field reversal ( at a magnetic domain wall ) can host confined snake state solutions . the dynamics responsible for these solutions is clearly different in these two situations . for the magnetic domain wall cyclotron orbits of opposite circulation are matched on a boundary to produce an unbound propagating excitation . for the electrostatic barrier a transverse electric field is introduced that reflects an incident mobile carrier to produce a skipping orbit drifting at velocity @xmath199 . despite these differences the pattern of boundary currents is the same demonstrating the common topological character of both interfaces . we found that these two problems can be mapped onto each other using a particle - hole doubled representation . in this extended basis the two problems are interconverted by a local gauge transformation using the particle and hole degrees of freedom in the nambu basis . an interesting and unavoidable consequence of this mapping is that a number - conserving version of one problem is the image of a bogoliubov de gennes problem for fermions with an interfacial pair field in the other . this relation offers the interesting possibility of generalizing coherent quantum transport phenomena such as andreev reflection or veselago lensing , to a new family of graphene - derived architectures . our results imply two no - go " theorems for this system . first , the gauge - transformed representation of the interfacial problem describes a one - dimensional particle / hole gas coupled by a pairing field with p - wave symmetry . in principle , this class of models can support majorana excitations in an appropriate parameter regime @xcite . however , the prospects for realizing such excitations in this setup are remote . we find that the doubled problem has the discrete symmetries of altland - zirnbauer symmetry class c so that its gapped ground state is indexed by an _ even _ integer - valued ( @xmath202 ) index . here topological domain wall solutions must appear in pairs , and the possibility of having an unpaired majorana excitation appearing on the boundary , or at its ends are excluded . this is understandable , since the appearance of such an excitation would not admit an interpretation in the original ( number conserving ) representation of the same problem . second , it is intriguing that under the gauge transformation the interfacial problem exhibits an apparent broken @xmath108 gauge symmetry due to the pseudo - pairing field . however , this seems not to be associated with any measurable collective effects in the interface : spatial variations of the phase of the order parameter are the images of momentum shifts of the spectrum in the original number - conserving representation of the problem , which is simply a gauge choice . again , any nontrivial property arising from the pairing field would require a dual interpretation in the number - conserving representation of the same problem . it remains an open question as to whether one might further break the symmetries of the original problem to identify measurable consequences of these symmetries in its gauge transformed image . our results highlight several directions for exploiting the structure of the domain wall solutions . it is possible that the sublattice and valley asymmetries found for two problems could be exploited to valley filter ballistic transport in patterned graphene . the two - channel interfacial solutions may also provide a venue for important one - dimensional interaction effects that are accessible in nonlocal transport measurements between reservoirs bridged by a domain wall . finally , we also note that while graphene provides a natural starting point for developing this formulation , the conclusions appear to be generally valid for ballistic transport in domain walls separating topologically gapped ground states in a wider class of semiconductor nanostructures . we would like to acknowledge fruitful discussions with p. makk , p. rickhaus , c. schnenberger , and c.l . the work of c.b . and was financially supported by the swiss snf and the nccr quantum science and technology . e.j.m . is supported by doe grant de - fg02-er45118 and a grant from the leverhulme foundation at loughborough university while this work was carried out .
snake states are open trajectories for charged particles propagating in two dimensions under the influence of a spatially varying perpendicular magnetic field . in the quantum limit they are protected edge modes that separate topologically inequivalent ground states and can also occur when the particle density rather than the field is made nonuniform . we examine the correspondence of snake trajectories in single - layer graphene in the quantum limit for two families of domain walls : ( a ) a uniform doped carrier density in an antisymmetric field profile and ( b ) antisymmetric carrier distribution in a uniform field . these families support different internal symmetries but the same pattern of boundary and interface currents . we demonstrate that these physically different situations are gauge equivalent when rewritten in a nambu doubled formulation of the two limiting problems . using gauge transformations in particle - hole space to connect these problems , we map the protected interfacial modes to the bogoliubov quasiparticles of an interfacial one - dimensional p - wave paired state . a variational model is introduced to interpret the interfacial solutions of both domain wall problems .
energy harvesting from environment resources ( e.g. , through solar panels , wind power , or geo - thermal power ) is a potential technique to reduce the energy cost of operating the base stations ( bss ) in emerging multi - tier cellular networks . while this solution may not be practically feasible for macrocell base stations ( mbss ) due to their high power consumption and stochastic nature of energy harvesting sources , it is appealing for small cell bss ( sbss ) that typically consume less power @xcite . providing grid power to all sbss may not always be feasible due to their possible outdoor / remote / hard - to - reach locations . wireless energy harvesting thus enables dense deployment of sbss irrespective of the availability of grid power connections . in general , wireless energy harvesting can be classified into the following two categories : ambient energy harvesting and dedicated energy harvesting . in the former case , energy harvested from renewable energy sources ( such as thermal , solar , wind ) as well as the energy harvested from the radio signals in the environment can be sensed by energy - harvesting receivers . in the latter case , energy from dedicated sources is transmitted to energy - harvesting devices to charge them wirelessly . designing efficient power control policies with different objectives ( e.g. , maximizing system throughput ) is among one of the major challenges in energy - harvesting networks . in @xcite , the authors proposed an offline power control policy for two - hop transmission systems assuming energy arrival information at the nodes . the optimal transmission policy was given by the directional water filling method . in @xcite , the authors generalized this idea to the case where many sources supply energy to the destinations using a single relay . a water filling algorithm was proposed to minimize the probability of outage . although the offline power control policies provide an upper bound and heuristic for online algorithms , the knowledge of energy / data arrivals is required which may not be feasible in practice . in @xcite , the authors proposed a two - state markov decision process ( mdp ) model for a single energy - harvesting device considering random rate of energy arrival and different priority levels for the data packets . the authors proposed a low - cost balance policy to maximize the system throughput by adapting the energy harvesting state , such that , on average , the harvested and consumed energy remain balanced . recently , in @xcite , the outage performance analysis was conducted for a multi - tier cellular network in which all bss are powered by the harvested energy . a detailed survey on energy harvesting systems can be found in @xcite where the authors summarized the current research trends and potential challenges . compared to the existing literature on energy - harvesting systems , this paper considers the power control problem for downlink transmission in two - tier macrocell - small cell networks considering stochastic nature of the energy arrival process at the sbss . in particular , we assume that ambient energy harvesting is exploited at a central energy storage ( ces ) from where energy can be transferred to the sbss , for example , by using dedicated power beacons ( pbs ) . pbs are low - cost devices that can potentially charge wireless terminals by transferring energy in a directional manner . note that the power control policies at the mbs and the sbss and their resulting interference levels directly affect the overall system performance . the design of efficient power control policies is thus of paramount importance . in the above context , we formulate a discounted stochastic game model in which all sbss form a coalition to compete with the mbs in order to achieve the target signal - to - interference - plus - noise ratio ( @xmath0 ) of their users through transmit power control . that is , the mbs and the ces ( which actually represents the set of sbss in the game ) compete to achieve the desired @xmath0 targets of macrocell users and small cell users , respectively . note that both the mbs and the sbss transmit in the same channel ( i.e. , a co - channel deployment scenario is considered ) . therefore , the competition ( or conflict ) arises due to the resulting cross - tier interference , i.e. , as the mbs uses more power / energy to increase the utility of macrocell users , it results in higher cross - tier interference to small cell users . similarly , the more energy / power the ces assigns to sbss , the higher would be the cross - tier interference to macrocell users . note that the energy harvesting component is an important factor that indirectly contributes to the conflict . if the energy arrival rate is large , the sbss will have a larger energy pool to spend and thus cause more interference . clearly , we need to take into the account the probability of energy arrival when determining the optimal transmit power policy for each sbs . the amount of available energy at the transmitter will vary according the amount of power transmitted and the energy arrival during each transmission interval . naturally , the competition above can be modeled and analyzed by game theoretic tools . however , unlike in traditional power control games , the actions and the payoffs of the transmitters at successive transmission intervals are correlated . this correlation is taken into account in the proposed stochastic game model . for this game model , the nash equilibrium power control policy is obtained as the solution of a quadratic programming problem . for the case when the number of sbss is very large ( i.e. , an ultra - dense small cell network ( scn ) @xcite ) , the stochastic game is approximated by a mean field game ( mfg ) . in general , mfgs are designed to study the strategic decision making in very large populations of interacting individuals . recently , in @xcite , the authors proposed an mfg model to determine the optimal power control policy for a finite battery powered scn with no energy replacements . however , the stochastic nature of energy arrival for small cells was not considered . in this paper , we will consider the case where the battery can be recharged using random energy arrivals . by solving a set of forward and backward partial differential equations , we derive a distributed power control policy for each sbs using a stochastic mfg model . the contributions of the paper can be summarized as follows . 1 . for a two - tier macrocell - small cell network , we consider a centralized energy harvesting mechanism for the sbss in which energy is harvested and then distributed to the sbss through a ces . unlike in @xcite , the ces can have any finite number of energy levels in its storage , not only 0 and 1 . note that the concept of ces is somewhat similar to the concept of dedicated power beacons for wireless energy transfer to users in cellular networks @xcite . moreover , in a cloud - ran architecture @xcite , where along with data processing resources , a centralized cloud can also act as an energy farm that distributes energy to the remote radio heads each of which acts as an sbs . note that the sbss are not restricted to indoor deployments . subsequently , we formulate the power control problem for the mbs and sbss as a discrete single - controller stochastic game with two players . also , in this paper , we use the signal - to - interference - plus - noise - ratio ( sinr ) model instead of an snr model which has been commonly used in other research works on energy harvesting communication . consideration of random energy arrivals along with both co - tier and cross - tier interferences in the downlink power control problem is the major novelty of the paper . 2 . the existence of the nash equilibrium and pure stationary strategies for this single - controller stochastic game is proven . the power control policy is derived as the solution of a quadratic - constrained quadratic programming problem . when the network becomes very dense , a stochastic mfg model is used to obtain the power control policy as a solution of the forward and backward differential equations . in this case , each sbs can harvest , store energy and transmit data by itself . an algorithm using finite difference method is proposed to solve these forward - backward differential equations for the mfg model . numerical results demonstrate that the proposed power control policies offer reduced outage probability for the users served by the sbss when compared to the power control policies using a simple stackelberg game wherein each sbs tries to obtain the target @xmath0 of its users without considering the distribution of energy arrivals . the rest of the paper is organized as follows . section ii describes the system model and assumptions . the formulation of the single - controller stochastic game model for multiple sbss is presented in section iii . in section iv , we derive the distributed power control policy using a mfg model when the number of sbss increases asymptotically . performance evaluation results are presented in section v before the paper is concluded in section vi . we consider a single macrocell overlaid with @xmath1 small cells . the downlink co - channel time - slotted transmission of the mbs and sbss is considered and it is assumed that each bs serves only a single user on a given transmission channel during a transmission interval ( e.g. , time slot ) . the mbs uses a conventional power source and its transmit power level is quantized @xcite into a discrete set of power levels @xmath2 , where the subscript @xmath3 denotes the mbs . on the other hand , the sbss receive energy from a centralized energy storage ( ces ) , which harvests renewable energies from the environment . we assume that only the ces can store energy for future use and each sbs must consume all the energy it receives from the ces at every time slot . the energy arrives at the ces in the form of packets ( one energy packet corresponds to one energy level in ces ) . the quantization of energy arrival was assumed in other research studies such as in @xcite . the number of energy packet arrivals @xmath4 during any time interval @xmath5 is discrete and follows an arbitrary distribution , i.e. , @xmath6 . we assume that the battery at the ces has a finite storage @xmath7 . therefore , the number of energy packet arrivals is constrained by this limit and all the exceeding energy packets will be lost , i.e. , @xmath8 . the statistics of energy arrival is known _ a priori _ at both the mbs and the ces . at time @xmath5 , given the battery level @xmath9 , the number of energy packet arrivals @xmath4 , and the energy packets @xmath10 that the ces distributes to the @xmath1 sbss , the battery level @xmath11 at the next time slot can be calculated as follows : @xmath12 given @xmath10 energy packets to distribute , the ces will choose the best allocation method for the @xmath1 sbss according to their desired objectives . denoting the slot duration as @xmath13 and the volume of one energy packet as @xmath14 , we have the energies distributed to the @xmath1 sbss at time @xmath5 as @xmath15 where @xmath16 is the transmit power of sbs @xmath17 at time @xmath5 . clearly we must have : @xmath18 from the causality constraint , @xmath19 , i.e. , the ces can not send more energy than that it currently possesses . note that @xmath9 is the current battery level which is an integer and has its maximum size limited by @xmath7 . since the battery level of the ces and the number of packet arrivals are integer values , it follows from ( [ energy ] ) that @xmath10 is also an integer . similar to @xcite and @xcite , in our system model , the conflict between the ces and the mbs arises due to the interferences between the mbs and the sbss . clearly , if the mbs transmits with large power to achieve the @xmath0 targets for macrocell users , it will cause high interference to the small cell users . this means , the sbss will need to transmit with larger power to combat this cross - tier interference . the ces and the mbs have different objective functions and are free to choose any actions that maximize their own objectives ( i.e. , non - cooperative game ) . since the sbss can only use renewable energy , it is crucial for them to use their harvested energy economically . also , unlike a traditional one - shot transmit power control game , the ces needs to take into the account the future payoff given the current battery size and the probability of energy arrivals . in summary , for our ces model , at each time slot , we will have a random battery size at the ces and our objective is to maintain the long - term average @xmath0 close to the target value as much as possible and thus improve the outage probability . without a centralized ces - based architecture , each sbs can have different amount of harvested energy and in turn battery levels at each time slot , which will make this problem a multi - agent stochastic game @xcite . although this kind of game can be heuristically solved by using q - learning @xcite , the conditions for convergence to a nash equilibrium are often very strict and in many cases impractical . by introducing the ces , the number of the possible states of the game is simplified into the battery size of the ces , and the multi - player game is converted into a two - player game . another benefit of the centralized ces - based architecture is that the energy can be distributed based on the channel conditions of the users served by the sbss so that the total payoff will be higher than the case where each sbs individually stores and consumes the energy . for ease of exposition , in this paper , we consider an ideal energy transfer from the ces to the sbss . however , to model a simple energy loss , we can add a fixed percentage of loss into the energy consumption of the ces at each time slot . all the symbols that are used in the system model and section iii are listed in * table i*. @xmath17 and its associated user & @xmath20 & average channel gain between bs @xmath21 and user of bs @xmath17 + @xmath22 ( @xmath23 ) & target sinr for mbs ( sbs ) & @xmath9 & ( discrete ) battery level of ces at time @xmath5 + @xmath13 & duration of one time slot in seconds & @xmath10 & number of quanta distributed by the ces at time @xmath5 + @xmath24 & average interference at the user served by the mbs at time @xmath5 & @xmath25 & average interference at the user served by sbs @xmath17 at time @xmath5 + @xmath7 & maximum battery level of the ces & @xmath26 & finite set of transmit power of the mbs + @xmath27 , @xmath28 & & @xmath29 , @xmath30 & + @xmath31 & probability that the mbs chooses power @xmath32 when @xmath33 & @xmath34 & probability that the ces sends @xmath17 quanta when @xmath33 + @xmath4 & energy harvested at time @xmath5 & @xmath35 & probability that the ces starts with battery level @xmath36 + @xmath37 & discount factor of the stochastic game & @xmath38 & utility function of the mbs and the ces , respectively + @xmath39 & payoff matrix for the mbs and the ces , respectively & @xmath40 , @xmath41 & + the received @xmath0 at the user served by sbs @xmath17 in the downlink at time slot @xmath5 is defined as follows : @xmath42 where @xmath43 is the interference caused by other bss . @xmath44 is the channel gain between mbs and the user served by sbs @xmath17 , @xmath45 represents the channel gain between sbs @xmath17 and the user it serves , and @xmath46 is the channel gain between sbs @xmath21 and the user served by sbs @xmath17 . finally , @xmath16 represents the transmit power of sbs @xmath17 at time @xmath5 . the transmit power of mbs @xmath47 belongs to a discrete set @xmath48 @xcite . we ignore the thermal noise assuming that it is very small compared to the cross - tier interference . similarly , the sinr at a macrocell user can be calculated as follows : @xmath49 where @xmath50 is the cross - tier interference from @xmath1 sbss to the macrocell user , @xmath51 denotes the channel gain between the mbs and its user , @xmath52 represents the channel gain between sbs @xmath17 and macrocell user , and @xmath53 is the thermal noise . the channel gain @xmath46 is calculated based on path - loss and fading gain as follows : @xmath54 where @xmath55 is the distance from bs @xmath21 to user served by bs @xmath17 , @xmath56 follows a rayleigh distribution , and @xmath57 is the path - loss exponent . we assume that the @xmath1 sbss are randomly located around the mbs and the users are uniformly distributed within their coverage radii @xmath58 . during a transmission interval ( i.e. , a time slot ) , only one user is served by each sbs in the downlink direction . a stochastic game is a multiple stage game where it can have different states at each stage . each player chooses an action from a finite set of possible actions ( which can be different at each stage ) . the players actions and the current state jointly determine the payoff to each player and the transition probabilities to the succeeding state . the total payoff to a player is defined as the discounted sum of the stage payoffs or the limit inferior of the averages of the stage payoffs . the transition of the game at each time instant follows markovian property , i.e. , the current stage only depends on the previous one . in our model , the mbs and the sbss try to maintain the average @xmath0 of their users to be close to some targets . note that a large target @xmath0 means that a high transmit power will be required for the sbs which could be limited by the energy arrival rate and the battery size of the ces . also , a higher transmit power means a higher level of interference to other users . similar to @xcite and @xcite , the utility function of the mbs at time @xmath5 is defined as : @xmath59 where @xmath60 is the average interference at the macrocell user at time @xmath5 , and @xmath22 is the target @xmath0 for the macrocell user . clearly , this utility function is maximized when the @xmath0 at the mbs is @xmath22 . if the @xmath0 is larger than the target @xmath0 , this implies that mbs transmits with a larger power than necessary and thus wastes energy . on the other hand , if the @xmath0 is smaller than the target @xmath0 , it implies that energy is not utilized effectively provided it has sufficient energy . similarly , the utility function of the ces is defined as follows : @xmath61 where @xmath62 is the average interference at the user served by sbs @xmath17 at time @xmath5 . the arguments of both the utility functions demonstrate that the action at time @xmath5 for the mbs is its _ transmission power _ @xmath47 while the action of the ces is the _ number of energy packets @xmath10 that is used to transmit data from the sbss_. later , in * remark 2 * , we will show that the interference and the transmit power of each sbs can be derived from @xmath63 and @xmath64 . the conflict in the payoffs of both the players arises from their transmit powers that directly impact the cross - tier interference . note that the proposed single - controller approach can be extended to consider a variety of utility functions ( average throughput , total network throughput , energy efficiency , etc . ) unlike a traditional power control problem , the action space of the ces changes at each time slot and is limited by its battery size . given the distribution of energy arrival and the discount factor @xmath37 , the power control problem can be modeled by a single - controller discounted stochastic game as follows : * there are two players : one mbs and one ces . * the state of the game is the battery level of the ces , which belongs to @xmath65 . * at time @xmath5 and state @xmath36 , the action @xmath47 of the mbs is its transmission power and belongs to the finite set @xmath2 . on the other hand , the action of the ces is @xmath10 , which is the number of energy packets distributed to @xmath1 sbss . @xmath10 belongs to the set @xmath66 . * let @xmath27 and @xmath28 denote the concatenated mixed - stationary - strategy vectors of the mbs and the ces , respectively . the vector @xmath27 is constructed by concatenating @xmath67 sub - vectors into one big vector as @xmath68,$ ] in which each @xmath29 is a vector of probability mass function for the actions of the mbs at state @xmath36 . for example , if the game is in state @xmath36 , @xmath31 gives the probability that the mbs transmits with power @xmath69 . therefore , the full form of @xmath27 will include the state @xmath36 and power @xmath69 . however , to make the formulas simple , in the later parts of the paper , we will use @xmath27 or @xmath29 to denote , respectively , the entire vector or a sub - vector at state @xmath36 , respectively . * similarly , for the ces , @xmath34 gives the probability that the ces distributes @xmath17 energy packets . note that the available actions of the ces dynamically vary at each state whereas the available actions for the mbs remain unchanged at every state . * * payoffs : * at state @xmath36 , if the mbs transmits with power @xmath64 and the ces distributes @xmath63 energy packets , the payoff function for the mbs is @xmath70 while the payoff function for the ces is @xmath71 . we omit @xmath5 since @xmath5 does not directly appear in @xmath72 and @xmath73 . * * discounted payoffs : * denoting by @xmath37 the discount factor ( @xmath74 ) , the discounted sum of payoffs of the mbs is given as : @xmath75,\ ] ] where @xmath76 $ ] is the average utility of macrocell user at time @xmath5 if the mbs and the ces are using strategy @xmath27 and @xmath28 , respectively . similarly , we define the discounted sum of payoffs @xmath41 at the ces . in ( * chapter 2 ) , it was proven that the limit of @xmath40 and @xmath41 always exist when @xmath77 . * * objective : * to find a pair of strategies @xmath78 such that @xmath40 and @xmath41 become a nash equilibrium , i.e. , @xmath79 , where @xmath80 and @xmath81 are the sets of strategies of mbs and ces , respectively . given the distribution of energy arrival at the ces , the transition probability of the system from state @xmath36 to state @xmath82 under action @xmath63 ( @xmath83 ) of the ces is given as follows : @xmath84 also , we assume that information about the average channel gains are available to all players . this implies that the single - controller stochastic game we present here will be a perfect information non - cooperative game . the states of the game can be described by a markov chain for which the transition probabilities are defined by ( [ markovstate ] ) . clearly , the ces controls the state of the game while the mbs has no direct influence . therefore , the single - controller stochastic game can be applied to derive the nash equilibrium strategies for both the mbs and the ces . the two main steps to find the nash equilibrium strategies are : * first , we build the payoff matrices for the mbs and the ces for every state @xmath36 , where @xmath85 . denote them by @xmath86 and @xmath87 , respectively . * second , using these matrices , we solve a quadratic programming problem to obtain the nash equilibrium strategies for both the mbs and the ces . to build @xmath86 and @xmath87 , we calculate @xmath73 and @xmath72 for every possible pair @xmath88 , where @xmath89 and @xmath90 . in this regard , we first derive the average channel gain @xmath20 . second , from the energy consumed @xmath63 and transmission power @xmath64 of the ces and the mbs , @xmath91 , and the user @xmath92 located within the disk centred at @xmath91.,width=192 ] respectively , we decide how the ces distributes this energy @xmath63 among the sbss . then , we calculate the transmit power at each sbs and obtain @xmath73 and @xmath72 . the next two remarks provide us with the methods to calculate @xmath73 and @xmath72 . [ rem1 ] given two bss @xmath93 and @xmath91 , assume that a user @xmath92 , who is associated with @xmath91 , is uniformly located within the circle centred at @xmath91 with radius @xmath58 ( fig . [ locations ] ) . assume that @xmath93 does not lie on the circumference of the circle centred at @xmath91 and @xmath94 . denote @xmath95 and @xmath96 , then the expected value of @xmath97 , i.e. , @xmath98 $ ] is @xmath99 . if @xmath100 , then @xmath98 = \frac{1-r^{-2}}{r^2}$ ] given that @xmath101 . for other values of @xmath57 , @xmath102 $ ] can be easily computed numerically using tools such as ` mathematica ` . see * appendix a. * recalling that @xmath103 and that the fading and path - loss are independent , we have @xmath104\mathbb{e}[r_{ij}^{-4}]$ ] , where @xmath105 = \lambda$ ] , if @xmath56 follows rayleigh distribution with scale parameter @xmath106 . next , we need to find how the ces distributes its energy to each sbs such that @xmath72 is maximized . ( optimal energy distribution at the ces)[rem2 ] if at time @xmath5 the ces distributes @xmath63 energy packets to @xmath1 sbss and the mbs transmits with power @xmath64 , then the transmit powers @xmath107 at the @xmath1 sbss are the solutions of the following optimization problem ( @xmath5 is omitted for brevity ) : @xmath108 where @xmath109 is the maximum transmit power of each sbs . since this problem is strictly concave , the solution @xmath110 always exists and is unique for each pair ( @xmath111 ) . thus , for each pair @xmath112 , where @xmath113 and @xmath114 , we have unique values for @xmath70 and @xmath71 . based on the remarks above , for each combination of @xmath63 and @xmath64 , we can find the unique payoff @xmath73 and @xmath72 of mbs and ces . since @xmath63 and @xmath64 belongs to discrete sets we can find the payoff for all of the possible combinations between them . thus , we can build the pay - off matrix @xmath86 for the mbs and @xmath87 for the ces . the matrix @xmath86 has the form of a block - diagonal matrix @xmath115 , where each sub - matrix @xmath116 , with @xmath89 and @xmath117 is the matrix of all possible payoffs for the mbs at state @xmath36 . similarly , we can build @xmath87 , which is the payoff matrix for the ces . a detailed explanation on how we use them will be given in the next subsection . by solving the pair of linear programs above , the probability that the sbs chooses action @xmath21 at state @xmath36 can be found as @xmath128 using some algebraic manipulations , we can convert the optimization problem in ( [ p ] ) into a matrix form as : @xmath129 and its dual as @xmath130 where @xmath87 is the payoff matrices of the ces . combining the primal and dual linear programs ( i.e. , ( p ) and ( d ) above ) and using the same notations , we have the following theorems . we can define different utility functions for the mbs and the sbs and apply the same method to achieve the nash equilibrium . as long as the number of states is finite and the transition probabilities and the payoff matrices are known , a nash equilibrium point always exists . [ best response strategy for the mbs ] [ mbs ] given a stationary strategy @xmath28 of the ces , there exists a pure stationary strategy @xmath27 as the best response for the mbs . similarly , for any stationary strategy @xmath27 of the mbs , there exists a pure stationary best response @xmath28 of the ces . see * appendix b*. because for every mixed strategy of the ces , the mbs can find a pure stationary strategy as a best response , we only need to find nash equilibrium where the strategy of the mbs is deterministic . from there , this problem can be converted to a mixed - integer program with @xmath27 as a vector of 0 and 1 . we can use a brute - force search to obtain an equilibrium point . for each feasible integer value of @xmath27 we insert it into ( [ quadr ] ) to obtain @xmath28 . if the objective is zero , then @xmath131 is the equilibrium point . this theorem implies that the optimization problem in ( [ quadr ] ) can be solved in a finite amount of time . notice that since @xmath136 is not a positive semi - definite matrix , there can be several solutions for the quadratic programming problem in ( [ quadr ] ) , i.e. , multiple nash equilibrium . therefore , to make the chosen nash equilibrium point more meaningful , we use the following lemma from @xcite . ( necessary and sufficient conditions for the nash equilibrium ) @xmath27 and @xmath28 constitute a pair of nash equilibrium policies for the mbs and the ces if and only if @xmath137 since @xmath138 is the probability that the ces starts with @xmath36 energy level in the battery at starting time , from ( [ p ] ) , @xmath139 is the average value function of the ces with respect to the energy arrival rate and the strategies @xmath140 . we are interested in the nash equilibrium that maximizes the average payoff of the ces ( i.e. , the sbss ) . this bias towards the sbss is crucial as the available energy of the ces is limited by the randomness of the energy arrival process and thus the sbss are more likely to suffer when compared to the mbs . using the lemma above , we change the problem in ( [ quadr ] ) to a quadratic - constrained quadratic programming ( qcqp ) as stated below . ( nash equilibriums that favor the sbss ) the nash equilibrium @xmath131 that has the best payoff for the ces is a solution of the following qcqp problem : @xmath141 by solving this qcqp ( i.e. , by using a brute - force search ) , we obtain a nash equilibrium that returns the best average payoff for the ces . again , we can still have multiple nash equilibrium in this case , but all of them must return the same payoff for the ces . because there may be multiple solutions , the ces and the mbs need to exchange information so that they agree on the same nash equilibrium . the mbs and the ces calculate their strategy @xmath27 and @xmath28 , respectively , by solving ( [ qcqp ] ) . at time @xmath5 , the ces sends its current battery level @xmath36 to the mbs . it also randomly chooses an action @xmath63 using the probability vector @xmath30 . the mbs then randomly picks power @xmath64 using distribution @xmath29 and sends it back to the ces . based on @xmath64 and @xmath63 , the ces searches its records and retrieves the corresponding tuple @xmath142 . the ces distributes energy @xmath143 , respectively , to the @xmath1 sbss . from * theorem 2 * , we know that there exists an equilibrium with pure stationary strategies for both the mbs and ces . recall that with pure strategy , the action of each player is a function of the state . thus , if we can obtain this equilibrium , the ces can predict which transmit power @xmath64 the mbs will use based on the current state , without exchanging information with the mbs and vice versa . for the discrete stochastic control game with the ces , each sbs first needs to send its location and average fading channel information @xmath105 $ ] of its user to the ces . then the ces and the mbs will exchange information so that the mbs can have complete knowledge about the average channel gains at each sbs . since we only use average value , the ces and the mbs only need to re - calculate the nash equilibrium strategies when either the locations of sbss change , e.g. , some sbss go off and some are turned on , or when the average of channel fading gain @xmath56 changes , or when distribution of energy arrival @xmath4 at the ces changes . note that the sbss and the mbs only need to send the channel gain information of the corresponding users ( to be served ) to the ces . the main problem of the two - player single - controller stochastic game is the curse of dimensionality " . the time complexity of * algorithm 1 * increases exponentially with the number of states @xmath7 or the maximum battery size . note that @xmath86 and @xmath87 have dimensions of @xmath144 , so the complexity increases proportionally to @xmath7 . moreover , unlike other optimization problems , we are unable to relax the qcqp in ( [ qcqp ] ) , because * theorem 1 * states that the nash equilibrium must be the global solution of the quadratic program in ( [ quadr ] ) . to tackle these problems , we extend the stochastic game model to an mfg model for a very large number of players . the main idea of an mfg is the assumption of similarity , i.e. , all players are identical and follow the same strategy . they can only be differentiated by their state " vectors . if the number of players is very large , we can assume that the effect of a specific player on other players is nearly negligible . therefore , in an mfg , a player does not care about others states but only act according to a mean field " @xmath145 , which usually is the probability distribution of state @xmath36 at time instant @xmath5 @xcite . in our energy harvesting game , the state is the battery @xmath146 and the mean field @xmath147 is the probability distribution of energy in the area we are considering . when the number of players @xmath1 is very large , we can assume that @xmath147 is a smooth continuous distribution function . we will express the average interference at an sbs as a function of the mean field @xmath148 . all the symbols used in this section are listed in * table ii*. [ cols="^,<,^ , < " , ] thanks to similarity , all the sbss have the same set of equations and constraints , so the optimal control problem for the @xmath1 sbss reduces to finding the optimal policy for only one generic sbs . mathematically , if an sbs has infinite available energy , i.e. , @xmath160 , it will act as an mbs . however , for simplicity , we will assume that only the sbss are involved in the game and the interference from the mbs is constant , which is included in the noise @xmath53 as in @xcite . except that , the system model and the optimization problem here are similar to those in the discrete stochastic game model . assuming that the optimal control above starts at time @xmath5 with @xmath172 , we obtain the bellman function @xmath173 as @xmath174.\ ] ] from this function , at time @xmath5 , we obtain the following hamilton - jacobi - bellman ( hjb ) @xcite equation : @xmath175 where @xmath176 is the average transmit power at a generic sbs . the hamiltonian @xmath177 is given by the bellman s principle of optimality . by applying the first order necessary condition , we obtain the optimal power control as follows : @xmath178^+ . \label{powermfg}\ ] ] the bellman @xmath179 , if exists , is a non - increasing function of time and energy . therefore , we have @xmath180 and @xmath181 . from equation ( @xmath182 ) , given the current interference @xmath183 at a user , the corresponding sbs will transmit less power based on the future prospect @xmath184 . if the future prospect is too small , i.e. , @xmath185 , it stops transmission to save energy . replacing @xmath186 back to the hjb equation , we have @xmath187^+\right)^2 = 0 , & \ ] ] which has a simpler form as follows : @xmath188 also , from ( [ mfg:2 ] ) , at time @xmath5 , we have the fokker - planck equation @xcite as : @xmath189 where @xmath190 is the probability density function of @xmath191 at time @xmath5 . combining all these information , we have the following proposition . [ mfgpower ] the average transmit power @xmath194 of a generic sbs is a derivative of the average energy available with respect to time and can be calculated as @xmath195 see * appendix c*. since @xmath196 is always non - negative , the average energy in an sbs s battery is a decreasing function of time . that is , the distribution @xmath148 should shift toward left when @xmath5 increases . this is because , we use the wiener process in ( [ mfg:1 ] ) . since @xmath197 has a normal distribution with mean zero , the energy harvested will be equal to the energy leakage . therefore , for the entire system , the total energy reduces when time increases . if @xmath198 and @xmath199 are two solutions of * proposition [ pdes ] * and @xmath200 , then we have @xmath201 . first , from ( [ mfg:2 ] ) we derive fokker - planck equation : @xmath202 since @xmath203 , we subtract the first equation from the second one to obtain @xmath204 this means @xmath205 is a function of @xmath5 . let us denote @xmath206 , then we have @xmath207 . from * lemma [ mfgpower ] * , if @xmath208 then @xmath209 . since @xmath210 , we have @xmath211 now we substitute @xmath207 that results in @xmath212 this means @xmath213 or @xmath214 . note that @xmath179 is a function of @xmath196 and @xmath69 . since @xmath215 and @xmath214 , it follows that @xmath216 . this lemma confirms that an sbs will act only against the mean field @xmath148 . thus only @xmath148 determines the evolution of the system . two systems with the same mean field will behave similarly . to obtain @xmath179 and @xmath148 , we use the finite difference method ( fdm ) as in @xcite and @xcite . we discretize time and energy coefficient @xmath191 into large intervals as @xmath217 $ ] and @xmath218 $ ] with @xmath219 and @xmath220 as the step sizes , respectively . then @xmath221 become matrices with size @xmath222 . to keep the notations simple , we use @xmath5 and @xmath191 as the index for time and energy coefficient in these matrices with @xmath223 and @xmath224 . for example , @xmath190 is the probability distribution of energy @xmath225 at time @xmath226 . using the fdm , we replace @xmath227 , @xmath228 , and @xmath229 with the corresponding discrete formulas as follows @xcite : @xmath230 by using them in ( [ valuef ] ) and after some simple algebraic steps , we have @xmath231 where @xmath232 similarly , discretizing ( [ discreteimfg ] ) , we have @xmath233 where @xmath234 to obtain @xmath221 , and @xmath196 using * proposition [ pdes ] * , we need to have some boundary conditions . first , to find @xmath148 , we assume that there is no sbs that has the battery level equal to or larger than @xmath235 so that @xmath236 . this is true if we assume that @xmath237 is the largest battery size of an sbs . also , when @xmath238 , from the basic property of probability distribution @xmath239 next , to find @xmath179 , again we need to set some boundary conditions . notice that @xmath240 for all @xmath191 . we further assume the following : * intutitively , if the battery level of a sbs is full , i.e. , when @xmath241 , this sbs should transmit something , or equivalently , @xmath242 . that means @xmath243 therefore , if we know @xmath244 and @xmath69 , we can calculate @xmath245 . * similarly , it must be true that when the available energy is @xmath3 , i.e. , @xmath238 , an sbs will stop transmission . therefore , we can assume @xmath246 again , if we know @xmath247 , we can calculate @xmath248 . * during simulations , in some cases when the density is very high , we obtain very large ( unrealistic ) values of transmit power . therefore , we must put an extra constraint for the upper limit . in this paper , we use @xmath249 , or @xmath250 , where @xmath13 is the duration of one time slot . this means , we have to limit the transmit power during one time step @xmath219 to be smaller than the maximum power that can be transmitted during one time interval @xmath13 . based on the above considerations , we develop an iterative algorithm ( * algorithm [ fdm ] * ) detailed as follows . set up @xmath251 matrices @xmath252 , @xmath253 , @xmath254 , and @xmath255 vector @xmath196 . guess arbitrarily initial values for power @xmath254 , i.e. , @xmath256 . initialize @xmath257 , @xmath258 , @xmath259 , @xmath260 , and @xmath261 . initialize @xmath220 and @xmath219 as the step size of energy and time with @xmath262 . set @xmath263 as the number of iteration . : solve the fokker - planck equation to obtain @xmath148 using ( [ discretem ] ) and ( [ mbound ] ) with given @xmath69 , @xmath264 . update @xmath194 for @xmath265 using discrete form of equation ( [ discreteimfg ] ) . calculate @xmath179 for all @xmath266 by using ( [ discreteu ] ) , ( [ urmax ] ) and ( [ -urmax ] ) with @xmath69 , @xmath196 . calculate new transmission power @xmath267 using ( [ pow ] ) . regressively update @xmath268 with @xmath269 . * for * @xmath270 * if * @xmath271 * then * @xmath272 . * end * @xmath273 . * end * at time slot @xmath5 , sbs with energy battery @xmath225 transmits with power @xmath274 . for the mfg , we do not need the location information for each sbs . however , we need information about the average channel gain @xmath155 , the number of sbss @xmath1 in one macrocell , and the initial distribution @xmath264 of the energy of the sbss . therefore , some central system should measure these information , solve the differential equations , and then broadcast the power policy @xmath69 to all the sbss . it is more efficient than broadcasting all the information to all sbss and let them solve the differential equations by themselves . again , the central system only needs to re - calculate and broadcast to all sbss a new power policy if there are changes in @xmath155 or @xmath1 . in this section , we quantify the efficacy of the developed stochastic policy in comparison to the simple stackelberg game - based power control policy . the stochastic policy is obtained from the qcqp problem . on the other hand , for the stackelberg policy , we follow a hierarchical method . let us assume that at time @xmath5 sbs @xmath17 has @xmath275 joules of energy in its battery . if the mbs transmits with power @xmath64 , then each sbs tries to transmit with a power such that @xmath276 where @xmath277 is the interference from both the sbss and the mbs . since this is a convex problem , although it is solved independently by each sbs , the results are the same . notice that the objective function for each sbs is similar to the utility function in ( [ u1 ] ) ; therefore , it can provide a fair comparison against our method . the constraint means that each sbs can not transmit more than the energy it has in its battery . next , knowing that sbs @xmath17 will solve the above optimization problem to find its @xmath278 , the mbs calculates its @xmath0 for different values of its transmit power and picks the optimal @xmath64 . knowing @xmath64 , each sbs solves the convex optimization above to obtain its transmit power @xmath279 . the tuple @xmath280 will be a nash equilibrium because no one can choose a better option given the others actions . to solve the qcqp in ( [ qcqp ] ) , we use the _ fmincon _ function from matlab . note that the _ fmincon _ function may return a local optimal instead . therefore , in order to obtain a good approximation for the optimal solution , in our simulations , we use the incremental method as described below . first , our qcqp problem is stated as follows : @xmath281 where @xmath282 is a group of linear functions of @xmath283 . by solving this using _ fmincon _ , we obtain a local optimal result @xmath284 . then we solve the updated qcqp problem as follows : @xmath285 where @xmath286 is a small positive constant . by solving this new qcqp , we obtain a local optimal solution that satisfies the constraints of the original qcqp and returns a better result . we keep repeating this step as long as _ fmincon _ is able to return a solution @xmath287 with @xmath288 that satisfies the constraints . then we say that @xmath289 is the optimal solution ( with the error of @xmath286 ) . note that the optimal solution for this qcqp can be found by using brute force search through all possible integer values of @xmath27 . therefore , the result can be double checked when the number of sbss is small . in the simulations , the ces has a maximum battery size of @xmath290 and the volume of one energy packet is @xmath291 j. the duration of one time interval is @xmath292 ms and the thermal noise is @xmath293 w. the small cell users are considered to be in outage if the @xmath0 falls below 0.02 . the mbs has two levels of transmission power [ 10 ; 20 ] watts and the sinr outage threshold is set to 5 . the energy arrival at each sbs follows a poisson distribution with unit rate . the volume of each energy packet arriving at the ces is @xmath294 times larger than the energy packet collected by each sbs . thus , the amount of energy in each packet at the ces will be @xmath295 @xmath296j . therefore , the ces should have a more efficient method to harvest energy than each sbs ( in the case of stackelberg method ) . however , as the maximum battery size of the ces is limited by @xmath7 , its total available energy is always limited by the product @xmath297 @xmath296j regardless of @xmath1 . for both the cases , each sbs can receive up to 1.5 mj of energy from either the ces or the environment . at the beginning , the ces is assumed to have its battery full . when @xmath290 states , @xmath298 , @xmath299 sbss , @xmath300.,width=288 ] when @xmath290 states , @xmath298 , @xmath299 sbss , @xmath300.,width=288 ] from fig . [ density ] , it can be seen that when the number of sbss is smaller than some value , the stochastic method gives better results . this is because , the ces can redistribute the harvested energy among the sbss based on their average channel gains , and also the qcqp in ( [ qcqp ] ) gives a nash equilibrium that favors the ces . however , at some point , its outage probability will be higher than that for the stackelberg approach . this is not surprising since the ces can only store at most @xmath301 @xmath296j of energy . therefore , when the number of sbss increases , the average allocated power per sbs by the ces reduces while the stackelberg method allows each sbs to harvest up to 1.5mj no matter how large @xmath1 is . this means that the stackelberg method can provide a better performance compared to using the ces when @xmath1 is large . following fig . [ threshold ] , by increasing the threshold sinr target @xmath23 while keeping the number of sbss fixed , we can reduce the outage probability of a user served by an sbs . this is understandable since the average @xmath0 will approach the higher target and thus reduce the outage probability . however , for the stochastic methods , the outage will start to increase when the target @xmath0 is larger than some value . to increase the average @xmath0 , the sbss need to transmit with higher power to at least mitigate the cross - tier interference however , a higher transmit power means a higher consumption of harvested energy , which can create shortages later . also , higher transmit powers from the sbss wil make the mbs to increase its own transmit power and thus create high cross - interference . when the target @xmath0 is higher than some value , the stochastic method will behave greedily by transmitting as much as possible and the outage will begin to increase . when the target @xmath0 is large enough , the ces distributes all of the energy it currently has and thus the outage probability will become flat . a similar observation can be made for the stackelberg approach . however , for this method , the energy can not be redistributed to the sbss with good channel gains to its users and the distribution of energy arrivals is ignored ; therefore , the results are worse than those for our stochastic method . fig . [ volume ] shows the outage probability when increasing the quanta volume by choosing a higher multiplier @xmath294 for the ces . it is easy to see that , with a higher @xmath294 , i.e. , choosing a more effective method to harvest energy at the ces , we can achieve a better performance . the stackelberg method does not use the ces , so the outage probability remains unchanged . note that , since the battery size of each sbs is limited to 1.5 mj , at some point , a higher @xmath294 does not improve the outage probability . states , @xmath299 sbss , @xmath298 , @xmath300.,width=288 ] states , @xmath299 sbss , @xmath298 , @xmath300.,width=288 ] fig . [ mbs_fbs ] shows the outage probability of the macrocell user when @xmath299 and @xmath298 . the stochastic method gives better results in this case since the sbss are more rational " in choosing their transmit powers in long term . also , unlike the stackelberg method , the ces has a fixed - energy battery , so when @xmath1 is large , the average amount of energy distributed to an sbs will be small , which in turn limits the cross - interference to the macrocell user . with the stackelberg method , the sbss only try to maximize their payoffs in the current time slot and ignore the distribution of energy ; therefore , it uses a higher transmit power to compete against the mbs when the target @xmath0 is increased ; therefore , it creates a larger cross - interference and in turn increases the outage probability of the macrocell user . in summary , we see that the centralized method using a ces can provide a better performance in terms of outage probability for both the mbs and sbss . the advantages of using ces are two folds : first , it allows the harvested energy to be distributed to the sbss which have good channel gains for the scheduled users , and second , it considers the probability distribution of energy arrivals when calculating the transmit power policy for both the mbs and sbss . however , since the ces has a fixed battery size , this centralized model performs poorer when it needs to support a large number of sbss . to improve this inflexibility , we can adjust other parameters as follows : change the target @xmath0 , increase the multiplier @xmath294 , or increase the battery size of each sbs . we assume that the transmit power at the mbs is fixed at 10w and it results in a constant noise at the user served by a generic sbs . the radius of the macrocell is @xmath302 meter , so we have constant cross - interference @xmath303 w. the target @xmath0 is @xmath304 and assume that @xmath305 . we discretize the energy coefficient @xmath191 into 80 intervals , i.e. , @xmath306 and @xmath307 intervals . similar to the discrete stochastic case , each sbs can hold up to 150 @xmath296j in the battery , so the maximum transmit power is 30 mw . we impose the threshold such that an sbs will not transmit at @xmath308 or @xmath309j . the intensity of energy loss / energy harvesting , @xmath161 is 1 . sbss / macrocell.,width=288 ] sbss / macrocell.,width=288 ] sbss / macrocell.,width=288 ] sbss / macrocell.,width=288 ] sbss / macrocell.,width=288 ] sbss / macrocell.,width=288 ] for @xmath310 sbss / cell , we have @xmath311 , so a generic sbs does not need to use a large amount of power in order to obtain the target @xmath0 . notice that @xmath196 is the average transmit power of a generic sbs . therefore , if a generic sbs reduces @xmath196 , the cost term @xmath312 also reduces . thus the difference between the cost and the received power @xmath313 will be smaller , which is desirable . it makes sense that a generic sbs will try to reduce its power as much as possible in this case . the power can not be zero though , because @xmath314 . moreover , from fig . [ powmfg401 ] and fig . [ powmfg402 ] , we see that , at the beginning , the sbs with higher energy ( i.e. , 100 @xmath296j ) will transmit with a high power and will gradually reduce to some value . the sbss with smaller battery will increase their transmit powers gradually . since the transmit power is small , we see that in fig . [ dismfg401 ] and fig . [ dismfg402 ] , the energy distribution shifts to the left slowly . on the other hand , when @xmath315 sbss / macrocell , we have @xmath316 . in this case , the effect is more complicated because reducing the transmit power may not reduce the gap between the received power @xmath313 and the cost term @xmath317 . again , as can be seen from fig . [ powmfg502 ] , the sbss with larger available energy will transmit with large power first and after sometime when there is less energy available in the system , all of them start to use less power . therefore , as can be seen in fig . [ dismfg502 ] , the energy distribution shifts toward the left with a faster speed than the previous case . for @xmath318 sbss / macrocell , we have @xmath319 . this means each sbs needs to transmit with a power larger than the average @xmath196 to achieve the target @xmath0 . in fig . [ powmfg602 ] , we see that the behavior of each sbs is the same as in the previous case . that is , the sbss with higher energy transmit with larger power first and then reduce it , while the poorer " sbss increase their transmit power over time . we compare the mfg model against the stochastic discrete model for different values of @xmath1 . for simplicity , we assume that each sbs has the same link gain to its user as @xmath320 . also , we assume that the channel gain from each sbs to the user of another sbs is @xmath321 . using * remark [ rem2 ] * , it can easily be proven that in this case , each sbs will transmit with the same power , i.e. , if the ces sends @xmath322 @xmath296j of energy to @xmath1 sbss , then each sbs receives @xmath323 @xmath296j . then , the interference at each sbs will be calculated as @xmath324 with multiplier @xmath325 and the maximum battery size of the ces as @xmath326 . because the mbs is not a player of the game , the simulation step becomes simpler , and we only need to solve a linear program for the mdp problem instead of a qcqp . therefore , we can call it as mdp method to accurately reflect the difference . for the discrete stochastic case , we discretize the gaussian distribution to model the energy arrivals at the ces . the battery size of each sbs is still 150 @xmath327 . the average sinr of a generic small cell user using both the mfg and mdp models with different density is plotted in fig . [ mfgvsdiscrete ] . we see that using the mfg model , the average @xmath0 increases at the beginning and then it starts falling at some point . this is because , when the density is low , the interference from the mbs is noticeable ( i.e. , @xmath328 w in our simulation ) . from the previous figures , it can be seen that an sbs will increase its power when the density is higher . therefore , after some point the co - tier interference becomes dominant and the average @xmath0 will begin to drop . it means at some value of the density , e.g. , @xmath310 sbss / macrocell in fig . [ mfgvsdiscrete ] , we obtain the optimal average @xmath0 . we notice that the mfg model performs better than the mdp model with the ces . this is due to the limited battery size of ces which makes it difficult to support a large number of sbss . in summary , we have two important remarks for the mfg model . first , if the density of small cells is high , the sbss will transmit with higher power . second , from fig . [ mfgvsdiscrete ] , we see that by choosing a suitable density of the sbss , we can obtain the highest average @xmath0 . from ( [ pow ] ) , it can be easily proven that the average @xmath0 at a user served by an sbs will always be smaller than the target @xmath0 ( because @xmath329 ) . therefore , the highest average @xmath0 is also the closest to the target @xmath0 , which is our objective in the first place . we have proposed a discrete single - controller discounted two - player stochastic game to address the problem of downlink power control in a two - tier macrocell - small cell network under co - channel deployment where the sbss use stochastic renewable energy source . for the discrete case , the strategies for both the macrocell and sbss have been derived by solving a quadratic optimization problem . the numerical results have shown that these strategies can perform well in terms of outage probability experienced by the users . we have also applied a mean field game model to obtain the optimal power for the case when the number of sbss is very large . we have also discussed the implementation aspects of these models in a practical network . in this paper , we have not explicitly considered the correlation in the energy arrival process . however , this correlation can be modeled by assuming that the energy arrival has markovian property . in this case , to calculate the transition probability , we will need to extend the definition of the state to a two - element vector , one is the current energy in the storage and the second is the energy arrival at this time slot . moreover , we have not considered the details of cost and latency analysis related to the information exchange to and from the ces and also the charge and discharge loss of the battery storage . these issues can be addressed in future . [ [ section ] ] denote the distance between the bs @xmath91 to its user @xmath92 as @xmath330 . if @xmath92 is uniformly located inside the disk centred at @xmath91 , the pdf of @xmath331 is @xmath332 . denote by @xmath333 the value of the angle @xmath334 , @xmath333 is uniformly distributed between @xmath335 . using the cosine law @xmath336 , we obtain @xmath337 = \int_0^{2\pi}\int_0^r ( r^2 + a^2 - 2ar \cos\theta)^{-2 } \frac{1}{2\pi } \frac{2a}{r^2 } \ , da \ , d\theta . \end{aligned}\ ] ] first , we solve the indefinite integral over @xmath333 as @xmath338 where @xmath339 is a constant and @xmath340 since @xmath341 , after integrating @xmath342 over @xmath343 $ ] , we can ignore it . thus @xmath344 next , we integrate the above result over @xmath345 to obtain the indefinite integral as : @xmath346 applying the upper and lower limits of @xmath345 , we complete the proof . first , we prove that given @xmath28 , there exists a pure stationary strategy @xmath27 which is the best response of the mbs against @xmath28 . since the action set of the mbs is fixed , at each state @xmath36 , given strategy @xmath30 of the ces , the mbs just needs to choose a mixed stationary strategy @xmath29 such that its average payoff is maximized . at state @xmath36 , the average utility function of the mbs is @xmath347 = \sum\limits_{p_0^s \in \mathcal{p}}\sum\limits^{s}_{j=0 } - ( p_0^s\bar{g}_0-\lambda_0\bar{i}(p_0^s , j))^2 \mathbf{m}(s , p_0^s)\mathbf{n}(s , j ) , \ ] ] where @xmath348 and @xmath117 are the transmit power of the mbs and the number of energy packets distributed at the ces at state @xmath36 , respectively . @xmath349 is the average interference from other sbss to the mbs if the ces distributes @xmath21 energy packets and the mbs transmits with power @xmath348 . we have @xmath350 where @xmath107 is the solution of * remark [ rem2 ] * , with @xmath63 and @xmath64 replaced by @xmath21 and @xmath348 , respectively . since @xmath351 and @xmath352 is a non - negative vector , we have @xmath353 \leq \max\limits_{p_0^s \in \mathcal{p } } \left\ { -\sum\limits^{s}_{j=0 } ( p_0^s\bar{g}_0-\lambda_0\bar{i}(p_0^s , j))^2 \mathbf{n}(s , j ) \right\}.\ ] ] since the set @xmath26 is fixed and finite , there always exists at least one value of @xmath348 that achieves the maximum for the right hand side . that means when the game in state @xmath36 , the mbs can choose this power level with probability of 1 . however , obtaining a closed - form @xmath348 is difficult because first we need to find @xmath142 in closed - form by solving @xmath354 . nevertheless , if the average channel gains from each sbs to the macrocell user ( say @xmath355 ) are same , we can obtain @xmath348 in closed form by defining @xmath356 the final equality is from ( [ power ] ) . replacing this result back into ( [ maxmbs ] ) , we have @xmath357 \leq \max\limits_{p_0^s \in \mathcal{p } } -\sum\limits^{s}_{j=0 } ( p_0^s\bar{g}_0-\lambda_0(\frac{k}{\delta t}j\bar{g}_{0,sbs}+n_0))^2 \mathbf{n}(s , j ) . \end{split}\ ] ] the right hand side of this inequality is a strictly concave function ( downward parabola ) with respect to @xmath348 . note that @xmath358 . the parabola will achieve the maximum value at its vertex given by @xmath359 if @xmath360 is not available in @xmath26 , since the right hand side of the inequality above is a parabola w.r.t . @xmath361 , the best response @xmath361 to @xmath30 is the one nearest to the vertex @xmath360 . on the other hand , given strategy @xmath27 of the mbs , the problem of finding the best response strategy @xmath28 for the ces is simplified into a simple mdp in ( [ p ] ) . then , there always exists a pure stationary strategy @xmath28 ( * ? ? ? * chapter 2 ) . this completes the proof . using the stochastic differential equation in ( [ mfg:1 ] ) at time @xmath5 , @xmath362 we obtain the integral form as follows : @xmath363 where @xmath364 . we obtain the second equality using the mean value theorem for integrals : if @xmath365 is a continuous function and @xmath366 is integrable function that does not change sign on the interval @xmath367 $ ] , then there exists @xmath368 $ ] such that @xmath369 since equation ( [ integralform ] ) is true for all sbss , taking expectation of this equality above for all sbss ( or all possible values of @xmath146 ) , we have @xmath370 -\mathbb{e}[e(t ) ] & = -\mathbb{e}[p(\bar{t},e(\bar{t}))]t ' + \sigma\mathbb{e}[w_{t+t'}-w_{t } ] , \\ \int\limits_{0}^{\infty}e m(t+t',e ) & de - \int\limits_{0}^{\infty}em(t , e)de = -t'\mathbb{e}[p(\bar{t},e(\bar{t } ) ) ] , \end{aligned}\ ] ] where @xmath147 is the distribution of @xmath146 in the system at time instant @xmath5 . using the fact that @xmath371 is a wiener process , @xmath372 follows a normal distribution with mean zero ( i.e. , @xmath373=0 $ ] ) . by dividing both sides by @xmath374 and letting @xmath374 to be very small ( or @xmath375 ) , we have @xmath376 and @xmath377 using @xmath378 , @xmath379 , and changing the variable @xmath146 to @xmath191 , we complete the proof . 99 k. t. tran , h. tabassum , and e. hossain , a stochastic power control game for two - tier cellular networks with energy harvesting small cells , " _ proc . ieee globecom14 _ , austin , tx , usa , 8 - 12 december , 2014 . n. michelusi , k. stamatiou , and m. zorzi , transmission policies for energy harvesting sensors with time - correlated energy supply , " _ ieee transactions on communications _ , vol . 61 , no . 7 , jul . 2013 , pp . 29883001 . p. semasinghe and e. hossain , downlink power control in self - organizing dense small cells underlaying macrocells : a mean field game , " _ ieee transactions on mobile computing , _ doi : 10.1109/tmc.2015.2417880 . k. huang and v. k. lau , enabling wireless power transfer in cellular networks : architecture , modeling and deployment , " _ ieee transactions on wireless communications _ 2014 , pp . 902912 . p. blasco , d. gunduz , and m. dohler , `` a learning theoretic approach to energy harvesting communication system optimization , '' _ ieee transactions on wireless communications , _ vol . 4 , pp . 18721882 , apr . 2013 . d. ngo , l. b. le , t. l .- ngoc , e. hossain , and d.i . kim , distributed interference management in two - tier cdma femtocell networks , " _ ieee transactions on wireless communications , _ vol . 2012 , pp . 979989 . a. y. al - zahrani , r. yu , and m. huang , a joint cross - layer and co - layer interference management scheme in hyper - dense heterogeneous networks using mean - field game theory , " _ ieee transaction of vehicular technology _ , doi : 10.1109/tvt.2015.2413394 . s. guruacharya , d. niyato , d. i. kim , and e. hossain , hierarchical competition for downlink power allocation in ofdma femtocell networks , " _ ieee transactions on wireless communications _ , vol . 2013 , pp . 15431553 . m. huang , member , p. e. caines , and r. p. malham , uplink power adjustment in wireless communication systems : a stochastic control analysis , " _ ieee transactions on automatic control _ 49 , no . 10 , oct . 2004 , pp . 16931708 .
energy harvesting in cellular networks is an emerging technique to enhance the sustainability of power - constrained wireless devices . this paper considers the co - channel deployment of a macrocell overlaid with small cells . the small cell base stations ( sbss ) harvest energy from environmental sources whereas the macrocell base station ( mbs ) uses conventional power supply . given a stochastic energy arrival process for the sbss , we derive a power control policy for the downlink transmission of both mbs and sbss such that they can achieve their objectives ( e.g. , maintain the signal - to - interference - plus - noise ratio ( sinr ) at an acceptable level ) on a given transmission channel . we consider a centralized energy harvesting mechanism for sbss , i.e. , there is a central energy storage ( ces ) where energy is harvested and then distributed to the sbss . when the number of sbss is small , the game between the ces and the mbs is modeled as a single - controller stochastic game and the equilibrium policies are obtained as a solution of a quadratic programming problem . however , when the number of sbss tends to infinity ( i.e. , a highly dense network ) , the centralized scheme becomes infeasible , and therefore , we use a mean field stochastic game to obtain a distributed power control policy for each sbs . by solving a system of partial differential equations , we derive the power control policy of sbss given the knowledge of mean field distribution and the available harvested energy levels in the batteries of the sbss . small cell networks , power control , energy harvesting , stochastic game , mean field game .
the standard model is famously plagued by the need for a light scalar . one possible solution is that the higgs is a pseudo - goldstone @xcite . this idea has received renewed attention with little higgs theories @xcite and certain 5d warped models @xcite . in the former , the higgs is protected by approximate , nonlinearly realized symmetries . the protection is redundant , such that breaking any individual symmetry will not generate a potential for the higgs . this collective symmetry breaking prevents the generation of a quadratically divergent higgs mass . in the 5d setup , the higgs is the bulk @xmath2 zero mode whose holographic dual is a 4d pgb arising from strong cft dynamics . the protection of the higgs potentials in these models is usually ascribed to the need for brane - to - brane propagators . in fact , it is just collective symmetry breaking . we can see this by deconstructing the fifth dimension @xcite and integrating out the heavy modes . this gives us an extreme deconstruction where the theory now lives on a small number of sites . counting global and gauged symmetries shows that we have produced a little higgs theory . as an example , we will study the minimal composite higgs model ( mchm ) and its descendants @xcite . this warped model has an so(5 ) bulk gauge symmetry and so(4 ) and @xmath3gauge symmetries on the ir and uv branes , respectively . by deconstructing this theory and running to a low scale , we recover the symmetry structure of the custodial minimal moose ( cmm ) @xcite . in this , there are two sites , one with so(4 ) and one with @xmath3gauge symmetry . pgb higgs models such as these represent a large and interesting class of beyond the standard model theories . it becomes an important question to decide which realistic models look the most promising . comparing mchm to the cmm will help to answer whether we do better to place a given symmetry pattern on branes in 5d or on sites in 4d . both models are suppressing contributions to @xmath4 and nonoblique corrections . mchm has a known moderate tuning for @xmath5 of @xmath6 , but were this the only difficulty we may forgive it in exchange for a uv complete theory ( to the planck scale ) with natural flavor hierarchy . in fact , we will see that the @xmath5 parameter and changes to the @xmath7rate force the theory into a corner of parameter space where reproducing @xmath1 conflicts with getting acceptable precision electroweak observables . the cmm has no large contributions to @xmath5 and uses custodial su(2 ) to suppress @xmath4 and the nonobliques . thus , its parameter space is free of tunings and tensions . it can even have the scale of new physics as low as 375 gev . we also have the freedom to add eight new quarks without causing problems for @xmath5 or tuning in the higgs potential . doing so allows us to generate weak scale dark matter within the cmm as discussed in section [ sec : dmv ] . the structure of the paper is the following : in section [ sec : appdecwarp ] , we will show how to deconstruct a 5d warped model and what one can learn from doing so . we will take an example from the higgs as a holographic pgb @xcite . in it , a possibly natural mechanism was proposed for generating an @xmath8 higgs quartic with no mass term . we show that the method used actually requires tuning . we also present an argument that in 5d what protects the higgs potential is just collective symmetry breaking . section [ sec : cmm ] is a review of the custodial minimal moose and a discussion of possible modifications to the third generation for protecting @xmath7or generating dark matter . in section [ sec : pewk ] we review the constraints on mchm and present a global analysis of the cmm . section [ sec : conc ] is the conclusion . the little higgs began as a deconstruction of higher dimensional theories @xcite . since deconstruction rewrites a theory into four dimensions , we can use it to clarify points which may seem obscure in the higher dimensional framework . furthermore , it allows us to systematically convert warped setups into little higgs models . by doing so , we can eliminate the large @xmath9 cft that forces tuning upon warped models by their contributions to the @xmath5 parameter . perturbativity is retained as our 4d field theory is now weakly coupled . as we shall see in this section , the viability of mchm as a pgb higgs theory without a hierarchy problem can be seen purely from its symmetry pattern . we can also decide how symmetry breakings in the uv will manifest themselves at low energies . in this 5d model @xcite , an @xmath10 gauge symmetry lives in the bulk . it is broken to @xmath11 by two separate su(3 ) triplet scalars , one on each brane . as the uv brane scalar gets a planck scale vev , the pgb higgs is almost entirely the ir field . this takes the form of excitations about the @xmath11 preserving vacuum configuration @xmath12 . we can thus parameterize the higgs field as follows : @xmath13 where the @xmath14 are the su(3)/su(2 ) generators . there was a proposal to generate a tree - level quartic and vanishing mass term with a bulk scalar @xmath15 transforming as a @xmath16 of @xmath17 .. ] using deconstruction , we can see that such a potential is necessarily tuned . the higgs couples to @xmath15 with the interaction , @xmath18 . under @xmath19 , the @xmath15 will decompose as triplet , doublet , and singlet . by integrating out @xmath15 , one induces a higgs potential . assuming an appropriate symmetry breaking pattern on the uv brane , the doublet and singlet contributions to the higgs mass can cancel , leaving just the quartic . by deconstructing the theory and introducing spurions , we can check if , in fact , there is a way to engineer such a cancellation . the collective symmetry breaking is exactly the same as in the simplest little higgs @xcite . we can express the theory at some scale @xmath20 with @xmath21 as the following moose ( the procedure for turning warped models into mooses is explained in the next subsection ) : @xmath22[f ] { } { \ar@{-}[rr ] |-{\selecttips{eu}{}\object@ { > } } } & & * = < 20pt>[o][f ] { } { \ar@{-}[rr ] |-{\selecttips{eu}{}\object@{>}}}^{\mbox{\raisebox{1.5ex}{$\sigma$ } } } & & * = < 20pt>[o ] [ ] { } \\ \mathrm{gauged : } & su(2 ) & & su(3 ) & & } \end{tabular}\ ] ] by the rightmost link , " we mean simply that @xmath23 transforms as a @xmath24 of su(3 ) under the symmetries of that site . the use of @xmath15 in the holographic higgs was to communicate a certain pattern of symmetry breaking on the uv brane to the higgs on the ir brane . in the deconstructed version , this communication will necessarily take the form : @xmath25 we can represent the above chain as an operator that transforms under the @xmath17 symmetry of the ir site . at low energies , it is reduced to being a spurion , and all interactions must remain invariant under the residual @xmath19 . in fact , there is only one type of nontrivial operator involving @xmath23 we can write down , @xmath26 , where @xmath27 expanding out the @xmath28 case , we get @xmath29 where @xmath30 is an overall coupling of mass dimension 1 . thus , with this term it is not possible to cancel the mass term without affecting the quartic as well . we could also include the @xmath31 contribution : @xmath32 with @xmath33 some heavy mass on the ir site that induces this operator . we can imagine tuning the two contributions against each other . however , that will not be radiatively stable . this is the exact problem one encounters in the simplest little higgs @xcite . in the original mchm papers , the absence of quadratic divergences in the higgs radiative potential is justified in five dimensions . in this version of the theory , the higgs is the zero mode of the bulk gauge field , @xmath2 . clearly , in the infinite bulk limit , it will have no self - interaction potential . after adding boundaries , we can still have the zero mode by imposing neumann boundary conditions on both branes . thus , contributions to the potential necessarily involve brane to brane propagators . the claim is that since these are long - distance and nonlocal , they are finite . however , this is bizarre from the point of view of holographic rg . integrating out high energy modes is equivalent to moving the uv brane toward the ir one , while inducing local operators on the former @xcite . thus , as we integrate out modes down to the weak scale , we lose the ability to credit the finiteness to a long - distance effect . the uv brane is now only a short distance from the ir . to perform the integration , we deconstuct the fifth dimension into of a large number of sites . for simplicity , let us consider a simple example with a @xmath34 gauge theory in the bulk and @xmath34 on both branes . the setup is the following . we use a constant lattice spacing , @xmath35 , and a uniform bulk gauge coupling , @xmath36 . the 4d gauge coupling , @xmath37 , will also be constant across sites . the link vevs , @xmath38 , change exponentially with scale , providing the remnant of the warp factor . following @xcite , we have @xmath39 the gauge boson mass matrix for this setup will be : @xmath40 the mass eigenstates will have @xmath41 etc . thus , a given massive particle is identified with a given site . we can integrate out particles simply by popping nodes off the uv end of the moose , then renormalizing @xmath37 and @xmath42 , ( fortunately , for arguments based on the symmetry structure , we will not need the exact running of @xmath37 . ) . the locality maintained by holographic rg thus manifests itself as theory space locality . such a simple trick would not work in flat space . the mass matrix in that setup would have a uniform value for @xmath42 , and we could no longer affiliate a given mass eigenstate with a certain site . in the @xmath34 case , the link vevs higgs the theory to the diagonal . all link scalars are eaten in the process , and there is no candidate for a pgb . going now to mchm , we give the leftmost site an @xmath43 symmetry , the rightmost @xmath44 , and everything in between becomes @xmath45 . @xmath46[f ] { } { \ar@{-}[rr ] |-{\selecttips{eu}{}\object@ { > } } } & & * = < 20pt>[o][f ] { } { \ar@{-}[r ] |-{\selecttips{eu}{}\object@ { > } } } & * = < 20pt>[o]{\cdots } { \ar@{-}[r ] |-{\selecttips{eu}{}\object@ { > } } } & * = < 20pt>[o][f ] { } { \ar@{-}[rr ] |-{\selecttips{eu}{}\object@ { > } } } & & * = < 20pt>[o][f ] { } \\ \mathrm{gauged : } & su(2)\otimes u(1 ) & & so(5 ) & & so(5 ) & & so(4 ) } \label{eqn:5deconstruct } \end{tabular}\ ] ] as before , integrating out " consists of removing nodes from the left side of the moose . the complication in this setup comes from the mismatch between bulk and brane symmetries . if we imagine starting at some high scale @xmath47 , we will have all gauge symmetries in the theory . as we descend beneath @xmath48 , the leftmost two gauge symmetries get higgsed to their diagonal , which is @xmath43 . thus , to integrate out the first set of heavy modes , we remove the leftmost node , but reduce the symmetry on the adjacent one to @xmath43 . since the @xmath43 site corresponds to the uv brane , we see that we have indeed moved it in toward the ir brane . we can now take the extreme deconstructed limit , replacing the entire extra dimensional setup with a three - site moose . the entire bulk of the extra dimension now consists of a single @xmath45 site . we can determine the existence of a pgb in mchm purely through goldstone counting . in limit that all gauge couplings turn off , the links have an @xmath49 global symmetry , thus there will be 20 exact ngbs in the theory when they condense . however , when the gauge groups break to the @xmath43 diagonal , only 16 gauge bosons get masses , and we get a @xmath50 of @xmath44 pseudo . furthermore , its radiative potential is protected by collective symmetry breaking . setting to zero the gauge coupling on any particular site , we once again get 20 exact goldstones . with three gauge spurions needed to have a nonzero potential , we have recovered the finiteness of the higgs mass . thus , the 5d warped model is protected by the same collective symmetry effect exploited in the little higgs . we can go further , reducing to two sites to recover the symmetry structure of the custodial minimal moose@xcite , was gauged , rather than an @xmath44 subgroup . this just has the effect of changing a quadruplet of heavy gauge bosons with no couplings to sm fermions into scalars . these extra pseudos make no one - loop contributions to the dimension six operators constrained by observables . ] which we will describe in detail below . working backward , we could have started with two sites , getting manifest locality in theory space . by integrating in additional nodes with the appropriate warp factor , we can grow the extra dimension to an arbitrary number of sites while maintaining locality . any such nonlinear sigma model local in theory space is equivalent to georgi s vector limit @xcite , and free from one loop ( at least ) quadratic divergences because of collective symmetry breaking @xcite . we will now review the setup of the custodial minimal moose ( cmm ) , a two - site , four - link little higgs model @xcite . in equation [ eqn:5deconstruct ] , we only have a single extra dimension , and thus each site is connected to its neighbors by just one link . deconstructing the theory to two sites , we can not write down a nonderivative term only involving the link , @xmath23 , without violating collective symmetry breaking . thus , the higgs potential is purely radiative and we get no parametric separation between the electroweak and nonlinear sigma scales . the solution for a two - site moose is to introduce three more links @xcite . this allows us to write plaquette operators that can generate an @xmath8 quartic . we give the details in the next subsection . the links transform under global @xmath51 symmetries with gauged @xmath43 and @xmath44 subgroups , respectively . @xmath52[f ] { } { \ar@/_1pc/@{-}[rr ] |-{\selecttips{eu}{}\object@{>}}}{\ar@/_0.3pc/@{-}[rr ] |-{\selecttips{eu}{}\object@{>}}}{\ar@/^0.3pc/@{-}[rr ] |-{\selecttips{eu}{}\object@{>}}}{\ar@/^1pc/@{-}[rr ] |-{\selecttips{eu}{}\object@ { > } } } & & * = < 20pt>[o][f ] { } \\ su(2)\otimes u(1 ) & & so(4 ) } \end{tabular}\ ] ] the nonlinear sigma fields higgs these groups to the diagonal . the advantage of this setup over the original minimal moose @xcite is the existence of an explicit custodial symmetry , @xmath53 . contributions to the @xmath4 parameter from the heavy gauge boson @xmath54 are partially canceled by the charged bosons in its @xmath19 multiplet . this cancellation allows us to take a limit in which @xmath55 and @xmath54 decouple from sm fermions , thus suppressing contributions to observables from 4-fermi and higgs - fermi operators . our motivations for choosing an @xmath45 global symmetry are the same as in mchm . it is the smallest example of a group which contains both @xmath56 as a subgroup and a field in its adjoint with the right quantum numbers for the higgs . we will also consider versions of the model , which differ from that in @xcite . the third generation quarks can be modified to implement the left - right parity proposed in @xcite to protect @xmath57 . this provides an example of translating features from 5d to a little higgs model . additionally , we will discuss the idea of @xcite , imposing an exact @xmath58 symmetry that allows for stable dark matter candidates . the link fields acquire vevs proportional to the identity at a scale @xmath42 , breaking the gauge symmetry to @xmath59 , giving masses to six gauge bosons . an approximate @xmath60 symmetry among the links ensures that symmetry breaking occurs for each one at the same scale . we can study interactions in the theory by expanding in linearized fluctuations , @xmath61 , about the vacuum . @xmath62 where the @xmath63 are @xmath45 generators in the vector representation , normalized as in @xcite . since the linear combination @xmath64 shifts under the broken generators , we can identify the @xmath16 of @xmath44 components as the eaten goldstones . this leaves 34 uneaten scalars that acquire radiative potentials @xcite . to determine the masses of the heavy gauge bosons , we start with the lagrangian for a gauged nonlinear sigma model : @xmath65 at lowest order in the pion expansion , we get a mass matrix for the gauge bosons , and the following eigenstates : .gauge bosons in the custodial minimal moose [ cols="^,^,^",options="header " , ] the original cmm paper @xcite did not perform a global analysis , and actually came up with overly conservative constraints . the authors proposed a lower bound for @xmath42 at 700 gev , and only in the limit that @xmath66 . in table [ tbl : fvalues ] we see that we can get values of @xmath42 below 700 gev ( significantly in some cases ) , and even where @xmath67 . the previous limit was determined by finding the @xmath68 that would satisfy limits for the most constraining operator . by only comparing to an individual operator , one can get an artificially severe bound as one necessarily ignores cancellations between operators contributions to observables . for example , the 95% cl on @xmath0 considered individually ( 11.8 tev , 6.5 tev ) are nearly twice those arrived at when fit jointly ( 6.6 tev , 3.7 tev ) . thus , a weakening of constraints can occur from the addition of more operators . figure [ fig : chi2thyb ] shows the cl in @xmath69 space for two different values of @xmath70 at @xmath42=700 gev . contours as a function of @xmath71 vs. @xmath72 for @xmath73 gev , @xmath74=0.12 ( left ) , and @xmath74=0.25 ( right ) . the contours from darkest to lightest are at cl of 68% , 95% and 99% . the regions above the thick black line are ruled out by a tachyonic physical higgs . ] we see that neither parameter is especially constrained even though we have imposed consistency with experiment , viable ewsb , correct value for @xmath75 , and a non - tachyonic physical higgs mass . there are similarly sized acceptable regions for both @xmath70 near its perturbative limit ( @xmath76 ) , and at @xmath67 . however , in accordance with our intuition , the small @xmath74 case has closer agreement to the data . we can increase the size of the regions of acceptance by increasing @xmath42 . figure [ fig : chi2thymb ] shows the limits in terms of @xmath77 and @xmath78 . for small values of @xmath74 , we have that @xmath79 contours as a function of @xmath77 vs. @xmath72 for @xmath80 gev , @xmath81 . the contours from darkest to lightest are at cl of 68% , 95% and 99% . the thin black curve is a plot of @xmath78 vs. @xmath72 for the same values of @xmath71 and @xmath42 . the left and lower right regions enclosed by the thick black lines are ruled out by having a tachyonic physical higgs mass . ] once again , we see no especially constrained parameters . for our region of interest , the new heavy particles can take masses in the 2 to 5 tev range . we must have @xmath824.5 tev for perturbativity in @xmath70 . for the values of @xmath71 , @xmath42 plotted , the masses of @xmath77 and @xmath78 roughly track each other . even with new particles at the upper end of these mass scales , our higgs potential is still natural , receiving contributions @xmath83200 gev thanks to loop factor suppression . in studying theories of a pgb higgs , we do well conceptually and numerically to go to four dimensions . by deconstrucing the extra dimension and using holographic rg to run down , we can get an extreme deconstruction and describe the low energy physics in terms of a small moose . questions involving the symmetry of the theory , whether those protecting the higgs mass or those in a particular breaking pattern become straightforward to answer . in particular , we see that 5d warped models are just using collective symmetry breaking in the same way as the little higgs to protect the higgs potential . in comparison with the data , the 4d model requires no tunings , nor was any individual parameter significantly constrained , nor did the data force us into a corner of parameter space where observables make opposite demands upon the theory . the only additional symmetry needed for this was a gauged so(4 ) to get an explicit custodial su(2 ) . the fermion sector is modified by the presence of just one additional particle . by contrast , the 5d warped analogue has all of the above difficulties , and requires the addition of a left - right parity and the placement of quarks into exotic representations of so(4 ) to fix @xmath7 . as we search for the best realistic pgb higgs model , little higgs models with explicit @xmath53 make a strong case ( @xcite offers a coset theory example ) . we can turn any 5d warped model into a theory space little higgs , instantly alleviating the @xmath5 problem and allowing for parametric separation in the higgs potential between quartic and mass terms . the cmm is the simplest possible way to include both @xmath53 and an @xmath8 quartic without theory in theory space . that being said , it is not without some aesthetic unpleasantness . there are the @xmath84 and @xmath85 plaquettes , as well as the spinor links to which fermions couple . it would be interesting to see whether there is a more elegant model that performs as well in comparison to the data . given its success , we believe the cmm offers a standard against which to judge other theories of a pgb higgs . the @xmath86 function for our analysis is parameterized as follows : @xmath87 in @xcite , the authors obtain the @xmath88 and @xmath89 from a global fit to the data of atomic parity violation , dis , lep , sld , and tevatron results . the @xmath90 are a subset of the 80 @xmath91 and @xmath92 conserving dimension six operators of the sm @xcite . one can eliminate several operators by demanding @xmath93 conservation , and remove all those with only colored particles . the latter step is justified by a lack of experimental constraints . lastly , we can impose various flavor symmetries . we will use @xmath94 , as our third generation makes the same contribution as the others at dimension six . this leaves us with 20 operators for comparison to data . operator , the triple gauge boson vertex correction , @xmath95 . however , since it is suppressed for the region of parameter space we are considering , we neglect it for the analysis . ] we can express the 20 operators in terms of four model parameters : @xmath96 and @xmath71 . the most complicated contributions are those from the higgs sector to @xmath5 and @xmath4 . the higgs potential is given in equation [ eqn : higgspotl ] . in it , @xmath97 and @xmath98 are given by the coleman - weinberg potential . their leading contribution comes from the quadratic divergences cutoff by the heavy gauge , fermion , and scalar particles . thus , they are estimated as follows : @xmath99 for our analysis , we set @xmath100 , as this is protected by the approximate @xmath60 link symmetry . following @xcite , minimization of the higgs potential allows us to solve for ewsb parameters . @xmath101 using equation [ eqn : solvelambda ] , we can replace @xmath30 in terms of our chosen four parameters . the higgs sector contributions to @xmath5 and @xmath4 are functions of the physical masses squared , which are given by : @xmath102 with @xmath103 the higgs contributions to @xmath5 ( the only ones considered for our fit ) are found in equation [ eqn : deltas ] . they contribute as follows to @xmath104 : @xmath105 where @xmath106 there is an additional one - loop contribution to @xmath104 from the top sector ( for our analysis we use @xmath107 = 173 gev ) : @xmath108 combining these with the higgs - fermion and fermion - fermion coefficients of equation [ eqn : dim6 ] , we can write down the 20 dimension six operator coefficients in terms of @xmath96 and @xmath71 . we have replaced @xmath109 in terms of @xmath74 , using @xmath110 . @xmath111 the terms @xmath112 and @xmath113 are the logarithmic contributions that come from having a higgs mass shifted from the global minimum of the precision data : @xmath114 for much of parameter space , they give the dominant contributions to @xmath86 . thank you to nima arkani - hamed for inspiring and advising the project . thanks to roberto contino for many useful insights into his papers , spencer chang and jay wacker for their feedback , and to clifford cheung for illuminating discussions on little higgs theories and deconstruction . d. b. kaplan and h. georgi , `` su(2 ) x u(1 ) breaking by vacuum misalignment , '' phys . b * 136 * , 183 ( 1984 ) ; `` composite higgs scalars , '' b * 136 * , 187 ( 1984 ) ; + h. georgi , d. b. kaplan and p. galison , `` calculation of the composite higgs mass , '' phys . b * 143 * , 152 ( 1984 ) ; + h. georgi and d. b. kaplan , `` composite higgs and custodial su(2 ) , '' phys . b * 145 * , 216 ( 1984 ) ; + m. j. dugan , h. georgi and d. b. kaplan , `` anatomy of a composite higgs model , '' nucl . b * 254 * , 299 ( 1985 ) . c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys . d * 67 * , 115002 ( 2003 ) [ arxiv : hep - ph/0211124 ] ; + c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys . d * 68 * , 035009 ( 2003 ) [ arxiv : hep - ph/0303236 ] ; + g. d. kribs , arxiv : hep - ph/0305157 ; + t. gregoire , d. r. smith and j. g. wacker , phys . d * 69 * , 115008 ( 2004 ) [ arxiv : hep - ph/0305275 ] ; + r. casalbuoni , a. deandrea and m. oertel , jhep * 0402 * , 032 ( 2004 ) [ arxiv : hep - ph/0311038 ] ; + c. kilic and r. mahbubani , jhep * 0407 * , 013 ( 2004 ) [ arxiv : hep - ph/0312053 ] ; + a. deandrea , arxiv : hep - ph/0405120 ; + m. c. chen , mod . a * 21 * , 621 ( 2006 ) [ arxiv : hep - ph/0601126 ] .
we examine the relationship between little higgs and 5d composite models with identical symmetry structures . by performing an extreme " deconstruction , one can reduce any warped composite model to a little higgs theory on a handful of sites . this allows us to use 4d intuition and the powerful constraints of nonlinear sigma models to elucidate obscure points in the original setup . we find that the finiteness of the higgs potential in 5d is due to the same collective symmetry breaking as in the little higgs . we compare a 4d and a 5d model with the same symmetry to the data . reviewing the constraints on models related to the minimal composite higgs ( hep - ph/0412089 ) , we see that it has difficulty in producing acceptable values for @xmath0 , and @xmath1 simultaneously . by contrast , in a global analysis , the minimal moose with custodial symmetry is viable in a large region of its parameter space and suffers from no numeric tunings . we conjecture that this result is generic for 4d and 5d models with identical symmetries . the data will less strongly constrain the little theory .
it is well - known that the distribution of the atomic hydrogen gas ( hi ) in many spiral galaxies is non - axisymmetric or lopsided . this lopsidedness was first studied by baldwin , lynden - bell , & sancisi ( 1980 ) . from the analysis of global profiles of hi using single dish radio telescopes , it was concluded that over 50 % of galaxies studied show such asymmetry ( richter & sancisi 1994 , haynes et al . 1998 , matthews , van driel , & gallagher 1998 ) . due to the lack of spatial resolution , such studies could only reveal the combined effect of spatial and velocity asymmetry . the lopsidedness has also been observed kinematically , as in terms of asymmetric rotation curves ( swaters et al . 1999 ) which can be directly used to obtain the lopsided perturbation potential ( jog 2002 ) . the kinematical asymmetry has been used to deduce the ellipticity of the potential from the analysis of the hi velocity fields ( franx , van gorkom , & de zeeuw 1994 ; schoenmakers et al . 1997 ) . many theories have been proposed for the observed lopsidedness . they include tidal interactions ( jog 1997 ) , minor mergers @xcite , asymmetric gas accretion @xcite , and offset stellar disc in the halo potential @xcite . it is not yet clear which of these possible scenarios work at scales of a galactic group . recently , in a first such study , the 2-d spatial distribution of hi was fourier - analysed to measure the lopsidedness for a sample of 18 galaxies in the eridanus group of galaxies ( angiras et al . 2006 , here onwards called paper i ) . in this paper , it was found that the mean amplitude of measured lopsidedness is large @xmath4 ( i.e. , a surface density contrast of 20% above an uniform disc ) , nearly twice of that seen in the stellar component of the field galaxies ( rix & zaritsky 1995 , bournaud et al . also , it was shown that the early - type spiral galaxies show a higher lopsidedness than the late - type spirals , which is contrary to that observed in field galaxies ( bournaud et al . 2005 ) . these two results suggest that tidal interactions play a dominant role in the generation of lopsidedness in the galaxies in groups , as was argued in paper i. it is not known whether these results are common to all group environments , especially since the groups are known to exhibit a large variety of properties @xcite , especially regarding the galaxy density and velocity dispersions . here we address this issue by analysing the 2-d hi data for galaxies in the ursa major group . the velocity dispersion , the fraction of early type galaxies and the lack of hi deficiency @xcite are the major differences between the ursa major and the eridanus groups . thus the galaxies in the ursa major group provide an opportunity to study lopsidedness in a different physical environment . we have selected 11 spiral galaxies belonging to the ursa major group for estimating the spatial lopsidedness and elongation . the spatial lopsidedness values are compared with the results of the eridanus group of galaxies ( paper i ) to see whether group environments give rise to similar values for the amplitudes and phases of the spatial asymmetries . the higher order fourier components ( m=1,2,3 ) have also been obtained . further , we have carried out an analysis of the kinematical data to estimate the elongation in the potential . this is compared with the values obtained from the spatial analysis.as in the eridanus case ( paper i ) , the use of hi as a tracer allows us to study lopsidedness to outer disks covering a radial distance larger than twice that studied earlier using the near - ir data on stars as in rix & zaritsky ( 1995 ) and bournaud et al . ( 2005 ) . this paper is organised as follows . in section 2 we discuss the hi and optical data used for the proposed analysis . details of the harmonic analysis , and the results are presented in section 3 . a discussion of these results in the context of the group environment is given in section 4 and in section 5 we give the conclusions . the ursa major group in the super galactic co - ordinate system lies between @xmath5 degrees and @xmath6 degrees @xcite . the systemic velocity of this group is @xmath7 km s@xmath8 with a dispersion of @xmath9 km s@xmath8 @xcite . seventy nine galaxies have been associated with this group @xcite and this group had a projected density of 3 galaxies per mpc@xmath10 . due to the small velocity dispersion , it is not yet clear whether it can be classified as a cluster as mentioned in tully et al.(1996 ) . in addition this group contains a smaller fraction of early - type galaxies ( @xmath11 ( e+s0 ) and a larger fraction of late - type galaxies @xmath12 ( sp+irr)@xcite ) this system does not show a central concentration @xcite which is atypical for a cluster and more similar to that found in a group . as we shall be using the results from a similar analysis carried out on eridanus group by angiras et al . ( 2006 ) in section 3&4 , a brief summary of the characteristics of this group is given . this group in super - galactic co - ordinate system lies between @xmath13 degrees and @xmath14 degrees at a mean distance of @xmath15mpc . approximately 200 galaxies are associated with this group with a velocity dispersion of @xmath16kms@xmath8 . in this case the projected density was higher , @xmath17 8 galaxies per mpc@xmath10 @xcite . unlike ursa major , in eridanus group there was a larger fraction of elliptical and s0 type galaxies ( @xmath18 ( e+s0 ) and a smaller fraction of the late - type galaxies @xmath19 ( sp+irr ) @xcite ) . in this case also , like ursa major , the group centre is not known . the selection criteria for the 18 galaxies that were used for the harmonic analysis by angiras et al . ( 2006 ) are similar to what is given in section 2.2 ( paragraph 1 ) . even though , these two groups are almost at the same distance from us they differ mainly in two aspects . firstly , in eridanus group , hi deficiency is seen which is ascribed to tidal interactions @xcite . hi deficiency is not seen in the case of ursa major @xcite . secondly , compared to eridanus , ursa major is a loose group @xcite . the higher number density and the higher velocity dispersion as seen in the eridanus group implies a higher rate of tidal interactions between the galaxies in the group . also , the higher fraction of early - type galaxies seen in the eridanus represent an earlier evolution of galaxies via tidal interactions . these agree with the higher amplitude of lopsidedness seen in the eridanus galaxies , if generated by tidal interactions . we caution , however , that the spatial distribution of the ursa major galaxies is that of an elongated filament @xcite . hence , it may not be so straightforward to calculate the dependence of the galaxy interaction rate on the number density in that case . out of the 49 galaxies observed using the westerbork synthesis radio telescope ( wsrt ) by verheijen and sancisi ( 2001 ) , we have selected 11 galaxies on the basis of their inclination and quality of hi maps ( galaxies , whose hi maps were patchy were rejected ) for further analysis . the right ascension ( @xmath20 ) , declination ( @xmath21 ) , systemic velocity ( @xmath22 ) , inclination ( @xmath23 ) and position angle ( pa ) of these galaxies are given in table 1 . all the galaxies selected were in the inclination range of 45 to 70 degrees . this was to ensure the availability of good resolution in velocity maps and hi maps , both of which were essential for the analysis @xcite . details of the observation and the preliminary data reduction are given elsewhere @xcite . [ cols="<,^,>,>,^,^,^ " , ] 1 . the ursa major group of galaxies show a typical lopsidedness of @xmath24 in the inner regions , that is comparable to the field case , and about half of what is seen in the eridanus group ( see section 3.1.1 for details ) . + we also measure the @xmath25 values in the outer parts between 1 - 2 @xmath26 , and find this to be @xmath27 . again this is smaller by a factor of @xmath28 compared to the eridanus case . we plot the mean @xmath25 ( @xmath29 ) value in the inner regions of the galaxies with respect to hubble type in figure 7 , where we also plot the corresponding values from the eridanus study for comparison ( paper i , figure 5 ) . note that the early - type galaxies in the eridanus group show a higher lopsidedness , this is opposite to what is seen in the field galaxies @xcite , and points to tidal interactions as the mechanism for the origin of lopsidedness as argued in paper i. in contrast , in field galaxies , gas accretion plays an important role in generating lopsidedness @xcite . the anti - correlation with galaxy hubble type is weaker in the ursa group case , perhaps because our sample here only covers a smaller subgroup of galaxy types from type 4 to 7 , whereas the eridanus study spans a much larger range from type 1 to 9 . + to address this issue , we obtained the r - band images of three more galaxies from the ursa major sample @xcite , namely ugc 6930 , ngc 4102 and ngc 3729 , from the canadian astronomy data centre ( cadc ) on which fourier analysis was carried out . ugc 6930 belonged to hubble type 7 and had an inclination of 32@xmath30 . ngc 4102 and ngc 3729 belonged to hubble type 2 and had inclinations 58@xmath30 and 48@xmath30 respectively . the a@xmath0 coefficient in the range 1.5 to 2.5 @xmath31 for ugc 6930 was estimated to be 0.02 . in the same range , the a@xmath0 coefficients for ngc 4102 and ngc3729 were 0.12 and 0.04 respectively . these galaxies were not included in our hi analysis because of lack of reliable hi data . it should be noted that if these three points are included in ks - test , the probability that the values of a@xmath0 come from the same distribution is @xmath32 while the maximum difference ( d ) between the cumulative distribution is 0.357 . the three resulting points are shown in fig . 7 ( denoted by symbol ) and they confirm that the distribution of a@xmath0 vs. r is nearly flat . thus there is no clear anti - correlation in this case unlike that seen in the eridanus group . however , this distribution does not show a positive correlation with the hubble type either , unlike the field case @xcite . + thus figure 7 confirms the different physical origins for the lopsidedness in the group and field cases . it also confirms that the anti - correlation seen in the eridanus group can be attributed to the group environment , and requires a higher galaxy number density as seen in the eridanus to be effective . 3 . the kinematical analysis gives a value for the elongation in the potential , showing all galaxies where such analysis could be carried out to be disturbed . this was found earlier for a group of five galaxies in the sculptor group galaxies ( schoenmakers 2000 ) . the above results show that the group environment is conducive to producing lopsidedness , with tidal interactions playing a major role in this . galaxy interactions can give rise both to lopsidedness and a secular evolution towards early - type galaxies as argued by bournaud et al . there are indeed some indications of tidal interactions in ursa major group of galaxies ( verheijen & sancisi 2001 ) . the ursa major group is a loose group , and has number density that is intermediate between the field and the eridanus values . hence tidal interactions are less important in the ursa major group , which could explain the lower amplitude of lopsidedness ( a@xmath0 ) . the lower values of asymmetry parameter ( @xmath29 ) observed for ursa major group of galaxies may also find a partial explanation , if we assume that it mainly falls on a filament and is in the process of forming a group . such a process is observed in zwcl 2341.1 + 0000 @xcite . the main conclusions drawn from this paper are as follows : 1 . the mean amplitude of disk lopsidedness in the ursa major group galaxies is measured and found to be comparable to the field sample , while the eridanus group showed a factor of two higher lopsidedness . + the smaller amplitudes of lopsidedness seen in the present study could be due to the lower galaxy number density and the lower velocity dispersion in the ursa major group ( see section 2 ) . the group environment and tidal interactions are shown to play a major role in generating lopsidedness , especially in a denser group like the eridanus . + the disk lopsidedness can thus be used as diagnostics to study the galaxy interactions and the halo properties in groups of galaxies . the values of elongation of potential as measured from spatial and kinematical studies gives comparable values , thus supporting the idea ( jog 1997 ) that both types of asymmetry arise due to the same perturbation potential . the values of the asymmetry as measured by the mean fractional fourier amplitudes a@xmath0 , a@xmath33 and a@xmath34 are found to be comparable , and also the derived perturbation potential parameters @xmath35 , @xmath36 and @xmath37 are found to be comparable . this is in contrast to the field galaxies where a@xmath0 and a@xmath33 are stronger than a@xmath34 and higher mode amplitudes . this indicates the importance of multiple tidal interactions that can occur under the special conditions of a group environment . we thank the referee , frederic bournaud , for a careful reading of the manuscript and for the critical comments and the suggestion of including the a@xmath0 values from r - band images in figure 7 . these have improved the presentation of the paper . raa takes great pleasure in thanking k. indulekha , school of pure and applied physics , m.g.university,for her constant encouragement during this project . he also thanks the university grants commission of india and st.josephs college , bangalore for granting study leave under the fip leave of 10th five year plan and raman research institute , bangalore for providing all the facilities to pursue this study.this research used the facilities of the canadian astronomy data centre operated by the national research council of canada with the support of the canadian space agency . angiras , r.a . , jog , c.j . , omar , a . , dwarakanath , k.s . , 2006 , mnras , 369 , 1849 ( paper i ) athanssoula , e . , makino , j . , bosma , a . , 1997 , mnras , 286 , 825 baldwin , j.e . , lynden - bell , d . , sancisi , r . , 1980 , mnras , 193 , 313 bagchi , j . , enlin,.a , miniati,.o , stalin , c . s. , singh , m . , raychaudhury,.s , humeshkar , n.b . , 2002 , 7 , 249 block , d. l. , bournaud , f. , combes , f. , puerari , i. , buta , r. a&a , 2002 , 394 , l35 bournaud , f . , combes , f . , jog , c.j . , puerari , i . , 2005 , a&a , 438 , 507 buta r. , alpert a.j . , cobb m.l.,crocker d.a . , purcell g.b . , 1998 , aj , 116 , 1142 buta r. , block d.l . , knapen j.h . , 2003 , aj , 126 , 1148 franx m. , van gorkom j.m . , de zeeuw p.t . , 1994 , apj , 436 , 642 haynes , m.p . , hogg , d.e . , maddalena , r.j . , roberts m.s . , van zee , l. 1998 , aj , 115 , 62 jog , c.j . , 1997 , apj , 488 , 642 jog , c.j . , 2000 , apj , 542 , 216 jog , c.j , 2002 , a&a , 391 , 471 jog , c.j . , maybhate , a. 2006 , mnras , 370 , 891 matthews , l.d . , van driel w. , gallagher , j.s . , 1998 , aj , 116 , 1169 noordermeer , e . , sparke , l.s . , levine , s.e . , 2001 , mnras , 328 , 1064 omar , a . , dwarakanth , k.s . , 2005a , ja&a , 26 , 1 omar , a . , dwarakanth , k.s . , 2005a , ja&a , 26 , 71 rasmussen j. , ponman t.j . , mulchaey j.s . , miles t.a . , raychaudhury s. , 2006 , mnras , 373 , 653 richter , o .- g . , sancisi , r . , 1994 , a&a , 290 , l9 rix , h .- w . , zaritsky , d . , 1995 , apj , 447 , 82 schoenmakers r.h.m . , franx m. , de zeeuw p.t . , 1997 , mnras , 292 , 349 schoenmakers , r.h.m.,1999 , in valtonen , m. , flynn , c. , eds . , asp conf . series , vol . 209 , small galaxy groups , astron . soc . pac . , san francisco , pg . 54 tully , r.b . , verheijen , m.a.w . , pierce , m.j . , huang , j.s . , wainscoat , r.j . , 1996 , aj , 112 , 2471 verheijen , m.a.w . , sancisi , r . , 2001 , a&a , 370 , 765 wong , t . , blitz , l. , bosma , a . , 2003 , apj , 605 , 183 zaritsky , d . , rix , h .- w . , 1997 , apj , 447 , 118
we have carried out the harmonic analysis of the atomic hydrogen ( hi ) surface density maps and the velocity fields for 11 galaxies belonging to the ursa major group , over a radial range of 4 - 6 disc scalelengths in each galaxy . this analysis gives the radial variation of spatial lopsidedness , quantified by the fourier amplitude a@xmath0 of the m=1 component normalised to the average value . the kinematical analysis gives a value for the elongation of the potential to be @xmath1 . the mean amplitude of spatial lopsidedness is found to be @xmath2 in the inner disc , similar to the field galaxies , and is smaller by a factor of @xmath3 compared to the eridanus group galaxies . it is also shown that the the average value of a@xmath0 does not increase with the hubble type , contrary to what is seen in field galaxies . we argue that the physical origin of lopsidedness in the ursa major group of galaxies is tidal interactions , albeit weaker and less frequent than in eridanus . thus systematic studies of lopsidedness in groups of galaxies can provide dynamical clues regarding the interactions and evolution of galaxies in a group environment . galaxies : kinematics and dynamics - galaxies : evolution - galaxies : ism - galaxies : spiral - galaxies : structure - galaxies : groups - galaxies : general
we used an selection algorithm based on assumption that a total energy of physical pair of galaxies must be negative @xcite . individual mass of the galaxies are estimated from their k - luminosity using appropriate value of m / l . it is chosen to reproduce the well known nearby groups of galaxies , like the centaurus a , m 81 , m 83 and ic 342 . the 5687 galaxies of 10571 ( about 54% ) with @xmath0 and @xmath1 have been gathered in 1127 groups . the median velocity dispersion in groups with @xmath2 members is 88 km s@xmath3 , and the median harmonic radius is 222 kpc . we find the median mass - to - k - luminosity ratio for the groups to be 26 in solar units using a projected mass estimator @xcite . that shows presence of moderate amount of dark matter in the considered groups . @xcite , expected m / l ratio for baryonic @xmath4 and dark @xmath5 matter in standard cosmology @xcite.,scaledwidth=65.0% ] based on our algorithm for group selection we derived the median mass - to - light ratio of groups @xmath6 that was used to trace the distribution of mass via light distribution in the local supercluster ( lsc ) . the mean k - band luminosity density within the considered volume is @xmath7 l@xmath8 mpc@xmath9 , being in good agreement with the global luminosity density @xmath10 l@xmath8 mpc@xmath9 from 2df survey @xcite and @xmath11 l@xmath8 mpc@xmath9 from 2mass and sdss surveys @xcite . thus , the mean total mass density inside the lsc volume is @xmath12 m@xmath8 mpc@xmath9 that corresponds to @xmath13 . this value amounts only a quarter of the global density @xmath14 in standard cosmology @xcite . even the total mass - to - light ratio of the virgo cluster ( @xmath15 ) is not enough to reach the global value @xmath16 expected for @xmath14 . to adjust the ensemble averaged value @xmath13 with the global density one needs to suppose that a significant part of the dark matter on the lsc scale is not associated with system of galaxies .
the local supercluster is an ideal laboratory to study distribution of luminous and dark matter in the nearby universe . the 1100 small groups have been selected using algorithm based on assumption that a total energy of physical pair of galaxies must be negative . the properties of the groups have been considered .
the famous yau - tian - donaldson s conjecture for the existence of khler - einstein metrics on fano manifolds asserts that the existence is equivalent to the k - stability . the conjecture has been recently solved by tian @xcite . chen , donaldson and sun also give an alternative proof @xcite . the notion of k - stability was first introduced by tian by using special degenerations @xcite and then reformulated by donaldson in algebraic geometry via test - configurations @xcite . for both special degenerations and test - configurations , one has to study an infinite number of possible degenerations of the manifold . a natural question is how to verify the k - stability by reducing it to a finite dimensional progress . the answer is known for fano surfaces by tian @xcite and for toric fano manifolds by wang and zhu @xcite ( see also @xcite ) . in fact , in both cases the existence is equivalent to the vanishing of futaki invariant . more recently , delcroix extends wang - zhu s result to a polarized compactification @xmath0 of a reductive lie group @xmath1 with @xmath4 @xcite . we call @xmath0 a _ ( bi - equivariant ) compactification of @xmath1 _ if it admits a holomorphic @xmath5 action on @xmath0 with an open and dense orbit isomorphic to @xmath1 as a @xmath5-homogeneous space . @xmath6 is called a _ polarized compactification _ of @xmath1 if @xmath7 is a @xmath5-linearized ample line bundle on @xmath0 . for more examples besides the toric manifolds , see @xcite . let @xmath8 be a @xmath9-dimensional maximal complex torus of @xmath1 with dimension @xmath10 and @xmath11 its group of characters . assume that @xmath12 is the root system of @xmath13 in @xmath11 and @xmath14 is a chosen set of positive roots . let @xmath15 be the polytope associated to @xmath16 , and @xmath17 the part of @xmath15 defined by @xmath14 . denote by @xmath18 its dilation at rate @xmath19 . let @xmath20 and @xmath21 be the relative interior of the cone generated by @xmath14 . then delcroix proved [ de ] let @xmath0 be a polarized compactification of @xmath1 with @xmath4 . then @xmath0 admits a khler - einstein metric if and only if @xmath22 where @xmath23 is the barycentre of @xmath24 with respect to the weighted measure @xmath25 and @xmath26 . it is pointed by delcroix that ( [ bar ] ) implies that the futaki invariant vanishes for holomorphic vector fields induced by @xmath5 , but the inverse is not true in general . thus one may ask if ( [ bar ] ) is related to the k - stability and is determined by a generalized futaki invariant for some test - configurations . in the present paper , we will answer this question . in fact , motivated by the study on toric manifolds @xcite , we investigate the k - energy on the space of @xmath3-invariant khler potentials through the reduced k - energy @xmath27 via legendre transformation . we show that condition ( [ bar ] ) comes from our formula of @xmath27 naturally when @xmath4 ( cf . proposition [ 5104 ] , proposition [ necessary ] ) . moreover , we give an alternative proof of theorem [ de ] by showing the properness of the k - energy ( cf . section 4 ) . the khler - ricci solitons case can be discussed similarly ( cf . section 5 ) . the main purpose of this paper is to give a criterion for the properness of the k - energy on a general polarized compactification @xmath16 of @xmath1 as done on a toric manifold in @xcite . we divide @xmath28 into several pieces @xmath29 such that for any @xmath30 , @xmath31 lies on an @xmath32-dimensional hyperplane defined by @xmath33 for some primitive @xmath34 , where @xmath35 is the @xmath36-dual of @xmath11 . define a cone by @xmath37,\,y\in f_a\}$ ] for any @xmath30 . it is clear that @xmath38 . let @xmath39 then the average of scalar curvature @xmath40 of @xmath41 is given by ) will be verified at the end of section 2 . ] @xmath42 define a weighted barycentre @xmath43 of @xmath18 by @xmath44 note that both @xmath45 and @xmath43 are in the dual space @xmath46 of @xmath47 , where @xmath47 is the non - compact part of lie algebra @xmath48 of @xmath8 . denote by @xmath49 and @xmath50 the projections of @xmath45 and @xmath43 on the semisimple part @xmath51 of @xmath46 , respectively . we prove [ lzz1 ] let @xmath16 be a polarized compactification of @xmath1 with vanishing futaki invariant , and @xmath52 a @xmath53-inariant khler metric . suppose that the polytope @xmath18 satisfies the following conditions , @xmath54 & & \left(\widetilde{bar}_{ss}-bar_{ss}\right)\in\bar{\xi } , \label{tildebar2}\\ & & ( n+1)\cdot\min_a\lambda_a-\bar s>0 . \label{bars}\end{aligned}\ ] ] then the k - energy @xmath55 is proper on @xmath56 modulo @xmath57 , where @xmath58 and @xmath57 is the centre of @xmath1 . in case that @xmath0 is fano and @xmath59 , then @xmath60 and @xmath61 for all @xmath30 . we have @xmath62 , thus , are automatically satisfied . moreover , ( [ bar ] ) is equivalent to the vanishing of futaki invariant and ( [ tildebar1 ] ) ( cf . corollary [ + 5205 ] ) . consequently , @xmath55 is proper modulo the action of @xmath57 . hence we get the an alternative proof for the sufficient part of theorem [ de ] @xcite . as mentioned above , we prove theorem [ lzz1 ] by using the reduced k - energy @xmath27 . one of the advantages of @xmath27 is that it can be defined on a complete space @xmath63 of convex functions on @xmath18 . following the argument in @xcite , we discuss the semi - continuity property of @xmath27 . as a consequence , we prove the following [ lzz2 ] @xmath64 is lower semi - continuous on @xmath63 . furthermore , if @xmath55 is proper on @xmath56 modulo @xmath57 , then there exists a minimizer of @xmath27 on @xmath63 . it is interesting to study the regularity of minimizers in theorem [ lzz2 ] . we guess that they are smooth in @xmath18 if the dimension of the torus @xmath8 is less than two . in case of toric surfaces , it is verified in @xcite . the paper is organized as following : in section 2 , we review some preliminaries on @xmath53-invariant metrics on @xmath0 , and then we give a formula of scalar curvature of such metrics in terms of legendre functions . the formula of @xmath27 is obtained in section 3 . in section 4 , we use the idea in @xcite for toric manifolds to prove theorem [ lzz1 ] , but there are new difficulties arising from energy estimates near the weyl walls to overcome . in section 5 , we focus on the fano case , and prove the properness of modified k - energy provided a modified barycentre condition ( [ bar4 ] ) ( cf . theorem [ thm40 ] ) . in section 6 , we prove theorem [ lzz2 ] . in this section , we first recall some preliminaries for @xmath53-invariant khler metrics on a polarized compactification @xmath6 of @xmath1 @xcite and the associated legendre functions , then we give a computation of scalar curvature in terms of legendre functions . let @xmath65 be the complex structure of @xmath1 and @xmath2 be one of its maximal compact subgroup such that @xmath66 . choose @xmath67 a maximal torus of @xmath2 . denote by @xmath68 , @xmath69 , @xmath70 the corresponding lie algebra of @xmath71 , respectively . then @xmath72 set @xmath73 and lie algebra of @xmath57 by @xmath74 . we decompose @xmath47 as a toric part and a semisimple part : @xmath75 where @xmath76 and @xmath77 $ ] . then for any @xmath78 , we have @xmath79 with @xmath80 and @xmath81 . we extend the killing form on @xmath82 to a scalar product @xmath83 on @xmath47 such that @xmath84 is orthogonal to @xmath82 . identify @xmath47 and its dual @xmath46 by @xmath83 . then @xmath46 also has an orthogonal decomposition @xmath85 denote by @xmath12 and @xmath86 the root system and weyl group with respect to @xmath13 , respectively . choose a system of positive roots @xmath14 . then it defines a positive weyl chamber @xmath87 , and a positive weyl chamber @xmath88 on @xmath46 , where @xmath89 which is also called the relative interior @xmath21 of the cone generated by @xmath14 . the weyl wall @xmath90 is defined by @xmath91 for each @xmath92 . let @xmath93 be the closure of @xmath8 in @xmath0 . it is known that @xmath94 is a polarized toric manifold with a @xmath86-action , and @xmath95 is a @xmath86-linearized ample toric line bundle on @xmath93 @xcite . let @xmath52 be a @xmath53-invariant khler form induced from @xmath6 and @xmath15 be the polytope associated to @xmath94 , which is defined by the moment map associated to @xmath96 . then @xmath15 is a @xmath86-invariant delzent polytope in @xmath46 . by the @xmath53-invariance , for any @xmath97 , the restriction of @xmath98 on @xmath93 is a toric khler metric . it induces a smooth strictly convex function @xmath99 on @xmath100 , which is @xmath86-invariant @xcite . by the @xmath101-decomposition ( @xcite , theorem 7.39 ) , for any @xmath102 , there are @xmath103 and @xmath78 such that @xmath104 . here @xmath105 is uniquely determined up to a @xmath86-action . this means that @xmath105 is unique in @xmath106 . then we define a smooth @xmath53-invariant function @xmath107 on @xmath1 by @xmath108 clearly @xmath107 is well - defined since @xmath99 is @xmath86-invariant . we usually call @xmath99 _ the function associated to @xmath107_. it can be verified that @xmath107 is a khler potential on @xmath1 such that @xmath109 on @xmath1 ( cf . lemma [ hessian ] below ) . the following @xmath101-integral formula can be found in @xcite , proposition 5.28 ( see also @xcite ) [ kak int ] let @xmath110 be a haar measure on @xmath1 and @xmath111 the lebesgue measure on @xmath112 . then there exists a constant @xmath113 such that for any @xmath53-invariant , @xmath110-integrable function @xmath107 on @xmath1 , @xmath114 where @xmath115 next we recall the local holomorphic coordinates on @xmath1 used in @xcite . by the standard cartan decomposition , we can decompose @xmath68 as @xmath116 where @xmath117 , the root space of complex dimension @xmath118 with respect to @xmath119 . by @xcite , one can choose @xmath120 such that @xmath121 and @xmath122=\alpha^{\vee},$ ] where @xmath123 is the cartan involution and @xmath124 is a dual of @xmath119 by the killing form . let @xmath125 and @xmath126 . denote by @xmath127 the real line spanned by @xmath128 , respectively . then we have the cartan decomposition of @xmath69 , @xmath129 choose a real basis @xmath130 of @xmath70 . then @xmath130 together with @xmath131 forms a real basis of @xmath69 , which is indexed by @xmath132 . @xmath132 can also be regarded as a complex basis of @xmath68 . for any @xmath102 , we define local coordinates @xmath133 on a neighborhood of @xmath134 by @xmath135 it is easy to see that @xmath136 , where @xmath137 is the dual of @xmath138 , which is a right - invariant holomorphic @xmath118-form . thus @xmath139 is also a right - invariant @xmath140-form , which defines a haar measure @xmath110 . the complex hessian of the @xmath53-invariant function @xmath107 in the above local coordinates was computed by delcroix as follows ( * ? ? ? * theorem 1.2 ) . [ hessian ] let @xmath107 be a @xmath53 invariant function on @xmath1 , and @xmath99 the associated function on @xmath112 . let @xmath141 . then for @xmath142 , the complex hessian matrix of @xmath107 in the above coordinates is diagonal by blocks , and equals to @xmath143 where @xmath144 by ( [ + 21 ] ) in lemma [ hessian ] , we see that @xmath99 is convex on @xmath112 . the complex monge - ampre measure is given by @xmath145 , where @xmath146 by the convexity of @xmath99 on @xmath47 , the gradient @xmath147 defines a diffeomorphism from @xmath112 to the interior of the dilated polytope @xmath148 , whose image is @xmath15 . ] . let @xmath149 , then by the @xmath86-invariance of @xmath99 and @xmath15 , the restriction of @xmath147 to @xmath150 is a diffeomorphism to the interior of @xmath18 . we note that one part of @xmath151 lies on @xmath152 ( which we call `` outer faces '' ) and the other part lies on weyl walls @xmath153 . for simplicity , we may assume that @xmath148 contains the origin @xmath154 in its interior . then @xmath148 can be described as the intersection of @xmath155 where @xmath156 and @xmath157 are primitive vectors in @xmath35 . recall that guillemin s function of @xmath148 is given by @xmath158 set @xmath159 and @xmath160 by @xcite , the legendre function @xmath161 of @xmath99 belongs to @xmath162 . the inverse is also true . this means that any @xmath163 corresponds to a khler potential in @xmath56 ( cf . * proposition 3.2 ) ) . by a direct computation , we have @xmath164 note that @xmath165 as @xmath166 , where @xmath167 and @xmath168 is the unit normal vector of face @xmath169 . similarly @xmath170 , where @xmath171 . thus we get [ bound - measure ] if @xmath163 , then for any @xmath172 , as @xmath166 , @xmath173 where @xmath174 and @xmath175 . we compute the ricci curvature of @xmath98 . clearly it is also @xmath53-invariant . as in lemma [ hessian ] , in the local coordinates in sect . 2.2 , @xmath176 can be expressed as @xmath177 for any @xmath178 , where @xmath179 then the scalar curvature @xmath180 by using the legendre function @xmath161 , we get @xmath181 where @xmath182 , @xmath183 and @xmath184 are the components of @xmath119 . by the relations @xmath185 we have @xmath186 substituting them into ( [ + s1 ] ) , we obtain ( [ 5103 ] ) immediately . note @xmath187 . since @xmath188 we can rewrite @xmath189 as @xmath190 by proposition [ kak int ] , it follows @xmath191 since @xmath192 on each @xmath90 , by integration by parts on ( [ + s ] ) , we get @xmath193 here we used lemma [ bound - measure ] and the fact that @xmath194 as @xmath195 . on the other hand , by proposition [ kak int ] , the volume of @xmath196 is given by @xmath197 hence , combining the above two relations , we get ( [ 5304 ] ) . let @xmath6 and @xmath41 be as before . denote by @xmath198 the space of khler potentials in @xmath199 $ ] . mabuchi s k - energy is defined on @xmath198 by @xmath200 where @xmath201 , @xmath40 is the average of @xmath202 and @xmath203 is a path of khler potentials joining @xmath204 and @xmath205 in @xmath198 . in this section , we give a formula of @xmath55 on @xmath56 in terms of the legendre function @xmath161 . define @xmath206 where @xmath207 for any @xmath208 then we have [ 5104 ] let @xmath97 and @xmath161 be the legendre function of @xmath209 . then @xmath210 where @xmath211 . note @xmath212 . by ( [ ma ] ) , it is easy to see @xmath213 then by , it suffices to compute the part @xmath214 by integration by parts , it follows @xmath215 note that @xmath194 as @xmath195 in @xmath216 and is away from weyl walls , and @xmath217 vanishes quadratically along any weyl wall . then the last term in becomes @xmath218 on the other hand , by the second relation in lemma [ bound - measure ] , we have @xmath219\pi\,dy\wedge dt + \int_0 ^ 1\int_{\partial(2p_+)}\dot{u}_{t}{u_{t}}^{ij}\pi_{,ij } \,d\sigma_0\wedge dt.\end{aligned}\ ] ] thus combining ( [ + 5103 ] ) and ( [ 5100 ] ) , we get from ( [ 5108 ] ) , @xmath220 by lemma [ bound - measure ] , we see @xmath221 hence , we obtain @xmath222\pi \,dy+const.\ ] ] recall that @xmath223 , the proof is finished . for convenience , we write @xmath64 as @xmath224 , where @xmath225[n ] @xmath226\pi\,dy.\end{aligned}\ ] ] by integration by parts , we can rewrite @xmath227 as @xmath228\pi \,dy,\ ] ] or @xmath229 in this subsection , we discuss the relationship between the futaki invariant @xmath230 and the linear part @xmath231 of @xmath27 . let @xmath232 be the identity component of the automorphisms group of @xmath0 with lie algebra @xmath233 . let @xmath234 be a reductive algebraic subgroup of @xmath232 . then @xmath234 is the complexification of a maximal compact subgroup @xmath235 ( with lie algebra @xmath236 ) . denote the lie algebra of @xmath234 by @xmath237 and its centre by @xmath238 . by a result of futaki @xcite , it suffices to consider @xmath239 for holomorphic vector fields @xmath240 . in our case @xmath16 , when @xmath241 is restricted on @xmath1 , @xmath242 with @xmath243 , for any @xmath244 , where @xmath245 are some constants , @xmath246 . if @xmath247 , @xmath248 are all real numbers . in particular , re@xmath249 . [ 5205]let @xmath250 be the linear function associated to @xmath240 . then the futaki invariant is given by @xmath251 let @xmath252 be the one - parameter group generated by re@xmath253 and @xmath254 be a family of induced khler potentials by @xmath255 since @xmath256 for any @xmath257 @xmath258 is @xmath53 invariant . then @xmath259 induces a family of @xmath86-invariant convex functions @xmath260 on @xmath47 . moreover , the legendre functions @xmath261 of @xmath262 are given by @xmath263 by proposition [ 5104 ] , we get @xmath264 note that @xmath243 for all @xmath244 , which implies @xmath265 and @xmath266 hence ( [ futaki ] ) is true . [ + 5205 ] @xmath0 has vanishing futaki invariant if and only if @xmath267 for any @xmath240 . the later is equivalent to @xmath268 by ( [ 5119 ] ) and ( [ 5304 ] ) , we have @xmath269 this proves the corollary . another explanation of @xmath227 for a @xmath86-invariant , convex piecewise linear @xmath161 can be described as the generalized futaki - invariant corresponding to a toric degeneration @xmath270 as done in ( * ? ? ? * theorem 3.3 ) . in fact , @xmath271 where @xmath272 . the coefficients @xmath273 arise from the homogeneous expression @xmath274 for the irreducible @xmath1-representation @xmath275 of highest weight @xmath276 . by the weyl character formula , @xmath277 and @xmath278 . thus by changing the integral variable @xmath279 to @xmath280 in ( [ generalized - futaki ] ) , we see that @xmath281 and @xmath282 for any @xmath86-invariant rational convex piecewise linear @xmath161 . in fano case , we have all @xmath61 . this is because there is a smooth @xmath53-invariant ricci potential @xmath283 on @xmath0 so that @xmath284 then it reduces to a bounded smooth @xmath285 on @xmath47 , @xmath286 by ( [ d of guillemin ] ) , the singular terms on the right hand side for @xmath182 is @xmath287 it follows @xmath288 thus @xmath61 . now , in fano case , we see that @xmath62 . then ( [ bar ] ) implies that @xmath289 . by corollary [ + 5205 ] , the futaki invariant vanishes . furthermore , by ( [ 5119 ] ) , we get @xmath290 the following proposition shows that ( [ bar ] ) is a necessary condition of the existence of khler - einstein metrics on @xmath16 from the view of k - stability . [ necessary ] let @xmath16 be a fano compactification of @xmath1 . then @xmath0 is not k - stable if @xmath291 . let @xmath292 be the simple roots in @xmath14 . since @xmath291 , without loss of generality we can write @xmath293 where @xmath294 and @xmath295 . let @xmath296 be the fundamental weights for @xmath292 such that @xmath297 . define a @xmath86-invariant rational piecewise linear function @xmath161 on @xmath148 by @xmath298 then @xmath161 defines a non - trivial toric degeneration . since @xmath299 is dominant , we have @xmath300 thus by ( [ l - fano ] ) , we get @xmath301 by ( [ + generalized - futaki ] ) , the proposition is proved . in this section , we study the properness of the k - energy associated to a general khler class @xmath96 . we reduce the problem to @xmath27 . let @xmath154 be the origin of @xmath46 . note that @xmath302 is the fixed point set of the @xmath86-action . then @xmath303 for any @xmath304 . we can normalize @xmath304 by @xmath305 then @xmath306 and @xmath307 the subset of normalized functions in @xmath162 and @xmath308 will be denoted by @xmath309 and @xmath310 , respectively . the following proposition gives a criterion for the properness of @xmath27 . [ proper - red ] under the assumption of theorem [ lzz1 ] , for any @xmath311 , there exists a uniform constant @xmath312 , such that @xmath313 we shall estimate both of the linear part @xmath231 and nonlinear part @xmath314 of @xmath315 below . the following lemma can be directly proved from the convexity of @xmath161 . [ norm - convex ] there is a uniform constant @xmath316 , such that @xmath317 now we prove [ 5301 ] under the assumption of theorem [ lzz1 ] , there exists a positive constant @xmath318 such that @xmath319 since @xmath161 is convex , we have @xmath320 by ( [ 5119 ] ) , we have @xmath321\pi dy\\ & \ \ \ + \sum_a\int _ { e_a}\langle\lambda_a\widetilde{bar}_{ss}-4\rho,\nabla u\rangle\pi dy + \sum_a\int _ { e_a}\lambda_a\langle\nabla{u}|_{\widetilde{bar}_{ss}},y-\widetilde{bar}_{ss}\rangle\pi dy\\ & \ \ \ + \sum_a\int _ { e_a}(\lambda_an-\bar s)\left(\langle\nabla{u}|_{\widetilde{bar}_{ss}},y-\widetilde{bar}_{ss}\rangle+u(\widetilde{bar}_{ss})\right)\pi dy.\notag \end{aligned}\ ] ] by , the last two terms equals @xmath322 \sum_a\int _ { e_a}\lambda_a\pi dy . \nonumber\ ] ] note that @xmath84 is orthogonal to @xmath82 . choosing re@xmath323 in corollary [ + 5205 ] , we have @xmath324 thus @xmath325\pi(y ) dy\nonumber\\ & \ \ \ + \sum_a\int _ { e_a}\langle\lambda_a\widetilde{bar}_{ss}-4\rho,\nabla u\rangle\pi \,dy\notag\\ & \ \ \ + n\left(\sum_a\int _ { e_a}\lambda_a\pi\ , dy\right)\langle\widetilde{bar}_{ss}-bar_{ss},\nabla{u}|_{\widetilde{bar}_{ss}}\rangle.\end{aligned}\ ] ] condition ( [ tildebar2 ] ) implies @xmath326 while ( [ tildebar1 ] ) implies @xmath327 moreover , each equality holds if and only if @xmath328 for all @xmath182 . hence the three terms in ( [ lower bounde of l ] ) are all nonnegative for @xmath329 . we want to use ( [ lower bounde of l ] ) to prove the lemma . suppose that it is not true . then there exists a sequence @xmath330 such that @xmath331 thus there is a subsequence ( still denoted by @xmath332 ) which converges locally uniformly to a convex function @xmath333 in @xmath18 . since the last two terms of ( [ lower bounde of l ] ) is nonnegative , we have @xmath334\pi(y ) dy\\&\leq\mathcal l(u_k)\to0.\end{aligned}\nonumber\ ] ] hence @xmath333 must be an affine linear function . by the fact @xmath335 , we have @xmath336 and so @xmath337 for some @xmath338 . substituting @xmath333 into ( [ l ] ) , we have @xmath339 note that @xmath340 and @xmath341 with `` = '' holds iff @xmath342 . by @xmath343 , we get @xmath344 . this implies that @xmath345 is a linear function depending only on @xmath346 , i.e. , the projection of @xmath279 in @xmath302 . since @xmath154 lies in the interior of @xmath347 and @xmath348 , we get @xmath349 . as a consequence , @xmath350 on the other hand , since all @xmath351 , there exists a uniform constant @xmath352 , such that for @xmath329 , @xmath353 note that @xmath354 on @xmath18 . hence , substituting ( [ 323 ] ) , ( [ 328 ] ) and the above equality for @xmath355 into ( [ + 5119 ] ) , we see @xmath356 , which contradicts to the second relation in ( [ 323 ] ) . the lemma is proved . we prove [ non - linear part ] there exist uniform constants @xmath358 such that for any @xmath329 , @xmath359 where @xmath360 first , we note that @xmath361 is strictly convex on @xmath216 ( cf . * lemma 3.7 ) ) . then by the convexity of @xmath362 , we have @xmath363 by ( [ n ] ) , it follows @xmath364 for some constant @xmath365 independent of @xmath161 . since @xmath366 vanishes at infinity away from weyl walls and @xmath367 vanishes quadratically along any weyl wall , @xmath368 thus by integration by parts for the first integral terms in ( [ 42 ] ) , and then by lemma [ bound - measure ] , we get @xmath369 on the other hand , by ( [ + 5119 ] ) and lemma [ norm - convex ] , we have @xmath370 moreover , @xmath371 @xmath372 since @xmath373 , @xmath374 are smooth up to the boundary , where @xmath375 are constants independent of @xmath161 . hence , substituting ( [ 43 ] ) , ( [ u-0-ij ] ) and ( [ u-0-ij-2 ] ) into ( [ 49 ] ) , we obtain ( [ + 4n ] ) . since @xmath376 is singular and @xmath217 vanishes along each @xmath90 , we shall give an explicit estimate for the singular order of @xmath376 . in the following , we will divide @xmath148 into two parts @xmath377 , where @xmath378 is a union of small neighborhoods of faces of codim@xmath379 which are contained in @xmath380 , and @xmath381 , where @xmath382 is a @xmath86-invariant polytope whose boundary intersects the weyl walls orthogonally . by proposition [ non - linear part ] , to finish the proof of proposition [ proper - red ] , it suffices to prove [ q int ] there are constants @xmath383 independent of @xmath161 such that @xmath384 it is easy to see that @xmath386 is uniformly bounded in @xmath18 . then @xmath387 in this subsection , we further prove [ 52 ] suppose that @xmath388 is a @xmath86-invariant polytope as above . then there exists a constant @xmath389 independent of @xmath161 such that @xmath390 set @xmath391 for @xmath392 . then @xmath393 since the number of elements of @xmath14 is @xmath394 consequently @xmath395 it suffices to estimate the second term for some fixed @xmath396 . let @xmath397 be a point which lies on the intersection of exactly @xmath398 weyl walls . for example , @xmath399 and @xmath400 is away from other walls . without loss of generality , we may assume that @xmath401 are simple roots in @xmath14 . then @xmath402 is an @xmath403-dim linear subspace in @xmath46 . take @xmath404 a cubic relative neighbourhood of @xmath405 in @xmath406 . consider the affine @xmath398-dim plane @xmath407 which is the unique @xmath398-plane passing through @xmath400 and orthogonal to all @xmath408 . by our assumptions , we can take a small relative neighbourhood @xmath409 of @xmath405 in @xmath410 , which is an @xmath398-dimensional polytope , such that @xmath411 intersects @xmath408 orthogonally and is away from other weyl walls . let @xmath412 be the shrinking of @xmath409 with centre at @xmath405 at rate @xmath413 . take @xmath409 small enough , one can assume @xmath414 and @xmath415 are contained in @xmath416 , whose closures are away from other weyl walls ( see figure 1 ) . ( a ) in a @xmath417-dimension @xmath18 . in ( a ) , @xmath405 lies on two walls @xmath418 and @xmath419 , @xmath420 is an outer face , the line segment @xmath421 stands for @xmath404 , and the deeper dark area presents a subpolytope @xmath422 of case ( 2 ) in ( a ) . in ( b ) , @xmath405 lies on three walls.,title="fig:",height=144 ] ( b ) in a @xmath417-dimension @xmath18 . in ( a ) , @xmath405 lies on two walls @xmath418 and @xmath419 , @xmath420 is an outer face , the line segment @xmath421 stands for @xmath404 , and the deeper dark area presents a subpolytope @xmath422 of case ( 2 ) in ( a ) . in ( b ) , @xmath405 lies on three walls.,title="fig:",height=144 ] let @xmath423 be any point in @xmath424 . fix a @xmath425 . since @xmath426 is a strictly convex function for @xmath427 , by the @xmath86-invariance of @xmath161 , it must attains its minima at @xmath428 , where @xmath429 is the coordinate component of @xmath405 in @xmath409 . by the convexity of @xmath161 , we have @xmath430 we divide @xmath431 into finitely many subpolytopes @xmath432 in two types : * @xmath422 is contained in some @xmath433 ; * @xmath434 intersects at most @xmath435 weyl walls and its outer faces are orthogonal to these walls . for @xmath422 of type @xmath436 , by ( [ 54 ] ) , we have @xmath437 for @xmath422 of type @xmath438 , we regard @xmath422 as @xmath439 for some @xmath440 which lies on at most @xmath435 weyl walls . then according to the above argument , there is a subset @xmath441 of @xmath439 such that as in ( [ 55 ] ) , @xmath442 moreover , we have finitely many subpolytopes @xmath443 , where @xmath444 is either contained in some @xmath445 for some @xmath446 , or intersects at most @xmath447 weyl walls such that @xmath448 . thus we can iterate the above progress for finite times until each @xmath449 in @xmath450 is of type ( 1 ) for some @xmath451 while @xmath452 is of type ( 2 ) . hence by the relations ( [ 55 ] ) and ( [ + 54 ] ) , we can find a small number @xmath453 such that @xmath454 since @xmath455 is compact , we can cover it by finitely many @xmath456 . choose @xmath457 such that @xmath458 . then ( [ 53 ] ) follows from ( [ 54 ] ) . if @xmath0 is a toroidal compactification of @xmath1 @xcite , we can take @xmath459 and then proposition [ proper - red ] follows from lemma [ 52 ] directly . lemma [ 52 ] will be also used in section 6 ( cf . lemma [ extend u ] ) . in general , a weyl wall @xmath90 could not intersect a @xmath32-dimensional face @xmath460 of @xmath148 orthogonally . in this case , if let @xmath461 be the reflection with respect to @xmath90 , then by the @xmath86-invariance of @xmath148 , @xmath462 is again a face of @xmath148 . for simplicity , we denote @xmath463 @xmath464 we note that @xmath460 may not intersect @xmath90 . in order to make the computation of the quantity @xmath376 more explicitly , associated to each @xmath90 , we relabel the @xmath32-dimensional faces of @xmath148 as follows : 1 . faces @xmath465 . we denote them by @xmath466 . by the convexity of @xmath148 , we have @xmath467 . since @xmath468 and @xmath469 , @xmath470 . 2 . faces @xmath471 with @xmath472 . we denote them by @xmath473 , where @xmath474 satisfies @xmath475 3 . faces @xmath476 which are orthogonal to @xmath90 . by the convexity of @xmath148 , @xmath477 . we denote them by @xmath478 . since @xmath479 , @xmath480 is invariant under @xmath481 . under the above notations , we rewrite guillemin s function @xmath482 in ( [ guillemin func ] ) as @xmath483 thus @xmath484 and @xmath485 [ u^-1 ] let @xmath486 . then @xmath487 & o(\alpha(y)),~\text{otherwise}. \end{aligned}\nonumber \right.\ ] ] since @xmath488 , we have @xmath489\notag\\ & \leq ( u_{0,ij})\leq{\frac12}\left[\sum_{a}{\frac{1}{l_{a}(y)}}\left(u_{a}^iu_{a}^j+{u_{a,\alpha}^iu_{a,\alpha}^j}\right ) + \sum_b{\frac{u_b^iu_b^j}{l_b(y)}}\right]=:m_2.\notag\end{aligned}\ ] ] it is easy to see @xmath490 thus @xmath491 moreover , if @xmath492 for all @xmath493 , by lemma [ appendix ] in appendix , @xmath494 & = o(\alpha(y)).\end{aligned}\ ] ] this implies @xmath495 the first case is proved . in the second case , there exists an @xmath496 such that @xmath497 and @xmath498 for some @xmath457 . then @xmath499 and @xmath500 thus @xmath501 the lemma is proved . [ m_2 ] let @xmath502 . suppose that @xmath400 also lies on another weyl wall @xmath503 . then as @xmath504 , it holds @xmath505[m2-beta ] @xmath506 ( [ m2-u-0 ] ) follows from the estimate in lemma [ u^-1 ] immediately . it remains to prove ( [ m2-u-0-beta ] ) . let @xmath507 be the group generated by the reflections @xmath481 and @xmath508 . we want to relabel faces of @xmath148 according to this @xmath509-action . in each orbit @xmath510 , where @xmath511 is a @xmath32-dimensional face , we take a face @xmath512 such that @xmath513 . let @xmath514 where @xmath515 set @xmath516 then we rewrite @xmath482 as @xmath517 thus it is easy to see @xmath518 note that @xmath519 is invariant under the reflection associated to @xmath503 . then as in the proof of lemma [ u^-1 ] , we relabel @xmath520 : faces @xmath521 ; faces @xmath522 with @xmath521 ; and faces @xmath476 which is orthogonal to @xmath503 . thus similar to ( [ eigenvalue ] ) , we get @xmath523 where @xmath524 is a constant with at least one @xmath525 since @xmath526 . as a consequence , @xmath527 and @xmath528 this means that @xmath529 is an eigenvector of both @xmath530 and @xmath531 . on the other hand , there are constants @xmath532 such that @xmath533 in particular , @xmath534 then as in the estimate ( [ large - matrix ] ) ( also see ( [ large - matrix-2 ] ) , ( [ large - matrix-3 ] ) ) , we get @xmath535 it follows @xmath536 the lemma is proved . [ remark - w - beta ] since ( [ d2 of guillemin func ] ) can be rewritten as @xmath537 for any weyl wall @xmath503 , as in the proof of lemma [ u^-1 ] for the second case , one can prove : if @xmath538 , then @xmath539 similar to ( [ m2-beta ] ) , @xmath540 from ( [ 59 ] ) , a direct computation shows @xmath541\nonumber\\ & + 2\sum_{\alpha\neq \beta\in\phi_+}\left[{\coth{\langle\alpha,\nabla u_0\rangle}}{\frac{\langle \alpha,\beta\rangle}{\beta(y)}}+{\coth{\langle\beta,\nabla u_0\rangle}}{\frac{\langle \alpha,\beta\rangle}{\alpha(y)}}-2u_0^{ij}{\frac{\alpha_i\beta_j}{\alpha(y)\beta(y)}}\right].\end{aligned}\ ] ] for simplicity , we denote each term in these two sums by @xmath542 and @xmath543 , respectively . we need to estimate them in the following key lemma . [ 521]let @xmath544 . then there exist @xmath545 , such that latexmath:[\[\label{ialpha } latexmath:[\[\label{ialphabeta } as @xmath504 . we consider the following three cases as @xmath504 : * @xmath547 , @xmath548 . * there is an @xmath549 such that @xmath550 for some @xmath551 , and @xmath552 . * there is an @xmath549 such that @xmath553 . * case ( i)*. in this case , @xmath554 . by ( [ d of guillemin func ] ) , it is easy to check @xmath555 then @xmath556 on the other hand , by ( [ d2 of guillemin func ] ) , one can show @xmath557 then @xmath558 thus @xmath559 hence , by the above relation and lemma [ u^-1 ] , we see that there exists a constant @xmath560 such that @xmath561 . * case ( ii)*. in this case , it is easy to see @xmath562 then by lemma [ u^-1 ] , we have @xmath563 , where the constant @xmath564 depends only on @xmath565 . * case ( iii)*. in this case , we may assume @xmath566 then by ( [ d of guillemin func ] ) , we have @xmath567 it follows @xmath568 and @xmath569 on the other hand , by ( [ d2 of guillemin func ] ) , it is easy to see @xmath570 thus @xmath571 here we used the fact that @xmath572 , hence @xmath573 . hence , combining ( [ o - nabla ] ) and ( [ second - derivative ] ) together with lemma [ u^-1 ] , we get @xmath574 . the proof of ( [ ialpha ] ) is completed . next , we prove ( [ ialphabeta ] ) . we may assume @xmath575 , otherwise , ( [ ialphabeta ] ) can be more easy to obtained ( cf . remark [ remark - w - beta ] ) . we note that @xmath576 and @xmath119 is an eigenvector of @xmath577 . then by the above discussion for @xmath578 in cases ( i)-(iii ) , we have @xmath579 on the other hand , by lemma [ m_2 ] , @xmath580 thus @xmath581 similarly , we have @xmath582 combining these two relations , we see that ( [ ialphabeta ] ) is true . by lemma [ 52 ] and lemma [ 521 ] , we begin to prove proposition [ q int ] . set a compact subset of @xmath151 by @xmath583 since @xmath584 if @xmath585 , each point in @xmath586 lies on a face of codimension greater than 2 . we claim : for any @xmath587 , there is a neighbourhood @xmath588 and a constant @xmath589 such that @xmath590 ) , we see there exists a uniform @xmath591 such that @xmath592 since @xmath593 for each @xmath119 . thus by lemma [ 521 ] , to prove the claim , it suffices to estimate @xmath594 with @xmath595 and @xmath596 with @xmath597 . the later can be easily settled . in fact , @xmath596 is bounded near @xmath400 . for @xmath594 , we observe that @xmath598 for any @xmath599 . as in the proof of lemma [ 521 ] for * case ( iii ) * , we have @xmath600 and @xmath601 hence , together with ( [ y - not ] ) in remark [ remark - w - beta ] , we get latexmath:[\ ] ] thus @xmath855\pi\ , dy \le { \frac{(c_1+c_2)\kappa}{\lambda}}+c(u_0)\delta.\ ] ] hence by ( [ scale - difference ] ) , we obtain @xmath856 choosing a sufficiently large @xmath316 such that @xmath857 , and @xmath632 small enough , we get @xmath858 . the proposition is proved . the first part follows from proposition [ semi - continuity ] . for the second part , we take a minimizing sequence @xmath332 of @xmath27 in @xmath308 . then by lemma [ 609 ] and the properness of @xmath859 , there exists a constant @xmath860 such that the normalized sequence @xmath861 is a subset of @xmath733 . moreover , @xmath809 for some @xmath365 . thus by lemma [ extend u ] , there is a limit @xmath161 of a subsequence of @xmath861 in @xmath862 proposition [ semi - continuity ] implies that @xmath161 is a minimizer of @xmath27 in @xmath863 [ appendix ] let @xmath864 be @xmath865 positive real numbers and @xmath866 @xmath865 positive functions . let @xmath867 be another positive function such that @xmath868 then @xmath869 denote @xmath870 . since @xmath871 @xmath872 by a direct computation , we have @xmath873 where @xmath874 are constants and @xmath875 is a @xmath876 product of @xmath877 of the form @xmath878 similarly , @xmath879 where @xmath880 are constants and @xmath881 is a @xmath882 product of @xmath877 of the form @xmath883 then one can show @xmath884 the lemma is proved .
in this paper , we study mabuchi s k - energy on a compactification @xmath0 of a reductive lie group @xmath1 , which is a complexification of its maximal compact subgroup @xmath2 . we give a criterion for the properness of k - energy on the space of @xmath3-invariant khler potentials . in particular , it turns to give an alternative proof of delcroix s theorem for the existence of khler - einstein metrics in case of fano manifolds @xmath0 . we also study the existence of minimizers of k - energy for general khler classes of @xmath0 .
the ionospheric effects on radio propagation , which are fairly small in the vhf frequency band during normal ionospheric conditions , are still significant enough that they need to be taken into account for a ground based measurement of the signature of the epoch of reionization ( eor ) [ _ bowman and rogers _ , 2010 ] . an overview of these ionospheric chromatic effects is given by _ vedantham et . al , _ [ 2010 ] . in this study , observational results of ionospheric effects in the 80 - 185 mhz frequency range are presented , including the first spectrometric detection of ionospheric radio emission , allowing the measurement of electron temperature in addition to ionospheric absorption . understanding and being able to measure these effects is not only important for future measurements of the spectral signature of the eor signature , but could also have applications in ionospheric remote sensing . currently electron temperatures in the ionosphere can be measured either using incoherent scatter radars , or satellite - borne in situ measurements with langmuir probes . a riometer [ _ little and leinbach _ , 1959 ] is an instrument that is used to measure ionospheric absorption . this instrument uses an antenna and a stable radio frontend to measure cosmic radio noise to infer the amount of absorption occurring in the ionosphere . the noise temperature measured by the instrument is equal to the sky noise brightness convolved with the antenna beam . this is a repeatable function of sidereal time if the variable effects of solar radiation , ionospheric absorption , and ionospheric emission are excluded . at frequencies below about 100 mhz , ionospheric absorption dominates . in this case observations of a decrease of the noise power from the maximum sky noise observed at that sidereal time can be used to measure the ionospheric absorption . the `` reference '' curve of maximum power as a function of sidereal time can be measured at the night when the ionosphere is at minimum and the radio noise from the sun is excluded . the receiver used in a riometer needs to be very stable in time , although accurate absolute calibration is not required . while typically a riometer uses only a single frequency , the measurements reported in this paper were made with a wideband instrument and cover 80 to 185 mhz . this instrument was designed as part of the experiment to detect the global eor signature ( edges ) for astronomical observations of the early universe [ _ bowman and rogers _ , 2010 ; _ bowman et al . _ , 2008 ; _ rogers and bowman _ , 2008 ] like a riometer , it is very stable and repeatable in time . further , its frequency response is very well calibrated ( see _ rogers and bowman _ [ 2012 ] for details of the calibration method and the instrument ) so that observed changes in the spectrum taken at the same time on successive days can be used to measure changes in ionospheric emission as well as absorption . observations reported here , which were made using the edges instrument deployed at the murchison radio - astronomy observatory ( mro ) at @xmath0 in western australia from 18 april 2014 to 6 may 2014 , may not be the first radiometric measurements of the electron temperature in the ionosphere . _ pawsey et al . _ [ 1951 ] report emission temperatures in the range 240 to 290 k at 1.6 and 2.0 mhz which could have been from the lower ionosphere . however , while at these low frequencies the emission from sky is completely reflected back into space during daytime , thermal emission from the earth can be propagated via a ground wave or be reflected from the ionosphere and atmospherics from lightning as well as man - made radio frequency interference ( rfi ) . this makes observations at these low frequencies very difficult . a broadband spectrometer , such as the edges instrument , has the advantage over a traditional riometer that contaminating changes in the observed sky power due to solar emission , rfi , and the effects of severe weather conditions can be distinguished based on their spectral properties from the absorption and emission due to the ionosphere . we describe our analysis method in section 2 . in section 3 , we present the data and principal results of the analysis , followed by interpretation and discussion in section 4 and concluding remarks in section 5 . the noise power temperature from an antenna at a given frequency is equal to the sky noise brightness , @xmath1 , convolved with the antenna beam directivity , @xmath2 , @xmath3 where @xmath4 is sky coordinate , @xmath5 is frequency , and @xmath6 is time [ _ rogers et al . _ , 2004 ] . in the absence of the ionosphere , rfi and solar flares the spectrum , @xmath7 , is a highly repeatable function of sidereal time . its spectral shape follows the beam averaged power law of the sky brightness provided the beam shape is constant with frequency . if an opacity change occurs in a region of uniform electron temperature , the change in the spectrum from the theory of radiative transfer [ _ chandrasekhar _ , 1960 ] is approximately @xmath8 where @xmath9 is @xmath10 evaluated at 150 mhz , @xmath11 is the electron temperature of the perturbed region , @xmath12 is the normalized frequency given by @xmath13 mhz ) , @xmath14 is the change in opacity at 150 mhz , and @xmath15 is the spectral index of the sky brightness power law spectrum . this approximation is valid for the small opacity and small changes in the ionosphere expected in the frequency range 80 to 185 mhz . the spectral index is approximately 2.5 in this frequency range [ _ rogers and bowman _ , typical attenuation through the ionosphere at 150 mhz is 0.015 db at night and 0.1 db during the day and is inversely proportional to frequency squared ( _ evans and hagfors _ , 1968 and references therein ) . two methods are used to analyze difference spectra for electron temperature and change in opacity . in the first method the average and standard deviation of the weighted electron temperature and average magnitude of the perturbations can be estimated from the difference spectra taken at the same sidereal time on different days . if there are @xmath16 days of data there will be @xmath17 difference spectra that can be analyzed . each difference spectrum is fitted to the function @xmath18 using weighted least squares . the perturbation in opacity and associated electron temperature are then derived from the best fit parameters @xmath19 and @xmath20 , resulting in @xmath21 and @xmath22 in order to improve the detection and measurement of weak perturbations , their average magnitude and associated electron temperature can be estimated by conducting incoherent averaging of the coefficients derived from the difference spectra . squaring equation ( 4 ) and multiplying equation ( 5 ) by @xmath19 eliminates the division by @xmath19 in equation ( 5 ) which results in @xmath23 and @xmath24 the bias in @xmath25 and @xmath26 due to gaussian noise in the spectra can be removed by subtracting the noise expected in the absence of perturbations , @xmath27 and @xmath28 where @xmath29 and @xmath30 are the values from the covariance matrix used to fit each difference spectrum and @xmath31 is the root mean square ( rms ) residual from the fit used to obtain the values of @xmath19 and @xmath20 . this method is similar to the method of bias removal used for the incoherent averaging of interferometer correlation amplitudes [ _ rogers , doeleman and moran _ , 1995 ] . a sequence of 16 days of data from day 108 to 126 ( excluding days 109 , 115 , 125 ) was analyzed in half hour integrations at the same sidereal time each day . the 16 days provide up to 120 difference spectra . a weighted average of the noise - corrected values of @xmath32 and @xmath33 is made to improve the signal - to - noise ratio ( snr ) of the measurement of the magnitude of opacity changes and the associated electron temperature . the average magnitude of the perturbations is computed using @xmath34^{1/2}\ ] ] and the weighted average electron temperature is computed using @xmath35 this averaging improves the snr by approximately @xmath36 , where @xmath37 is the number of difference spectra used in equations ( 10 ) and ( 11 ) . the second method is to take the difference between the spectrum on a given day from the average of the spectra on all the other days at the same sidereal time . in this case the change in opacity and electron temperature are calculated from equations ( 4 ) and ( 5 ) and errors in these quantities can be estimated from the covariance matrix and the rms residuals to the weighted least squares fit of the difference spectrum to function ( 3 ) . the optimal weighting is equal to @xmath38 where @xmath39 is the expected noise as a function of frequency . figures 1 and 2 show the magnitude of the perturbations and the associated electron temperature derived by the first method . the error bars in these figures are computed from the weighted rms deviation of the noise - corrected values of @xmath14 and @xmath11 obtained from equations ( 6 ) and ( 7 ) from the mean values derived from equations ( 10 ) and ( 11 ) , divided by @xmath36 . the incoherent averaging of results from many difference spectra is needed to obtain a sufficient snr for measurements of the electron temperature during the night . examples of these differences for large perturbations are shown in figure 3 along with the least squares fit to these spectra and the corresponding values of @xmath14 and electron temperature derived from equation ( 4 ) and ( 5 ) . in this case the snr for a single difference spectrum is sufficient to obtain the electron temperature with the errors from the least squares analysis of each spectrum . the second method allows the variation in ionospheric opacity to be shown as a function of time for each day as shown in figure 4 . in this plot only the points with a good fit to the ionosphere are plotted . the missing points correspond to data corrupted by solar emission , rfi , or rain . each point is from a 15 minute integration from independent data . the high correlation between adjacent points shows that , in most cases , the time scale for opacity changes in the ionosphere is longer than 15 minutes and on average is about an hour . the variation of opacity shown in figure 1 is similar to the variations in the radiometric measurements of _ steiger and warwick _ [ 1961 ] at 18 mhz , who show a plot of attenuation vs local time on 16 december 1958 . their plot shows variations of ionospheric attenuation of about 2 db , which corresponds to 0.7% at 150 mhz , on what they call a `` typical day '' . the plot also shows the attenuation rising sharply at 6 hours local time and declining at 21 hours in a manner similar to that shown in figure 1 . the time scale for the variations is about one hour , consistent with those of figure 4 . the measurements of electron temperature shown in figure 2 indicate that absorption is occurring at a broad range of altitudes . figure 5 shows electron temperatures on 2014 - 04 - 20 at mro based on the msise-90 model of _ hedin _ [ 1991 ] and the iri 2007 model of _ bilitza and reinisch _ [ 2008 ] . this indicates that the measured electron temperatures during the day time are comparable with model - based f - region electron temperatures . figure 6 shows absorption of a single 150 mhz ray at 45@xmath40 zenith angle . this also supports the conclusion that slightly over half of the absorption occurs at altitudes over 150 km . the shape of the total absorption also agrees relatively well with figure 1 , which is reasonable , assuming that variation of opacity is to first order proportional to opacity . the absorption was computed using a variant of the appleton - lassen equation that ignores the magnetic field [ _ hargreaves _ , 1969 ] @xmath41 where @xmath42 is the attenuation in db , @xmath43 is altitude dependent electron density and @xmath44 is the angular frequency of the radio wave . here , @xmath45 is the electron collision frequency , including not only electron - neutral , but also electron - ion collisions , which are important in the f - region [ _ mitra and shain _ , 1953 ] . the electron collision frequency height distribution @xmath46 function shown in figure 7 of _ aggarwal et.al.,_ [ 1979 ] was used to derive the attenuation curves plotted in figure 6 . the simple model of a single layer , considered in section 2.1 can be extended to include many layers . in this case , if the modeled absorption is small , the spectrum of a ray after passage through layers on the ionosphere , @xmath47 , can be written as , @xmath48 where @xmath49 is the absorption coefficient for a narrow altitude range interval at range @xmath50 , @xmath51 , is the spectrum of sky noise above the ionosphere in the direction from which the ray originates and @xmath52 is electron temperature at range @xmath50 . the sum is formed across the range of all layers though which the ray propagates . this ignores the change in sky noise spectrum through the ionosphere owing to the small losses involved . for spectra @xmath47 and @xmath53 taken on different days at the same sidereal time , the difference spectrum , @xmath54 , is given by @xmath55 here the primed variables correspond to changed quantities in the spectrum @xmath53 . if the temperatures on the two different days are assumed equal @xmath56 , then equation ( 14 ) can be simplified to @xmath57 in this case for a single ray path @xmath58 and @xmath59 if the day to day perturbations in the f - region dominate , the electron temperature derived from the difference spectrum would be an estimate of the electron temperature in the f - region . conversely , if the perturbations in the d - region dominate , the electron temperature derived from the difference spectrum would be an estimate of the d - region temperature . as we currently do not have a way to model the day - to - day variation in absorption , we have attempted to model the observed electron temperatures using the approximation that assumes there is a constant fractional change from day to day in attenuation for each layer , i.e. , @xmath60 , where @xmath61 is an arbitrary constant . in this case , the measured electron temperature is the weighted average of electron temperature , weighted with attenuation in each layer . this simplified model , computed for a single @xmath62 slanted ray going through the ionosphere is shown in figure 2 , overlayed on the measurements . we used values of @xmath63 and @xmath52 obtained for the mro location from the iri-2007 and msise-90 models for our computations . a more complete modeling of the electron temperature from the difference spectra requires estimates of changes in loss for each layer of the ray path , horizontal structure , refraction , as well as convolution over all the ray paths with the beam of the dipole antenna used for the measurements . while the measurements of electron temperature agree reasonably well with this model during the daytime , the night temperatures are consistently lower in the measurements . the reason for this is unknown , but it could be caused by the model underestimating the d - region electron density or overestimating the f - region electron density during the night . this study shows a novel method for measuring the electron temperature using a relatively inexpensive passive ground based radio remote sensing method . at present the instrument can only measure the electron temperature of changes in the ionosphere so the results presented here are weighted by the perturbations in the attenuation through the ionosphere . as data is acquired over a full year it will be possible to obtain the electron temperature weighted by the attenuation relative the minimum values observed at each sidereal time but it will require models or data from other instruments to obtain the electron temperature as a function of altitude . the use of a well calibrated broadband spectrometer in the frequency range 80 to 185 mhz allows the radiometric measurement of the electron temperature of ionospheric perturbations in addition to the measurement of the attenuation changes through the ionosphere . the measurements agree relatively well with theory and previously published measurements . this scientific work makes use of the murchsion radio - astronomy observatory , operated by csiro . we acknowledge the wajarri yamatji people as the traditional owners of the observatory site . we thank the mro support facility team , especially michael reay and suzy jackson , for maintenance of the instrument . we also acknowledge dr . phillip erickson of mit haystack observatory for helpful discussions on the analysis used in this paper . the raw spectral data and the data format can be made available upon request to alan rogers at [email protected] . this work was supported by the u.s . national science foundation through research awards for the experiment to detect the global eor signature ( ast-0905990 and ast-1207761 ) . bowman , j. d. , rogers , a. e. e. , and hewitt , j. n. ( 2008 ) , toward empirical constraints on the global redshifted 21 cm brightness temperature during the epoch of reionization , _ the astrophysical journal , _ , _ 676 _ , 1 . mitra , a. p. , and shain , c. a. , ( 1953 ) , the measurement of ionospheric absorption using observations of 18.3 mc / s cosmic radio noise , _ journal of atmospheric and terrestrial physics _ , _ 4 _ no . 4 , 204 - 218 . rogers , a. e. , and bowman , j. d. ( 2012 ) , absolute calibration of a wideband antenna and spectrometer for accurate sky noise temperature measurements , _ radio science , _ _ 47_(6 ) , doi:10.1029/2011rs004962 . vedantham , h. k. , l. v. e. koopmans , a. g. de bruyn , s. j. wijnholds , b. ciardi , and m. a. brentjens . , ( 2013 ) . chromatic effects in the 21 cm global signal from the cosmic dawn , _ monthly notices of the royal astronomical society _ , stt1878 . . the electron temperature was not well determined around 5 hours local time because of a combination of small perturbations and large sky noise from the transit of the galactic center at this time . the solid line represents the weighted sum of electron temperatures obtained from the iri and msis model . the sum is weighted with absorption across altitudes 60 and 400 km . ]
changes in the sky noise spectrum are used to characterize perturbations in the ionosphere . observations were made at the same sidereal time on multiple days using a calibrated broadband dipole and radio spectrometer covering 80 to 185 mhz . in this frequency range , an ionospheric opacity perturbation changes both the electron thermal emission from the ionosphere and the absorption of the sky noise background . for the first time , these changes are confirmed to have the expected spectral signature and are used to derive the opacity and electron temperature associated with the perturbations as a function of local time . the observations were acquired at the murchison radio - astronomy observatory in western australia from 18 april 2014 to 6 may 2014 . they show perturbations that increase at sunrise , continue during the day , and decline after sunset . magnitudes corresponding to an opacity of about 1 percent at 150 mhz with a typical electron temperature of about 800 k , were measured for the strongest perturbations .
within the framework of hierarchical formation , dwarf ( @xmath3 @xmath4 10@xmath5 m@xmath6 ) galaxies are often considered the first systems to collapse , supplying the building blocks for the formation of more massive galaxies through merging and accretion . so present - day dwarfs may have been sites of the earliest star - formation ( sf ) activity in the universe . however , this hypothesis is challenged by the physical properties of blue compact dwarf ( bcd ) galaxies , which have blue colors indicative of a young stellar population . bcds are experiencing intense sf ( 0.01 - 10 m@xmath6 yr@xmath7 ; thuan 1991 ) but still have a high neutral gas content ( @xmath8 10@xmath9 m@xmath6 ; thuan & martin 1981 ) . oxygen abundances in h ii regions generally imply a low metal content , with values in the range @xmath10@xmath11 ( izotov & thuan 1999 ; hereafter it99 ) . some bcds contain much less heavy elements than the majority of high-@xmath12 galaxies . the most metal - poor bcds ( @xmath13 ) have therefore been pointed out as particularly good candidate `` primeval '' galaxies in the nearby universe that started to form stars no more than @xmath14 ago ( it99 ) . this would support the view that sf in low - mass systems has been inhibited until the present epoch ( babul & rees 1992 ) . in any case , chemically poorly evolved star - forming dwarfs are the best laboratory where to study sf processes similar to those occurring in the early universe , but with a spatial resolution and sensitivity that are impossible to achieve in high-@xmath12 galaxies . the most direct way to infer the age of a nearby galaxy is to resolve it into individual stars and study the color - magnitude diagram ( cmd ) . the red giant branch ( rgb ) sequence , formed by evolved stars with ages in excess of @xmath15 , is of particular interest . an unambiguous rgb detection implies that sf was already active more than a gyr ago , whereas absence of the rgb indicates that the system has started forming stars only recently . here we report results from a new @xmath16 and @xmath17-band imaging study of the metal - poor bcd sbs 1415 + 437 , performed with the hubble space telescope ( hst ) advanced camera for surveys ( acs ) . together with i zw 18 ( see section [ s : conc ] ) , sbs 1415 + 437 may be one of the best candidate primeval galaxies in the local universe . it has an elongated shape with a bright h ii region at its sw tip ( see figure [ f : acsimage ] ) . the systemic velocity corrected for local group infall towards virgo is @xmath18 ( from the lyon extragalactic database ) . ignoring any possible peculiar velocity , this implies a distance @xmath19 and an intrinsic distance modulus @xmath20 for @xmath21 . the galaxy has an oxygen abundance @xmath22 ( it99 ) and was studied in considerable detail by thuan , izotov , & foltz ( 1999 , hereafter t99 ) . it has a blue integrated color , @xmath23 throughout the main body , indicating that young stars are dominating its light . its sf regions were studied through uv and optical spectroscopy , stellar population synthesis modeling and chemical abundance determination . t99 concluded from all the combined evidence that sbs 1415 + 437 is a young galaxy that did not start to make stars until @xmath24 ago . t99 also presented hst / wfpc2 data that resolved the brightest young stars , but these data were too shallow ( 1800 sec in @xmath16 and 4400 sec in @xmath17 ) to reach evolved asymptotic giant branch ( agb ) or rgb stars . sbs 1415 + 437 was observed in hst go program 9361 using the acs wide field channel ( wfc ) . the full field of view is @xmath25 arcsec@xmath26 with a pixel size of @xmath27 arcsec@xmath26 . total integration times of @xmath28 sec were obtained for each of the filters f606w and f814w . the 16 optimally dithered single exposures per filter were reprocessed with the most up - to - date version of the acs calibration pipeline . the exposures were registered , corrected for geometric distortion , and co - added with cosmic - ray rejection using the multidrizzle software . figure [ f : acsimage ] shows a true color image created by combining the data from the two filters . point source photometry was performed with the daophot / allstar package ( stetson 1987 ) . a spatially variable point - spread function ( psf ) was inferred from the most isolated stars in the target vicinity and was fitted to the detected stars within the galaxy . a master catalog was created by matching peaks in excess of @xmath29 in a sum of the @xmath16 and @xmath17 images , yielding @xmath30 detected objects . because of the large number of dithered exposures , the master catalog is essentially free of instrumental artifacts such as cosmic rays or hot pixels . the vast majority of the detected sources are individual stars in the galaxy , with a small potential contamination from stars blends and background galaxies ( foreground star contamination is predicted to be negligible ) . we did not remove potentially contaminating sources from our catalog by applying cuts in sharpness or @xmath31 , so as to not run the risk of rejecting bona fide stars in sbs 1415 + 437 as well . however , we did experiment with rejection strategies of various kinds and found that none of the results presented here depend on this aspect of the analysis . photometry was performed at the positions of the objects in the master catalog separately in the @xmath16 and @xmath17 band images . aperture corrections were measured from stars in the frame , and applied to all photometry . corrections for imperfect charge transfer efficiency ( cte ) were applied based on extrapolation of the calibration of riess & mack ( 2004 ) . contamination to f606w by extended h@xmath32 emission was corrected in the photometric analysis through local background subtraction . count rates were transformed to the johnson - cousins @xmath16 and @xmath17 magnitudes using the color - dependent synthetic transformations given by sirianni et al . ( 2005 ) , which have been shown to be accurate to @xmath33 mag . figure [ f : cmds]a shows the inferred @xmath34 cmd for the full area of the sky covered by the galaxy . the median random errors in the photometry are indicated in the figure . these errors do not account for the possible influence of star blending , which will generally add additional systematic uncertainties . an estimate of the 50% completeness limit is also shown in the figure . this estimate was obtained by fitting the combination of a parameterized completeness function and a model sf history to the observed cmd . the resulting estimate is an average over the galaxy , given that the actual completeness varies spatially depending on the level of crowding . the simple estimates of the errors and completeness in figure [ f : cmds]a suffice for the purposes of the present paper . in a forthcoming follow - up paper we will present artificial star tests to estimate these quantities more robustly , and also as function of position in the galaxy . that paper will also contain more details about the data reduction and photometric analysis , and will discuss narrow - band imaging obtained in the context of the same hst program . at bright magnitudes in the cmd ( @xmath35 ) we find the same star types detected previously with wfpc2 ( t99 ) : main sequence stars , and their more evolved counterparts , the young supergiants . the combined feature of main sequence stars and blue supergiants ( evolved stars at the hot edge of their core he - burning phase ) shows up near @xmath36 , while red supergiants appear near @xmath37 . our deeper acs data show , as expected , that these sequences extend down to much fainter magnitudes . but more importantly , the acs data reveal the unmistakable presence of a much older population . agb and carbon stars are seen at @xmath38@xmath39 , with colors extending from @xmath40 to as much as @xmath41 . below @xmath42 there is a prominent rgb with a pronounced rgb tip ( trgb ) . we used the software developed by one of us ( rvdm ) and described in cioni et al . ( 2000 ) to determine the trgb magnitude : @xmath43 . we also determined the color of the trgb , using a histogram of the star colors in a magnitude range just below the trgb : @xmath44 . the systematic errors include the uncertainty introduced by the transformations from the acs filters to the johnson - cousins bands , the algorithmic uncertainties associated with finding a discontinuity in the luminosity function , and the fact that this discontinuity is not infinitely steep due to observational uncertainties . to fit the rgb we use theoretical isochrones from the padua group , as compiled and transformed to the johnson - cousins system by girardi et al . ( 2002).-enhanced models are more appropriate for bcds , but are not generally available at the metallicities relevant for bcds . ] the models include a simple synthetic scheme for tp - agb evolution . their metallicity @xmath45 ( the mass fraction of metals heavier than helium ) samples the range from @xmath46 to @xmath47 . to constrain the age of the rgb stars with the help of these models we need some independent knowledge of the metallicity and the internal reddening @xmath48 if sbs 1415 + 437 has the same ratio of oxygen to metals as the sun , then the observed nebular oxygen abundance can be combined with the most up - to - date solar values @xmath49 and @xmath50 ( asplund , grevesse & sauval 2005 ) to obtain @xmath51 . however , if the galaxy has the same ratio of iron to metals as the sun then the implied metallicity is only @xmath52 , because the hii regions in sbs 1415 + 437 , like other bcds , are @xmath32-enhanced compared to the sun ( t99 ) . either way , the rgb stars probably formed from gas that was more pristine and less metal - enriched than the present - day gas for which nebular abundances are available . we therefore adopt @xmath53 as a fairly firm upper limit to the metallicity of sbs 1415 + 437 . t99 used the balmer decrement in a nuclear spectrum to estimate an internal reddening @xmath54 , corresponding to @xmath55 . we adopt this as an upper limit to the average extinction suffered by rgb stars , for three different reasons : ( i ) the nebular gas is usually associated with sf regions , which tend to be inherently more dusty than the regions in which older stars reside ( as demonstrated explicitly for the case of the lmc ; zaritsky 1999 ) ; ( ii ) the nuclear regions of galaxies often tend to be the most dusty ones ; and ( iii ) the t99 estimate did not account for possible balmer absorption contamination from an underlying evolved population . given the upper limit to @xmath56 , and taking into account the random and systematic errors , we infer that the true unreddened color of the trgb must be in the range @xmath57 . figure [ f : trgbcolor ] shows contours of the predicted trgb color in the parameter space of metallicity vs. age . the trgb becomes redder if either age or metallicity is increased . the filled green area shows the region bounded by the available constraints on @xmath45 and @xmath58 . it shows that the age of the observed red giants might be anywhere between @xmath2 and the hubble time . to obtain the youngest ages , one must assume that the dust extinction @xmath48 and metallictity @xmath45 are at the upper extremes of their allowed ranges . for reference it is useful to consider one particular model in more detail . let us choose the model with @xmath53 and @xmath59 and call this the `` standard '' model . to reproduce the observed color of the trgb in this model , the age of the rgb stars must be @xmath60 . the absolute magnitude of the trgb at this age is @xmath61 . this implies a distance modulus @xmath62 ( @xmath1 ) . figure [ f : cmds]a overplots in red the rgb for the @xmath63 isochrone on top of the data , showing an excellent fit . figure [ f : cmds]b plots isochrones for other ages in the standard model and shows good qualitative agreement also with the other sequences in the observed cmd . only the carbon star sequence is not well fit , which is a known shortcoming of these stellar evolution models ( marigo , girardi & chiosi 2003 ) . the inferred trgb distance modulus for the standard model has a systematic error of at least @xmath64 mag due to uncertainties in the evolutionary calculations ( e.g. , bellazzini et al . there is an additional systematic uncertainty associated with the unknown dust extinction @xmath65 . so overall the result agrees well with the value @xmath20 implied by straightforward application of the hubble law . it also agrees with a simple estimate that can be obtained from the carbon stars . their modal i - band magnitude is @xmath66 ( see fig . [ f : cmds ] ) . if we assume that these carbon stars are on average equally luminous as those in the lmc at @xmath67 ( van der marel & cioni 2001 ) , then sbs 1415 + 437 must have @xmath68 . the age and metallicity dependence of the carbon star magnitudes is poorly known , but probably adds at least @xmath69 mag of systematic error to this estimate . either way , these agreements provide an independent reason why the rgb stars in sbs 1415 + 437 can not be younger than @xmath70 . not only would such young rgb stars have a trgb that is bluer than observed , but @xmath71 would also be up to 2 mag fainter than the usual value ( barker , sarajedini & harris 2004 ) since in the youngest rgb stars he ignites in non - degenerate cores . the implied trgb distance modulus of sbs 1415 + 437 would then be quite inconsistent with the values implied by the hubble law and the observed carbon star brightnesses . moreover , for ages below @xmath70 the rgb and its tip cease to be well - defined features in the cmd , in conflict with the observed cmd morphology in figure [ f : cmds]a . at @xmath72 the observed rgb has a gaussian @xmath73 dispersion of @xmath74 ( see fig . [ f : cmds]a ) , which is twice the median photometric error . this might be because the photometric random errors underestimate the true errors . alternatively , the observed rgb width may indicate an intrinsic spread in properties . for the standard model , the observed rgb width can be reproduced either with a gaussian dispersion of @xmath75 dex in @xmath45 or @xmath76 dex in age . a spread in the extinctions @xmath56 towards individual stars might also contribute towards the observed width . the mass in evolved stars can be estimated using the best - fitting ( @xmath63 ) isochrone in the standard model . for a salpeter initial mass function ( imf ) from @xmath64@xmath77 , only 1 in @xmath78 randomly drawn stars is observed in a 1 magnitude range below the trgb . since we actually observe @xmath79 stars there ( with @xmath80 ) the galaxy must have @xmath81 stars on this isochrone . the associated mass is @xmath82 ( the average mass per star is @xmath83 for the adopted imf ) . the mass increases if the rgb stars are assumed to be older . t99 estimated the mass in young stars ( @xmath0 ) to be only @xmath84 . detailed modeling of the sf history will be presented in a forthcoming paper , but preliminary results broadly confirm that at least 80% of the stellar mass of sbs 1415 + 437 resides in stars with ages @xmath85 . the follow - up paper will also address the variations in the sf history along the main body of the galaxy , which was previously found by t99 to be quite non - homogeneous . here we simply note that the rgb stars seen in the cmd of figure [ f : cmds ] are in fact spread over the entire galaxy , and are not just found exclusively in one particular region . we have used hst / acs to detect agb and rgb stars in the metal - poor bcd sbs 1415 + 437 . the data imply that most of the stellar mass of this galaxy resides in stars older than @xmath2 . it was proposed previously that this galaxy ( t99 ) , and others like it ( it99 ) , did not form stars more than @xmath24 ago . this was based primarily on interpretation of integrated spectra and heavy element abundances in hii regions . our results show that such data should be used with caution when addressing the sf history of the underlying stellar population . our results add to the growing list of low - metallicity galaxies in which an rgb has been detected with hst , including the bcds i zw 36 ( @xmath86 ; @xmath87 ; schulte - ladbeck et al . 2001 ) , vii zw 403 ( @xmath88 ; @xmath89 ; schulte - ladbeck et al . 1999 ) and ugc 4483 ( @xmath90 ; @xmath91 ; dolphin et al . 2001 ; izotov & thuan 2002 ) , and the local group dwarf irregular galaxies leo a ( @xmath92 ; tolstoy et al . 1998 , schulte - ladbeck et al . 2002 ) and sagdig ( @xmath93@xmath94 ; momany et al . 2005 ) . the situation for the most metal - poor bcd , i zw 18 , has been less clear - cut , possibly because it is more metal - poor than any other galaxy studied ( @xmath95 ; it99 ) or because of its larger distance ( @xmath96 ) . early hst images did reveal asymptotic giant branch ( agb ) stars in i zw 18 ( e.g. , aloisi , tosi & greggio 1999 ; stlin 2000 ) , but izotov & thuan ( 2004 ) did not detect an rgb in more recent deeper imaging with acs . however , the latter result has now been challenged by momany et al . ( 2005 ) based on a better photometric analysis of the same data . they show that many red sources do exist at the expected position of an rgb , and that their density in the cmd drops exactly where a trgb would be expected . additional hst / acs data of i zw 18 may be needed for a conclusive understanding of its sf history . but either way , the preponderance of the evidence now seems to suggest that the local universe simply may not contain any galaxies that are currently undergoing their first burst of star formation . support for proposal 9361 was provided by nasa through a grant from stsci , which is operated by aura , inc . , under nasa contract nas 5 - 26555 . e. smith and t. brown are acknowledged for their advice on the photometric reduction of crowded fields observed with acs .
we present results from an hst / acs imaging study of the metal - poor blue compact dwarf galaxy sbs 1415 + 437 . it has been argued previously that this is a very young galaxy that started to form stars only @xmath0 ago . however , we find that the optical color - magnitude diagram prominently reveals asymptotic giant branch and red giant branch ( rgb ) stars . the brightness of the rgb tip yields a distance @xmath1 . the color of the rgb implies that its stars must be older than @xmath2 , with the exact age depending on the assumed metallicity and dust extinction . the number of rgb stars implies that most of the stellar mass resides in this evolved population . in view of these and other hst results for metal - poor galaxies it seems that the local universe simply may not contain any galaxies that are currently undergoing their first burst of star formation .
finite - size effects in critical phenomena have been the object of theoretical studies for a long time.@xcite finite - size scaling ( fss ) describes the critical behavior around a critical point , when the correlation length @xmath0 of the critical modes becomes comparable to the size @xmath1 of the system . for large sizes , this regime presents universal features , shared by all systems whose transition belongs to the same universality class . although formulated in the classical framework , fss also holds at zero - temperature quantum transitions,@xcite in which the critical behavior is driven by quantum fluctuations . the fss approach is one of the most effective techniques for the numerical determination of the critical quantities . while infinite - volume methods require that the condition @xmath2 is satisfied , fss applies to the less demanding regime @xmath3 . more precisely , fss theory provides the asymptotic scaling behavior when both @xmath4 keeping their ratio @xmath5 fixed . however knowledge of the asymptotic behavior may not be enough to estimate the critical parameters , because data are generally available for limited ranges of parameter values and system sizes , which are often relatively small . under these conditions , the asymptotic fss predictions are affected by sizable scaling corrections . thus , reliably accurate estimates of the critical parameters need a robust control of the corrections to the asymptotic behavior . this is also important for a conclusive identification of the universality class of the quantum critical behavior when it is _ a priori _ uncertain . an understanding of the finite - size effects is also relevant for experiments , when relatively small systems are considered , see , e.g. , ref . , or in particle systems trapped by external ( usually harmonic ) forces , as in recent cold - atom experiments , see , e.g. , refs . . in this paper we study fss at quantum transitions , @xcite within the framework of the renormalization - group ( rg ) theory . we generalize wegner s scaling ansatz @xcite to quantum systems . this allows us to characterize the corrections to the asymptotic fss behavior . we predict nonanalytic scaling corrections due to the irrelevant rg perturbations and analytic contributions which are due to regular backgrounds and to the expansions of the nonlinear scaling fields in terms of the hamiltonian parameters . to verify the rg predictions , we consider the quantum xy chain in a transverse field , which represents a standard theoretical laboratory for the understanding of quantum transitions . its hamiltonian can be rewritten as a quadratic hamiltonian of spinless fermions.@xcite using this representation , several quantities can be computed either exactly or very precisely by numerical methods . this allows us to check the fss predictions for the scaling corrections of several observables . we consider the free energy , the energy differences between the lowest - energy levels , and the correlation functions of the order parameter , confirming the rg results in all cases . finally , we discuss the fss behavior of the bipartite entanglement entropies of one - dimensional systems with an isolated critical point with @xmath6 . we perform a detailed study in the xy model , verifying the presence of further peculiar corrections , @xcite beside those associated with the usual bulk and boundary rg irrelevant perturbations . the paper is organized as follows . in sec . [ generalfss ] we discuss the general rg approach to the study of fss at quantum transitions , considering , in particular , the case of isolated quantum critical points between different quantum phases . in sec . [ xychain ] we present a thorough fss analysis of the quantum xy chain in a transverse field at the ising transition , checking the general asymptotic fss predictions for several physically interesting quantities . [ entang ] is devoted to a fss analysis of the bipartite entanglement entropy at the quantum transition of the quantum xy chain , focusing again on the nature of the scaling corrections . finally , in sec . [ conclusions ] we summarize our main results and draw some conclusions . a few appendices report some formulas which are used in the paper . in this section we summarize the main rg ideas behind fss , providing the framework to analyze critical finite - size effects in continuous quantum transitions . as their classical counterparts , they are characterized by a diverging length scale @xmath0 and show universal scaling properties , which can be analyzed in the framework of rg theory . guided by the quantum - to - classical mapping , we generalize wegner s scaling ansatz @xcite to quantum systems , and then use it to predict the type of subleading corrections that are expected in finite systems and/or at finite @xmath7 , close to a continuous transition . we consider the standard case in which the quantum zero - temperature transition of a @xmath8-dimensional system is characterized by two relevant parameters @xmath9 and @xmath10 , defined so that they vanish at the critical point . therefore , the quantum critical point is at @xmath11 we assume the presence of a parity - like @xmath12 symmetry , as it occurs , for instance , in ising or o@xmath13 transitions which separate a paramagnetic phase with @xmath14 from a ferromagnetic phase with @xmath15 . the parameter @xmath9 is coupled to a rg perturbation that is invariant under the symmetry , while @xmath10 is associated with the leading odd perturbation , generally related to the order parameter of the transition . as usual , we express the rg dimensions of the perturbations associated with @xmath9 and @xmath10 in terms of the critical exponents @xmath16 and @xmath17 , as @xmath18 where @xmath19 is the dynamic critical exponent associated with time and temperature . at the critical point the low - energy scales vanish : the gap @xmath20 behaves as @xmath21 at @xmath22 and @xmath23 . the length scale @xmath0 associated with the critical modes diverges as @xmath24 at @xmath22 and @xmath23 , and as @xmath25 at @xmath26 and @xmath23 . under the quantum - to - classical mapping , the inverse temperature corresponds to the system size in an imaginary time direction . thus , the temperature scaling at a quantum critical point in @xmath8 dimensions is analogous to fss in a corresponding @xmath27 classical system . if @xmath6 , which holds for transitions described by 2d conformal field theories ( cfts ) @xcite and for paramagnetic - ferromagnetic transitions in @xmath8-dimensional o(@xmath28 ) symmetric spin systems @xcite ( ising systems correspond to @xmath29 ) , the quantum transition corresponds to a classical @xmath30-dimensional equilibrium transition , in which @xmath31 plays the role of an additional spatial dimension . there are also interesting cases in which @xmath32 . for instance , superfluid - to - vacuum and mott transitions of lattice particle systems described by the hubbard and bose - hubbard models have @xmath33 when driven by the chemical potential.@xcite in this case , the classical system is strongly anisotropic since @xmath34 , @xmath35 under a rg rescaling . also for this class of transitions , which include many dynamic off - equilibrium transitions,@xcite fss is quite well established . we can , therefore , extend those results to the quantum case . according to rg , close to a continuous transition the free energy satisfies a general scaling law . extending the classical fss ansatz , @xcite we generally write the free - energy density as @xmath36 where @xmath37 is a nonuniversal function which is analytic at the critical point , and @xmath38 bears the nonanalyticity of the critical behavior and its universal features . the arguments of @xmath38 are the so - called nonlinear scaling fields . @xcite they are analytic nonlinear functions of the model parameters , which are associated with the eigenoperators that diagonalize the rg flow close to the rg fixed point . the scaling fields @xmath39 and @xmath40 are the relevant scaling fields related to the model parameters @xmath9 and @xmath10 . the scaling fields @xmath41 and @xmath42 are also relevant , with rg dimensions @xmath43 respectively , and are associated with the finite spatial size @xmath1 ( @xmath44 ) and with the temperature ( @xmath45 ) . beside the relevant scaling fields , we should also consider an infinite number of irrelevant scaling fields with negative rg dimensions . we distinguish them into two families , the bulk scaling fields @xmath46 and the surface scaling fields @xmath47 , with rg dimensions @xmath48 and @xmath49 , respectively . the first set is the only one that occurs in the infinite - volume limit and whenever there are no boundaries in the system , for instance , for periodic boundary conditions ( pbc ) . they are responsible for the scaling corrections to the leading critical behavior in the infinite - volume limit . using standard notation , assuming that they are ordered so that @xmath50 in the presence of a surface , an additional set of boundary rg perturbations should be included . their rg dimensions depend on the type of boundary conditions and , in particular , on the type of surface critical behavior one is considering . as before , we set @xmath51 where @xmath52 is the dimension of the leading boundary operator . the singular part of the free energy is expected to satisfy the scaling equation @xmath53 where @xmath54 is arbitrary . in the fss case it is useful to take @xmath55 , obtaining @xmath56 . \label{scalsing}\end{aligned}\ ] ] an important question concerns the universality of the function @xmath57 . since scaling fields are arbitrarily normalized , universality holds apart from a normalization of each argument and an overall constant . therefore , given two different models , if @xmath58 and @xmath59 are the corresponding scaling functions , we have @xmath60 where all constants @xmath61 , @xmath62 , @xmath63 , @xmath64 , are nonuniversal . to go further we must discuss how the nonlinear scaling fields depend on the control parameters @xmath9 , @xmath10 , @xmath1 , and @xmath7 . first , it is natural to assume that the bulk scaling fields @xmath39 , @xmath40 , and @xmath65 do not depend on the temperature and the size of the system , i.e. they do not mix with @xmath66 and @xmath7 . this hypothesis is quite natural for systems with short - range interactions . under a rg transformation , the transformed bulk couplings only depend on the local hamiltonian , hence they are independent of the boundary . taking into account the assumed @xmath12 symmetry and the even / odd properties of @xmath9/@xmath10 , close to the critical point the relevant scaling fields @xmath39 and @xmath40 can be generally expanded as @xmath67 where @xmath68 and @xmath69 are nonuniversal constants . as for the irrelevant scaling fields , they are usually nonvanishing at the critical point . let us now discuss the scaling fields @xmath41 and @xmath42 , which are associated with the size of the @xmath30 dimensional system . for classical systems in a box of size @xmath70 with pbc and , more generally , for translation - invariant boundary conditions , it is usually assumed that @xmath71 , exactly . this assumption , which has been verified in many instances for instance , in the two - dimensional ising model and extensively discussed in ref . , can be justified as follows . consider a lattice system and a decimation transformation which reduces the number of lattice sites by a factor @xmath72 . in the absence of boundaries and for short - range interactions , the new ( translation - invariant ) couplings are only functions of the couplings of the original lattice and are independent of @xmath1 , while @xmath73 . since the flow of @xmath1 is independent of the flow of the couplings , we expect @xmath74 this condition does not generally hold for non - translation invariant systems . we thus assume that @xmath41 is an arbitrary function of @xmath66 . for @xmath75 it can be expanded as @xmath76 note that , if we define an effective size @xmath77 the scaling field becomes @xmath78 hence , by using @xmath79 , all subleading corrections due to @xmath80 are eliminated in _ any _ observable . of course , this does not imply that @xmath66 corrections are absent in _ any _ observable , as such type of corrections may have other sources ( we will come back to this point in sec . [ iii.e ] ) . such an observable - independent shift is often considered in fss studies of systems with boundary conditions that differ from the periodic ones , see , e.g. , refs . , which also provide some evidence of the presence of @xmath81 , @xmath82 corrections in the scaling field @xmath41 . let us now consider the thermal scaling field @xmath42 . to clarify the issue , let us first assume that @xmath6 , so that the quantum system is equivalent to a classical @xmath30-dimensional system . the classical system is , however , weakly anisotropic : couplings in the thermal direction differ from those in the spatial one . moreover , the anisotropy depends on the model parameters . in classical weakly anisotropic systems universality is obtained only after transforming to an isotropic system by means of a scale transformation , see refs . and references therein . therefore , we define @xmath83 , \label{utt}\ ] ] where @xmath84 is an appropriate nonuniversal function , @xmath85 and @xmath86 is a constant . the function @xmath87 may be identified with the speed of sound . more precisely , if @xmath88 is the dispersion relation of the model , which is assumed to be spatially isotropic , we define @xmath89 , where @xmath90 is the value of @xmath91 where the energy has an absolute minimum . relation ( [ utt ] ) is expected to hold also when @xmath32 , although in this case the rescaling factor is not related to the sound velocity . the scaling variables @xmath92 and @xmath93 , corresponding to the irrelevant scaling fields , vanish for @xmath75 since @xmath48 and @xmath49 are negative . thus , provided that @xmath38 is finite and nonvanishing in this limit , we can expand the singular part of the free energy as @xmath94 where we retain only the contributions of the dominant ( least ) irrelevant bulk and surface scaling fields , of rg dimensions @xmath95 and @xmath96 , respectively . note that the expansion ( [ fsing - rg ] ) is only possible below the upper critical dimension.@xcite above it , @xmath38 is singular and can not be expanded as in eq . ( [ fsing - rg ] ) . the breakdown of this expansion causes a breakdown of the hyperscaling relations and allows us to obtain the mean - field exponents . the scaling functions @xmath97 , @xmath98 , and @xmath99 are expected to be universal , apart from multiplicative normalizations and normalizations of the scaling fields . this implies that , within the given universality class , they are independent of the microscopic features of the model . however , they depend on the nature of the boundary conditions . note also the presence of the variable @xmath100 , which corresponds to the so - called shape factor in classical transitions : the universal scaling functions depend on the shape of the finite system that is considered . finally , we should also take into account the regular part @xmath101 of the free energy , see eq . ( [ gsing - rg-1 ] ) . for classical systems , in the absence of boundaries , e.g. , for pbc , @xmath37 is assumed to be independent of @xmath1 , or , more plausibly , to depend on @xmath1 only through exponentially small corrections.@xcite extending this result to the quantum case , we assume that @xmath37 does not depend on @xmath7 . instead , we see no reason why @xmath37 should not depend on @xmath1 for generic spatial boundary conditions . therefore , we assume a regular expansion in powers of @xmath66 such as @xmath102 where @xmath103 is the bulk contribution , the only one present when pbc are considered . expansions ( [ fsing - rg ] ) and ( [ favb ] ) allow us to compute all scaling corrections . as usual , we introduce the scaling variables @xmath104 and @xmath105 where @xmath79 is defined above in eq . ( [ defleff ] ) . then , we have @xmath106 where we have included the leading scaling correction . if @xmath107 and @xmath108 , @xmath109 , and @xmath110 are used as fss variables , the leading correction is of order @xmath66 . if instead , one uses @xmath111 , @xmath112 , and @xmath113 , the leading correction decreases faster , as @xmath114 . collecting all terms and using @xmath79 as basic length scale , we can write @xmath115 where @xmath116 and @xmath117 are computed at the critical point . the missing corrections are of order ( relative to the leading singular term @xmath118 ) @xmath119 , @xmath120 , @xmath121 ( they are due to the singular part of the free energy ) , and of order @xmath122 ( they are due to the regular part of the free energy ) . the last three terms appearing in eq . ( [ f - expansion - final ] ) represent boundary contributions , hence they should not be present for pbc . morever , in this case we also have @xmath123 . finally , note that , since the corrections of order @xmath119 are due to the expansion of the scaling fields , they are always proportional to @xmath108 , see eq . ( [ espansione - rapporti ] ) , thus they vanish for @xmath124 . to summarize , the rg expansion ( [ f - expansion - final ] ) provides information on the corrections to the asymptotic behavior . there are several different sources of scaling corrections : * the irrelevant rg perturbations which give generally rise to @xmath125 corrections , where @xmath126 is a universal exponent associated with the leading irrelevant rg perturbation . * corrections arising from the expansion of the scaling fields @xmath39 , @xmath40 , and @xmath42 in terms of the hamiltonian parameters . they give rise to corrections of order @xmath114 and are absent for @xmath124 . * corrections arising from the analytic background term of the free energy . * the irrelevant rg perturbations asssociated with the boundary conditions , which are of order @xmath127 . they are absent in the absence of boundaries , such as pbc . * the @xmath128 boundary corrections arising from the nontrivial analytic @xmath1-dependence of the scaling field @xmath41 , eq . ( [ ull ] ) . they are absent in the absence of boundaries . the leading correction can be taken into account by simply redefining the length scale @xmath1 , i.e. , by using @xmath79 instead of @xmath1 , cf . ( [ defleff ] ) . ( [ scalsing ] ) and ( [ fsing - rg ] ) give the generic scaling form of the free - energy density . however , in certain cases the behavior is more complex due to the appearance of logarithmic terms.@xcite they may be due to the presence of marginal rg perturbations , as it happens in berezinskii - kosterlitz - thouless transitions in u(1)-symmetric systems,@xcite or to resonances between the rg eigenvalues , as it occurs in transitions belonging to the 2d ising universality class @xcite or to the 3d o@xmath13-vector universality class in the large-@xmath28 limit.@xcite we should also mention that peculiar fss behaviors , for instance , a modulation of the leading amplitudes , are observed in quantum particle systems at fixed chemical potential when an infinite number of level crossings occurs as the system size varies , and in the so - called xx chain in a transverse external field . @xcite several interesting quantities can be obtained by taking derivatives of the free energy . for example , in particle systems whose relevant parameter @xmath9 is a linear function of the chemical potential , the fss of the particle density is obtained by differentiating eq . ( [ f - expansion - final ] ) with respect to @xmath9 , i.e. @xmath129 . therefore , for @xmath23 , we obtain @xmath130 we note that the regular term represents the leading term when @xmath131 , which is the case for most physically interesting systems . the compressibility can be obtained by taking an additional derivative with respect to @xmath9 . we can also write down general scaling laws for the quantum critical behavior in the infinite - volume limit . we start again from eq . ( [ fsing - scaling ] ) , setting in this case @xmath132 and @xmath133 . the free - energy density scales as @xmath134 where , as explained above , we assume that @xmath135 is @xmath7 independent . for @xmath23 and @xmath136 , we can expand the free energy as @xmath137 the specific heat is obtained by differentiating the previous expression : @xmath138 \nonumber\end{aligned}\ ] ] notice that there are no contributions from the regular part of the free energy . at the critical point @xmath124 , eq . ( [ cviv ] ) predicts @xmath139 . \label{cvnbeh}\ ] ] at @xmath22 and @xmath23 , any low - energy scale , and , in particular , the energy difference of the lowest - energy levels , is expected to show the asymptotic fss behavior @xmath140 , \label{deltasca}\end{aligned}\ ] ] where @xmath87 is the function providing the relation of @xmath42 with @xmath7 , cf . ( [ utt ] ) . such a factor is needed to take into account that energies are expressed in terms of the temperature @xmath7 , while the right - hand side contains the spatial dimension @xmath79 . the neglected corrections are of order @xmath81 , @xmath114 , @xmath141 , @xmath142 . the scaling functions @xmath143 are universal , apart from multiplicative factors and a normalization of their argument . for @xmath144 , @xmath145 to ensure @xmath146 for @xmath147 ( paramagnetic phase ) in the infinite - voume limit . we now consider the correlation functions of the order - parameter field @xmath148 , for example , the equal - time two - point function , @xmath149 for vanishing magnetic field , the leading scaling behavior is given by @xmath150 eq . ( [ gxysca ] ) is only valid for @xmath151 , @xmath152 with @xmath153 fixed . instead , if one takes the limit at fixed @xmath154 , no singular behavior is observed in the fss limit . corrections to eq . ( [ gxysca ] ) arise from two different sources . first of all , there are the corrections due to the scaling fields with negative rg dimensions . moreover , there are corrections which we will call _ field - mixing _ terms . indeed , the order - parameter field @xmath155 is in general a linear combination , @xmath156 of the odd fixed - point operators @xmath157 , which satisfy @xmath158 at the critical point . the associated dimensions @xmath63 ( we assume here @xmath159 ) are related to the rg dimensions defined before by @xmath160 . the leading odd operator @xmath161 is associated with the leading nonlinear scaling field of rg dimension @xmath162 given in eq . ( [ ymuh ] ) . ( [ gxysca ] ) represents the contribution of the leading operator @xmath163 since @xmath164 beside , we should also consider the contributions of all subleading operators that have the same symmetry properties of the order parameter . hence , we end up with the expansion @xmath165 in the case of obc , also boundary operators should be considered . let us consider the space integral of the correlation function ( [ gxy ] ) , defined as @xmath166 in the case of pbc , since translation invariance holds , @xmath167 is independent of @xmath168 . in the presence of a boundary , this is no longer the case . as long as @xmath168 is fixed in the fss limit , the leading scaling behavior is always the same , while scaling corrections are expected to depend on @xmath168 . the asymptotic fss expansion of @xmath167 for @xmath23 and @xmath22 is expected to be @xmath169 \nonumber \\ & & \ ; + b_\chi(\mu , l ) , \label{chisclaw}\end{aligned}\ ] ] where @xmath108 is defined in eq . ( [ scalvar ] ) , the scaling functions @xmath170 are universal apart from multiplicative factors and a normalization of the argument , and @xmath171 is the rg dimension of the next - to - leading operator which is odd under @xmath172 ( this term is due to the field mixing ) . the corrections of order @xmath114 arise from the expansion ( [ umu ] ) of the scaling field @xmath39 . finally , @xmath173 is an analytic background term which represents the contribution to the integral of points @xmath174 such that @xmath175 . it is the analogue of the analytic part of the free energy , see eq . ( [ gsing - rg-1 ] ) . therefore , the leading scaling corrections for @xmath176 scale as @xmath177 with @xmath178 . \label{expzeta}\ ] ] it is important to note that @xmath167 should not be confused with the magnetic susceptibility , which is a macroscopic quantity obtained by differentiating the free energy with respect to the magnetic field . one can also consider a correlation length @xmath0 associated with the critical modes . since @xmath0 has rg dimension 1 , in the fss limit we obtain an expansion analogous to eq . ( [ chisclaw ] ) , i.e. @xmath179 \nonumber \\ & & \ ; + b_\xi(\mu , l ) . \label{xisca}\end{aligned}\ ] ] here @xmath180 is a background term depending on the explicit definition of the correlation length . for example , we may define a second - moment correlation length by using the two - point function ( [ gxy ] ) , as @xmath181 where the point @xmath182 is at the center of the system . in the case of pbc , one may consider the more convenient definition @xmath183 where @xmath184 , @xmath185 , and @xmath186 is the fourier transform of @xmath187 . for these definitions there are two background contributions . one contribution is due to @xmath188 and scales as @xmath189 . a second one is due the sum appearing in the numerator of expression ( [ xidef ] ) and scales as @xmath190 . this second contribution is subleading with respect to the first one , hence @xmath191 we thus conclude that scaling corrections are analogous to those for @xmath176 , i.e. scale as @xmath177 , where @xmath192 is given in eq . ( [ expzeta ] ) . dimensionless rg invariant quantities are particularly useful to investigate the critical region . examples of such quantities are the ratio @xmath193 where @xmath0 is any length scale related to the critical modes , for example the one defined in eq . ( [ xidef ] ) , and ratios of the correlation function @xmath194 at different distances , e.g. , @xmath195 \label{rgxy}\ ] ] where the point @xmath196 is at the center of the system . we denote them generically by @xmath197 . according to fss , at @xmath22 and @xmath23 , they must behave as @xmath198 where @xmath199 . note the presence of the corrections of order @xmath200 , which are related to the fact that @xmath1 is used as a normalizing length scale in eqs . ( [ rxidef ] ) and ( [ rgxy ] ) . one could have equally used @xmath79 or @xmath201 , obtaining rg invariant quantities that have the same universal scaling behavior , but that differ by corrections of order @xmath66 . the scaling function @xmath202 is universal apart from a trivial normalization of the argument . in particular , the limit @xmath203 is universal within the given universality class , i.e. , it is independent of the microscopic details of the model , although it depends on the shape of the finite volume and on the boundary conditions . since @xmath204 arises from the next - to - leading @xmath205 term of the expansion ( [ umu ] ) of the scaling fields , we have @xmath206 ( with an unknown coefficient because the expansion of the scaling field is usually unknown ) . thus , this term does not contribute at @xmath26 . note also that the boundary term is absent for pbc . moreover , in the case of @xmath207 with @xmath0 defined as in eq . ( [ xidef ] ) , there is also a @xmath208 correction due to the background @xmath209 [ this term is absent in the case of @xmath210 as defined in eq . ( [ rgxy ] ) ] . a popular method to determine the critical point uses the finite - size behavior of @xmath197 as a function of @xmath1 and @xmath9 . indeed , if @xmath211 or viceversa , one can define @xmath212 by requiring @xmath213 the crossing point @xmath212 converges to @xmath26 with corrections of order @xmath214 . here @xmath215 $ ] for generic boundary conditions breaking translation invariance and @xmath216 $ ] for pbc . in the presence of backgrounds , we should also include the background corrections . for instance , in the case of @xmath207 , cf . ( [ rxidef ] ) , we have @xmath217 . \label{zetarxi}\ ] ] the quantum xy chain in a transverse field is a standard theoretical laboratory for quantum transitions . its hamiltonian is @xmath218 + g \sigma^{(3)}_x , \nonumber\end{aligned}\ ] ] where @xmath219 are the pauli matrices . we set @xmath220 and consider chains with open and periodic boundary conditions ( obc and pbc , respectively ) . we always take @xmath1 even , setting the origin at the center of the domain , more precisely at one of the two central sites , so that @xmath221 . for @xmath222 we recover the so - called xx chain in a transverse external field . for any @xmath223 the model undergoes a quantum transition at @xmath224 separating a quantum paramagnetic phase for @xmath147 from a quantum ferromagnetic phase for @xmath15 . the transition belongs to the 2d ising universality class , hence its critical behavior is associated with a 2d conformal field theory ( cft ) with central charge @xmath225 . the critical exponents take the values @xmath6 , @xmath226 , and @xmath227 . the structure of the subleading corrections for ising systems was discussed in refs . . in particular , reinicke@xcite analyzed the subleading corrections for the xy chain at the critical point . if the finite system is translation invariant this is the case of pbc the most relevant subleading operators have rg dimension @xmath228 . they belong to the identity family and can be expressed by using the virasoro generators as @xmath229 and @xmath230 , where @xmath231 , latexmath:[$q_4 = ( l_{-2}^2 - { 3\over 5 } l_{-4 } ) @xmath229 ( which can be related to the energy - momentun tensor ) is absent , as it also occurs in the classical 2d ising model.@xcite the second ( spin - four ) operator gives instead rise to scaling corrections that are proportional to @xmath233 . the primary field associated with the energy family controls the off - critical behavior . the corresponding scaling field is @xmath234.@xcite according to the analysis of ref . , the next subleading operator ( @xmath235 in their notations ) gives correction of order @xmath82 . such an operator is odd under duality transformations , which also hold for the xy model to some extent , as we discuss below . hence , we expect it to contribute only at quadratic order ( hence it gives corrections of order @xmath236 ) , as in occurs in the 2d ising model . @xcite if this term is absent , the next - to - leading correction is related to the leading spin-6 operator in the identity family , which has rg dimension @xmath237 . the quantum xy hamiltonian ( [ isc ] ) can be mapped onto a quadratic hamiltonian of spinless fermions by a jordan - wigner transformation,@xcite which can be straightforwardly diagonalized . one obtains@xcite @xmath238 where @xmath239 and @xmath240 are fermionic creation - annihilation operators and @xmath241^{1/2}. \label{ek - xy}\ ] ] the set of values of @xmath91 which must be summed over and the allowed states depend on the boundary conditions.@xcite in the limit @xmath242 , the relevant modes are those with the lowest energy , i.e. , those with @xmath243 . for @xmath244 , the energy @xmath245 can be expanded as @xmath246 , \label{dispersion - e}\ ] ] where @xmath247 as we shall see , @xmath248 and @xmath249 play the role of the nonlinear scaling fields associated with @xmath9 and with the leading irrelevant operator . note that @xmath250 therefore , provided that @xmath249 is the correct scaling field , no corrections of order @xmath251 due to the leading bulk irrelevant operator are expected for the _ improved _ value @xmath252 in any observable ( note , however , that corrections of order @xmath253 do not cancel out ) . the identification of @xmath39 and @xmath116 as scaling fields is in full agreement with the cft results of refs . , but it goes beyond that , since it conjectures the expression of the scaling field also outside the critical point , i.e. for @xmath254 . finally , let us discuss duality in the xy model.@xcite an exact transformation can be defined for the model with @xmath255 . in this case , one should consider slightly modified hamiltonians with free boundary conditions . one can consider@xcite @xmath256 it differs from hamiltonian ( [ isc ] ) because of the absence of the magnetic field on site @xmath257or @xmath258 in which there is an additional magnetic field along the @xmath259 direction at site @xmath260 . for these hamiltonians one can show that there exists a transformation @xmath261 , such that @xmath262 it follows that there exists an exact correspondence between the energy level for @xmath263 and those for @xmath264 ( again we set @xmath220 ) . the presence of a symmetry can be guessed from the expression of @xmath245 , see eq . ( [ ek - xy ] ) . indeed , for @xmath255 the energy levels satisfy @xmath265 where we have written explicitly the @xmath266 dependence . it is important to note that exact duality holds only for hamiltonians ( [ ham - dual ] ) and ( [ ham - dual2 ] ) . for different types of boundary conditions , boundary terms break duality , hence there is no direct correspondence between the states with @xmath264 and those with @xmath263 . it is interesting to understand physically why boundary conditions break duality . this is due to the different nature of the ground states for @xmath263 and @xmath264 . indeed , if @xmath266 is large , we expect hamiltonian ( [ isc ] ) to have a nondegenerate ground state with the spins aligned in the @xmath19 direction . on the other hand , if @xmath266 is small we expect a doubly degenerate ground state , with the spins aligned either in the @xmath259 direction or in the @xmath267 direction . since the degeneracy of the ground state is different for @xmath263 and @xmath268 , there can not be an exact duality symmetry . in order to have exact duality , one must therefore change the model so that ( at least ) the degeneracy of the ground state does not depend on @xmath266 . if we consider the hamiltonian ( [ ham - dual2 ] ) , this condition is realized by lifting the degeneracy of the ground state for @xmath264 : the magnetic field along the @xmath259 direction at site @xmath260 makes the ground state nondegenerate , with all spins pointing in the @xmath269 direction . if we instead consider the hamiltonian ( [ ham - dual ] ) , duality is obtained at the price of making the ground state doubly degenerate for any value of @xmath266 . to show this , note that @xmath270 = 0 $ ] . thus , the hilbert space can be decomposed into two subspaces @xmath271 , such that @xmath272 . if we restrict @xmath273 to @xmath274 we obtain hamiltonian ( [ ham - dual2 ] ) for a chain of length @xmath275 . hence , the ground state in @xmath274 is nondegenerate for all values of @xmath266 . if we restrict @xmath273 to @xmath276 we obtain @xmath277 , where @xmath278 , hence we obtain the same spectrum as that of @xmath273 restricted to @xmath279 . thus , for hamiltonian ( [ ham - dual ] ) duality is obtained by making each state doubly degenerate . finally , let us note that a remnant of duality is also present for @xmath280 . this guarantees that the transition always occurs at @xmath281 . indeed , consider the transformation @xmath282 then , we have @xmath283 so that , at points that only differ by the sign of @xmath39 , the low-@xmath91 behavior of @xmath245 is the same , apart from a change of normalization . the free energy of the quantum xy model can be directly related to the finite - size free energy of the 2d ising model . if we consider a strip of width @xmath284 , the ising free energy density is given by ( we use @xmath285 instead of @xmath286 to avoid confusion with the quantum case and write @xmath287)@xcite @xmath288 , \nonumber\end{aligned}\ ] ] where @xmath289 , \nonumber \\ & & \zeta(k ) = \cosh 2 k \coth 2 k - \cos { k/2}.\end{aligned}\ ] ] for large values of @xmath284 , the leading behavior of the finite - size correction term is obtained by expanding @xmath290 for @xmath244 , since @xmath290 is positive and has an absolute minimum at @xmath291 . close to the critical point @xmath292 , if @xmath293 ( paramagnetic phase ) , we obtain @xmath294 this expression allows us to rewrite @xmath295 where @xmath296\ } , \nonumber \end{aligned}\ ] ] and @xmath194 is catalan s constant . for the xy model we obtain a similar result . defining @xmath297 we obtain @xcite @xmath298 , \nonumber \end{aligned}\ ] ] where @xmath245 is given in eq . ( [ ek - xy ] ) . also in this case the large @xmath286 behavior is obtained by expanding @xmath245 around @xmath291 , i.e. , by using eq . ( [ dispersion - e ] ) . at leading order , we reobtain eq . ( [ fscalising ] ) with some different normalization constants : @xmath299 where @xmath300 with @xmath87 defined in eq . ( [ cmugamma ] ) , and @xmath301 note the presence of the logarithmic terms , which are due to a resonance between the identity operator of rg dimension 2 and the thermal operator of rg dimension 1.@xcite in principle , logarithmic terms should appear in all observables and both at leading and at subleading order . however , extensive analyses of the 2d ising model have identified logarithmic corrections only in a very few cases . we wish now to compute the corrections to eq . ( [ fscalxy ] ) . for this purpose we set @xmath302 and consider the expansion of @xmath303 where @xmath304 , in powers of @xmath42 , keeping @xmath259 and @xmath54 fixed . we obtain @xmath305 since @xmath306 is independent of @xmath42 and @xmath307 , we can write @xmath308 \nonumber \\ & & + \ln \bigl[1 - { e^{-{e}_c(x ) } { e}_c(x,\lambda ) b_c(x,\lambda , t ) \over 1 + e^{-{e}_c(x,\lambda ) } } + \ldots \bigr]\bigr\}\end{aligned}\ ] ] each term @xmath309 of eq . ( [ bcn ] ) increases as a power of @xmath259 for @xmath310 . therefore , the integrand vanishes exponentially as @xmath310 order by order , and we can extend the upper limit of integration to @xmath311 , making an exponentially small error . the second term in braces can then be expanded in powers of @xmath42 , proving that the free energy admits an expansion in powers of @xmath42 at @xmath54 fixed . let us now compute the first correction of order @xmath312 . proceeding as discussed above , we obtain @xmath313 = \int_0^{\infty } dx \ln\left[1 + e^{-{e}_c(x,\lambda)}\right ] \nonumber \\ & & \ ; - { u^2_t\over2 } v_1(\mu,\gamma ) \int_0^{\infty } dx { x^4 ( x^2 + \lambda^2)^{-1/2 } \over 1 + \exp \sqrt{x^2 + \lambda^2 } } .\end{aligned}\ ] ] note that the corrections of order @xmath312 are proportional to @xmath249 . hence , it is natural to identify this quantity as the scaling field associated with the leading irrelevant operator . we will confirm this conjecture in the next sections . we wish now to compute the finite - size behavior of the difference @xmath20 between the energy of the lowest excited states and that of the ground state , extending the results of ref . . for pbc we shall show that @xmath20 admits an expansion of the form @xmath314 \label{expansion - delta - wtilde}\ ] ] where @xmath315 an analogous expansion holds also for antiperiodic boundary conditions ( abc ) . such a result confirms the identification of @xmath39 and @xmath116 as nonlinear scaling fields . note that , if @xmath316 is used as scaling variable and @xmath87 is replaced by its leading behavior @xmath317 , then we have @xmath318 . \label{expansion - delta - w}\ ] ] the corrections of order @xmath200 , which vanish at the critical point @xmath319 , are due to the expansion of the nonlinear scaling field @xmath39 and of the sound velocity @xmath87 ( in the general case , they would be of order @xmath114 ) . to compute the energy levels , we use the results of katsura.@xcite they are obtained by using eq . ( [ h - harmonic ] ) , with a proper identification of the allowed values of @xmath91 . the energy levels can be divided in two sectors : the even one in which @xmath320 , @xmath321 , and the odd one in which @xmath322 , @xmath321 . the corresponding ground - state energies are @xmath323 where @xmath245 is given in eq . ( [ ek - xy ] ) . note that , for @xmath223 , we have @xmath324 . half of the states belong to the odd sector . they can be written as @xmath325 , where @xmath326 , @xmath327 , @xmath328 , @xmath329 belong to the odd sector , @xmath330 is even for pbc and odd for abc . the allowed states in the even sector can also be written as @xmath331 , but now @xmath330 depends both on the boundary conditions and on the value of @xmath266 . for @xmath332 , @xmath330 is odd ( even ) for pbc ( abc ) . for @xmath333 the opposite condition holds : @xmath330 is even for pbc , odd for abc . for @xmath281 the parity of @xmath330 can be chosen arbitrarily , obtaining the same spectrum in all cases as a result of the fact that @xmath334 . therefore , for @xmath263 and pbc , the lowest energy states are @xmath335 while for @xmath333 we obtain @xmath336 for abc we have for @xmath337 @xmath338 while for @xmath333 @xmath339 note that the first two excited states are doubly degenerate . then , @xmath20 and @xmath340 , the energy gaps for the first and second excited state , respectively , are given by @xmath341 with @xmath342 for @xmath343 , @xmath344 for @xmath345 . the behavior of these quantities in the fss limit , in which @xmath346 , @xmath75 at @xmath347 fixed , was considered in ref . . we have performed again the calculation , using the general method discussed in app . b of ref . . we obtain @xmath348 \nonumber \\ & & - { w^2\over 4 \pi l } \left ( \ln { w^2\over 16\pi^2 } + 2 \gamma_e - 1\right ) , \\ { 1\over \gamma } { \cal e}_0^{\rm even } & \approx & - { l\over \gamma } j + { 1\over l}\left [ - { \pi\over6 } + 4\pi g_1(w/2\pi ) - 2 \pi g_1(w/\pi ) \right ] \nonumber \\ & & - { w^2\over 4 \pi l } \left ( \ln { w^2\over \pi^2 } + 2 \gamma_e - 1\right),\end{aligned}\ ] ] where @xmath349 , @xmath350 is euler s constant , @xmath351 and @xmath352 is a remnant function:@xcite @xmath353 for @xmath354 , @xmath355 , while for @xmath356 we have @xcite @xmath357 where @xmath358 is a modified bessel function . using these results we obtain @xcite @xmath359 these results are consistent with eq . ( [ expansion - delta - wtilde ] ) since @xmath360 and @xmath361 for @xmath362 . the pbc curve is shown in fig . [ delta1-as ] . for @xmath363 , eqs . ( [ delta1p - scal ] ) and ( [ delta1a - scal ] ) give @xmath364 and @xmath365 , in agreement with ref . . for @xmath366 , using eq . ( [ g1as - large ] ) we obtain @xmath367 , \nonumber \end{aligned}\ ] ] which shows that @xmath368 for @xmath369 and @xmath370 for @xmath371 , with exponentially small corrections . in the same limit , @xmath372 behaves as @xmath373 , but corrections are now of order @xmath374 . versus @xmath375 in the scaling limit for pbc ( @xmath123 for pbc ) , obc , and for the hamiltonian ( [ ham - dual2 ] ) . ] versus @xmath376 for @xmath377 and 0.8 . the two sets of data appear to follow a unique curve . ] let us now focus on the corrections . for this purpose , in the pbc case we consider the combination @xmath378.\ ] ] if eq . ( [ expansion - delta - wtilde ] ) is correct , such a combination should converge to @xmath379 as @xmath151 at fixed @xmath380 . moreover , the limiting curve should be independent of @xmath381 . the results for @xmath377 and 0.8 shown in fig . [ corr - nmeno2-pbc ] are in full agreement , confirming eq . ( [ expansion - delta - wtilde ] ) at order @xmath82 . these curves are obtained by computing @xmath20 using high - precision arithmetics for values of @xmath1 in the range @xmath382 at fixed @xmath380 . on the scale of the figures the results fall on top of each other , providing the limiting scaling curve . to verify that the neglected corrections in eq . ( [ expansion - delta - wtilde ] ) decay as @xmath383 , we consider the case @xmath384 , for which @xmath385 , and compute @xmath386 . \label{delta4def}\end{aligned}\ ] ] if corrections are of order @xmath383 , this quantity should have a finite limit as @xmath151 at fixed @xmath380 . if corrections are instead of order @xmath82 , this quantity should diverge linearly in @xmath1 as @xmath151 . the results shown in fig . [ delta14 ] are consistent with a finite limit , hence confirm that corrections decay as @xmath383 . defined in eq . ( [ delta4def ] ) versus @xmath380 for two values of @xmath1 , @xmath387 and @xmath388 . the two sets of data are hardly distinguishable , showing that @xmath389 approaches a nontrivial large-@xmath1 limit . ] let us now consider the obc case . this case is more complex than the pbc one , since we must take into account boundary irrelevant corrections and the fact that @xmath41 is a nontrivial function of @xmath1 . the latter type of corrections can be taken into account by using an effective size @xmath79 . boundary irrelevant operators give instead rise to corrections of order @xmath81 , hence @xmath390 . in practice we will show that @xmath391 , \label{expan - delta - obc}\ ] ] where @xmath392 the function @xmath393 is not proportional to @xmath394 ( for instance , it does not vanish for @xmath395 ) , indicating the presence of boundary contributions with @xmath396 . for @xmath255 the effective length @xmath79 is equal to @xmath397 . from the analysis of the numerical data , we will conjecture a general expression for @xmath79 , valid for all values of @xmath381 , see below . we first consider the case @xmath255 , for which we can use the analytic expressions reported in ref . . the gap @xmath20 is given by @xmath398 , where @xmath399 is the smallest ( in absolute value ) @xmath91 that satisfies the secular equation @xcite @xmath400 in the scaling limit at fixed @xmath401 , the solution of the secular equation depends on @xmath108 . for @xmath402 , we have @xmath403 where @xmath404 is the solution in @xmath4050,\pi[$ ] of the equation @xmath406 and @xmath407 is given by @xmath408 for @xmath409 , we have instead @xmath410 where @xmath404 is the solution in @xmath4050,+\infty[$ ] of the equation @xmath411 and @xmath412 for @xmath413 , we have @xmath414 and @xmath415 ; for @xmath416 , we have @xmath417 and @xmath418 ; for @xmath419 , @xmath404 and @xmath407 both vanish , while for @xmath420 we obtain @xmath421 $ ] and @xmath422 . using the expansion of @xmath399 and eq . ( [ dispersion - e ] ) we obtain in the limit @xmath151 at @xmath108 fixed @xmath423 where the upper signs should be used for @xmath424 and the lower signs for @xmath425 . as in the pbc case , the gap @xmath404 vanishes as @xmath426 , while for @xmath369 we have @xmath427 . the resulting curve is reported in fig . [ delta1-as ] . to get rid of the analytic corrections , we express @xmath428 as a function of @xmath380 . since @xmath429 , we obtain @xmath430 where @xmath404 is now a function of @xmath380 and we used @xmath431 the @xmath66 correction can be eliminated by rescaling the size @xmath1 . indeed , if we define @xmath432 we obtain @xmath433 where @xmath404 is now a function of @xmath434 . this result can be derived immediately if we rewrite the secular equation in terms of @xmath39 and @xmath79 : @xmath435^{1/2 } } .\ ] ] this equation is symmetric under @xmath436 and @xmath437 , implying the @xmath91 has an expansion in odd powers of @xmath438 . no even powers appear , confirming the absence of corrections of order @xmath439 in the expansion of the gap . the previous analysis was restricted to the first correction . it is important to stress that it is not possible to eliminate the corrections of order @xmath82 in the expansion of @xmath399 by redefining @xmath440 , with a suitable @xmath441 . indeed , at the critical point the secular equation gives @xmath442 exactly . therefore , for @xmath443 , there are no @xmath444 corrections only if @xmath445 . but , if we take @xmath445 , corrections are present for @xmath446 . let us now consider the energy gap for @xmath447 . in the absence of analytic results , we compute the difference @xmath20 of the two lowest energy levels numerically for @xmath448 and for @xmath449 , 0.8 , and 0.4 . also for these values of @xmath381 we find that the leading scaling correction can be eliminated using appropriate @xmath381-dependent @xmath79 . an accurate numerical guess of @xmath79 is @xmath450 with this choice @xmath20 has an expansion of the form @xmath451 we can estimate the correction term by considering @xmath452 , \nonumber\\ \label{delta2obc}\end{aligned}\ ] ] where @xmath453 and @xmath454 correspond to @xmath1 and @xmath455 , respectively , and @xmath456 and @xmath457 are obtained by solving @xmath458 and @xmath459 . the resulting quantity has a finite limit for @xmath75 at fixed @xmath460 , reported in fig . [ delta12-obc ] . note that corrections do not vanish for @xmath461 , where @xmath462 , hence they can not be only due to the bulk subleading operator with @xmath463 . moreover , there is no rescaling that allows us to obtain a collapse of all data onto a single curve . therefore the data show the presence of corrections due to a boundary subleading operator with exponent @xmath396 . defined in eq . ( [ delta2obc ] ) versus @xmath460 for obc . results for @xmath464 , @xmath465 , 0.4 , and @xmath466 ( for @xmath467 we have @xmath123 ) . ] finally , it is interesting to consider hamiltonian ( [ ham - dual2 ] ) . the secular equation turns out to be particularly simple . the allowed values of @xmath91 are simply @xmath468 , @xmath469 for all values of @xmath266 . therefore , if we define @xmath470 , we have @xmath471 the scaling function for @xmath151 is reported in fig . [ delta1-as ] . note that @xmath472 does not vanish for @xmath473 , a consequence of the fact that the degeneracy for @xmath264 is lifted by the added magnetic field . a second peculiarity of the result is the absence of boundary corrections , once length scales are expressed in terms of @xmath474 . , defined in eq . ( [ rdef2 ] ) , versus @xmath475 for pbc ( top ) and versus @xmath375 for obc ( bottom ) . we report data for @xmath377 , 0.8 , and @xmath461 . they approach a universal curve with increasing @xmath1 . the horizontal dotted lines correspond to the exact value at @xmath26 . the dashed lines show the asymptotic behavior @xmath476 for @xmath477 . , title="fig : " ] , defined in eq . ( [ rdef2 ] ) , versus @xmath475 for pbc ( top ) and versus @xmath375 for obc ( bottom ) . we report data for @xmath377 , 0.8 , and @xmath461 . they approach a universal curve with increasing @xmath1 . the horizontal dotted lines correspond to the exact value at @xmath26 . the dashed lines show the asymptotic behavior @xmath476 for @xmath477 . , title="fig : " ] in sec . [ iii.b ] , [ iii.c ] , and [ iii.d ] we have shown that the data for the free energy and energy gap are consistent with the assumption that @xmath39 and @xmath116 are nonlinear scaling fields . moreover , in the case of obc the leading boundary correction can be eliminated by redefining @xmath478 . we wish now to verify these conjectures by studying different observables related to the correlation function of the order parameter @xmath479 . we consider the equal - time correlation function @xmath480 then we define @xmath481 and the rg invariant quantities @xmath482 . \label{rdef2}\ ] ] we compute @xmath210 for pbc and obc , for several values of @xmath1 and @xmath380 ( @xmath79 and @xmath460 in the obc case ) and extrapolate the results to @xmath75 . the results are reported in fig . [ rgfss ] . note that in the scaling limit , all scaling variables are equivalent , i.e. , @xmath483 , but this is not true when considering the scaling corrections . the value of @xmath210 at the critical point for @xmath75 can be computed by using the exact expression of the two - point function at the critical point in the scaling limit . the numerical values are reported in app . [ cftfo ] . we can also predict the large-@xmath108 behavior by using the known expression of @xmath484 in the infinite - volume limit for @xmath14 . since @xmath485 for @xmath486,@xcite where @xmath487 is a modified bessel function , we obtain @xmath488 for @xmath489 . , ( [ whrg2 ] ) , versus @xmath376 for pbc . we report data for @xmath490 , and some values of @xmath1 . data clearly converge toward an asymptotic large-@xmath1 curve . ] let us now discuss the leading corrections in the pbc case . according to the general analysis , in the limit @xmath151 , @xmath491 at fixed @xmath380 , we expect corrections of order @xmath492 due to the leading irrelevant operator and corrections due to field mixings . we will show that the latter also scale as @xmath81 , obtaining an expansion of the form @xmath493 with @xmath494 where @xmath249 is the nonlinear scaling field reported in eq . ( [ v1mugamma ] ) . to verify this expansion , we consider the combination @xmath495 . \label{whrg2}\end{aligned}\ ] ] if eq . ( [ expan - rg - pbc ] ) holds , @xmath496 converges to @xmath497 with corrections of order @xmath81 . if instead eq . ( [ expan - rg - pbc ] ) does not hold and corrections to the leading scaling behavior are of order @xmath66 , @xmath496 diverges as @xmath75 . in fig . [ rgcpbc ] we show the results for @xmath498 for which @xmath462 . the combination @xmath496 has a finite limit for @xmath151 , confirming that the leading scaling corrections decay as @xmath499 . since @xmath500 , the corrections we observe can not be due to the operator @xmath501 which controls the leading scaling correction for the free energy and the spectrum . corrections are instead a field - mixing effect . the lattice operator is a combination of conformal fields : @xmath502 where @xmath503 is the primary cft field and @xmath504 are the secondary fields that belong to the @xmath505 family , the leading one being @xmath506 and @xmath507 . to provide additional evidence for the validity of eq . ( [ expan - rg - pbc ] ) , we consider @xmath508 \label{rg21def}\\ & - & { 4 l^2\over 3 } { v_2(0,\gamma)\over v_2(0,\gamma_i ) } [ r_g(\mu , l,\gamma_i ) - r_g(\mu,2 l,\gamma_i ) ] . \nonumber \end{aligned}\ ] ] if eq . ( [ expan - rg - pbc ] ) holds , then @xmath509 since @xmath249 is known , this relation gives us a recipe to identify @xmath510 . we determine @xmath510 by requiring @xmath511 to be independent of @xmath381 . by using numerical results for @xmath377 and @xmath465 , we find that this condition is satisfied by simply taking @xmath512 this is shown by the data in fig . [ srg - pbc ] . the scaling field @xmath513 is independent of @xmath381 for @xmath491 . versus @xmath376 for pbc , cf . ( [ rg21def ] ) . we use eq . ( [ v2cond ] ) . the data for different values of @xmath381 appear to approach the same curve with increasing @xmath1 . ] , defined in eq . ( [ def - t ] ) , for @xmath514 and some values of @xmath381 . ] let us now consider the scaling corrections in the obc case . in this case we find that scaling corrections of order @xmath66 are present even if the limit @xmath151 is taken at fixed @xmath460 . indeed , the numerical data are consistent with @xmath515 to estimate the @xmath66 correction , we consider @xmath516 . \label{def - t}\end{aligned}\ ] ] for @xmath151 , we have @xmath517 with corrections of order @xmath82 . @xcite we have computed @xmath518 for @xmath519 , obtaining , for all values of @xmath381 a nonzero result . the function @xmath518 for @xmath520 ( it is essentially asymptotic ) is reported in fig . [ plot - t ] . note that it has a nontrivial dependence on @xmath381 : no rescaling exists that makes the curves corresponding to different values of @xmath381 fall one on top of the other . this implies that such correction can not be ascribed to a single subleading operator . we can also exclude that the @xmath66 correction can be eliminated by using @xmath79 in the definition of @xmath210 , i.e. , by defining @xmath521.\ ] ] indeed , for @xmath522 , we have @xmath123 , hence @xmath523 . but also in this case @xmath66 corrections are present . they may be explained by the presence of field mixings with the boundary operators . versus @xmath376 for pbc ( top ) and versus @xmath375 for obc ( bottom ) . we report data for @xmath377 and 0.8 . for both obc and pbc the data for different values of @xmath381 approach a universal curve with increasing @xmath1 . the horizontal dotted lines correspond to the exact value at @xmath26 . the dashed lines show the asymptotic behavior @xmath524 for @xmath525 ( for obc we should replace @xmath380 with @xmath460 ) . , title="fig : " ] versus @xmath376 for pbc ( top ) and versus @xmath375 for obc ( bottom ) . we report data for @xmath377 and 0.8 . for both obc and pbc the data for different values of @xmath381 approach a universal curve with increasing @xmath1 . the horizontal dotted lines correspond to the exact value at @xmath26 . the dashed lines show the asymptotic behavior @xmath524 for @xmath525 ( for obc we should replace @xmath380 with @xmath460 ) . , title="fig : " ] at @xmath124 vs @xmath526 . the dotted lines are fits to @xmath527 , while the dot - dashed one is a fit to @xmath528 only . results for obc and pbc and for @xmath377 and 0.8 . ] let us finally consider @xmath207 . its behavior in the scaling limit is shown in fig . [ rxifss ] . the finite - size behavior of @xmath207 is more complex , since one must also take into account the background term which gives corrections of order @xmath529 , independent of the type of boundary conditions . for obc next - to - leading corrections are of order @xmath200 , while for pbc , if the scaling limit is taken at fixed @xmath380 , they are of order @xmath530 and are due to the @xmath1 dependence of the background term . these predictions are well confirmed by the data shown in fig . [ scalcorxi ] . in a quantum system the reduced density matrices of spatial subsystems , and , in particular , the corresponding entanglement entropies and spectra , provide effective probes of the nature of the quantum critical behavior , see , e.g. , refs . . their dependence on the finite size of the system may be exploited to determine the critical parameters of a quantum transition.@xcite in this section we discuss the finite - size behavior of the entanglement entropy of spatial bipartitions of the system . we restrict the discussion to zero temperature and to one - dimensional systems with an isolated quantum critical point with @xmath6 and central charge @xmath87 . the general fss behavior is then compared with exact and numerical results for the xy chain . we divide the chain into two connected parts of length @xmath531 and @xmath532 , and consider the rnyi entropy ( @xmath533 ) @xmath534 where @xmath535 is the reduced density matrix of one of the two subsystems . for @xmath536 , the rnyi entropy coincides with the von neumann ( vn ) entropy @xmath537 the asymptotic behavior of bipartite entanglement entropies is known at the critical point @xmath26.@xcite we have @xmath538 , \label{ccfo}\ ] ] where @xmath87 is the central charge , @xmath539 counts the number of boundaries between the two parts , thus @xmath540 in the case of pbc and @xmath541 in the case of obc . the constant @xmath542 is nonuniversal and depends on the boundary conditions.@xcite the corrections to eq . ( [ ccfo ] ) may have various origins . beside the corrections discussed in sec . [ generalfss ] , there are additional corrections . within cft they are related to the operators associated with the conical singularities at the boundaries between the two parts , which appear in the @xmath543-sheeted riemann surface introduced to compute @xmath544 . @xcite in the limit @xmath545 at fixed @xmath546 , these new operators give rise to terms of order @xmath547 in the case of obc@xcite and of order @xmath548 in the case of pbc.@xcite here @xmath549 is the rg dimension @xcite of the leading conical operator . the results for a number of 1d models suggest that the energy operator plays a major role in this respect,@xcite hence @xmath550 . moreover , the analysis of exactly solvable models shows the presence of other corrections suppressed by integer powers of @xmath1 . @xcite the general predictions are confirmed by the exact results for the xy chain at the critical point , for both obc and pbc . they are summarized in app . [ app.b ] . the asymptotic behavior of the bipartite entanglement entropies is also known in the thermodynamic limit close to the transition point,@xcite i.e. , for @xmath551 , where @xmath0 is the length scale of the critical modes . one obtains @xcite @xmath552 where again @xmath540 in the case of pbc and @xmath541 in the case of obc , and @xmath553 is a nonuniversal constant . the corrections to the asymptotic behavior ( [ salthl ] ) are expected to be@xcite of order @xmath554 , where @xmath555 is the same exponent controlling the finite - size corrections at the critical point . additional corrections of order @xmath556 should also be present , see e.g. ref . in the general fss regime , the bipartite entanglement entropy has been conjectured to satisfy the asymptotic scaling equation @xcite @xmath557 consistency with eqs . ( [ ccfo ] ) and ( [ salthl ] ) implies that for @xmath477 @xmath558 if we include the scaling corrections , we expect @xmath559 where the dots correspond to other corrections of order @xmath560 , @xmath561 , @xmath328 , which may be more relevant than the conical ones in some cases . starting from the entanglement entropies , one can define rg invariant quantities , which can be used to determine the critical behavior in a finite volume . for this purpose , we consider @xmath562 \label{qalpha}\ ] ] with @xmath563 . according to eq . ( [ ccfo ] ) , at the critical point @xmath26 @xmath564 on the other hand , for @xmath254 and @xmath565 , since @xmath566 is independent of @xmath531 in this limit , we have @xmath567 . the quantity @xmath568 may be used to determine the transition point and critical exponents , as the rg invariant quantities @xmath197 considered in sec . [ drgi ] . for any boundary condition , eq . ( [ fsssal ] ) implies @xmath569 with @xmath570 , where the dependence on the interval coordinates @xmath571 is understood . the scaling functions @xmath572 and @xmath573 depend only on @xmath574,@xmath575 , and the boundary conditions , apart from a trivial normalization of their argument , while @xmath576 is a nonuniversal constant . in the pbc case , we have @xmath577 since corrections decay as @xmath578 at the critical point . beside the corrections of order @xmath547 , one should also consider the standard corrections related to the usual bulk and boundary irrelevant operators , and analytic corrections . , derived from the vn entanglement entropy using eqs . ( [ qalpha ] ) and ( [ qadef ] ) , for @xmath579 and obc for several values of @xmath1 . the dotted lines connecting the data corresponding to the same value of @xmath1 are only meant to guide the eye . ] ( bottom ) and @xmath580 ( top ) for obc vs @xmath581 , for @xmath579 and @xmath465 . in both cases the data for different values of @xmath381 approach a universal large-@xmath1 curve.,title="fig : " ] ( bottom ) and @xmath580 ( top ) for obc vs @xmath581 , for @xmath579 and @xmath465 . in both cases the data for different values of @xmath381 approach a universal large-@xmath1 curve.,title="fig : " ] ( bottom ) and @xmath580 ( top ) for obc vs @xmath582 , for @xmath579 and @xmath465 . in both cases the data clearly converge toward an asymptotic large-@xmath1 curve which is independent of @xmath381.,title="fig : " ] ( bottom ) and @xmath580 ( top ) for obc vs @xmath582 , for @xmath579 and @xmath465 . in both cases the data clearly converge toward an asymptotic large-@xmath1 curve which is independent of @xmath381.,title="fig : " ] to verify the general fss behaviors presented in sec . [ eprobes ] , we consider again the xy chain . in this case @xmath583 , so that @xmath584 . therefore , for @xmath585 the corrections associated with the rnyi entanglement entropies of order @xmath586 are stronger than the standard ones discussed in the previous sections , which scale as @xmath66 at least . we consider the quantity @xmath587 i.e. , we take @xmath588 and @xmath589 in eq . ( [ qalpha ] ) . in the following we present results derived from the renyi and vn entanglement entropies of xy chains , for several values of @xmath381 , obc and pbc , and lattice sizes up to @xmath590 . fig . [ q1vsmu ] shows @xmath591 for @xmath579 and several values of @xmath1 with obc . the curves show a maximum for @xmath15 and cross each other approximately at @xmath26 . using eq . ( [ qalfss ] ) , one can easily establish that the crossing point @xmath592 , defined by @xmath593 = q_\alpha[\mu_{\rm cross}(l),2l ] , \label{crossp}\ ] ] approaches the critical point as @xmath594 this is confirmed by the results for @xmath568 , see e.g. fig . [ q1vsmu ] . [ qfssobc ] and [ qpbcfss ] show plots of @xmath591 and @xmath580 , for obc and pbc respectively , versus the scaling variable @xmath595 for @xmath579 and @xmath465 . the data appear to approach universal curves with increasing @xmath1 , clearly supporting the universality of the asymptotic function @xmath596 , cf . ( [ qalfss ] ) , for both obc and pbc . note that the maximum of the pbc scaling curve is at @xmath597 and equals @xmath225 , while the obc maximum is larger than @xmath225we obtain @xmath598 and @xmath599and it is located in the region @xmath600 . the scaling curves vanish exponentially for @xmath601 . as expected , scaling corrections appear larger for @xmath580 than @xmath591 . limit of @xmath602 for obc , defined in eq . ( [ q1corr - pbc - def ] ) , vs @xmath434 , for several values of @xmath381 , in particular @xmath461 . the different curves are hardly distinguishable : the small differences are within the accuracy of the large-@xmath1 extrapolation of the data up to @xmath603 . ] limit of @xmath602 for pbc , defined in eq . ( [ q1corr - pbc - def ] ) , vs @xmath604 , for several values of @xmath381 . here on the scale of the figure , the different curves are hardly distinguishable . the small differences are within the extrapolation errors . for @xmath319 the numerical data are consistent with zero , i.e. , with the absence of @xmath605 corrections , in agreement with the exact results at the critical point . ] let us now investigate the corrections to the leading term . to begin with , we consider the renyi entanglement entropies for @xmath606 , whose leading corrections are expected to be due to the conical singularities , i.e. the @xmath607 term explicitly reported in eq . ( [ qalfss ] ) . we use the asymptotic formulas reported in app . [ app.b ] to derive the finite - size behavior of @xmath568 at @xmath26 . we obtain @xmath608 where @xmath609 ( 2^{1/(2\alpha ) } - 1 ) \over ( 1+\alpha ) \gamma[3/2 - 1/(2\alpha ) ] \ln 2}. \nonumber\end{aligned}\ ] ] in particular @xmath610 for @xmath611 . instead , for pbc at @xmath26 , we find @xmath612 ^ 2 \over \alpha ( \alpha+1 ) \gamma[3/2 - 1/(2\alpha)]^2 { \rm ln}2 } \nonumber \end{aligned}\ ] ] for @xmath486 . in particular @xmath613 . the fss limit is taken at fixed @xmath434 for obc and @xmath376 for pbc ( see eqs . ( [ umugamma ] ) and ( [ legamma ] ) for the definitions of @xmath39 and @xmath474 ) , to avoid analytic corrections due to the expansion of the scaling fields . the numerical data of the @xmath611 renyi entropy are in full agreement with eq . ( [ qalfss ] ) : for both pbc and obc scaling corrections decay as @xmath605 . moreover , for both obc and pbc , the corrections are proportional to @xmath614 as found at the critical point , cf . ( [ gdepba ] ) . this is clearly demonstrated by the analysis of the large-@xmath1 behavior of the quantity @xmath615 . \label{q2corr - pbc - def}\end{aligned}\ ] ] if corrections are of order @xmath616 , in the limit @xmath75 at fixed @xmath380 or @xmath617 , @xmath618 converges to a nontrivial @xmath381-independent scaling function , i.e. to the function @xmath619 appearing in eq . ( [ qalfss ] ) . figs . [ q2corrobc ] and [ q2corrpbc ] show the extrapolation of @xmath618 for obc and pbc , respectively . they are obtained by using results for chains of length @xmath620 . the resulting tiny differences that are hardly visible in figs . [ q2corrobc ] and [ q2corrpbc ] are plausibly due to tiny numerical errors affecting the raw data and to the extrapolation uncertainty . the curves for different values of @xmath381 appear to approach a unique curve , thus supporting our general scenario . analogous results are expected for any @xmath606 . the analysis of the leading corrections for the vn entanglement entropy is more complicated , essentially because , in the limit @xmath536 , the leading corrections may have different origins . this is already shown by the results at the critical point . the asymptotic expansion of the vn entanglement entropy at the critical point for obc and @xmath486 is reported in app . [ app.b ] . this allows us to derive @xmath621 note that @xmath622 does not coincide with the @xmath536 limit of the coefficient @xmath576 appearing in eq . ( [ gdepba ] ) . thus other corrections contribute at order @xmath66 . to understand better the subleading fss behavior , we computed the corrections of order @xmath200 at fixed @xmath434 . they can be estimated by considering @xmath623 . \label{q1corr - obc - def}\end{aligned}\ ] ] this quantity is constructed so that it approaches a nontrivial function if the leading corrections are of order @xmath200 . [ q1corrobc ] shows the large-@xmath1 extrapolations @xcite of @xmath624 for several values of @xmath381 . we verify that the @xmath381 dependence can not be eliminated by rescaling @xmath625 by a function of @xmath381 . hence , beside the conical contribution , there must be other corrections due to the boundaries . they may be interpreted as analytic corrections related to the length @xmath626 of the domain . analogously to the nonlinear scaling field @xmath41 associated with @xmath66 which has an expansion in powers of @xmath66 , cf . ( [ ull ] ) , it is natural to introduce a scaling field @xmath627 associated with @xmath626 , with @xmath628 . the expansion of @xmath627 would contribute additional boundary corrections of order @xmath66 , when the limit is taken at fixed @xmath629 . this is confirmed by the asymptotic behavior of the vn entanglement entropy at the critical point , see app . [ app.b ] . indeed , for @xmath486 it can be written as @xmath630 \nonumber\\ & & - { \pi \over 16\sin(\pi \ell_e / l_e ) } { 1\over l_e } + o(l_e^{-2 } ) , \label{salpha1 t } \end{aligned}\ ] ] where @xmath631 and @xmath632 , @xmath633 is a constant , and the term of order @xmath200 is the @xmath536 limit of the corrections of order @xmath586 occurring for generic @xmath606 . ( [ salpha1 t ] ) allows us to identify the origin of the correction terms : there are conical corrections that give rise to the @xmath634 term appearing in eq . ( [ salpha1 t ] ) , and boundary terms that can be allowed for by introducing @xmath474 and @xmath635 . in the case of pbc , the results of app . [ app.b ] at the critical point lead to @xmath636 the constant @xmath637 is unrelated to the constant @xmath638 defined in eq . ( [ qacritpbc ] ) . indeed , @xmath638 vanishes for @xmath639 . these results apparently indicate that conical singularities are not related to the @xmath81 corrections at the critical point . let us now extend the analysis outside the critical point , computing @xmath591 in the fss limit at fixed @xmath380 . a detailed numerical analysis shows that there are no scaling corrections of order @xmath66 . the function @xmath640 appearing in eq . ( [ qalfss ] ) vanishes identically : corrections for @xmath641 decay as @xmath81 as it occurs at the critical point . a detailed analysis of the numerical data for several values of @xmath381 shows that we can write @xmath642 + o(l^{-3 } ) . \nonumber \end{aligned}\ ] ] where @xmath249 , defined in eq . ( [ v1mugamma ] ) , is the scaling field associated with the leading bulk subleading corrections . notice that if we replace @xmath643 with its linear approximation @xmath581 in eq . ( [ scal - q1-pbc ] ) , the leading term does not change but now the corrections are of order @xmath200 . they are due to the next - to - leading term appearing in the expansion of @xmath380 in powers of @xmath108 . to verify eq . ( [ scal - q1-pbc ] ) , we consider @xmath644 , \label{q1corr - pbc - def}\ ] ] where @xmath490 ( we remind the reader that @xmath500 ) . if eq . ( [ scal - q1-pbc ] ) holds , @xmath645 should converge to @xmath646 , hence it should be independent of @xmath381 for large values of @xmath1 . as shown in fig . [ q1corrpbc ] , a straightforward extrapolation @xcite of data up to @xmath647 supports it . the second correction term in eq . ( [ scal - q1-pbc ] ) is associated with the bulk irrelevant operator . the origin of the first term , which is independent of @xmath381 , is instead less clear . as we have already discussed , at the critical point the conical corrections of order @xmath648 vanish , hence @xmath649 can only be an analytic correction . it is natural to conjecture that the same is true outside the critical point . indeed , if conical and analytic corrections were both present , one would expect them to have different @xmath381 dependencies . hence , one would expect two different scaling functions with different @xmath381-dependent coefficients . limit of @xmath650 defined in eq . ( [ q1corr - obc - def ] ) vs @xmath434 , for several values of @xmath381 . here @xmath461 and @xmath651 . ] limit of @xmath652 defined in eq . ( [ q1corr - pbc - def ] ) for various values of @xmath381 vs @xmath376 . the data collapse along a unique curve ; the apparent oscillations , that are particularly visible for @xmath653 and @xmath654 , are essentially due to numerical errors in the computation of the raw data for @xmath655 . ] we study fss at quantum zero - temperature transitions , focusing on the corrections to the leading asymptotic behavior . this issue is relevant for numerical and experimental studies of quantum transition , where the data are generally available for a limited range of system sizes , which are often relatively small . in these cases , the fss predictions are subject to sizable scaling corrections , which must be taken into account to obtain reliably accurate estimates of the critical parameters and , if needed , to identify the universality class of the transition . we present a rg analysis of fss at quantum zero - temperature transitions of @xmath8-dimensional systems characterized by two relevant parameters @xmath9 and @xmath10 , which are respectively even and odd with respect to an assumed parity - like symmetry . well known examples of such quantum transitions are those occurring in quantum xy ( ising ) systems and general o@xmath13-symmetric spin models , superfluid or metallic transitions in particle systems , etc . ; see , e.g. , ref . . to characterize the scaling corrections , we generalize wegner s scaling ansatz @xcite to quantum transitions . this allows us to predict the type of subleading corrections that are expected in finite systems and/or at finite temperature . first , there are corrections associated with the bulk and boundary irrelevant rg perturbations , that decay as @xmath656 , where @xmath109 is generally a noninteger exponent ( for example , in the case of the quantum transitions of two - dimensional quantum ising and heisenberg models , the leading bulk @xmath125 corrections have @xmath657 , see e.g. ref . ) . then , one should consider analytic corrections due to the regular backgrounds . finally , since the rg predictions are expressed in terms of the nonlinear scaling fields , one should also consider the correction terms arising from their expansion in powers of the hamiltonian parameters , the spatial size @xmath1 , and the temperature @xmath7 . these corrections are also named analytic , though they decay as @xmath658 , with noninteger @xmath659 . to check the general predictions , we consider the quantum xy chain in a transverse field with hamiltonian ( [ isc ] ) , which is a standard theoretical laboratory to understand issues concerning quantum transitions . in particular , it is an ideal testing ground , since its hamiltonian can be exactly diagonalized,@xcite allowing us to compute several interesting quantities either exactly or very accurately by using numerical methods . the analytic computation of the finite - size behavior of the energy spectrum and of the free energy allows us to infer the exact form of the nonlinear scaling fields related to the relevant hamiltonian parameter @xmath660 and to the leading irrelevant operator with rg dimension @xmath228 . moreover , we can also determine the _ speed of sound _ @xmath87 which enters the relation between the temperature @xmath7 and the corresponding scaling field . we provide a complete analysis of the asymptotic fss behavior of the energy gap @xmath20 ( i.e. , the difference between the energies of the two lowest levels ) up to @xmath661 for pbc and to @xmath662 for obc , cf . ( [ expansion - delta - wtilde ] ) and eq . ( [ expan - delta - obc ] ) , respectively . in the pbc case , we show that all terms up to @xmath383 are due to the expansion of the nonlinear scaling field @xmath39 associated with @xmath9 and to the leading irrelevant rg perturbation . in the obc case the corrections of order @xmath200 in the expansion of @xmath20 are due to the @xmath1 dependence of the nonlinear scaling field @xmath41 associated with the spatial size @xmath1 ; they can be eliminated by introducing an effective spatial size @xmath663 , cf . ( [ legamma ] ) . instead , the corrections of order @xmath81 show contributions associated with boundary irrelevant rg perturbations of rg dimension @xmath664 . then , we perform an analogous analysis for some rg invariant quantity derived from the two - point function of the order parameter , i.e. @xmath665 , pointing out the presence of further corrections , arising from mixings of the operator @xmath666 with other odd subleading operators . these results for the xy chains are in full agreement with the general rg framework put forward in sec . [ generalfss ] , which generalizes wegner s scaling theory to quantum transitions . finally , we discuss the fss behavior of bipartite entanglement rnyi and vn entropies of one - dimensional systems with an isolated quantum critical point with @xmath6 and central charge @xmath87 . they presents further peculiar corrections to the asymptotic fss behavior predicted by cft , arising from operators associated with conical singularities in the corresponding conformal mapping.@xcite the fss predictions are compared with results for the xy chain . we show that the leading fss corrections for the rnyi entropies with @xmath606 are always of order @xmath586 , for any boundary conditions , see eq . ( [ fsssal ] ) . in particular , in the pbc case corrections are of order @xmath667 only at the critical point @xmath26 . the behavior of the vn entanglement entropy is more complex . in the obc case , the leading correction of order @xmath200 is the sum of terms of different origin : we find contributions from the conical operators and boundary corrections as well . in the pbc case , the expected corrections of order @xmath200 vanish : the leading fss corrections are of order @xmath81 also for @xmath254 . apparently , they are the sum of an analytic contribution and of a term due to the bulk irrelevant rg operator . in our fss study of the entanglement properties we introduce the rg invariant quantity @xmath568 . it is defined in terms of the rnyi entanglement entropy @xmath668 , see eq . ( [ qalpha ] ) , in such a way to have a universal fss behavior ( in particular , it approaches the central charge @xmath87 at the critical point ) . the quantity @xmath568 may be useful to investigate 1d quantum transitions exploiting entanglement properties . _ acknowledgements . we thank pasquale calabrese for useful discussions . the 2d ising universality class can be associated with a cft with central charge @xmath225 . cft provides the asymptotic fss behavior of the two - point function at the critical point.@xcite we report some useful cft formulas for the critical two - point function which are used in the paper . we consider strips @xmath669 with pbc and obc , i.e. with coordinates @xmath670 and @xmath671 . setting @xmath672 , and @xmath673 the critical two - point function on a strip with obc reads @xcite @xmath674^{1/8 } } \times \label{gcftobcis}\\ & & \left [ { |\sin\pi(z_++iy)/2|^{1/2}\over |\sin\pi(z_-+iy)/2|^{1/2 } } - { |\sin\pi(z_-+iy)/2|^{1/2}\over |\sin\pi(z_++iy)/2|^{1/2 } } \right]^{1/2}. \nonumber\end{aligned}\ ] ] the two - point function @xmath675 at the critical point is obtained by setting @xmath676 . this result allows us to exactly compute the universal large-@xmath1 limit of the rg invariant quantities @xmath677 and @xmath210 , defined in eqs . ( [ xidef2 ] ) and ( [ rdef2 ] ) , respectively , at the critical point . we obtain the critical value @xmath678 for the xy chain we may also consider the connected equal - time two - point function of the operator @xmath679 , i.e. @xmath680 for @xmath681 , we obtain @xmath682 note that @xmath683 for @xmath684 . therefore , the integral of @xmath685 with respect to @xmath259 is infinite . in the case of pbc we have @xmath686 again , setting @xmath676 , we obtain the two - point function @xmath675 at the critical point , from which we can compute @xmath687 for @xmath0 defined as in eq . ( [ xidef2 ] ) . if instead the correlation length is defined as @xmath688 where @xmath689 is the fourier transform of @xmath194 , and @xmath690 , we obtain @xmath691 in this appendix we report some exact results for the entanglement entropies of the xy chain at the critical point . for this purpose , we also exploit known results for the xx model , @xmath692 , \label{xxmodel}\end{aligned}\ ] ] and the exact relation @xcite @xmath693 between the entanglement entropies of the xy model ( [ isc ] ) with @xmath281 and @xmath486 , and those of the xx model ( [ xxmodel ] ) . some results for the corrections to the leading behavior within obc were already reported in ref . . using also the results of ref . for the xx model , we can write the large-@xmath1 behavior of the rnyi entropy with @xmath606 at fixed @xmath629 as @xmath694 \nonumber \\ & & \;\ ; - { \gamma[1/2 + 1/(2\alpha ) ] \over 2 \alpha \gamma[3/2 - 1/(2\alpha ) ] } \left [ { \pi \over 8 \gamma l \sin(\pi x)}\right]^{1/\alpha } \nonumber \\ & & \;\;+\,o(l^{-2/\alpha } ) \,+ \ , o(l^{-1 } ) , \label{salpha } \end{aligned}\ ] ] where @xmath695 . \label{egamma}\end{aligned}\ ] ] note that scaling corrections are proportional to @xmath696 , a property which also holds outside the critical point , see sec . [ xychainent ] . ( [ salpha ] ) does not allow us to compute the corrections for the vn entropy . indeed , for @xmath697 , there are two sources of @xmath200 terms : the conical corrections and the analytic boundary corrections . for @xmath486 we have @xcite @xmath698 \nonumber \\ & & \ ; - { \pi \over 16 l \sin(\pi x ) } + o(l^{-2 } ) , \label{salpha1 } \end{aligned}\ ] ] where @xmath631 , and @xmath699 with @xmath700 . thus , after appropriately shifting @xmath626 and @xmath1 to @xmath635 and @xmath474 respectively , the remaining @xmath200 correction term turns out to be equal to the limit @xmath536 of the correction of order @xmath586 appearing in eq . ( [ salpha ] ) . therefore , at the critical point the leading @xmath701 correction in the vn entropy shows both conical and boundary contributions . however , the latter can be reabsorbed be redefining both length scales @xmath1 and @xmath626 . actually , for @xmath486 , replacing @xmath702 and @xmath703 in eq . ( [ salpha ] ) , we also obtain the @xmath701 corrections for general @xmath668 . in the case of pbc , using the results for the xx model reported in refs . , we obtain for @xmath606 and @xmath486 @xmath704 \nonumber \\ & & \ ; - ( \alpha-1){\gamma[1/2 + 1/(2\alpha)]^2\over 4 \alpha^2 \gamma[3/2 - 1/(2\alpha)]^2 } \left [ { \pi \over 4 l \sin(\pi x)}\right]^{2/\alpha } \nonumber \\ & & \ ; + \ ; o(l^{-4/\alpha } ) \;+ \ ; o(l^{-2 } ) , \label{salphap } \end{aligned}\ ] ] where @xmath705 . for the vn entropy we instead obtain @xmath706 \nonumber\\ & & \quad - { \pi^2\over 480 l^2 \sin^2(\pi x ) } + { \pi^2\over 144 l^2 } + o(l^{-4 } ) . \label{salpha1p } \end{aligned}\ ] ] note that the limit @xmath536 of the corrections of order @xmath667 in eq . ( [ salphap ] ) vanishes , hence in the pbc case the leading conical singularities do not contribute at the critical point . j. m. kosterlitz and d. j. thouless , j. phys . c : solid state * 6 * , 1181 ( 1973 ) ; v. l. berezinskii , zh . . fiz . * 59 * , 907 ( 1970 ) [ sov . jetp * 32 * , 493 ( 1971 ) ] ; j. m. kosterlitz , j. phys . c * 7 * , 1046 ( 1974 ) ; j. v. jos , l. p. kadanoff , s. kirkpatrick , and d. r. nelson , phys . b * 16 * , 1217 ( 1977 ) . ref . reports @xmath707 , a result which can not be correct since @xmath39 must be an analytic function of the model parameters . the absolute value is a consequence of an erroneous term @xmath708 appearing in eq . ( 5.3 ) of ref . . the correct result does not have the absolute value , hence @xmath709 . ( [ delta1p - scal ] ) is essentially in agreement with the results of ref . , see his eq . the only difference concerns the linear term , which is reported erroneously as @xmath710 . this is probably due to the fact that the relevant states were incorrectly identified for @xmath264 .
we develop the finite - size scaling ( fss ) theory at quantum transitions , considering generic boundary conditions , such as open and periodic boundary conditions , and also the corrections to the leading fss behaviors . using renormalization - group ( rg ) theory , we generalize wegner s scaling ansatz to the quantum case , classifying the different sources of scaling corrections . we identify nonanalytic corrections due to irrelevant ( bulk and boundary ) rg perturbations and analytic contributions due to regular backgrounds and analytic expansions of the nonlinear scaling fields . to check the general predictions , we consider the quantum xy chain in a transverse field . for this model exact or numerically accurate results can be obtained by exploiting its fermionic quadratic representation . we study the fss of several observables , such as the free energy , the energy differences between low - energy levels , correlation functions of the order parameter , etc . , confirming the general predictions in all cases . moreover , we consider bipartite entanglement entropies , which are characterized by the presence of additional scaling corrections , as predicted by conformal field theory .
along with the improvement of the measurement precision of physical experiments , we need more and more radiative - correction calculations beyond one - loop level . the fermion mass renormalization scheme has been discussed enough at one - loop level @xcite , but is rarely discussed beyond one - loop level . at present there are two mass renormalization schemes : the conventional on - shell scheme @xcite and the recently proposed pole scheme @xcite . people have pointed out that the mass definition of pole scheme is gauge invariant compared with the on - shell scheme @xcite , but this conclusion is only based on a native hypothesis that the location of the complex pole of the propagator is gauge independent at any order in perturbation theory @xcite and a strict proof of the hypothesis hasnt been present . so we need to further investigate this problem . in this paper we will investigate the gauge dependence of fermion mass definition . in order to obtain the fermion mass renormalization scheme beyond one - loop level we firstly investigate the fermion inverse propagator i s^-1(p)=p - m_0+(p ) p(a_l+b_r)+c_l+d_r , where @xmath1 is the bare fermion mass , @xmath2 and @xmath3 are the left- and right- handed helicity operators , and ( p)=p_l^l(p^2)+p_r^r(p^2 ) + m_l^s , l(p^2)+m_r^s , r(p^2 ) . when we renormalize fermion mass , the fermion propagator should be changed into @xcite s(p)==. the on - shell mass renormalization scheme requires the real part of the propagator s denominator equal to zero at the physical mass point , i.e. @xcite re[m^2(1+^l(m^2))(1+^r(m^2))-(m+m - m^s , l(m^2 ) ) ( m+m - m^s , r(m^2))]=0 . solving this equation about @xmath4 to two - loop level we obtain m&=&re[^l(m^2)+^r(m^2)+^s , l(m^2 ) + ^s , r(m^2)]-re^2 [ ^l(m^2)-^r(m^2 ) ] + & + & im^s , l(m^2)im^s , r(m^2 ) -im^l(m^2)im^r(m^2 ) . in eq.(5 ) we have used the fact @xmath5 at one - loop level . different from the on - shell scheme , the pole mass renormalization scheme requires both the real part and the imaginary part of the propagator s denominator equal to zero at the location of the propagator s complex pole @xcite |s(1+^l(|s))(1+^r(|s ) ) -(m_2+m_2-m_2^s , l(|s))(m_2+m_2-m_2^s , r(|s ) ) = 0 , where @xmath6 is the physical mass and @xmath7 is the location of the propagator s complex pole . solving this equation about @xmath8 to two - loop level we obtain m_2&=&re[^l(m_2 ^ 2)+^r(m_2 ^ 2)+^s , l(m_2 ^ 2 ) + ^s , r(m_2 ^ 2)]-re^2 [ ^l(m_2 ^ 2)-^r(m_2 ^ 2 ) ] + & + & im^s , l(m_2 ^ 2)im^s , r(m_2 ^ 2 ) -im^l(m_2 ^ 2)im^r(m_2 ^ 2 ) + & + & m_2 ^ 2im[^l(m_2 ^ 2)+^r(m_2 ^ 2 ) + ^s , l(m_2 ^ 2)+^s , r(m_2 ^ 2 ) ] + im[^l(m_2 ^ 2)+^r(m_2 ^ 2 ) ] , where @xmath9 et al and the one - loop fermion decay width @xmath10 is _ 2=m_2im[^l(m_2 ^ 2)+^r(m_2 ^ 2)+^s , l(m_2 ^ 2 ) + ^s , r(m_2 ^ 2 ) ] . in order to investigate the gauge dependence of the two fermion mass definitions we calculate the gauge dependence of the branch cut of top quark s two - loop mass counterterm in the following section ; in section iii we calculate the cross section of the physical process @xmath0 , i.e. a pair of gauge boson w decaying into a pair of bottom quarks , to investigate the gauge dependence of physical result under the on - shell mass renormalization scheme ; lastly we give the conclusion in section iv . if fermion mass is defined gauge independently , its counterterm must be also gauge independent @xcite . this can be used to test whether the fermion mass definition is gauge independent . in the following calculations we will investigate whether eq.(5 ) and eq.(7 ) are gauge independent . for convenience we only consider w gauge parameter s dependence in the @xmath11 gauge and only introduce physical parameter s counterterms . note that we have used the computer program packages _ feynarts _ and _ feyncalc _ @xcite in the following calculations . we give top quark s mass counterterm as an example to investigate this problem . since nobody has discussed this problem before , we begin the discussion from the one - loop level . the one - loop top quark s @xmath12-dependent self - energy and tadpole diagrams are shown in fig.1 and fig.2 , where @xmath12 is w gauge parameter . we note that in order to investigate the gauge dependence of mass counterterm the tadpole diagrams of fig.2 must be included @xcite . + + we firstly obtain the @xmath12-dependent contribution of fig.1 to top quark s one - loop mass counterterm ( see eqs.(5,7 ) ) re[^l(m_t^2)+^r(m_t^2)+^s , l(m_t^2 ) + ^s , r(m_t^2)]__w=_w 1+b_0(0 , _ w m_w^2 , _ w m_w^2 ) , where the subscript @xmath12 denotes the @xmath12-dependent part of the quantity , @xmath13 and @xmath14 are the masses of top quark and gauge boson w , @xmath15 is electron charge , @xmath16 is the sine of weak mixing angle , and @xmath17 is the passarino - veltman function @xcite . on the other hand , the @xmath12-dependent contribution of fig.2 to top quark s one - loop mass counterterm is re[^l(m_t^2)+^r(m_t^2)+^s , l(m_t^2 ) + ^s , r(m_t^2)]__w=-_w 1+b_0(0 , _ w m_w^2 , _ w m_w^2 ) . from eqs.(9,10 ) one readily see that the top quark s one - loop mass counterterm is gauge independent . now we calculate the gauge dependence of top quark s two - loop mass counterterm . for convenience we only calculate the gauge dependence of the branch cut of top quark s mass counterterm . since the gauge - dependent branch cut of top quark s mass counterterm comprises gauge - dependent heaviside function only ( see below ) and the other part of top quark s mass counterterm does nt contain such function , this simplified treatment is reasonable and will not break our final conclusion . the branch cut of top quark s two - loop mass counterterm will be calculated by the _ cutting rules _ @xcite . the concrete algorithm used here can be found in ref.@xcite . from eqs.(5,7 ) we find at two - loop level the top quark s mass counterterms are different in the two mass renormalization schemes . we firstly discuss the case of on - shell mass renormalization scheme . it is obvious that at two - loop level the second term of the r.h.s . of eq.(5 ) does nt contain branch cut . so we do nt need to consider this term . through the imaginary part of fig.1 we obtain the @xmath12-dependent branch cut of the last term of the r.h.s . of eq.(5 ) & & im^s , l(m_t^2)im^s , r(m_t^2 ) -im^l(m_t^2)im^r(m_t^2 ) _ _ w - cut + & = & -_i = d , s , b|v_3i|^2 a_ic_i_j = d , s , b|v_3j|^2 b_j(x_t-_w+ x_j ) + & + & _ i , j = d , s , b|v_3i|^2 |v_3j|^2 b_ib_j ( x_t-_w)(x_t-_w-3 x_i)+(x_t-_w+x_i)x_j + & & , where the subscript @xmath18 denotes the @xmath12-dependent branch cut of the quantity , @xmath19 and @xmath20 are ckm matrix elements @xcite , @xmath21 and @xmath22 are the masses of down - type @xmath23 and @xmath24 quarks , @xmath25 , @xmath26 and @xmath27 , @xmath28 is the heaviside function , and a_i&=&(x_t^2 - 2(x_i+1)x_t+(x_i-1)^2)^1/2 , + b_i&=&(x_t^2 - 2(_w+x_i)x_t+(_w - x_i)^2)^1/2 , + c_i&=&x_t^2+(1 - 2 x_i)x_t+x_i^2+x_i-2 . in order to investigate the gauge dependence of the branch cut of the first term of the r.h.s . of eq.(5 ) we need to calculate top quark s two - loop self energy . the top quark s two - loop self energy can be classified into two kinds : one contains the one - loop counterterm , the second does nt contain any counterterm . since except for ckm matrix elements all the one - loop physical parameter s counterterms are real numbers and free of branch cut @xcite , the first kind self energy does nt contribute to the branch cut of eq.(5 ) , because except the one - loop counterterm the left part of the self energy is the one - loop self energy which real part is free of branch cut . here we do nt need to worry about the problem of the ckm matrix elements and their counterterms being complex numbers , because the total contribution of them to eq.(5 ) is real number ( see the appendix for example ) . so we only need to consider the second kind self energy . according to the _ cutting rules _ of ref.@xcite , the second kind self energy can also be classified into three kinds : one does nt contain branch cut , the second contains branch cut , but its branch cut does nt contribute to eq.(5 ) , the third contains branch cut and its branch cut contributes to eq.(5 ) @xcite . obviously only the third kind self energy needs to be calculated . the topologies of it are shown in fig.3 @xcite . + one notices in fig.3 that an one - particle - reducible diagram has been included . although a self energy is usually defined to comprise one - particle - irreducible diagrams only , no physical principle demonstrates the one - particle - reducible diagram can not be included in a correct self energy . furthermore in the following calculations of top quark s mass counterterm one will see the one - particle - reducible diagram in fig.3 guarantees the gauge independence of the branch cut of top quark s mass definition under the ple mass renormalization scheme . in fig.3 we also draw the possible cuts of the first four topologies which contribute to eq.(5 ) , where the propagator with an arrow on it denotes it is cut and the arrow represents the _ positive - on - shell _ momentum propagating direction of the cut propagator @xcite ) . the possible cuts of the left two topologies which contribute to eq.(5 ) are shown in fig.4 and fig.5 . + + the concrete calculations of the @xmath12-dependent contribution of the possible cuts of fig.3 to eq.(5 ) are shown in the appendix . here we list the final result & & re[^l(m_t^2)+^r(m_t^2)+^s , l(m_t^2 ) + ^s , r(m_t^2)]__w - cut + & = & _ i = d , s , b|v_3i|^2 a_ic_i_j = d , s , b|v_3j|^2 b_j(x_t-_w- x_j ) + & -&_i , j = d , s , b|v_3i|^2 |v_3j|^2 b_ib_j ( x_t-_w - x_i)(x_t-_w - x_j ) + & & . from eq.(5 ) and eqs.(11,13 ) we obtain the gauge dependence of the branch cut of top quark s two - loop mass counterterm under the on - shell mass renormalization scheme m_t|__w - cut= _ i = d , s , b|v_3i|^2 a_ic_i_j = d , s , b|v_3j|^2 b_j(x_t-_w-3 x_j ) . according to the above discussion above eqs.(11 ) eq.(14 ) manifests top quark s two - loop mass counterterm is gauge dependent under the on - shell mass renormalization prescription . this means the conventional on - shell mass definition is gauge dependent according to the discussion at the begining of this section . then we calculate the gauge dependence of top quark s two - loop mass counterterm under the pole mass renormalization scheme . the result of the first three terms of eq.(7 ) is same as the result of eq.(5 ) . so we only need to calculate the last term of eq.(7 ) . the concrete result is & & m_t^2im[^l(m_t^2)+^r(m_t^2 ) + ^s , l(m_t^2)+^s , r(m_t^2 ) ] + im[^l(m_t^2)+^r(m_t^2 ) ] |__w - cut + & = & -_i = d , s , b|v_3i|^2 a_ic_i_j = d , s , b|v_3j|^2 b_j(x_t-_w-3 x_j ) . from eq.(7 ) and eqs.(11,13,15 ) we readily have m_t|__w - cut=0 . this means in the pole mass definition we do nt find the gauge dependence present in the on - shell mass definition . from the above discussion one readily sees that the conventional on - shell mass renormalization scheme is unreasonable . in fact the on - shell mass renormalization condition of eq.(4 ) does nt exist , because it requires the bare fermion mass @xmath1 is gauge dependent which is disallowed . in this section we will further discuss the effect of the gauge dependence of the on - shell fermion mass definition on physical result . we give the physical process @xmath0 as an example to illustrate this problem . note that we only consider the gauge dependence of the branch cut containing the function @xmath29 $ ] of the two - loop cross section of the physical process . for convenience we hypothesize the higgs mass and the total energy of the incoming particles are less than top quark s mass . this will not break the following conclusion . under the hypothesis only the diagram containing a two - loop @xmath30 needs to be consider . + the contribution of fig.6 to the two - loop amplitude @xmath0 is - @xmath31 is bottom quark s mass , @xmath32 , @xmath33 and @xmath34 are the momentums of @xmath35 , bottom and anti - bottom quarks , and @xmath36 and @xmath37 are the polarization vectors of @xmath38 and @xmath35 . the contribution of eq.(17 ) to two - loop @xmath39@xmath40 is & & -3 x_e(2 x_p-1)(2 x_p+1)^2 - 4 x_p^2(4(x_p-1)x_p+5 ) _ i = u , c , t + & + & -8(x_e-2)x_p^3 + 4(x_e-3)(x_e+2)x_p^2 + 4 x_e^2 x_p-3 x_e(x_e+2 ) , where @xmath41 and @xmath42 ( @xmath43 is the total energy of the incoming particles in the center of mass frame ) . for convenience in eq.(18 ) we have approximated @xmath44 and averaged the result over the incoming particles polarization vectors and summed the result over the outgoing particles helicity states . since there is no other origin of function @xmath29 $ ] in the two - loop @xmath39@xmath40 , eq.(14 ) and eq.(18 ) mean the on - shell mass renormalization scheme makes the cross section of the physical process @xmath0 gauge dependent . from eqs.(14,18 ) we also find that the gauge dependence caused by the on - shell scheme can not be neglected at two - loop level . the concrete numerical result is shown in fig.7 . + in fig.7 we have used the data : @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 and @xmath60 @xcite . contrarily , from eq.(16 ) one readily find that the branch cut containing the function @xmath29 $ ] of the two - loop cross section of the physical process @xmath0 is gauge independent under the pole mass renormalization scheme . we have discussed the gauge dependence of fermion mass definition under the on - shell and pole mass renormalization schemes . through the calculation of top quark s two - loop mass counterterm we strictly prove that the fermion mass definition is gauge dependent under the on - shell mass renormalization scheme at the first time . this leads to the physical result obtained by the on - shell mass renormalization scheme also gauge dependent . on the contrary , we do nt find such gauge dependence in the pole mass renormalization scheme . besides , the gauge - dependent difference of the physical result between the two mass renormalization schemes can not be neglected at two - loop level . therefore we should use the pole mass renormalization scheme rather than the on - shell mass renormalization scheme to calculate physical results beyond one - loop level . in this appendix we list the @xmath12-dependent contributions of the possible cuts of fig.3 to eq.(5 ) . we firstly calculate the @xmath12-dependent contribution of the possible cut of the first topology of fig.3 to eq.(5 ) . using the cutting rules of ref.@xcite we obtain the @xmath12-dependent contribution of the cuts of fig.8 to eq.(5 ) : & & re[^l(m_t^2)+^r(m_t^2)+^s , l(m_t^2 ) + ^s , r(m_t^2)]__w - cut + & = & _ i = d , s , b|v_3i|^2 b_i(x_t-_w - x_i ) b_i x_t(x_t-_w - x_i)-2 a_ic_i + 2_j = d , s , b|v_3j|^2 a_jc_j + & & - _ i , j = d , s , b|v_3i|^2 |v_3j|^2 b_ib_j(x_t-_w - x_i)(x_t-_w - x_j ) + & & . since there is nt four - particle - interaction vertex including fermion in standard model , the second , 3th and 4th topologies of fig.3 can not contribute to eq.(5 ) . so we do nt need to consider their contributions . using the cutting rules we obtain the @xmath12-dependent contribution of the cuts of fig.9 to eq.(5 ) : & & re[^l(m_t^2)+^r(m_t^2)+^s , l(m_t^2 ) + ^s , r(m_t^2)]__w + & = & -(_w^3 - 3_w^2 - 9_w+11 ) _ i = d , s , b|v_3i|^2 a_ic_i + & + & ( _ w-1)^3 _ i = d , s , b|v_3i|^2 b_i + & & x_i^2-(2 x_t+_w)x_i+x_t(x_t-_w ) + & + & d^3 _ i = d , s , b|v_3i|^2 b_i + & & x_i^2-(2 x_t+_w)x_i+x_t(x_t-_w ) , where d= ( _ w-1)^2 c_w^4 - 2(_w+1)c_w^2 + 1 ^1/2 . here we have restricted ourselves to @xmath61 @xcite and used the present experiment result @xmath62 @xcite . we will calculate the contributions of the five cuts of fig.5 one by one . firstly we calculate the @xmath12-dependent contribution of the first cut of fig.5 to eq.(5 ) . the result is & & re[^l(m_t^2)+^r(m_t^2)+^s , l(m_t^2 ) + ^s , r(m_t^2)]__w - cut + & = & _ i = d , s , b|v_3i|^2 ( x_t-_w - x_i ) 2 a_ib_ic_i - x_t(x_t-_w - x_i)b_i^2 . we note that in the calculation of eq.(22 ) we have given the photon an infinitesimal mass @xmath63 in order to regularize the result of the diagrams the 5th propagator of which is photon ( see fig.3 ) . the result of each such diagram contains the term @xmath64 , but the total contribution of them to eq.(5 ) is zero as shown in eq.(22 ) . because we find that none of the diagram in fig.10 satisfies the cutting condition of the second cut of fig.5 , the contribution of the second cut of fig.5 to eq.(5 ) is zero . similarly , the contribution of the 4th cut of fig.5 to eq.(5 ) is also zero duo to the same reason . for the contribution of the third cut of fig.5 to eq.(5 ) we have & & re[^l(m_t^2)+^r(m_t^2)+^s , l(m_t^2 ) + ^s , r(m_t^2)]__w - cut + & = & _ i = d , s , b|v_3i|^2 a_ic_i(_w^3 - 3_w^2 - 9_w+11 ) + & -&(_w-1)^3 _ i = d , s , b x_t^4-(3_w+4 x_i)x_t^3 + & + & 3(_w^2+x_i_w+2 x_i^2)x_t^2-(_w^3 - 2 x_i_w^2 - 3 x_i^2_w+4 x_i^3)x_t -(_w - x_i)^3 x_i + & -&d^3 _ i = d , s , b x_t^4-(3_w+4 x_i)x_t^3 + & + & 3(_w^2+x_i_w+2 x_i^2)x_t^2-(_w^3 - 2 x_i_w^2 - 3 x_i^2_w+4 x_i^3)x_t -(_w - x_i)^3 x_i . for the contribution of the 5th cut of fig.5 to eq.(5 ) we obtain the same result as eq.(23 ) . one can also see this point from the left - and - right symmetry of the 3th and 5th cuts of fig.5 . summing up all of the above results we obtain the @xmath12-dependent contribution of the cuts of fig.3 to eq.(5 ) . the result has been shown in eq.(13 ) . from eqs.(19 - 23 ) one also readily see that the total contribution of ckm matrix elements of each topology of fig.3 to eq.(5 ) is real number . m. veltman , physica * 29 * ( 1963 ) 186 ; + d. bardin and g. passarino , the standard model in the making precision study of the electroweak interactions , oxford science pub . , clarendon press , oxford,1999 ; + d. espriu , j. manzano and p. talavera , phys . d66 ( 2002 ) 076002 . m. passera , a. sirlin , phys . 77 ( 1996 ) 4146 ; + b.a . kniehl , a. sirlin , phys . * 81 * ( 1998 ) 1373 ; + p.a . grassi , b.a . kniehl and a. sirlin , phys . 86 ( 2001 ) 389 ; + m.l . nekrasov , phys . b * 531 * ( 2002 ) 225 ; + b.a . kniehl , a. sirlin , phys . b * 530 * ( 2002 ) 129 . j. kublbeck , m. bohm , a. denner , comput . phys . commun . * 60 * ( 1990 ) 165 ; + g.j . van oldenborgh , j.a.m . vermaseren , z. phys . c*46 * ( 1990 ) 425 ; + t. hahn , m. perez - victoria , comput . * 118 * ( 1999 ) 153 .
we discuss the gauge dependence of fermion mass definition and physical result under the conventional on - shell mass renormalization scheme and the recently proposed pole mass renormalization scheme in standard model . by the two - loop calculations of top quark s mass counterterm and the cross section of physical process @xmath0 we strictly prove that the on - shell mass renormalization scheme makes fermion mass definition and the physical result gauge dependent at the first time , while we do nt find such gauge dependence in the pole mass renormalization scheme . besides , our calculation also shows the difference of the physical result between the two mass renormalization schemes can not be neglected at two - loop level .
the pion form factor , @xmath9 , in the @xmath4 resonance region has been the subject of several recent experiments @xcite which , together , provide the most precise picture yet of the structure of the @xmath4 resonance , including the prominent @xmath0 interference ` shoulder ' . the markedly improved precision of the new data ( e.g. a statistical uncertainty of 1% versus 4 - 5% for earlier studies @xcite ) enables high - precision studies of isospin breaking in the @xmath0 sector and an improved determination of the dominant @xmath10 contribution to the leading order hadronic vacuum polarization term in standard model ( sm ) estimates of the anomalous magnetic moment of the muon , @xmath11 . isospin breaking in the @xmath0 sector is of particular and ongoing importance to studies of charge symmetry breaking in the @xmath8 interaction ( see ref . @xcite and references therein for more details ) . significant unresolved questions include _ i ) _ the precise scale of the @xmath0 mixing matrix element , @xmath12 , in the region of the @xmath4 and _ ii ) _ the scale of the contribution associated with the unavoidable direct ( i.e. non - mixing induced ) isospin - breaking @xmath6 transition , which has often been neglected in determining @xmath13 . previous analyses of these issues ( e.g. in ref . @xcite ) were hindered by data of relatively lower precision , making a precise determination of the phase between the @xmath4 and @xmath14 contributions impossible . without a well - constrained phase , it is not possible to obtain a well - constrained estimate of the direct @xmath6 contribution , and the uncertainty in @xmath15 remains large ( in fact much greater than what is often quoted , for example in ref . the recent wealth of high - precision @xmath16 data in the resonance region represents a good first step in addressing these concerns . in this paper we revisit @xmath0 mixing in @xmath9 and use the recent data to effect a separation of @xmath6 and mixing contributions , from which we determine the ratio of isospin - breaking ( ib ) to isospin - conserving ( ic ) couplings , @xmath17 , and the off - diagonal vector meson self - energy matrix element @xmath15 . we perform this analysis using a selection of models for @xmath18 in order to clarify the numerical significance of the unavoidable model dependence of such a separation . one important reason for the recent interest in @xmath18 is the dominant role it plays in theoretical estimates of the leading order ( lo ) hadronic contribution to @xmath19 , @xmath20^{lo}_{had}$ ] . @xmath21 is now known experimentally to a remarkable 0.5 parts - per - million precision as a result of bnl experiment e821 @xcite . it is well known that @xmath22_{had}^{lo}$ ] which , after the purely leptonic contribution , is the largest of the sm contributions to @xmath21 , can be obtained from a weighted integral of the @xmath23 cross - section @xcite , with the dominant contribution coming from the broad @xmath4 resonance . precise data on the @xmath24 cross - section ( or , equivalently , @xmath18 ) below 1 gev is , therefore , a crucial ingredient in the determination of @xmath22_{had}^{lo}$ ] . use of recent @xmath25 data as input for the @xmath10 part of @xmath22_{had}^{lo}$ ] leads to sm estimates for @xmath19 which are consistent with one another ( even though the form factor measurements themselves do not all agree on the precise @xmath26-dependence of @xmath18 @xcite ) , but which deviate from the experimental value by @xmath27 @xcite . ( it should be noted , however , that preliminary radiative return results from babar differ from previous electroproduction results @xcite . ) as is well known @xcite , in the isospin limit , cvc allows the @xmath28 isovector contribution to @xmath20^{lo}_{had}$ ] to be computed using hadronic @xmath2 decay data @xcite in place of the corresponding isovector electroproduction ( em ) data . at the present level of precision , additional small isospin - breaking corrections to the cvc relation @xcite must be taken explicitly into account . we denote the collection of such corrections by @xmath29_{had}^{lo}$ ] in what follows . incorporating the isovector @xmath2 decay data in this manner , one obtains estimates for the sm contribution to @xmath19 which deviate from the experiment by @xmath30 @xcite . the low - energy region from which the bulk of the sm contribution to @xmath22_{had}^{lo}$ ] is generated is dominated by the @xmath10 component of the em cross - section . the ib corrections in this region , which have been extensively studied in refs . @xcite , are believed to be well understood . however , the component of these corrections associated with @xmath0 mixing , which in general is model dependent , has , to date , usually been estimated from fits to @xmath18 which employ only a single model for @xmath18 ( usually that of refs . this procedure turns out to lead to a significant underestimate of the resulting uncertainty , and hence also of the uncertainty in the @xmath2 decay - based value of the two - pion contribution to @xmath22^{lo}_{had}$ ] . in ref . @xcite several models were used to examine the model - dependence of @xmath31^{lo}_{had}$ ] . it was found that the variation across models was significantly greater than the uncertainty ( associated with errors in the experimental data ) obtained for any individual model , though not to such an extent that it would resolve the @xmath2 vs. @xmath16 discrepancy . it should be borne in mind in this regard that finite energy sum rule studies of the electromagnetic current - current correlator @xcite , based on the then - current em database , yielded values of @xmath32 , which are @xmath33 low compared to recent high - precision lattice determinations @xcite . in contrast , analogous studies of the charged isovector current correlator @xcite , based on the final version of the aleph hadronic @xmath2 decay data @xcite , yield a value of @xmath34 , in excellent agreement with the new lattice results . the preliminary babar results for the @xmath35 cross - sections reported by davier at tau08 @xcite , moreover , are in better agreement with @xmath2 expectations than with previous em results , also favoring the @xmath2 determination , though the @xmath2-em discrepancy has been reduced by the recent re - assessment of ib corrections performed in ref . the question of whether the determination of @xmath20^{lo}_{had}$ ] incorporating @xmath2 data , or that based solely on em cross - sections , is the most reliable thus remains an open one . the goal of achieving as accurate as possible an understanding of the uncertainty on the ib correction , @xmath36^{lo}_{had}$ ] , to be applied to the raw @xmath2-based version of @xmath20^{lo}_{had}$ ] thus remains a highly relevant one . in the present paper , we update the analysis of the @xmath0 mixing contribution to @xmath37^{lo}_{had}$ ] contained in ref . @xcite , taking into account the corrected version of the snd @xmath38 data @xcite , the new cmd-2 @xmath38 results @xcite , and the latest initial state radiation ( isr ) determination of @xmath9 , from kloe @xcite . in addition , we provide details of the fit results associated with the various models of @xmath9 , and new results for the separation of the ib part of the em cross - sections into pure @xmath0 mixing and direct @xmath7 coupling induced contributions . we find that , taking the model dependence of the @xmath0 contribution to @xmath31_{had}^{lo}$ ] into account , the uncertainty assigned to this quantity should be increased by @xmath39 . the improved data also allows for a more reliable separation of @xmath0 mixing and direct non - mixing @xmath6 contributions to @xmath9 . the uncertainty in the scale of the non - mixing contribution , while still large due to weak constraints on the relative phase of the @xmath14 and @xmath4 contributions to the amplitude , is reduced by a factor of two compared to the analysis of ref . @xcite and favors a non - zero value . as a result , the central value of @xmath1 is significantly different from that obtained using analyses in which direct @xmath14 contributions are ignored . the rest of this paper is organized as follows : section [ models ] reviews the different models of @xmath9 considered in our analysis ; section [ data ] discusses the four recent @xmath24 data sets and the results of fits to each set using the models of section [ models ] ; section [ gminus2ib ] updates estimates of the @xmath0 contribution to @xmath31_{had}^{lo}$ ] ; and , finally , section [ mow ] reviews the ( model - dependent ) separation of the direct @xmath6 and mixing contributions to @xmath18 and presents the results of that analysis for the models and data sets of sections [ models ] and [ data ] . many recent estimates of the @xmath0 mixing contribution to @xmath37^{lo}_{had}$ ] , which we denote by @xmath37^{lo}_{had;mix}$ ] in what follows , have often been based on a model of @xmath18 which supplements the chpt - based formulation of guerrero and pich ( gp ) @xcite with a @xmath0 mixing contribution as proposed in refs . @xcite ( cen ) . we refer to this model as the gp / cen model . however , @xmath18 is well - described by many other models in the literature . various analyses have used the khn - santamaria ( ks ) model @xcite , the hidden local symmetry ( hls ) model @xcite , and the gounaris - sakurai ( gs ) model @xcite . each model differs in the manner in which the dominant @xmath4 contribution to @xmath9 is represented , with the gs and gp / cen models explicitly taking account of strong final state interactions in @xmath24 . the use of models for @xmath18 is unavoidable at present if one wishes to separate the ic and ib contributions to the em cross - sections in the interference region , this separation being necessary for an estimate of @xmath37^{lo}_{had;mix}$ ] . the ks model @xcite combines contributions from the @xmath4 , @xmath40 , and @xmath41 resonances in such a way as to satisfy the charge constraint @xmath42 . an @xmath14 contribution of the nominal @xmath0 mixing form is added to account for the narrow interference shoulder . the resulting expression for @xmath9 is @xcite @xmath43 where the vector meson propagator is given by @xmath44 and @xmath45 , @xmath46 and @xmath47 are complex - valued constants . unless otherwise stated , @xmath48 is the standard @xmath26-dependent width implied by @xmath49-wave phase space for the decay of the vector meson @xmath50 , explicitly for the @xmath4 , @xmath51 where @xmath52 , and @xmath53 with @xmath54 approximated by a constant . an alternate version of the ks model , referred to in what follows as ks@xmath55 , is also sometimes employed in the literature . the ks@xmath55 form for @xmath9 is @xmath56 it is this alternate version which is employed in ref . because of strong cancellations in the region of the @xmath57 interference shoulder , the ks and ks@xmath55 models produce significantly different results for @xmath37^{lo}_{had;mix}$ ] . we return to this point below . the hls model @xcite of @xmath9 has the form @xmath58 with @xmath59 a constant associated with a non - resonant @xmath60 coupling . despite having no explicit contribution from higher resonances such as the @xmath40 , the model provides a good quality fit to the data below @xmath61 gev . it also turns out to produce the correct @xmath9 phases in the elastic @xmath10 scattering region after the model parameters have been fitted @xcite , though the phase constraint is not imposed in the basic structure of the model . the gs model @xcite , as adopted by cmd-2 @xcite and used here , is given by @xmath62 + \beta { \rm bw}_{\rho^\prime}^{(gs)}(s)\right ) \label{gsmodel}\ ] ] where @xmath63 with @xmath64 + ( m_v^2-s)p_{\pi}^2(m_v^2){\frac{dh}{ds}}(m_v^2)\right ) \nonumber\\ h(s ) & = & { \frac{2 p_{\pi}(s)}{\pi\sqrt{s}}}\ , \ell n\left ( { \frac{\sqrt{s } + 2 p_{\pi}(s)}{2 m_{\pi}}}\right ) \end{aligned}\ ] ] where @xmath65 is the pion cm momentum for squared invariant mass @xmath26 and @xmath66 . the elastic @xmath10 scattering phase constraint on @xmath9 is reproduced by the gs model by explicit construction . for the ks , hls , and gs models , the constant @xmath45 , which parameterizes the strength of the narrow ib amplitude , is taken to be complex since a non - zero phase is , in general , unavoidable in the presence of an ib direct ( non - mixing ) @xmath7 contribution @xcite . resonance chiral effective theory provides the basis for the gp model of @xmath9 , which is given by @xcite @xmath67\right)\ , \label{gpmodel}\ ] ] where @xmath68\ ] ] with @xmath69 and the @xmath26-dependent width , @xmath70 appearing in @xmath71 ( eq . [ propagator ] ) replaced by the resonance chiral effective theory expression @xmath72 because the gp model implements the constraints of chiral symmetry explicitly at next - to - leading order in the chiral expansion , it also ensures the correct phase for @xmath9 in the elastic region to this same order . in ref . @xcite , a small rescaling of the coefficient appearing on the rhs of eq . ( [ gpwidth ] ) is allowed in order to account for the @xmath73 mev contribution of @xmath74 decays to the total width of the @xmath4 @xcite . we implement such a rescaling by multiplying the rhs of eq . ( [ gpwidth ] ) by @xmath75 with @xmath76 a fit parameter . some ib effects are incorporated into @xmath77 if one evaluates the phase space factors in the @xmath26-dependent width using the physical charged @xmath78 and @xmath79 masses . the cen modification of @xmath77 , designed to incorporate the @xmath0 mixing contribution not included in the original gp model , has the form @xmath80 the parameter @xmath81 was assumed real in ref . @xcite . as discussed in ref . @xcite , the original version of the gp / cen model does not provide a good fit to the corrected version of the earlier cmd-2 data @xcite . as already noted , the fact that the @xmath0 `` mixing '' signal is actually a combination of mixing and direct @xmath7 effects means that an effective representation of this combination using the form given by eq . ( [ gpcenmodel ] ) is not generally possible without allowing @xmath81 to have a non - zero phase @xcite . in ref . @xcite the gp / cen model was extended in this way , treating the phase of @xmath82 as a fourth parameter to be fit to the data , and an acceptable fit became possible . we refer to this version of the gp / cen model as gp / cen@xmath83 below . in most of what follows , however , we will focus instead on an alternate version of the gp / cen model , referred to below as gp / cen@xmath84 , which incorporates the effects of strong @xmath10 rescattering not only in the isospin - conserving @xmath85 contribution ( the first term on the rhs of eq . [ gpcenmodel ] ) , but also in the @xmath4 propagator factor of the isospin - breaking @xmath57 mixing contribution , leading to @xmath86 where , as in the gp / cen@xmath83 version , @xmath81 is allowed to have a non - zero phase . recent information on @xmath9 in the interference region has been provided by the cmd-2 and snd experiments at novosibirsk and kloe at dafne . results are available from a total of five separate data - taking runs . preliminary isr results from babar have also been reported , but are not yet publicly available . cmd-2 has reported data on @xmath38 in the centre of mass energy range from @xmath87 to @xmath88 gev , taken in two separate runs , one in 1994 - 95 @xcite and one in 1998 @xcite . we refer to these as cmd-2(94 ) and cmd-2(98 ) , respectively . the 1998 run has five times the luminosity of the earlier run , and correspondingly smaller statistical errors , but a less precise beam energy calibration . whereas the beam energy was measured very precisely using resonant depolarization for cmd-2(94 ) , this technique was not available for cmd-2(98 ) @xcite with the result that the beam energy for the latter is only known to a few parts in @xmath89 @xcite . a significant part of the beam energy uncertainty , moreover , is fully correlated @xcite . the combination of increased statistics but poorer beam energy calibration resulted in no overall improvement in the determination of the @xmath10 contribution to @xmath22_{had}^{lo}$ ] from cmd-2(98 ) as compared to cmd-2(94)@xcite . indeed , the beam energy contribution to the systematic error in @xmath22_{had}^{lo}$ ] increased from 0.1% in cmd-2(94 ) to 0.3% in cmd-2(98)@xcite . the less precise energy calibration of cmd-2(98 ) also has an impact on studies of isospin breaking in @xmath18 because such studies require a precise experimental characterization of the interference shoulder . snd results for the @xmath38 cross - sections are based on data taken from 1996 - 2000 with centre of mass energies in the range @xmath90 to @xmath91 gev @xcite . the snd data carries a slightly larger systematic error than cmd-2(94 ) , but much smaller statistical uncertainty and slightly better beam energy calibration . the cmd-2 and snd data sets are in good overall agreement . the most recent , kloe , measurements of the @xmath38 cross - section are obtained using the isr technique @xcite . it is well known that the earlier , 2004 , kloe results for @xmath9 @xcite , based on data collected in 2001 , deviated in shape from both the cmd-2 and snd results , lying systematically higher than either below @xmath92 mev , and systematically lower beyond this @xcite . we refer to these results as kloe(01 ) below . a new analysis has recently been completed using data collected in 2002 and an improved analysis strategy @xcite . the new data represents twice as many events as before and corresponds to an integrated luminosity that is 1.7 times greater , and the analysis yields significantly reduced statistical and systematic uncertainties , with the systematic uncertainties now dominant . in addition , part of the previous shape discrepancy has been resolved , kloe now being in better agreement with cmd-2(98 ) below the @xmath4 peak . a reduced disagreement persists above the @xmath4 peak . in view of the fact that the results of the new analysis supercede those of kloe(01 ) @xcite , our study of ib in the interference region , reported below , is based on the former , which we refer to for clarity as kloe(02 ) in what follows . our earlier analysis of @xmath31^{lo}_{had;mix}$ ] , reported in ref . @xcite , was based on the cmd-2(94 ) and uncorrected snd data sets . the kloe(01 ) results were also used , though , without full covariance information , the resulting fits were unreliable . the fits presented below reflect the new cmd-2(98 ) data , the corrected snd data , and the new kloe(02 ) data , including full covariance information , where available . we have performed fits to each data set using the models described previously in section [ models ] . in all cases the results shown below correspond to the bare form factor , that is the form factor with the effects of vacuum polarization removed @xcite . where applicable , the following input values have been used : @xmath93 mev , @xmath94 mev , @xmath95 mev @xcite , @xmath96 mev ( 310 mev for the gs model @xcite ) , @xmath97 mev , @xmath98 mev , and @xmath99 mev . although a contribution from the @xmath41 appears in the ks model , we have found that , because the data we consider is limited to @xmath100 below 1 gev , it provides insufficient constraints on this term . we manually set the @xmath41 contribution to zero in the ks model to avoid it being used by the minimization algorithm to ` fine tune ' the @xmath4 region of the fit without regard for the consequences at higher @xmath100 . for similar reasons the coefficient of the @xmath40 contribution , @xmath46 , is taken to be real . fit parameters and their covariances were obtained using minuit . the results for each model and for each data set are shown in tables [ table1 ] , [ table2 ] , [ table3 ] , and [ table4 ] . a blank entry indicates that a fit parameter is inapplicable to that particular model . for the gp / cen@xmath83 and gp / cen@xmath84 models , the effective value of @xmath101 is shown in brackets to highlight that it is in fact @xmath76 which is the fit parameter . note also that for both of these models the quantity @xmath102 is replaced by @xmath103 in the tables to facilitate comparison with the other models . .1 in .[table1 ] results of fits to the cmd-2(94 ) data . [ cols="^,^,^,^,^,^",options="header " , ] as figures [ cmd294mow][kloemow ] show , it is possible to constrain @xmath104 rather well . with the prescription above for handling averages and uncertainties , we obtain @xmath105 the consistency across different models for each data set is actually quite good . the determination of @xmath106 is less precise , with the prescription above for determining the average across models and model - dependence uncertainty yielding @xmath107 for comparison , the uncorrected snd analysis @xcite , using a model very similar in form to the ks model described in section [ models ] , reported @xmath108 , consistent with our result , @xmath109 , for the snd+ks case . the present analysis favors a non - vanishing @xmath106 , as expected on general grounds , with a vanishing @xmath106 lying @xmath110 away from our central value . an improved determination of the orsay phase , particularly in the case of the kloe data , is required to make further progress on this issue . to illustrate this point , and also because of the lingering kloe shape discrepancy @xcite , if one omits the kloe data from the experimental averages one obtains results with a slightly reduced model - dependence uncertainty , @xmath111 and @xmath112 the values for @xmath106 and @xmath104 obtained in tables [ table5][table8 ] imply , using the earlier definitions , and our prescription for combining the results corresponding to different models , the result @xmath113 for the real part of @xmath15 . the imaginary part of @xmath15 is , similarly , found to be @xmath114 . the result is an estimate of the off - diagonal part of the physical @xmath0 self - energy matrix _ at the @xmath4 mass _ @xmath115 where we have combined the model - dependence and experimental errors _ modestly reduced uncertainties are again obtained if the kloe(02 ) data is excluded in forming the combined result . explicitly : @xmath116 this quantity is not to be confused with the sometimes quoted ` effective ' @xmath0 mixing matrix element , which contains direct @xmath7 contributions . the separation described here is possible only because the two contributions enter the form factor with different ( but known ) phases , as shown by eq . [ ff2 ] . experimental determination of the combined phase ( the orsay phase ) then allows the magnitude of each contribution to be determined separately . any improvement in the separation of the @xmath0 mixing and direct @xmath7 ib contributions thus requires corresponding improvements in the determination of the orsay phase . recent experiments provide a much - improved characterization of the prominent @xmath0 interference shoulder in the @xmath38 cross - section , making precision analyses of isospin breaking in the @xmath0 system possible . such an analysis is inherently dependent on the model chosen to represent the broad @xmath4 resonance contribution . by studying a range of models , all of which provide good fits to the data , we have shown that the model sensitivity of most fit parameters is generally greater than the statistical uncertainty associated with any given model . the only exception is the phase , @xmath117 , of the parameter describing the strength of the @xmath14 contribution , for which the statistical uncertainty tends to be greater than the model sensitivity . any model sensitivity of course propagates to all quantities obtained from the fit parameters . the example discussed in section [ gminus2ib ] is that of the @xmath0 interference contribution to the ib correction , @xmath31_{had}^{lo}$ ] , required for incorporating hadronic @xmath2 decay data into the determination of the sm expectation for @xmath22_{had}^{lo}$ ] . using our result , eq . [ cvall ] , the full ib correction obtained in ref . @xcite using the method of ref . @xcite ( see also @xcite ) , @xmath31^{lo}_{had } = ( -13.8 \pm 2.4)\times 10^{-10}$ ] ( or @xmath31^{lo}_{had } = ( -16.6 \pm 1.6)\times 10^{-10}$ ] based on the recent analysis of ref . @xcite ) , receives only a small shift of @xmath118 ( @xmath119 ) in its central value . the uncertainty in the size of the correction , however , increases by a much more significant @xmath39 . in arriving at the combined error , we have added the component associated with the model - dependence of the @xmath0 mixing contribution _ linearly _ to the other errors , for the reason discussed above . while the increased error does not serve to resolve the @xmath120 @xcite ( @xmath121 @xcite ) @xmath2 vs. em discrepancy which existed prior to the preliminary babar isr results , it is nonetheless important to take into account in assessing the uncertainty in the ib corrections which must be applied if one wishes to use @xmath2 decay data in evaluating the sm contribution to @xmath22^{lo}_{had}$ ] . the precision of the recent electroproduction data also allows new improved results for the ( model - dependent ) separation of contributions to @xmath18 from @xmath0 mixing and direct ( non - mixing ) ib @xmath6 transitions . we estimate the ratio of the two - pion couplings of the isospin - pure @xmath14 and @xmath4 to be @xmath122 where the uncertainty is conservatively chosen to encompass the experimental error intervals of all of the models considered . further improvements in the determination of @xmath106 may be possible in future . as can be seen in figures [ cmd294mow][kloemow ] , the off - diagonal part of the vector meson self - energy matrix evaluated at the @xmath4 mass is very well constrained by the analysis . the central value we obtain for the real part is given in eq . [ repiroresult ] , where the uncertainty again conservatively encompasses the full ranges corresponding to all models and all data sets . our central value of @xmath123 which is free of contamination from direct @xmath7 coupling contributions , and thus differs from the usually quoted ` effective ' @xmath0 mixing matrix element obtained by ignoring such a direct coupling , should be used in place of the ` effective ' matrix element results in meson - exchange models of ib nn scattering . further improvement of the separation of direct and mixing contributions requires a reduction in the uncertainty of the orsay phase , which in turn requires a further increase in the precision of the @xmath24 data in the @xmath0 interference region . this requires not only increased statistics but also very precise beam energy calibration such as that expected from vepp-2000 . .2 in _ note added : subsequent to the posting of this paper , the babar collaboration posted its results for the @xmath38 cross - sections , obtained using the isr method @xcite . we will report the results of the analysis discussed here , applied to that data set , as soon as the data becomes publicly available . _ we thank g. venanzoni for clarification on some of the kloe systematic errors and z. zhang , c. yuan , p. wang and x. mo for extensive interchanges on the relation between our results and those of ref . cw thanks s. eidelman and i. logashenko for valuable discussions about the cmd-2(98 ) data . km would like to acknowledge the hospitality of the cssm , university of adelaide , and the ongoing support of the natural sciences and engineering research council of canada . et al . _ ( the cmd-2 collaboration ) , phys . b * 527 * , 161 ( 2002 ) ; phys . b * 578 * , 285 ( 2004 ) . r. r. akhmetshin _ et al . _ [ cmd-2 collaboration ] , phys . b * 648 * , 28 ( 2007 ) . a. aloisio _ et al . _ ( the kloe collaboration ) , phys . b * 606 * , 12 ( 2005 ) . f. ambrosino _ [ kloe collaboration ] , phys . b * 670 * , 285 ( 2009 ) . m. n. achasov _ et al . _ , j. exp . * 101 * , 1053 ( 2005 ) [ zh . * 101 * , 1201 ( 2005 ) ] . m. n. achasov _ et al . _ , j. exp . phys . * 103 * , 380 ( 2006 ) [ zh . * 130 * , 437 ( 2006 ) ] . l. m. barkov _ et al . _ , nucl . b * 256 * , 365 ( 1985 ) . v. e. balakin _ et al . _ , phys . b * 41 * , 205 ( 1972 ) . d. benaksas _ et al . _ , phys . b * 39 * , 289 ( 1972 ) . g. barbiellini _ et al . _ , lett . nuovo cim . * 6s2 * , 557 ( 1973 ) [ lett . nuovo cim . * 6 * , 557 ( 1973 ) ] . g.a . miller , a.k . opper and e.j . stephenson , ann . sci . * 56 * , 253 ( 2006 ) . k. maltman , h.b . oconnell , and a.g . williams , phys . lett . b * 376 * , 19 ( 1996 ) . et al . _ ( the muon @xmath124 collaboration ) , phys . . lett . * 92 * , 161802 ( 2004 ) . m. gourdin and e. de rafael , nucl . b * 10 * , 667 ( 1969 ) . while estimates of @xmath19 based on the new data are consistent with one another , a discrepancy still exists between the kloe @xcite results for @xmath18 and those obtained by cmd-2 @xcite and snd @xcite in the region above the @xmath4 . the fact that this discrepancy is not yet understood @xcite strikes a cautionary note . m. passera , j. phys . g * 31 * , r75 ( 2005 ) . m. davier , nucl . * 169 * , 288 ( 2007 ) . m. passera , nucl . suppl . * 169 * , 213 ( 2007 ) . k. hagiwara , a. d. martin , d. nomura and t. teubner , phys . b * 649 * , 173 ( 2007 ) . f. jegerlehner and a. nyffeler , phys . rep . * 477 * , 1 ( 2009 ) . m. davier _ et . arxiv:0906.5443 [ hep - ph ] . m. davier ( babar coll . ) , nucl . suppl . * 189 * , 222 ( 2009 ) for the powerpoint slides of this talk , follow the link at /tau08.inp.nsk.su / prog.php . r. alemany , m. davier and a. hocker , eur . j. c * 2 * , 123 ( 1998 ) . m. davier and a. hocker , phys . b * 419 * , 419 ( 1998 ) ; phys . b * 435 * , 427 ( 1998 ) . g. colangelo , nucl . proc . suppl . * 131 * , 185 ( 2004 ) . j.f . de troconiz and f.j . yndurain , phys . d * 71 * , 073008 ( 2005 ) . a. hocker , arxiv : hep - ph/0410081 r. barate _ et al . _ ( the aleph collaboration ) , z. phys . c * 76 * , 15 ( 1997 ) . k. anderson _ _ ( the opal collaboration ) , eur . j. c * 7 * , 571 ( 1999 ) . s. anderson _ et al . _ ( the cleo collaboration ) , phys . d * 61 * , 112002 ( 2000 ) . s. schael _ et al . _ [ aleph collaboration ] , phys . rept . * 421 * , 191 ( 2005 ) . m. davier , a. hocker and z. zhang , rev . * 78 * , 1043 ( 2006 ) . m. fujikawa _ et al . _ [ belle collaboration ] , phys . d * 78 * , 072006 ( 2008 ) . m. davier , s. eidelman , a. hocker and z. zhang , eur . j. c * 27 * , 497 ( 2003 ) . m. davier , s. eidelman , a. hocker and z. zhang , eur . j. c * 31 * , 503 ( 2003 ) . v. cirigliano , g. ecker and h. neufeld , phys . b * 513 * , 361 ( 2001 ) . v. cirigliano , g. ecker and h. neufeld , jhep * 0208 * , 002 ( 2002 ) . s. ghozzi and f. jegerlehner , phys . b * 583 * , 222 ( 2004 ) . f. flores - baez , a. flores - tlalpa , g. lopez castro and g. toledo sanchez , phys . d * 74 * , 071301 ( 2006 ) . a. flores - tlalpa , f. flores - baez , g. lopez castro and g. toledo sanchez , nucl . * 169 * , 250 ( 2007 ) . k. maltman and c. e. wolfe , phys . d * 73 * , 013004 ( 2006 ) . k. maltman , phys . b * 633 * , 512 ( 2006 ) . et al . _ ( the hpqcd collaboration ) , phys . d * 78 * , 114507 ( 2008 ) ; k. maltman , d. leinweber , p. moran and a. sternbeck , phys . rev . d*78 * , 114504 ( 2008 ) . k. maltman and t. yavin , phys . d * 78 * , 094020 ( 2008 ) . f. guerrero and a. pich , phys . b * 412 * , 382 ( 1997 ) . kuhn and a. santamaria , z. phys . c * 48 * , 445 ( 1990 ) . * 54 * , 1215 ( 1985 ) . gounaris and j.j . sakurai , phys . * 21 * , 244 ( 1968 ) . m. benayoun _ et al . _ , j. c * 2 * , 269 ( 1998 ) . we thank vincenzo cirigliano for clarifying this point . s. eidelman , private communication . see ref . @xcite for further details on the vacuum polarization corrections necessary to convert the cross - sections reported by various collaborations in the literature to the corresponding `` bare '' cross - section values . davier @xcite , working with the gs model , has noted that the @xmath125 and @xmath126 values obtained from fits to the bare and vacuum - polarization - dressed versions of the cmd-2 cross - sections differ by @xmath127 mev . similar shifts also occur for the other models . these shifts produce corresponding shifts in the @xmath0 interference contribution to @xmath37_{had}^{lo}$ ] . the versions associated with the fits to the bare cross - sections , quoted in the text , lie higher than those associated with the fit to the vacuum - polarization - dressed values by @xmath128 , depending on the model . the relevant versions for comparison to the @xmath2 data are , as explained by davier , those obtained from the fit to the bare cross - sections @xcite . these are also the versions relevant to the determination of @xmath3^{lo}_{had}$ ] . m. davier , nucl . . suppl . * 131 * , 123 ( 2004 ) . use of the recently posted belle values for the mass and width of the @xmath40 @xcite results in no significant difference in fit quality or in the values of the fit parameters . the older pdg average value for the width of the @xmath40 is used with the gs model to ease comparison with the cmd-2 analyses . the fits are largely insensitive to the precise value of the @xmath129 width . when quoting the earlier central value and uncertainty for @xmath130^{lo}_{had;mix}$ ] , we use the result obtained in ref . @xcite , based on the same formalism as elaborated in ref . @xcite , but with improved experimental inputs . coon and r.c . barrett , phys . c * 36 * , 2189 ( 1987 ) .
we study the implications of several recent high - precision measurements of the pion form factor in the region of the @xmath0 interference `` shoulder '' for ( i ) the extraction of the @xmath0 mixing matrix element , @xmath1 , and ( ii ) the evaluation of the isospin - breaking ( ib ) correction needed to incorporate hadronic @xmath2 decay data into the determination of the standard model expectation for the leading order hadronic contribution , @xmath3_{had}^{lo}$ ] , to the anomalous magnetic moment of the muon , focussing , in the latter case , on the model - dependence of the @xmath0 mixing component of the ib correction . we consider a range of different models for the broad @xmath4 contribution to the @xmath5 amplitude , applying these models to each experimental data set , and find that the model dependence of the @xmath0 mixing correction is significantly larger than the uncertainty induced by experimental errors for any individual model . we also find that , for each such model , the recent data allows one to separate @xmath0 mixing and direct @xmath6 coupling contributions to the amplitude , and hence to obtain a reasonably precise extraction of @xmath1 , uncontaminated by direct @xmath7 coupling effects , for use in meson exchange model calculations of charge symmetry breaking in @xmath8 scattering . .
protoplanetary discs are a natural outcome of low mass star formation , providing a reservoir of material from which the star itself continues to accrete and from which planets may later form . the accretion history of a newly formed star , the evolution of its disc and the formation of a planetary system are all intimately linked and affected by irradiation from the central star . a lot of attention has recently been paid to the final dispersal of protoplanetary discs as this sets the time - scale over which planets may form . observationally , the study of the dust content of these discs through the analysis of their continuum spectral energy distributions ( seds ) , in the infra - red ( ir ) , has enormously advanced over the past few decades . these observations have indicated discs evolve in a way that suggest ` standard ' viscous evolution ( hartmann et al . 1998 ) for the first few myrs ( e.g. haisch et al . 2001b ) , but then rapidly evolve from a disc - bearing ( primordial ) to disc - less status ( e.g. luhman et al . 2010 ; muzerolle et al . this rapid clearing of the inner disc indicated from ir observations has been supplemented with complementary observations in the sub - millimetre that show most non - accreting stars ( wtts ) are devoid of emission out to @xmath11 au ( duvert et al . this suggests the removal of material material close to the star ( @xmath12 au ) is correlated with the removal of material at larger radius , ( @xmath13 au ) ; and is hence , associated with the entire destruction of the protoplanetary disc . furthermore , there have been several observations of discs thought to be caught in the act of clearing ( e.g calvet et al . 2002 ; najita et al . 2007 ) ; these ` transition ' discs show a deficit in emission at near - ir ( nir ) wavelengths compared to a primordial optically thick disc ( one that is optically thick all the way to the dust destruction radius ) , while returning to the emission levels expected from an optically thick disc at longer wavelengths . currently , there is no clear consensus as to what constitutes a transition disc and both accreting ( e.g. hughes et al . 2009 ; espaillat et al . 2010 ) and non - accreting ( e.g. cieza et al . 2010 ; mern et al . 2010 ) objects have been discovered . the frequency of these inner - hole sources is approximately 10 to 20% of the total disc population for young solar - type stars in nearby star forming regions ( e.g strom et al . 1989 ; skrutskie et al . 1990 ; luhman et al . if one assumes that these inner hole objects are indeed ` transition ' discs ( i.e. discs that are caught in the act of dispersal ) and that the gas and dust evolution go hand - in - hand , then a 10% frequency of transition discs suggest a dispersal mechanism that operates over a tenth of the optically thick disc lifetime ( kenyon & hartmann 1995 ) . moreover the colours of such discs imply that they clear from the inside out ( ercolano , clarke & hall , 2010 ) . the observations described above have encouraged theorists to propose dispersal mechanism that operate over two separate time - scales : a first time - scale of a few million years which allows discs to remain optically thick at ir wavelengths , followed by a brief dispersal phase that leaves the discs optically thin with colours consistent with those of bare photospheres or debris discs . the proposed mechanisms include planet formation itself ( e.g. armitage & hansen 1999 ) , grain growth ( dullemond & dominik 2005 ) , photophoresis ( krauss et al 2007 ) , mri - driven winds ( suzuki & inutsuka 2009 ) , and photoevaporation due to extreme ultraviolet radiation -euv- ( clarke et al 2001 ; alexander et al 2006a , b ; richling & yorke 2000 ) , x - ray radiation ( alexander et al . 2004 ; ercolano et al 2008b , 2009 ; gorti & hollenbach 2009 , owen et al . 2010 ) and far ultraviolet radiation -fuv- ( gorti & hollenbach 2009 ; gorti et al . however there is still no consensus over which mechanism ( or which combination ) may dominate . the recent developments of x - ray photoevaporation ( xpe ) models for t - tauri discs have yielded encouraging results , suggesting that this may be the dominant player in disc dispersal . in particular , we have shown that x - ray photoevaporation rates exceed the euv photoevaporation rates by two orders of magnitude ( ercolano et al 2009 ) and can easily reproduce the two time - scale behaviour suggested by the observations ( owen et al 2010 ) . this model does not suffer from the uncertainties in the heating rates that plague the fuv model , due to the unknown abundance of pahs in discs and the very uncertain photoelectric yields ( see discussion in ercolano & owen 2010 ) . furthermore , ercolano & clarke ( 2010 ) discussed the role of metallicity in disc dispersal , and found that photoevaporation predicts shorter disc lifetimes at lower metallicity , in contrast to planet formation which would predict a very strong negative correlation . yasui et al ( 2009 ) have presented observation that favour a positive correlation between disc lifetimes and metallicity and hence which favour xpe ; however , it is too early to draw any definitive conclusions about the role metallicity plays in disc dispersal . finally , the slightly blue - shifted forbidden line spectrum of discs seen almost face on ( e.g. hartigan et al 1995 ; pascucci et al 2009 ) is well reproduced by the xpe model for all of the low ionisation and atomic diagnostics considered ( ercolano & owen , 2010 ) , an improvement over the euv photoevaporation model which can not reproduce the oi 6300 line intensities ( e.g. font et al 2004 ) . more recent observations have focused on understanding the nature of some known transition discs ( tds ) by investigating their gas content and accretion properties . evidence for accretion and gas inside the dust inner hole has sometimes been used to argue against photoevaporation being the agent of inner - hole clearing ( e.g. espaillat et al 2010 ) . this is however a misconception , perhaps based on earlier euv - based photoevaporation models ( alexander et al . 2006b ) . in this work we show that the xpe model of owen et al ( 2010 ) clearly predicts a transition phase where there is a detectable quantity of accreting gas inside the dust inner hole and is indeed consistent with the accretion versus inner hole size properties of most observed tds . in this paper we use radiation - hydrodynamics simulations to develop a x - ray photoevaporation model which is then used to construct a synthetic disc population to compare with available observed disc statistics . we discuss the x - ray luminosities of tts in section 2 , while section 3 provides a description of the x - ray photoevaporation model . in section 4 we discuss a viscously evolving photoevaporating disc model and the construction of our disc population . in section 5 we present the results from our population synthesis model while our final conclusions are given in section 6 . the x - ray luminosity ( @xmath14 ) , is a crucial input for the x - ray photoevaporation model , where we take the x - ray luminosity to span the range @xmath15kev . as discussed in section 3 , the choice of @xmath14 fully describes the mass - loss rate due to photoevaporation in a given system . the input x - ray spectrum is identical to the spectrum used in previous work ( ercolano et al . 2008b ; ercolano et al . 2009 ; owen et al . 2010 ; ercolano & owen 2010 ) . it is a synthetic spectrum generated by the plasma code of kashyap & drake ( 2000 ) and fits to _ chandra _ spectra of t tauri stars ( e.g. maggio et al . it includes an extreme ultraviolet component - euv - ( 13.6 - 100 ev ) , soft x - ray component ( 0.1 - 1 kev ) and a hard x - ray component ( @xmath16 kev ) . ercolano et al . ( 2009 ) studied the effect of attenuating this x - ray spectrum with neutral columns . they found that at columns of order @xmath17 @xmath18 the soft x - ray component was screened out and the photoevaporation rates dropped significantly . however , for columns @xmath19 @xmath18 the photoevaporation rates remained unaffected . furthermore , ercolano & owen ( 2010 ) showed that the x - ray wind itself is optically thick to euv photons : thus when considering photoevaporation it is only the strength of the soft x - rays that is relevant . we assume that the soft x - rays are able to reach the disc surface as is consistent with the explanation of the oi 6300 line emission presented in ercolano & owen ( 2010 ) . therefore , we adopt the unattenuated spectrum shown in owen et al . ( 2010 ) and expect our results to generalise to cases of moderate screening . we will discuss the implications of this assumption further in section 5 . it has been known for some time that there is a large scatter in the x - ray luminosity of t - tauri stars for a given stellar mass ( and bolometric luminosity ) . therefore , we have used the data for the taurus cluster ( gdel et al . 2007 ) and the orion nebular cluster ( preibisch et al . 2005 ) to build cumulative x - ray luminosity functions for the orion sample functions ( xlfs ) for all pre - main sequence stars ( including both ctts and wtts ) in the 0.5 to 1 m@xmath5 range . the data for these two regions shows good agreement at low luminosity , but they differ at high luminosities , in the sense that the orion data contains a higher fraction of sources with high @xmath14 . the difference is most probably _ not _ intrinsic to the x - ray properties of these two regions , but due to the different treatment of strong flares in the two samples . strong flares were excluded in the taurus data , in contrast to the orion sample where luminosities were averaged over the whole observational period . strong flares are relatively rare and albacete colombo et al . ( 2007 ) found that non - flaring sources have a median @xmath20kev , compared to flaring sources with @xmath21kev . therefore , due to their higher x - ray temperature flares tend to emit most of their radiation in the hard x - ray region ( as shown by observations of objects in a ` flare ' state compared to in a ` quiescent ' state , e.g. imanishi et al . 2001 ) , where the thermal impact is low , as discussed above . for this reason , the taurus sample is the most appropriate for use in the photoevaporation model , since it should better approximate the quiescent and therefore softer x - ray luminosity , which as discussed above is the relevant input for calculations of the photoevaporation rate . therefore , we adopt the taurus xlf for the remainder of this work . we assume that the x - ray luminosity function remains invariant throughout the stars pre - main sequence evolution . while it is known that the median of the stellar x - ray luminosity function does in fact decrease with age due to stellar spin down ( e.g. hemplemann et al . 1995 ; gdel et al . 1997 ; gdel 2004 ) , the evolution of the x - ray luminosity during the disc dispersal phase is much smaller ( see figure 41 , gdel 2004 ) . certainly , any evolution of the x - ray luminosity for ages up to several tens of myrs is smaller than the observed spread in x - ray luminosities at @xmath22myr . furthermore , there is observational evidence that the x - ray luminosities of ctts and wtts are significantly different , with wtts in general being more luminous than ctts ( e.g. neuhauser et al 1995 ; stelzer & neuhauser 2001 ; flaccomio et al . 2003 ; preibisch et al . this has lead to a discussion of x - ray emission being ` disturbed ' by accretion , in terms of either x - ray absorption in accretion columns ( gregory et al . 2007 ) or confinement of the x - ray producing corona in accreting systems ( preibisch et al . 2005 ) . recently , drake et al . ( 2009 ) suggested that x - rays may modulate accretion through photoevaporation ( something they called ` photoevaporation starved accretion ' ) . such a scenario may be able to account for the difference in the observed x - ray luminosities of ctts and wtts since more luminous stars will lose their disc s first . in order to assess the effect of ` photoevaporation starved accretion ' in explaining the x - ray observations we adopt here the null stance ; the x - ray luminosity of an individual young stellar object ( yso ) remains constant in time as our models evolve from ctts to wtts ( i.e. we assume that the x - ray luminosity function is fixed ) . m@xmath9 , including both wtts and ctts . ] while the first self - consistent numerical simulations performed by owen et al . ( 2010 ) were a significant step forward in understanding x - ray photoevaporation , the models were calculated for only one value of the x - ray luminosity ( @xmath23erg s@xmath2 ) . a variety of underlying disc models were considered , which showed that x - ray photoevaporation is fairly insensitive to the details of the x - ray ` dark ' region , due to the fact that the sonic surface is located at least several flow scale heights from the x - ray ` dark'/`bright ' transition . however , the large observed spread in x - ray luminosities means that the dependence of photoevaporation on the x - ray luminosity needs to be considered . before we discuss the results of a detailed numerical investigation of parameter space , we can use simple fluid mechanics to predict the variation of photoevaporation rates with x - ray luminosity . namely , any axi - symmetric inviscid steady state flow must satisfy the following conditions : conservation of mass ; conservation of specific angular momentum ; hydrostatic equilibrium perpendicular to the streamlines ; and the conservation of a bernoulli type potential along each streamline that allows us to construct an energy integral equivalent to the bernoulli potential for a baratropic flow with the identical @xmath24 . by uniqueness , this bernoulli - type energy integral must be a conserved quantity along the streamline in the x - ray heated flow . ] along each streamline . as discussed in ercolano & owen ( 2010 ) the velocity and temperature in a thermally driven wind is intimately linked the value of the effective potential , and unlikely to be greatly affected by changes in the x - ray luminosity . furthermore , as shown in owen et al . ( 2010 ) the gas temperature can be roughly described in terms of the ionization parameter ( @xmath25 ) alone . thus if we consider any x - ray heated wind ( primordial or a disc with an inner hole ) and vary the x - ray luminosity , spatially fixing the temperature and velocity requires that the ionization parameter remains constant . this means that the density will scale _ globally _ in the flow as @xmath26 . it is then a trivial matter to show that the new wind solution we have constructed , with a different x - ray luminosity will still satisfy the required conditions described previously . hence , we can use the conservation of mass to show that the x - ray photoevaporation model predicts : @xmath27 for all x - ray flows including both primordial discs and disc with inner holes . also , since the only change is to the global density , the normalised mass - loss profiles should also remain approximately independent of the x - ray luminosity . furthermore , owen et al . ( 2010 ) also showed that there was very little variation in mass - loss rate with inner hole radius . we have performed radiation - hydrodynamic models to study the form of the mass - loss , under the assumption that it is driven by x - ray photoevaporation ( xpe ) . we consider xpe from discs around solar - type stars for both primordial ( optically thick discs extending into the dust destruction radius ) and discs with inner - holes ( in both gas and dust ) . we run primordial disc models at six equal logarithmic intervals between @xmath28 $ ] . we further calculate four inner hole models at three different x - ray luminosities @xmath29 ( in addition to the eight models already computed by owen et al . 2010 at @xmath30 ) , at radii of @xmath31 au . we start from an initial density structure of a protoplanetary disc surrounding a 0.7m@xmath9 star with @xmath32=4000 k and @xmath33=2.5@xmath34 , taken from the set of dalessio et al . we adopt the following elemental abundances , given as number densities with respect to hydrogen : he / h@xmath35 , c / h@xmath36 , n / h@xmath37 , o / h@xmath38 , ne / h@xmath39 , mg / h@xmath40 , si / h@xmath41 , s / h@xmath42 . these are solar abundances ( asplund , grevesse & sauval , 2005 ) depleted according to savage & sembach ( 1996 ) . we use the zeus2d code ( stone & norman 1992 ) to calculate the hydrodynamical evolution of the disc , where the gas temperature due to x - ray irradiation is parametrised as a function of ionisation parameter using the 3d radiative transfer code mocassin ( ercolano et al 2003 , 2005 , 2008 ) . the temperature of gas that is not heated by the x - rays is fixed to the dust temperature and the transition point between the x - ray bright and the x - ray dark region occurs at a column of 10@xmath43 @xmath18 . such an approximation obviously results in a temperature and density discontinuity at this point ; however , since this transition occurs well below the sonic surface ( where the mass - loss rate is effectively set ) it does not affect the resulting mass - loss rate . the distribution evolves to a steady state , with a bound x - ray dark disc and a thermally driven , x - ray bright , transonic photoevaporative wind . the numerical methods employed here and briefly described above are similar to those of owen et al . ( 2010 ) and we refer the reader to that article for a more detailed description of the model setup . the numerical models described above were used to calibrate the analytical relations derived at the beginning of section 3 , which we can use to build a synthetic disc population . the main results are summarised as follows : as expected , for a star of a given mass the total mass loss rate scales almost linearly with the x - ray luminosity . this result applies to both primordial and inner hole sources . specifically , @xmath44 where @xmath45 is a constant taking the value unity for primordial discs and 0.75 for discs with inner holes . figure [ fig : mdotlx ] shows the mass loss rates calculated by our grid of radiation - hydrodynamics models for primordial discs . equation [ eqn : mdotlx ] is also a good fit for inner - hole sources , irrespective of hole size , for the cases studied here ( inner hole radii between 5 and 70 au ) . + ) , the dashed line represents the linear analytical prediction . ] + figure [ fig : hole_mdot ] shows the results for inner hole sources , where the left and right panels show the mass - loss rate as a function of hole radius and x - ray luminosity respectively . 2 . the cumulative mass - loss rates ( i.e. the radial profiles of the surface mass - loss rates ) for different x - ray luminosities are fairly self - similar for primordial discs , and also for inner hole sources when one considers the profile as a function of @xmath46 , where r@xmath47 is the inner hole radius to refer to cylindrical co - ordinates and @xmath48 to refer to spherical polar co - ordinates . ] . figure [ fig : mdotr ] ( left panel ) shows the normalised cumulative surface mass - loss rates as a function of radius for different x - ray luminosities for our primordial disc models . it is clear from the figure that higher x - ray luminosities produce a broader mass - loss profile ; however , this has negligible effect on the global disc evolution and we adopt a mean profile , shown by the solid black line in the figure . figure [ fig : mdotr ] ( right panel ) shows the normalised cumulative surface mass loss rates as a function of r - r@xmath47 for the inner hole models . the mean profile adopted for the inner hole models is shown again as the solid black line . from the results presented in this section it is immediately apparent that the mass - loss properties for a star of a given mass are completely controlled by the x - ray luminosity . we can therefore construct a population synthesis model that includes only viscosity and xpe ( with appropriate initial conditions , see section 4.1 ) . the evolution of the surface density of a photoevaporating and viscously evolving disc can be described in one dimension using the formalism of lynden - bell & pringle ( 1974 ) : @xmath49-\dot{\sigma}_w(r , t)\ ] ] where @xmath50(r ) describes the viscosity term and @xmath51 represents the mass - loss due to photoevaporation calculated in the previous section . before moving on to discuss our choice of initial conditions and viscosity law , it is worthwhile discussing the qualitative evolution of a photoevaporating viscously evolving disc . equation [ eqn : big ] is perhaps easiest to understand in its integral form ( i.e @xmath52 [ eqn : big ] ) , which tells us : @xmath53 where @xmath54 is the disc mass and @xmath55 is the accretion rate onto the star due to viscous transport . equation [ eqn : bigm ] tells us immediately that there are two phases of disc evolution : when @xmath56 the disc evolution is dominated by the viscous transport of angular momentum and hence associated accretion rather than the removal of gas through photoevaporation . during this stage the disc will behave in a similar manner to a standard viscously evolving disc , without photoevaportion especially when @xmath57 . however , when @xmath58 it is now photoevaporation that dominates the evolution of the disc , opening a gap in the disc and then finally removing the remaining disc material until the disc is dispersed . the accretion rate of a disc evolves on the disc s viscous time @xmath59 at the outer radius , which is of order the disc s lifetime ( lynden - bell & pringle , 1974 ; pringle , 1981 ; ruden , 1993 ) . the transition from a viscously evolving disc to a clearing disc occurs when a gap opens in the inner disc and the material inside the gap drains onto the central star . since , the time - scale for the inner disc to drain is the viscous time at the point the gap opens ( which is typically much less than the global viscous time of the disc ) , the transition from a viscously accreting disc ( i.e. one that would observationally be classified as an optically thick accreting primordial disc ) to one that is being dispersed , occurs on a time - scale much shorter than the disc s current lifetime ( clarke et al . 2001 ; ruden 2004 ) . hence , the evolution of a viscously evolving photoevaporating disc can naturally explain the observations of discs which appear to evolve under the effect of viscous transport alone , but are then rapidly dispersed . a key test for a model of viscously evolving photoevaporating discs is to check whether it can self consistently explain the observational statistics for _ both _ the evolution of primordial discs and those thought to be in the act of clearing ( i.e. ` transition ' discs ) and the time - scales associated with their evolution . as such , any model of disc evolution depends on the choice of viscosity law and the initial conditions associated with disc evolution . we now turn our attention to the choice of initial conditions and viscosity laws that are required to solve equation [ eqn : big ] . we adopt the form @xmath60 , where @xmath61 , which we always evaluate at 1au . such a choice is appealing both observationally and theoretically , since it predicts a surface density scaling as @xmath62 which is supported by observations ( e.g. hartmann et al 1998 ; andrews et al . 2010 ) and is consistent with a constant @xmath63 disc which is mildly flaring i.e. @xmath64 . furthermore , we adopt the zero - time similarity solution of lynden - bell & pringle ( 1974 ) for which the initial surface density distribution takes the form : @xmath65 where m@xmath66 and r@xmath67 are the initial disc mass and a scale radius describing the exponential taper of the disc s outer region . these initial conditions , together with the viscosity parameter @xmath63 and the x - ray luminosity fully determine the evolution of the disc through equation [ eqn : big ] . at this point it is useful to construct a ` null model ' of viscous evolution without photoevaporation that when combined with the observed x - ray luminosity function can explain the observed decline in disc fractions with age . in effect we are asking whether there is a universal set of disc viscous parameters which can explain the variation in disc lifetime from cluster to cluster purely in terms of the observed spread in x - ray luminosity . as discussed above the basic principle of all photoevaporation models is that discs should evolve viscously , hardly noticing the effects of photoevaporation , until the mass accretion rates in the discs have fallen to a value that is comparable to the photoevaporation rate , at which point the remaining disc material is rapidly cleared . viscous evolution alone with @xmath68 predicts that accretion rates evolve as : @xmath69 where @xmath70 is the viscous timescale at @xmath71 . this evolution should therefore be observed in discs before xpe sets in . we can equate the fraction of disc - bearing pre - main sequence stars at a given time ( @xmath72 ) , with the fraction of stars in the x - ray luminosity function that have luminosities less than a cut - off x - ray luminosity @xmath73 . in order for objects with x - ray luminosities equal to @xmath74 to be about to lose their discs , the viscous accretion rate must be equal to @xmath75 at this point . we have performed this simple exercise using the disc fractions for nearby clusters compiled by mamajek ( 2009 ) , the taurus x - ray luminosity function and the xpe theory developed above . the result is shown in figure [ fig : null ] , where each point represents the current accretion rate cut - off in a cluster implied by xpe . we scale our results on disc fractions assuming an initial close binary fraction of 14% by considering that the 0.3myr old cluster ngc 2024 ( haisch et al . 2001a ) , which shows a disc fraction of 86% , is too young for any disc to have been destroyed by photoevaporation or planet formation , but only through binary interactions . the solid line in the plot represents a suitable fit of equation [ eqn : mdot ] to the data , from this fit we can extract an initial accretion rate of @xmath76m@xmath9 yr@xmath2and a viscous time of @xmath77yr . from these two values we can calculate an initial disc mass of @xmath78m@xmath9 , which is similar to the canonical value ( 10% of the stellar mass ) at which viscous angular momentum transport takes over from self - gravity . errors in the disc fractions . the solid line shows a fit of equation [ eqn : mdot ] used to determine suitable values of @xmath79 and @xmath59 . ] along with giving us appropriate initial conditions for our disc population model , the above also provides a stringent test of the hypothesis that x - rays are key to disc evolution and dispersal . if the x - rays were not the dominant dispersal mechanism , there is no a priori reason to expect a ` null ' model constructed only using knowledge of the x - ray luminosity function and observed disc fractions to reproduce a plausible evolution of the accretion rates seen in ctts , both in terms of the time exponent ( in equation [ eqn : mdot ] ) and its initial value . in fact , increasing or decreasing the spread about the median of the taurus x - ray luminosity function by a factor of @xmath75 or greater makes a fit of equation [ eqn : mdot ] to the ` null ' model impossible . although this agreement could be fortuitous , it is reassuring that the x - ray luminosity function , disc lifetimes and accretion histories are consistent with our xpe hypothesis . furthermore , in order to uniquely specify the viscous evolution we must pick suitable values of @xmath63 and @xmath71 which in turn specifies @xmath80 . while any combination of @xmath71 and @xmath63 that give the required viscous time will reproduce the same ` null ' viscous model ( i.e. non - photoevaporating ) , a disc evolution model that includes photoevaporation mildly breaks this degeneracy . by performing a fit ( by eye ) of a viscously evolving photoevaporating disc population to the disc fractions used to derive the ` null ' model we obtain values of @xmath81 and @xmath82au although due to the large scatter in the disc fractions and only mild breaking of the @xmath63 , @xmath71 degeneracy , we estimate errors on these values of a factor of 2 - 3 . in figure [ fig : df ] we show the obtained fit to the disc fractions compiled by mamajek ( 2009 ) for our viscously evolving photoevaporating disc model with a single set of initial conditions : @xmath78m@xmath9 , @xmath83 and @xmath82au . finally , figure [ fig : mdott ] shows the predicted evolution of accretion rate with time for individual disc models ( each line shows the evolution of accretion rate with time for one disc model , for a given value of the x - ray photoevaporation rate ) . this clearly shows that a spread in accretion rates at late times @xmath16myr does not necessary require a spread in initial conditions as some authors have suggested is necessary outside the xpe framework ( e.g. armitage et al . 2003 ; alexander & armitage , 2009 ) . observationally there is however , a spread in accretion rates seen at early times @xmath841myr ( e.g. hartmann et al . 1998 ) . this variability can not be fit with a single set of initial conditions since photoevaporation has had no time to act on the disc . the failure to match the spread in accretion rates at early times @xmath12myr may not be surprising , since at early times the angular momentum transport mechanism may be dominated by self - gravity and accretion may be episodic ( e.g. lodato & rice , 2005 ) . in reality the zero time point in our models corresponds to the point at which the transition from a self - gravitating to a viscous disc occurs , as indicated by the determination of @xmath85 . this may explain , why our initial accretion rate is significantly lower than some of the accretion rates measured for the youngest objects ( hartmann et al . 1998 ) . furthermore , better agreement for the early time data could easily be obtained by assuming a range of initial surface density profiles . indeed our choice of a self - similar surface density distribution at zero time has no physical basis other than convenience ( in fact it would be extremely surprising for discs to be born with the surface density distribution of the zero time similarity solution of equation [ eqn : big ] given viscous angular momentum transport is not a key process during the disc formation stage ) . we emphasise , that any initial surface density distribution with the same initial viscous parameters @xmath79 , @xmath59 and @xmath63 will tend to the same evolution after a few viscous times as shown in lynden - bell & pringle ( 1974 ) . for this reason , we do not attempt to explain disc evolution at early times here , since we are mainly interested in the question of disc dispersal , for which only the viscous evolution phase at @xmath16myr is relevant . we have used the methods and initial conditions derived in the previous section to construct a population synthesis model for the evolution of discs dominated by viscosity and xpe . we have calculated 500 disc models based on a random sampling of the taurus x - ray luminosity function . our disc evolution models are computed by solving equation [ eqn : big ] numerically using the method set out in owen et al . ( 2010 ) , following the evolution of the disc until the disc is cleared to 100au . at radii larger than 100au the photoevaporation rates are extremely uncertain and we can not be confident in results that continue the evolution beyond this radius . however , for the sake of completeness we will discuss the possible qualitative evolution of these remnant discs in section 5.3 . we now turn our attention to some specific predictions from these models and , where possible , compare them with observations . as discussed in section 2 , drake et al . ( 2009 ) suggested that coronal x - rays suppress the accretion flow onto young solar - type stars through the driving of a photoevaporating wind . this photoevaporation starved accretion phase can explain the tentative negative correlation between mass accretion rate and stellar x - ray luminosity reported by drake et al ( 2009 ) . moreover the reduction in disc lifetime in strong x - ray sources can explain the observation that the x - ray luminosities of accreting t tauri stars are systematically lower than those of non - accretors . the xpe models of owen et al ( 2010 ) clearly show that there is indeed a phase in the disc evolution ( before the opening of the gap ) where the effects of this ` starving ' are apparent in the radial dependence of the accretion rate . in figure [ fig : psa ] we compare the accretion rate and surface density profiles of the median disc model undergoing xpe , 0.5myr before the gap opens , against those of a disc which is only subject to viscous evolution . inside @xmath86 au the accretion rate drops before it reaches the star compared to the standard case where the accretion rate tends to a constant throughout the entire disc . this can be compared to the euv photoevaporation model : in this case the mass - loss profile is narrowly peaked between @xmath87au ( for solar type stars ) and the total mass - loss rate is considerably less ( @xmath88m@xmath9 yr@xmath2 ) . this results in a shorter and much less pronounced period of ` starving ' i.e. the disc is only affected inside a few au . in contrast , the photoevaporation starved accretion lasts for @xmath720 - 30% of the disc lifetime in the x - ray model , with significant consequences for global disc evolution : i.e. a flattening of the surface density profile and a significant drop in the accretion rate through the disc . myr before a gap opens in the disc due to xpe . ] in figure [ fig : median ] we show the evolution of the surface density for the median disc model ( i.e. the disc with the median x - ray luminosity of @xmath89 erg s@xmath2 which corresponds to a photoevaporation rate of @xmath90 m@xmath9 yr@xmath2 ) undergoing the stages of gap - opening and final clearing . this shows the drop in surface density through the disc between @xmath91au before the gap opens ( as shown in figure [ fig : psa ] ) . moreover the broad photoevaporation profile also causes the steady erosion of the disc during the draining of the inner hole , so that the hole ( though opening at @xmath92au ) roughly doubles in size during inner hole draining . once the inner disc has completely drained there is a rapid clearing of the disc out to @xmath93au , because this region was previously depleted during the photoevaporation starved accretion phase . the fast clearing phase slows down once the inner hole reaches radii less affected by photoevaporation starved accretion . this results in a a total clearing time of roughly @xmath93% of the disc lifetime , which is consistent with transition disc statistics . this can be compared to the euv models of alexander et al . ( 2006b ) , which resulted in a transition phase approximately 3% of the disc s lifetime , and those of clarke et al . ( 2001 ) which clear the outer disc on time - scales of the order of the disc lifetime . myr ) and the remaining lines show the surface density at 1% steps in disc lifetime . ] the negative @xmath94-@xmath14 correlation reported by drake et al ( 2009 ) is a simple consequence of the fact that discs with higher x - ray luminosities produce more vigorous winds causing the disc s accretion rates to be lower than those for discs with a less vigorous wind . thus a negative @xmath94-@xmath14 correlation is expected for clusters with a relatively narrow age range . this effect is however counteracted by the fact that discs with lower x - ray luminosities take longer to evolve , spending more time at lower accretion rates compared to high luminosity objects . therefore , if the x - ray luminosity was compared to accretion rates for an entire disc population s lifetime , in clusters with very large age spreads a positive correlation might be expected . in figure [ fig : syn_obs ] we show plots of @xmath94-@xmath14 for our synthetic ` cluster ' members selected to give clusters with various narrow age ranges ( upper panels and lower left panel ) along with the a plot of the total disc population ( bottom right panel ) . as expected , the xpe model predicts a positive @xmath94-@xmath14 correlation for clusters with large age spread and a negative correlation for clusters with a narrow age spread . as a comparison , the age spread in orion is roughly 2 - 3myr ( haisch et al . 2001b ) , which explains the observation of a negative correlation by drake et al . ( 2009 ) . the observation of systematically higher x - ray luminosities of non - accreting tts ( wtts ) compared to accreting tts ( ctts ) can also be explained by our models in terms of photoevaporation starved accretion . figure [ fig : cttvwtt ] shows the time evolution of the median x - ray luminosity of the ctts ( solid line ) and wtts ( dashed line ) populations for our model compared to the data compiled by gdel et al . ( 2007 ) for the taurus cluster ( black circles and red squares ) . we note the good agreement between the model predictions and observations . the dotted line represents the critical x - ray luminosity as a function of time which in our model separates ctts and wtts and corresponds to the x - ray luminosity of objects which have just opened a gap and have begun the clearing phase . the number of anomalous objects ( i.e. ctts above the line and wtts below the line ) is small at ages @xmath95 1 myr . given uncertainties in age determinations ( particularly at @xmath84 a myr ) the agreement between the observations and predictions is very encouraging . the recent observations of a class of transition ( inner hole ) discs with residual gas inside the inner dust radius and with signature of accretion has prompted some authors to question the viability of photoevaporation as the formation mechanism for the inner hole in some sources ( e.g. cieza et al 2010 ) . previous euv - driven photoevaporation models ( e.g. alexander & armitage 2009 ) indeed predicted that at the time of gap opening the surface density of the gas in the inner disc and the accretion rates due to the inner disc draining onto the star should be undetectable . this is however not the case for xpe , mainly due to the fact that the wind rates can be two orders of magnitude higher than the euv - driven rates , meaning that at the time of gap opening the mass of the draining inner disc and the accretion rate onto the star of the inner disc material remain detectable for a non - negligible amount of time . for the disc population generated in this work , figure [ fig : prob_holes ] , shows that the accreting inner holes and non - accreting inner - holes ( @xmath94 @xmath96m@xmath9 yr@xmath2 ) are in general equally likely out to a radius of @xmath97au . clearly as the transition disc is further photoevaporated and its inner radius moves out the accretion signatures onto the star become less evident and non - accreting inner holes dominate at radii larger than 20 au . the total integrated ratio out to 10au ( the radius probed by 24@xmath98 m emission around solar - type stars ) is found to be @xmath99 accreting and @xmath100 non - accreting for the entire population . we caution that this is _ not _ equivalent to the observed fraction of accreting to non - accreting objects in a individual cluster , where the cluster age should also be accounted for . in young clusters the transition disc population is dominated by high x - ray luminosity objects which give rise to a considerably longer accreting inner hole phase . in contrast this ratio is much lower in old clusters where the transition disc population is dominated by low x - ray luminosity objects that have very short accreting inner - hole phases . it is perhaps worth noting at this point that the detection of a ` transition ' disc observationally is made through observation of the dust continuum spectral energy distribution ( sed ) . alexander & armitage ( 2007 ) examined the behaviour of dust in a photoevaporating disc , finding that , under the action of dust drag , the time - scale for dust grains to drain onto the star is of order 10@xmath101yrs , after the gap opens , approximately two order of magnitudes faster than the gas draining time - scales . this means that an observer would certainly see a significant drop in opacity in the inner disc immediately after a gap has opened , while the gas will still linger in the inner dust disc for the duration of its viscous draining time - scale of @xmath102yrs . we have used our population synthesis model to investigate the accretion rate versus inner hole size evolution for transition discs created by xpe under the assumption of immediate dust clearing at the time of gap opening . figure [ fig : innerholes ] shows the probability distribution of the disc models in the @xmath94-@xmath103 plane . the symbols represent a sample of observations of solar - type objects classified as ` transition ' discs by espaillat et al . ( 2007a , b,2008,2010 ) - red circles , hughes et al . ( 2009,2010 ) - red squares , kim et al . ( 2009 ) - red diamonds , calvet et al . ( 2005 ) - black diamonds , mern et al . ( 2010 ) - black squares and cieza et al . ( 2010 ) - black triangles ( although cieza et al . 2010 do not fit for the inner - hole radius they list as transitional sources those discs that have a deficit of emission in the _ spitzer _ irac bands ; therefore we conservatively estimate an inner hole radius of @xmath104au for all their sources ) . it is immediately clear from the figure that there is a population of large inner hole , strongly accreting transition discs that can not have been created by xpe . gap opening by a giant planet or grain growth is perhaps the most plausible explanation for these objects . however , there is a significant number of discs with inner holes that are consistent with an xpe origin . furthermore we note the lack of observations of non - accreting ` transition ' discs with holes at radii greater than 20au , where our model predicts a significant populations ( although several non - accreting discs with large holes have been detected in different mass ranges e.g. mern et al . 2010 , where the observations probe different radial scales ) . the observations are still rather sparse and it is currently not possible to say whether the observed population of transition discs is a true representation or an artifact of observational selection effects . -@xmath103 plane . we have applied an observational cut - off of @xmath105m@xmath9 yr@xmath2 in accretion rate and objects below the cut - off are all shown at the bottom of the diagram . we show a representative sample of solar - type objects classified as ` transition ' discs ; data from taken from espaillat et al . ( 2007a , b,2008,2010 ) - red circles , hughes et al . ( 2009,2010 ) - red squares , kim et al . ( 2009 ) - red diamonds , calvet et al . ( 2005 ) - black diamonds , mern et al . ( 2010 ) - black squares and cieza et al . ( 2010 ) - black triangles , where error - bars are listed they are shown . we also show two model tracks corresponding to a 1dex spread in x - ray luminosities about the median model in magenta . ] one obvious consequence of an x - ray photoevaporation mechanism is that the properties of transition discs should be correlated with the x - ray luminosity , something no other model of photoevaporation or ` transition ' disc origin would predict . in figure [ fig : inner_corr ] , we show two such correlations namely the inner hole radius ( left panel ) and accretion rate ( right panel ) against x - ray luminosity , considering only accreting ` transition ' discs ( i.e. those with an accretion rate @xmath106m@xmath9 yr@xmath2 ) . the plots have been generated by randomly sampling ( in time ) the accreting transition phase of each disc model several times , and should therefore provide a reasonable estimate of both the general form of the correlation plus the associated scatter . clearly , since discs with higher x - ray luminosities open gaps earlier and at higher accretion rates a strong positive correlation between @xmath14 and @xmath94 would be expected and is reproduced by the models as shown . furthermore , we also recover a positive correlation between inner hole radius and x - ray luminosity for _ accreting _ transition discs as shown in the figure . we can understand this by remembering that while the inner disc is still draining the inner hole radius is also being eroded outwards as shown in figure [ fig : median ] . since , for more x - ray luminous objects the region being eroded is more depleted due to photoevaporation starved accretion and the magnitude of the mass - loss is higher , the inner - hole radius is able to be eroded to larger radius during the accreting phase . this explains the positive slope of the right hand extent of the symbols in the left panel of figure [ fig : inner_corr ] . the x - ray luminosity function is a crucial input into the xpe model . in this work we have used the taurus x - ray luminosity function as this best represents the quiescent x - ray flux the disc sees throughout its lifetime . however , here we discuss the consequences of a different x - ray spectrum that may be incident on the disc . if the incident spectrum the disc sees is systematically harder than the one used ( through attenuation of the x - ray spectrum by large neutral columns close to the star ) or softer than the spectrum used , the qualitative behaviour of the disc population would remain the same , since one can vary the initial condition to fit . therefore , it is important to assess the qualitative changes relating to some of the predictions relating to transition discs . as discussed in section 2 it is only the soft x - rays ( 0.1 - 1kev ) that have any thermal impact , then the result of a changing spectrum can just be considered to be a change in the soft x - ray luminosity incident on the disc . therefore , overall mass - loss rates can simply be scaled to this new harder / softer spectrum . in section 4.1 we argued that any x - ray luminosity function with a similar spread can have a ` null ' model constructed to fit the observations of disc fraction . thus , a harder or more attenuated spectrum would result in lower - mass loss rates , which would require a lower initial accretion rate , compared to a softer spectrum , which would require a higher initial accretion rate to explain the observed disc fractions . provided this effect is systematic across all values of @xmath14 ( i.e. in does nt change the spread of x - ray luminosities ) then the consequences for the predicted ` transition ' disc population can be considered . for the harder / more attenuated spectrum a lower initial accretion rate is required implying a lower initial disc mass , and therefore a smaller population of accreting transition discs , with lower accretion rates and smaller inner holes . the converse is true for a softer spectrum which requires a higher initial accretion rate and hence larger initial disc mass , giving rise to a larger population of accreting transition discs , with higher accretion rates and larger holes . our mass loss rates are only accurate out to @xmath7100 au , since at this point other clearing mechanisms ( e.g. fuv photoevaporation ) may become dominant . for this reason , we have stopped our viscous evolution models once the inner hole reaches a size of 100 au . if we were to extrapolate our models to clearing beyond 100au using equation 2 , we would find that in about 10% of cases ( sources irradiated by a low x - ray flux ) the final clearing time - scales would exceed 10myr , resulting in a population of long - lived discs with large inner holes . these discs may not survive long if fuv photoevaporation is efficient at large radii as suggested by gorti & hollenbach ( 2009 ) . however , given the uncertainties in the fuv rates , we also consider the case where these objects survive for a long enough time for their dust continuum emission to be observed . the spectral energy distribution ( sed ) of these cold massive discs should be similar to that of young debris discs , where a debris disc model is normally considered to be a single temperature belt of optically thin dust ( thought to be constantly replenished by collisions between planetesimals ) at a given radius from the star ( wyatt 2008 ) . this suggests that some of the sources that are currently classified as debris discs may in fact be ` xpe relics ' . we have used the radiative transfer code of whitney et al ( 2003a , b ) to calculate the sed of a typical xpe relic using the standard input disc structure and dust properties of the code . the properties of these xpe relics were taken from the end point of the lowest x - ray luminosity ( and hence the longest lived ) model with an inner radius of 100 au , outer e - folding radius of 310au and a mass of @xmath107m@xmath9 in gas ( and a dust to gas mass ratio of 0.01 ) . while such a disc is likely to be settled in the dust , we computed three models : flat dust distribution , fully mixed dust distribution and a disc where @xmath108 . figure [ fig : debris ] shows a plot of the fractional excess of the disc compared with the stellar photosphere at 24@xmath98 m and @xmath109 m , where we compare our models with observations of objects classified as young debris discs around solar - type stars ( wyatt 2008 ) . it is clear from this figure that xpe relics with a degree of dust settling share the same space in the excess - excess plot as sources currently classified as young debris discs . the fully - mixed discs show a @xmath109 m excess which is probably too large to be classified as a debris disc , although it is extremely unlikely that any disc could survive for @xmath13myr without undergoing dust settling . we also note that the predicted 850@xmath98 m emission from the evolved xpe relics falls below the current detection limits at 50 pc ( andrews & williams 2005 ) , and thus these relics would not contradict previous sub - mm observations of wtts which show that most are devoid of emission out to 500 au ( duvert et al 2000 ) . only alma will be able to separate these large massive discs from canonical debris discs , and hence confirm or dismiss the existence of this proposed class of objects , placing constraints on the role of fuv photoevaporation at large radius . we have used radiation - hydrodynamic calculations of x - ray photoevaporated discs coupled to a viscous evolution model to construct a population synthesis model , with which we have studied the physical properties of primordial and transition discs . the initial conditions and viscosity law are constrained using recent observations of disc fractions in nearby clusters ( mamajek 2009 ) . we require a viscosity coefficient of approximately @xmath110 to match the observational constraints . m excess above the photosphere against @xmath111 m excess for disc models with a large inner hole ( 100 au ) and low x - ray luminosities and therefore a predicted remaining lifetime @xmath13myr the circles show discs close to edge - on and increasing @xmath111 m excess represents the evolution from a flat to fully mixed dust distribution . similarly the diamonds show the disc models close to face - on . also shown as open squares is observations of young solar - type stars ( @xmath112myr ) with ir excesses classified as debris discs , taken from ( wyatt 2008 ) . ] our main conclusions can be briefly summarised as follows : 1 . x - rays play a major role in the evolution and dispersal of discs around solar - type stars , driving vigorous photoevaporative winds whose rates scale linearly with the x - ray luminosity , which are in the range of observed accretion rates for t - tauri stars . we have constructed a ` null ' accretion disc model using only knowledge of the observed disc fractions and x - ray luminosity functions under the assumption that discs are dispersed through x - ray photoevaporation . this ` null ' model shows very good agreement with observed accretion rates in ysos as well as their evolution with time , providing further independent confirmation of the viability of x - ray photoevaporation as a dominant dispersal mechanism . x - rays suppress accretion by preventing accreting material from reaching the star , since this material is removed through photoevaporation . this ` photoevaporation starved accretion ' ( drake et al 2009 ) produces a negative correlation between x - ray luminosity and accretion rate for clusters with a relatively narrow age spread , in agreement with the observational correlation reported by drake et al . ( 2009 ) in the orion data . 4 . our models successfully reproduce the observation that wtts are systematically brighter in x - rays than ctts . whereas this has previously been interpreted as modification of x - ray emission or detectability by the presence of discs , our models support drake et al . ( 2009 ) in reversing the causal link - i.e. that disc lifetimes are regulated by xpe and hence discless stars are on average those with higher @xmath14 . a large fraction ( @xmath113% ) of observed transition discs can be easily explained by x - ray photoevaporation . there is however a population of strongly accreting transition discs with large inner holes @xmath114au that lies outside the @xmath115 region predicted by our models , suggesting that alternative mechanisms are responsible for their inner hole ( e.g. binary interaction , grain - growth and/or planet formation ) . . a fraction of currently observed objects classified as young debris discs ( on the basis of their excesses at 24@xmath98 m and @xmath109 m ) may in fact be the relics of x - ray photoevapored discs , which are predicted to be long - lived ( @xmath1010 myr ) for low x - ray luminosity sources . future mass determination with _ alma _ are necessary to shed light onto the nature of these objects . we are grateful to the anonymous referee who helped improve the clarity of this work . the authors would like to thank thomas preibisch and manuel gdel for providing their data on x - ray luminosities and helpful discussions regarding x - ray luminosity functions . we also thank jeremy drake for insightful discussions on photoevaporation starved accretion , x - ray observations and the x - ray luminosity function . we would also like to thank mark wyatt and mark booth for helpful discussions regarding debris disc observations and to mark wyatt for providing the sample of debris disc observations used in this work . jeo acknowledges support of a stfc phd studentship and is indebted to the university of exeter astrophysics department for hospitality during the completion of this work . be is supported by a science and technology facility council advanced fellowship . 100 albacete colombo , j. f. , flaccomio , e. , micela , g. , sciortino , s. , & damiani , f. 2007 , , 464 , 211 alexander r. d. , clarke c. j. , pringle j. e. , 2004 , mnras , 354 , 71 alexander r. d. , clarke c. j. , pringle j. e. , 2006 , mnras , 369 , 216 alexander r. d. , clarke c. j. , pringle j. e. , 2006 , mnras , 369 , 229 alexander , r. d. , & armitage , p. j. 2007 , , 375 , 500 alexander , r. d. , & armitage , p. j. 2009 , , 704 , 989 andrews , s. m. , & williams , j. p. 2005 , , 631 , 1134 andrews , s. m. , wilner , d. j. , hughes , a. m. , qi , c. , & dullemond , c. p. 2010 , arxiv:1007.5070 armitage p. j. , hansen b. m. s. , 1999 , natur , 402 , 633 armitage p. j. , clarke c. j. , palla f. , 2003 , mnras , 342 , 1139 asplund m. , grevesse n. , sauval a. j. , 2005 , aspc , 336 , 25 ercolano b. , clarke c. j. , 2010 , mnras , 402 , 2735 ercolano b. , owen j. e. , 2010 , mnras , 406 , 1553 ercolano b. , clarke c. j. , hall a. c. , 2010 , arxiv , arxiv:1008.0866 espaillat , c. , et al . 2007 , , 664 , l111 espaillat , c. , calvet , n. , dalessio , p. , hernndez , j. , qi , c. , hartmann , l. , furlan , e. , & watson , d. m. 2007 , , 670 , l135 espaillat , c. , et al . 2008 , , 689 , l145 espaillat c. , et al . , 2010 , apj , 717 , 441 flaccomio e. , micela g. , sciortino s. , 2003 , a&a , 402 , 277 font a. s. , mccarthy i. g. , johnstone d. , ballantyne d. r. , 2004 , apj , 607 , 890 gregory s. g. , wood k. , jardine m. , 2007 , mnras , 379 , l35 gorti u. , hollenbach d. , 2009 , apj , 690 , 1539 gorti u. , dullemond c. p. , hollenbach d. , 2009 , apj , 705 , 1237 gdel m. , guinan e. f. , skinner s. l. , 1997 , apj , 483 , 947 gdel m. , 2004 , a&arv , 12 , 71 gdel m. , et al . , 2007 , a&a , 468 , 353 haisch , k. e. , jr . , lada , e. a. , pia , r. k. , telesco , c. m. , & lada , c. j. 2001 , aj , 121 , 1512 hughes a. m. , et al . , 2009 , apj , 698 , 131 kim , k. h. , et al . 2009 , , 700 , 1017 imanishi k. , koyama k. , tsuboi y. , 2001 , apj , 557 , 747 kashyap v. , drake j. j. , 2000 , basi , 28 , 475 krauss o. , wurm g. , mousis o. , petit j .- m . , horner j. , alibert y. , 2007 , a&a , 462 , 977 kenyon s. j. , hartmann l. , 1995 , apjs , 101 , 117 lynden - bell , d. , & pringle , j. e. 1974 , , 168 , 603 lodato , g. , & rice , w. k. m. 2005 , , 358 , 1489 luhman k. l. , allen p. r. , espaillat c. , hartmann l. , calvet n. , 2010 , apjs , 186 , 111 maggio a. , flaccomio e. , favata f. , micela g. , sciortino s. , feigelson e. d. , getman k. v. , 2007 , apj , 660 , 1462 najita j. r. , strom s. e. , muzerolle j. , 2007 , mnras , 378 , 369 owen j. e. , ercolano b. , clarke c. j. , alexander r. d. , 2010 , mnras , 401 , 1415 neuhaeuser r. , sterzik m. f. , schmitt j. h. m. m. , wichmann r. , krautter j. , 1995 , a&a , 297 , 391 pascucci i. , sterzik m. , 2009 , apj , 702 , 724 preibisch t. , et al . , 2005 , apjs , 160 , 401 pringle , j. e. 1981 , , 19 , 137 richling s. , yorke h. w. , 2000 , apj , 539 , 258 ruden , s. p. 1993 , planets around pulsars , 36 , 197 ruden , s. p. 2004 , , 605 , 880 savage b. d. , sembach k. r. , 1996 , apj , 470 , 893 skrutskie m. f. , dutkevitch d. , strom s. e. , edwards s. , strom k. m. , shure m. a. , 1990 , aj , 99 , 1187 stelzer b. , neuhuser r. , 2001 , a&a , 377 , 538 stone j. m. , norman m. l. , 1992 , apjs , 80 , 753
we explore the role of x - ray photoevaporation in the evolution and dispersal of viscously evolving t - tauri discs . we show that the x - ray photoevaporation wind rates scale linearly with x - ray luminosity , such that the observed range of x - ray luminosities for solar - type t - tauri stars ( 10@xmath0 - 10@xmath1 erg s@xmath2 ) gives rise to vigorous disc winds with rates of order 10@xmath3 - 10@xmath4 m@xmath5 yr@xmath2 . these mass - loss rates are comparable to typically observed t - tauri accretion rates , immediately demonstrating the relevance of x - ray photoevaporation to disc evolution . we use the wind solutions from radiation - hydrodynamic models , coupled to a viscous evolution model to construct a population synthesis model so that we may study the physical properties of evolving discs and so - called ` transition discs ' . current observations of disc lifetimes and accretion rates can be matched by our model assuming a viscosity parameter @xmath6 . our models confirm that x - rays play a dominant role in the evolution and dispersal of protoplanetary discs giving rise to the observed diverse population of inner hole ` transition ' sources which include those with massive outer discs , those with gas in their inner holes and those with detectable accretion signatures . to help understand the nature of observed transition discs we present a diagnostic diagram based on accretion rates versus inner hole sizes that demonstrate that , contrary to recent claims , many of the observed accreting and non accreting transition discs can easily be explained by x - ray photoevaporation . however , we draw attention to a smaller but still significant , population of strongly accreting ( @xmath710@xmath8 m@xmath9 yr@xmath2 ) transition discs with large inner holes ( @xmath10 20 au ) that lie outside the predicted x - ray photoevaporation region , suggesting a different origin for their inner holes . finally , we confirm the conjecture of drake et al . ( 2009 ) , that accretion is suppressed by the x - rays through ` photoevaporation starved accretion ' and predict this effect can give rise to a negative correlation between x - ray luminosity and accretion rate , as reported in the orion data . we also demonstrate that our model can replicate the observed difference in x - ray properties between accreting and non - accreting t - tauri stars . [ firstpage ] accretion , accretion discs - circumstellar matter - planetary systems : protoplanetary discs - stars : pre - main - sequence - x - rays : stars .
in the standard model ( sm ) , @xmath0 violation arises from an irreducible complex phase , the kobayashi - maskawa ( km ) phase @xcite , in the weak - interaction quark - mixing matrix . in particular , the sm predicts @xmath0 asymmetries in the time - dependent rates for @xmath9 and @xmath10 decays to a common @xmath0 eigenstate @xmath11 @xcite . recent measurements of the @xmath0-violation parameter @xmath12 by the belle @xcite and babar @xcite collaborations established @xmath0 violation in @xmath13 and related decay modes @xcite , which are governed by the @xmath14 transition , at a level consistent with km expectations . here @xmath15 is one of the three interior angles of the unitarity triangle @xcite . despite this success , many tests remain before it can be concluded that the km phase is the only source of @xmath0 violation . the @xmath16 decay , which is dominated by the @xmath17 transition , provides an additional test of the sm . within the sm , measurements of @xmath0 violation in this mode should yield the @xmath18 value to a good approximation if the contribution from the penguin diagram is neglected . the correction from the penguin diagram is expected to be small @xcite . thus , a significant deviation in the time - dependent @xmath0 asymmetry in these modes from what is observed in @xmath19 decays would be evidence for a new @xmath0-violating phase . in the decay chain @xmath20 , where one of the @xmath7 mesons decays at time @xmath21 to a final state @xmath22 and the other decays at time @xmath23 to a final state @xmath24 that distinguishes between @xmath25 and @xmath10 , the decay rate has a time dependence given by @xcite @xmath26 \biggr\}.\ ] ] here @xmath27 and @xmath28 are @xmath0-violation parameters , @xmath29 is the @xmath25 lifetime , @xmath30 is the mass difference between the two @xmath25 mass eigenstates , @xmath31 = @xmath21 @xmath32 @xmath23 , and @xmath33 = + 1 ( @xmath34 ) when the tagging @xmath7 meson is a @xmath25 ( @xmath10 ) . the parameter @xmath27 corresponds to the mixing - induced @xmath0 violation and is related to @xmath18 , while @xmath28 represents direct @xmath0 violation that normally arises from the interference between tree and penguin diagrams . in @xmath16 decays the final state @xmath35 mesons may be in a state of @xmath36- , @xmath37- or @xmath38-wave relative orbital angular momentum . since @xmath36- and @xmath38-waves are even under the @xmath0 transformation while the @xmath37-wave is odd , the @xmath0-violation parameters in eq . ( [ eq : psig ] ) are diluted . in order to determine the dilution , one needs to measure the @xmath0-odd fraction . this can be accomplished with a time - integrated angular analysis . the babar collaboration has measured the polarization and @xmath0 asymmetries @xcite , and find the @xmath0-odd contribution to be small , consistent with theoretical expectations @xcite . the @xmath0 asymmetries are found to differ slightly from the expectation that neglects the contribution from the penguin diagram . in this letter we report measurements of the branching fraction , the polarization parameters and @xmath0 asymmetries in @xmath1 decays based on a 140 fb@xmath2 data sample , which corresponds to 152 million @xmath39 pairs . at the kekb energy - asymmetric @xmath4 ( 3.5 on 8.0 gev ) collider @xcite , the @xmath3 is produced with a lorentz boost of @xmath40 antiparallel to the positron beamline ( @xmath41 ) . since the @xmath25 and @xmath10 mesons are approximately at rest in the @xmath3 center - of - mass system ( cms ) , @xmath42 can be determined from the displacement in @xmath41 between the @xmath22 and @xmath24 decay vertices : @xmath43 . the belle detector @xcite is a large - solid - angle spectrometer that includes a three - layer silicon vertex detector ( svd ) , a 50-layer central drift chamber ( cdc ) , an array of aerogel threshold cherenkov counters ( acc ) , time - of - flight ( tof ) scintillation counters , and an electromagnetic calorimeter comprised of csi(tl ) crystals ( ecl ) located inside a superconducting solenoid coil that provides a 1.5 t magnetic field . an iron flux - return located outside of the coil is instrumented with resistive plate chambers to detect @xmath44 mesons and to identify muons ( klm ) . we reconstruct @xmath45 decays in the following @xmath35 final states ; @xmath46 , @xmath47 and @xmath48 . for the @xmath49 decays we use @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 and @xmath55 . for the @xmath56 decays we use @xmath57 , @xmath58 , @xmath59 , @xmath60 and @xmath61 . we allow all combinations of @xmath62 decays except for cases where both @xmath62 decays include neutral kaons in the final state . charged tracks from @xmath62 meson decays are required to be consistent with originating from the interaction point ( ip ) . charged kaons are separated from pions according to the likelihood ratio @xmath63 $ ] , where the likelihood function @xmath64 is based on the combined information from the acc , cdc @xmath65 and tof measurements . we require @xmath66 for kaons in 2-prong ( 4-prong ) @xmath62 meson decays . the kaon identification efficiency is @xmath67 , and @xmath68 of pions are misidentified as kaons . candidate charged pions are required to satisfy @xmath69 , which provides a pion selection efficiency of @xmath70 with a kaon misidentification probability of @xmath71 . neutral pions are formed from two photons with invariant masses above 119 mev/@xmath72 and below 146 mev/@xmath72 . to reduce the background from low - energy photons , we require @xmath73 gev for each photon and @xmath74 gev/@xmath75 , where @xmath76 and @xmath77 are the photon energy and the @xmath78 momentum in the laboratory frame , respectively . candidate @xmath79 decays are reconstructed from oppositely charged track pairs that have invariant masses within 15 mev/@xmath72 of the nominal @xmath80 mass . a reconstructed @xmath80 is required to have a displaced vertex and a flight direction consistent with that of a @xmath80 originating from the ip . candidate @xmath62 mesons are reconstructed from the selected kaons and pions , and are required to have invariant masses within 6@xmath81 ( 3@xmath81 ) of the @xmath62 meson mass for 2-prong ( 3- or 4-prong ) decays , where @xmath81 is the mass resolution that ranges from @xmath82 to @xmath83 mev/@xmath72 . in this selection @xmath81 is obtained by fitting the monte carlo ( mc ) simulated @xmath62 meson mass . these @xmath49 ( @xmath56 ) candidates are then combined with @xmath84 ( @xmath78 ) to form @xmath85 candidates , where the ip and pion identification requirements are not used to select @xmath84 candidates . the mass difference between @xmath85 and @xmath49 ( @xmath56 ) is required to be within 3.00 ( 2.25 ) mev/@xmath72 of the nominal mass difference . we identify @xmath7 meson decays using the energy difference @xmath86 and the beam - energy constrained mass @xmath87 , where @xmath88 is the beam energy in the cms , and @xmath89 and @xmath90 are the cms energy and momentum , respectively , of the reconstructed @xmath7 candidate . the @xmath7 meson signal region is defined as @xmath91 gev and @xmath92 within 3@xmath81 of the @xmath7 meson mass , where @xmath81 is @xmath93 mev/@xmath72 . in order to suppress background from the @xmath94 , or @xmath95 continuum , we require @xmath96 , where @xmath97 ( @xmath98 ) is the second ( zeroth ) fox - wolfram moment @xcite . after applying this requirement , we find that the contributions to the background from @xmath99 , @xmath100 and continuum are approximately equal . figure [ fig : mb ] shows the @xmath92 and @xmath101 distributions for the @xmath45 candidates that are in the @xmath101 and @xmath92 signal regions , respectively . in the @xmath92 and @xmath101 signal regions there are 194 events . to determine the signal yield , we perform a two - dimensional maximum likelihood fit to the @xmath92-@xmath101 distribution ( @xmath102 gev/@xmath103 gev/@xmath72 and @xmath104 gev ) . we use a gaussian signal distribution plus the argus background function @xcite for the @xmath92 distribution . the signal shape parameters are determined from mc . the background parameters are obtained simultaneously in the fit to data . the @xmath101 distribution is modeled by a double gaussian signal function plus a linear background function . we obtain shape parameters separately for candidates with and without @xmath105 decays to account for small differences between the two cases . the fit yields @xmath106 signal events , where 20% include @xmath105 decays . to obtain the branching fraction @xmath107 , we use the reconstruction efficiency and the known branching fraction for each subdecay mode . we obtain an effective efficiency of @xmath108 \times 10^{-3}$ ] from the sum of the products of mc reconstruction efficiencies and branching fractions for each of the subdecays . small corrections are applied to the reconstruction efficiencies for charged tracks , neutral pions and @xmath80 mesons to account for differences between data and mc . we obtain @xmath109\times10^{-3}},\ ] ] where the first error is statistical and the second is systematic . the result is consistent with the present world - average value @xcite . the dominant sources of the systematic error are uncertainties in the tracking efficiency ( 11% ) and in the subdecay branching fractions ( 7% ) . other sources are uncertainties in the fit parameters and methods ( 1% ) , in the reconstruction efficiencies of @xmath110 and @xmath111 , particle identification @xmath112 , polarization parameters @xmath113 , the number of @xmath7 mesons @xmath112 , and mc statistics @xmath112 , where each value in parentheses is the total contribution . the time - dependent @xmath0 analysis requires knowledge of the @xmath0-odd fraction . to obtain the @xmath0-odd fraction without bias , we must take into account the efficiency difference between the two @xmath0-even components . therefore , we perform a time - integrated two - dimensional angular analysis to obtain the fraction of each polarization component . we use the transversity basis @xcite where three angles @xmath114 , @xmath115 and @xmath116 are defined in fig . the angle @xmath114 is the angle between the momentum of the slow pion from the @xmath117 in the @xmath117 rest frame and the direction opposite to @xmath7 momentum in the @xmath117 rest frame . the angle @xmath115 is the polar angle between the normal to the @xmath117 decay plane and the direction of flight of the slow pion from the @xmath85 in the @xmath85 rest frame . the angle @xmath116 is the corresponding azimuthal angle , where @xmath118 is the direction antiparallel to the @xmath117 flight direction . integrating over time and the angle @xmath116 , the two - dimensional differential decay rate is @xmath119 where @xmath120 or @xmath121 denotes longitudinal , transverse parallel , or transverse perpendicular components , @xmath122 is its fraction that satisfies @xmath123 and @xmath124 is its angular distribution defined as @xmath125 the fraction @xmath126 corresponds to the @xmath0-odd fraction . ( [ eq : root_angular ] ) is affected by the detector efficiency , in particular due to the correlations between transversity angles and slow pion detection efficiencies . to take these effects into account , we replace @xmath127 with distributions of reconstructed mc events @xmath128 , which are prepared separately for candidates with and without @xmath105 decays as is done in the branching fraction measurement . we also introduce effective polarization parameters @xmath129 , where @xmath130 is a total reconstruction efficiency for each transversity amplitude . as a result , the signal probability density function ( pdf ) for the fit is defined as @xmath131 we determine the following likelihood value for each event : @xmath132 where @xmath133 is the signal probability calculated on an event - by - event basis as a function of @xmath101 and @xmath92 . the background pdf @xmath134 is determined from the sideband region ( @xmath135 gev/@xmath136 gev/@xmath72 , @xmath137 gev ) . a fit that maximizes the product of the likelihood values over all events yields @xmath138 figure [ fig : pol ] shows the angular distributions with the results of the fit . we study the uncertainties of the following items to determine the systematic errors : background shape parameters , angular resolutions , and slow pion detection efficiencies . also included are a possible fit bias , mc histogram bin size dependence and misreconstruction effects . these systematic errors are much smaller than the statistical errors . we perform an unbinned maximum likelihood fit to the three dimensional @xmath139 , @xmath140 and @xmath141 distributions for @xmath45 candidates to measure the @xmath0-violation parameters . the @xmath9 meson decay vertices are reconstructed using the @xmath62 meson trajectory and an ip constraint . we do not use slow pions from @xmath85 decays . we require that at least one @xmath62 meson has two or more daughter tracks with a sufficient number of the svd hits to precisely measure the @xmath62 meson trajectory . the @xmath142 vertex determination is the same as for other @xmath0-violation measurements @xcite . the @xmath143-flavor of the accompanying @xmath7 meson is identified from inclusive properties of particles that are not associated with the reconstructed @xmath144 decay @xcite . we use two parameters , @xmath33 and @xmath145 , to represent the flavor tagging information . the first , @xmath33 , is already defined in eq . ( [ eq : psig ] ) . the parameter @xmath145 is an event - by - event , mc - determined flavor - tagging dilution factor that ranges from @xmath146 for no flavor discrimination to @xmath147 for unambiguous flavor assignment . this assignment is used only to sort data into six @xmath145 intervals . the wrong tag fractions for the six @xmath145 intervals , @xmath148 , and differences between @xmath9 and @xmath10 decays , @xmath149 , are determined from the data ; we use the same values that were used for the @xmath18 measurement @xcite . the signal pdf is given by @xmath150 , \label{func : cp_pdf}\end{aligned}\ ] ] where @xmath0 parity @xmath151 is @xmath152 for @xmath153 and @xmath154 , and @xmath34 for @xmath155 . we assume universal @xmath0-violation parameters in eq . ( [ func : cp_pdf ] ) , i.e. @xmath156 and @xmath157 . the distribution is convolved with the proper - time interval resolution function @xmath158 @xcite , which takes into account the finite vertex resolution . we determine the following likelihood value for the @xmath159-th event : @xmath160 d(\dt ' ) + \fol p_{\rm ol}(\dt_i),\end{aligned}\ ] ] where @xmath161 is a broad gaussian function that represents an outlier component @xcite with a small fraction @xmath162 . the @xmath133 calculation is explained in the previous section . the pdf for background events , @xmath163 , is expressed as a sum of exponential and prompt components , and is convolved with @xmath164 that is a sum of two gaussians . all parameters in @xmath165 and @xmath164 are determined by a fit to the @xmath139 distribution of a background - enhanced control sample ; i.e. events outside of the @xmath101-@xmath92 signal region . we fix @xmath166 and @xmath30 to their world - average values @xcite . the only free parameters in the final fit are @xmath27 and @xmath28 , which are determined by maximizing the likelihood function @xmath167 , where the product is over all events . the fit yields @xmath168 where the first errors are statistical and the second errors are systematic . these results are consistent with the sm expectations for small penguin contributions . we define the raw asymmetry in each @xmath139 bin by @xmath169 , where @xmath170 is the number of observed candidates with @xmath171 . figure [ fig : asym ] shows the raw asymmetries in two regions of the flavor - tagging parameter @xmath145 . while the numbers of events in the two regions are similar , the effective tagging efficiency is much larger and the background dilution is smaller in the region @xmath172 . note that these projections onto the @xmath42 axis do not take into account event - by - event information ( such as the signal fraction , the wrong tag fraction and the vertex resolution ) , which are used in the unbinned maximum - likelihood fit . the sources of the systematic errors include uncertainties in the vertex reconstruction ( 0.05 for @xmath27 and 0.03 for @xmath28 ) , in the flavor tagging ( 0.04 for @xmath27 and 0.02 for @xmath28 ) , in the resolution function ( 0.05 for @xmath27 and 0.01 for @xmath28 ) , in the background fractions ( 0.04 for @xmath27 and 0.02 for @xmath28 ) , in the tag - side interference @xcite ( 0.01 for @xmath27 and 0.03 for @xmath28 ) , and in the polarization parameters ( 0.06 for @xmath27 and 0.01 for @xmath28 ) . other contributions for @xmath27 come from a possible fit bias ( 0.04 ) and from uncertainties in @xmath29 and @xmath30 ( 0.02 ) . we add each contribution in quadrature to obtain the total systematic uncertainty . we perform various cross checks . a fit to the same sample with @xmath28 fixed at zero yields @xmath173(stat ) . we check with an ensemble of mc pseudo - experiments that the fit has no sizable bias and the expected statistical errors are consistent with the measurement . we also select the following decay modes that have similar properties to the @xmath45 decay : @xmath174 , @xmath175 , @xmath176 , @xmath177 , and @xmath178 , @xmath179 , @xmath180 and @xmath181 . fits to the control samples yield @xmath182 = -0.12 \pm 0.08 $ ] , @xmath183 = + 0.02 \pm 0.05 $ ] , @xmath184 = -0.10 \pm 0.07 $ ] , and @xmath185 = -0.001 \pm 0.050 $ ] , where errors are statistical only . all results are consistent with zero . we also measure the @xmath7 meson lifetime using @xmath45 candidates as well as the control samples . all results are consistent with the present world - average values . a fit to the @xmath42 distribution of the @xmath45 without using polarization angle information yields @xmath186 , @xmath187 ; this result suggests that the @xmath0-odd component is small , supporting our polarization measurement . although the statistics are not sufficient to provide tight constraints , we also consider polarization - dependent values for @xmath27 and @xmath28 , which may arise from possible differences in the contributions of the penguin diagrams . we assume that the @xmath0 asymmetries for the @xmath0-odd component are consistent with the sm expectations , and fix @xmath188 at the world - average value of @xmath8 @xcite and @xmath189 at zero . a fit with this assumption yields @xmath190 and @xmath191 for the @xmath0-even component , also consistent with the sm expectations . in summary , we have performed measurements of the branching fraction , the polarization parameters and the @xmath0-violation parameters for @xmath45 decays . the results are @xmath192\times10^{-3 } } , \nonumber \\ { r_{\perp}}&= & { { 0.19}\pm{0.08}{\rm(stat)}\pm{0.01}{\rm(syst ) } } , \nonumber \\ { r_{0}}&= & { { 0.57}\pm{0.08}{\rm(stat)}\pm{0.02}{\rm(syst ) } } , \nonumber \\ \sdsds & = & { { -0.75}\pm { 0.56}{\rm(stat)}\pm { 0.12}{\rm(syst ) } } , \nonumber \\ \adsds & = & { { -0.26}\pm { 0.26}{\rm(stat)}\pm { 0.06}{\rm(syst)}}.\end{aligned}\ ] ] the polarization parameters and @xmath0-violation parameters are consistent with the sm expectations and theoretical predictions for small penguin contributions @xcite . we thank the kekb group for the excellent operation of the accelerator , the kek cryogenics group for the efficient operation of the solenoid , and the kek computer group and the nii for valuable computing and super - sinet network support . we acknowledge support from mext and jsps ( japan ) ; arc and dest ( australia ) ; nsfc ( contract no . 10175071 , china ) ; dst ( india ) ; the bk21 program of moehrd and the chep src program of kosef ( korea ) ; kbn ( contract no . 2p03b 01324 , poland ) ; mist ( russia ) ; mess ( slovenia ) ; nsc and moe ( taiwan ) ; and doe ( usa ) . belle collaboration , k. abe _ et al . _ , hep - ex/0408111 . babar collaboration , b. aubert _ et al . . rev . lett . * 87 * , 091801 ( 2001 ) ; phys . d * 66 * , 032003 ( 2002 ) ; phys . * 89 * , 201802 ( 2002 ) .
we present measurements of the branching fraction , the polarization parameters and @xmath0-violating asymmetries in @xmath1 decays using a 140 fb@xmath2 data sample collected at the @xmath3 resonance with the belle detector at the kekb energy - asymmetric @xmath4 collider . we obtain @xmath5\times10^{-3}},~ { r_{\perp}}= { { 0.19}\pm{0.08}{\rm(stat)}\pm{0.01}{\rm(syst)}},~ { r_{0}}= { { 0.57}\pm{0.08}{\rm(stat)}\pm{0.02}{\rm(syst)}},~ \sdsds = { { -0.75}\pm { 0.56}{\rm(stat)}\pm { 0.12}{\rm(syst)}}$ ] and @xmath6 consistency with standard model expectations is also discussed . belle preprint 2004 - 41 + kek preprint 2004 - 82 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , and @xmath7 decay , @xmath0 violation , @xmath8 11.30.er , 12.15.ff , 13.25.hw
the globular cluster ( gc ) terzan 8 is one of the faintest gcs populating the outer halo . discovered by terzan ( 1968 ) , it is located at a distance of 19.4 kpc from the galactic center in the constellation of sagittarius ( @xmath2 ; harris 1996 , 2010 edition ) . the photometric studies of this gc ( ortolani & gratton 1990 ; montegriffo et al . 1998 ) suggested that it is an old , metal - poor object with a blue horizontal branch ( hb ) . the first spectroscopic study of this cluster analysed 3 red giant stars in the caii triple region and derived a very low metallicity ( [ fe / h]=-2.3 ; da costa & armandroff 1985 ) . more recently , mottini , wallerstein & mcwilliam ( 2008 ) used high - resolution spectroscopy to determine abundances of fe , o , na , @xmath3- , fe - peak , and neutron - capture elements for 3 red giants . the small number of targets of the above studies did not allow a relevant analysis of the cluster kinematics . recently , we performed an high - resolution spectroscopic survey of this cluster ( carretta et al . 2014 ) providing accurate radial velocities for 82 stars in the central region of the cluster and abundances of 16 members . from the dynamical point of view , terzan 8 is the less concentrated gc of the milky way ( c=0.41 ) and it appears very extended ( with an half - light radius of 15.3 pc ; salinas et al . such a peculiar structure implies a half - mass relaxation time comparable with its age , suggesting that terzan 8 could be dynamically young . the interest for this cluster increased because of its probable association with the sagittarius dwarf galaxy ( ibata , gilmore & irwin 1994 ) . indeed , the location in the sky , radial velocity and distance of terzan 8 are consistent with that of this satellite galaxy ( da costa & armandroff 1995 ; dinescu et al . 2001 ; palma , majewski & johnston 2002 ; bellazzini et al . 2003 ; majewski et al . 2004 ; carraro , zinn & moni bidin 2007 ) . in particular , n - body simulations indicate that terzan 8 could be immersed in the trailing arm of the sagittarius stream ( edelsohn & elmegreen 1997 ; law & majewski 2010 ) . moreover , the presence of the sagittarius stream has been detected as an overdensity of stars in a wide region surrounding this cluster ( giuffrida et al . 2010 ; siegel et al . 2011 ) . in this paper we use the radial velocities obtained in the survey by carretta et al . ( 2014 ) to investigate the kinematics of terzan 8 . in sect . 2 we describe the observational material and the data reduction procedure . sect . 3 is devoted to the comparison of the derived velocity dispersion with a set of dynamical models . in sect . 4 we discuss the spectroscopic signature of the surrounding sagittarius stream . we summarize and discuss our results in sect . ) . in the right panel the distribution of radial velocities is shown . the prediction of the galactic model of robin et al . ( 2003 ) is shown with red histograms ( grey in the printed version of the paper).,width=328 ] spectra have been obtained using the multi - object spectrograph vlt / flames ( pasquini et al . a more detailed description of the observations and analysis is presented in carretta et al . we give here only a short description . we used the ultraviolet and visual echelle spectrograph ( uves ) 580 nm setup ( @xmath4 ; with a spectral resolution of @xmath5 ) and the giraffe high - resolution setups hr11 and hr13 , providing a spectral resolution of r@xmath624200 and 22500 , respectively . the target stars have been selected from the photometric catalog by montegriffo et al . ( 1998 ) covering a region of @xmath7 centered on the cluster . we selected seven among the brightest red giant branch ( rgb ) stars for the uves fibres ( @xmath5 ) while the giraffe fibres were allocated to objects on or near the rgb ( 84 stars ) or the hb ( 17 stars ; see fig . [ cmd ] ) . a set of @xmath8 sec exposures have been obtained . unfortunately , because of the instrumental setup , the low metallicity and s / n , we could not measure radial velocities for 2 rgb and all hb stars . the observations were obtained in service mode under seeing conditions of @xmath9 . the spectra were reduced ( bias and flat - field corrected , 1d extracted , and wavelength calibrated ) using the eso pipeline . we applied sky subtraction and division by an observed early - type star ( uves ) , or a synthetic spectrum ( giraffe ) to correct for telluric features near the [ oi ] line using the iraf routine telluric . the latter correction was applied only to the uves and bright giraffe samples . we shifted all spectra according to their heliocentric velocity and combined the individual exposures . the uves final spectra have s / n in the range 45 - 80 while the giraffe spectra have s / n values ranging from 5 to 80 . the heliocentric radial velocities and their uncertainties were then measured on the combined spectra of 82 stars using the iraf task rvidlines . uncertainties have been calculated from the r.m.s . of the velocities measured using different lines . they are listed in table 2 of carretta et al . ( 2014 ) . in fig . [ rv ] the radial velocities of the observed stars are plotted as a function of the distance from the cluster center . the signal of terzan 8 is clearly visible as a group of stars with velocity @xmath10 km / s and a small ( @xmath11 km / s ) dispersion . the galactic field population is also clearly identifiable at velocities @xmath12 km / s . we note that at @xmath13 km / s , besides the group of cluster members , other 8 stars have velocities similar to that of terzan 8 but with a larger spread . seven of them are located in the color - magnitude diagram ( cmd ) close to the cluster rgb , while one of them is clearly a bright field star at @xmath14 ( see fig . [ cmd ] ) . curiously , all these 7 stars are located in the same half of the observed field of view . all these stars were observed with giraffe . for 5 out these 7 stars we derived the abundance of fe ( from both fei and feii lines ) , mg , si , ca , ti , ba from their spectra . the other two stars have a faint magnitude ( @xmath15 ) and the low s / n ( @xmath16 ) of their spectra do not allow a reliable estimation of their abundances . we performed the analysis as described in carretta et al . temperatures have been derived using the v - i color - temperature relation by alonso , arribas & martnez - roger ( 1999 ) and the reddening e(b - v)=0.12 ( schlegel , finkbeiner & davis 1998 ; harris catalog ) . gravities have been derived using the universal gravitation law assuming a mass of 0.85 @xmath17 ( suitable for a red giant ) and a luminosity calculated converting the v band into bolometric magnitude using the corrections by alonso et al . ( 1999 ) , and assuming the apparent distance modulus of terzan 8 @xmath18 ( from the harris catalog ) , the above mentioned reddening and a solar bolometric magnitude of @xmath19 . microturbulence velocities have been calculated as a function of gravities using the relation by worley et al . we interpolated within the kurucz ( 1993 ) grid of model atmospheres ( with the option for overshooting turned on ) and tuned the abundances of the various elements to match the equivalent widths of several lines . to study the structure and kinematics of terzan 8 we derived its overall velocity dispersion , its density profile and its luminosity function ( mf ) . as a first step , we selected a sample of bona fide cluster members on the basis of their locations in the cmd and their radial velocities . in particular , we considered only those stars with * a location in the cmd within @xmath20 around the cluster rgb mean ridge line and a magnitude within @xmath21 ; * a radial velocity within 5@xmath22 from the mean systemic velocity . according to the above cirteria we defined a sample of 48 members . the average radial velocity turns out to be @xmath23 km / s ( where the reported uncertainty is the error on the mean , while r.m.s=2.46 km / s ) . for comparison , da costa & armandroff ( 1995 ) reported an average radial velocity @xmath24 from the analysis of three stars . the star - to - star comparison with our catalog gives a perfect agreement and the difference in the average velocity is probably due to their very small sample . we can not compare our values with the study of mottini et al . ( 2008 ) because they do not provide the radial velocities for their targets . the overall velocity dispersion has been calculated using the algorithm proposed by pryor & meylan ( 1993 ) , where the velocity dispersion is calculated as the quantity @xmath25 that maximizes the likelihood @xmath26}{2\pi\sigma_{v}\delta_{i}^{2 } } dv'\nonumber\\ & = & -\frac{1}{2}\sum_{i}\left(\frac{(v_{i}-\bar{v})^{2}}{\sigma_{v}^{2}+\delta_{i}^{2}}+ln[2\pi(\sigma_{v}^{2}+\delta_{i}^{2})]\right ) \label{sig_eq}\end{aligned}\ ] ] where @xmath27 and @xmath28 are the velocity of the i - th star and its associated uncertainty . the calculated velocity dispersion turns out to be @xmath29 km / s . unfortunately , because of the small number of target stars it is not possible to construct a reliable velocity dispersion profile . in fig . [ rot ] the radial velocities of the bona fide cluster members are plotted as a function of the position angle . the distribution appears homogeneous and no clear trend is apparent . a fit with a sinusoidal curve indicates a maximum rotation amplitude of @xmath30 km / s consistent with no significant rotation along the line - of - sight within terzan 8 . the density profile has been constructed from the photometric catalog by montegriffo et al . ( 1998 ) by dividing the number of stars brighter than @xmath31 and contained in circular annuli by the annulus area . we adopted the above magnitude limit to ensure a significant number of stars with a completness level @xmath190% , as estimated through artificial star experiments . the derived density profile is shown in fig . [ dens ] . the luminosity function of the cluster has been derived using the deep acs@hst photometry provided by the acs treasury project ( sarajedini et al . these data cover a region of @xmath32 around terzan 8 and provide a deep cmd in the f606w and f814w filters down to @xmath33 . a detailed description of the photometric reduction , astrometry , and artificial star experiments for this database can be found in anderson et al . we selected main sequence stars by selecting stars within 3 times the standard deviation of the median f666w - f814w color at a given f606w magnitude . the f606w luminosity function has been then calculated by counting stars in bins of magnitudes of 0.5 mag width and it is shown in fig . [ lf ] . [ cols="<,^,^,^,^,^,^,^,^,^ , > " , ] as already reported in sect . [ obs_sec ] , a group of 7 stars located close to the rgb of terzan 8 with a similar systemic velocity but a larger spread are present . these stars could be either _ i ) _ binary stars of terzan 8 , _ ii ) _ galactic field interlopers , or _ iii ) _ stars belonging to the sagittarius stream surrounding terzan 8 . we compared the observed distribution of velocities with the galactic model of robin et al . ( 2003 ) . for this purpose , a synthetic catalog covering 1 sq.deg . centered at the cluster location has been retrieved . we selected only stars which lie in the cmd brighter than @xmath34 and within @xmath20 about the rgb mean ridge line and normalized the sample to reproduce the observed number of stars at @xmath12 km / s ( where only the galactic field population is expected to be present ) . the model predicts a contamination of 3.7 stars at @xmath13 km / s . note that all the synthetic field stars in this velocity range are main sequence stars in the foreground of the cluster . unfortunately , the time coverage of our spectra does not allow to identify binaries from their velocity variation , so the only chance to distinguish them from cluster members and the field population is to use their chemistry . the derived abundances of fe , @xmath3-elements and ba for the 5 stars of this sample with sufficient s / n are listed in table 2 . in fig . [ alpha ] they are compared with the sample of field giants analysed by gratton et al . ( 2003 ) and venn et al . ( 2004 ) , terzan 8 stars by carretta et al . ( 2014 ) and sagittarius stars by carretta et al . ( 2010 ) and sbordone et al . ( 2007 ) . in spite of the large uncertainties on the derived abundances it is clear that the 5 stars have a metallicity which is not consistent with that of terzan 8 : they span a wide range @xmath35<-0.83 $ ] being significantly more metal - rich than the cluster members ( at [ fe / h]@xmath6 - 2.3 ) . the comparison with the field milky way and sagittarius stars is less clear and is hampered by the large uncertainties in the derived abundances . however , we note that while the abundances of si and ca are consistent with those of both groups , the mean ti i and mg abundance are significantly lower than those of the galactic giants while still compatible with those of sagittarius ( @xmath36 ) below the mean locus of sgr stars . given the large involved uncertainties it is not clear whether this difference is real or due to measurement errors . ] . moreover , ba appears to be enhanced by @xmath37 in 4 of them , in agreement with what observed in sgr stars ( sbordone et al . even more important is the excellent agreement between the abundance of fe measured from the neutral and ionized lines in all the analysed stars . as stated in sect . [ obs_sec ] we adopted a gravity derived assuming these objects as rgb stars at the same distance of terzan 8 . were these stars foreground milky way dwarfs their gravity would be larger by @xmath38 . according to the sensibility reported in carretta et al . ( 2014 ) this would produce a striking mismatch between the abundances of fe i and fe ii as a consequence of the wrong adopted ionization equilibrium . on the basis of the above considerations we conclude that at least 4 out these 7 stars are likely member of the sagittarius stream . we performed a detailed study of the structure and kinematics of the gc terzan 8 by means of a set of accurate radial velocities obtained through high - resolution spectroscopy in the inner part of the cluster . the cluster density profile is well fitted by very extended models with the largest half - mass radius among galactic gcs ( @xmath39 ) . by considering the estimated mass and the galactocentric distance of this gc we estimate an half mass - to - jacobi radii ratio of @xmath40 . in comparison with the other galactic gcs it situates in the upper branch of the @xmath41 diagram constituted by extended low - mass gcs ( baumgardt et al . the position of these clusters has been interpreted as a consequence of different cloud pressure and background tidal force at their formation ( elmegreen 2008 ) . it is interesting to note that many of these clusters have been associated to merging dwarf galaxies . it is not clear if terzan 8 was born with such an extended structure or this is due to its dynamical evolution , although its long relaxation time suggests the former hypothesis . in this context , if terzan 8 spent most of its evolution within a dwarf galaxy ( like the sagittarius dsph ) it experienced a mild tidal field and could therefore form and evolve with a more extended structure with respect to other gcs formed within the milky way halo . this could explain also the small fraction of na - enhanced stars found in this cluster ( see carretta et al . 2014 ) . these stars are indeed expected to form in the central part of the cluster after some @xmath42 after the cluster formation being more resistant than the first stellar generation to the tidal stripping process ( dercole et al . 2008 ; decressin et al . 2007 ) . for this reason , while in gcs the first stellar generation generally constitute only a small fraction of the cluster content , in terzan 8 it is almost entirely retained in terzan 8 because of the mild tidal stress experienced by this cluster . we found no significant radial component of rotation within the cluster , although slow rotations ( with @xmath43 ) would not be detected by our study and can not therefore be excluded . the derived overall velocity dispersion appears to be larger than what predicted by dynamical models assuming kinetic energy equipartition and a moderate amount of binaries . on the other hand , models without equipartition can reproduce the observed velocity dispersion regardless of the adopted fraction of binaries . the half - mass relaxation times of the considered models are @xmath44 , shorter than the estimated cluster age ( @xmath45=12.2 gyr ; marn - franch et al . it is therefore unlikely that collisional effects have not altered the kinematics of this gc . by means of nbody simulations we estimated only a negligible contribution of tidal effects to the cluster velocity dispersion , at least when quasi - circular orbits are considered . in principle , highly eccentric orbits could produce a more significant inflation of the velocity dispersion even in the innermost regions of the clusters . such a hypothesis requires a specific study with extensive nbody simulations . the observed discrepancy could be eliminated if a large fraction ( @xmath46 ) of binaries is present within terzan 8 . unfortunately , the time coverage of our spectra does not allow to identify binaries from their velocity variation . an estimate of the binary fraction in this cluster has been provided by milone et al . ( 2012 ) who found @xmath47 within the acs field of view . it is worth noting that the uncertainties in the adopted distribution of periods , eccentricity and mass - ratios of binaries affect the determination of the binary fraction through photometric and spectroscopic studies in different ways and can produce a mismatch between these two estimates . it is also interesting to note that the masses of the models have been calculated from the number counts on the cmd and can be therefore considered as an estimate of the luminous mass of this cluster . the large observed velocity dispersion could be viewed as an excess of dynamical mass within terzan 8 , as indicated by the large dynamical mass - to - light ratios @xmath48 . a systematic difference between the luminous and dynamical masses of many gcs have been already noticed by sollima et al . ( 2012 ) who addressed the observed difference to the large uncertainties on the mf and the retention fraction of dark remnants . also in this case , different assumptions on the dark remnants could alleviate or even eliminate the difference between the observed and predicted velocity dispersion of this gc . we detected a sample of 7 stars with radial velocities in the range @xmath49 km / s apparently clustered around the systemic velocity of terzan 8 but with a spread significantly larger than the cluster velocity dispersion . these stars are spatially segregated with respect to the other field and cluster stars occupying the same half of the observed field of view . the probability that this occurs by chance is @xmath60.8% . the abundance analysis performed on these stars indicates that they have a metallicity which is inconsistent with that of the cluster . the abundance patterns of the @xmath3-elements and ba indicate a deficiency of mg and ti i and an overabundance of ba in these stars with respect to galactic field stars and more similar to those measured in sagittarius stars . moreover , the gravities estimated for these stars are consistent with a distance similar to that of terzan 8 . the comparison with the galactic model of robin et al . ( 2003 ) indicates that it is unlikely that a significant number of galactic field stars could be present at such a large galactocentric distance . we therefore conclude that they are likely member of the sagittarius stream surrounding terzan 8 . the presence of the sagittarius stream has been already detected in some previous studies based on wide field photometric surveys ( giuffrida et al . 2010 ; siegel et al . 2011 ; salinas et al . this however represents the first spectroscopic detection of sagittarius stars around this gc , confirming its belonging to the stream of this satellite galaxy . this research has been founded by prin inaf 2011 `` multiple populations in globular clusters : their role in the galaxy assembly '' ( pi e. carretta ) and prin miur 2010 - 2011 `` the chemical and dynamical evolution of the milky way and local group galaxies '' ( pi f. matteucci ) . vd is an arc super science fellow . we thank the anonymous referee for his / her helpful comments and suggestions . alonso a. , arribas s. , martnez - roger c. , 1999 , a&as , 140 , 261 anderson j. , et al . , 2008 , aj , 135 , 2055 baumgardt h. , parmentier g. , gieles m. , vesperini e. , 2010 , mnras , 401 , 1832 bellazzini m. , ibata r. , ferraro f. r. , testa v. , 2003 , a&a , 405 , 577 carraro g. , zinn r. , moni bidin c. , 2007 , a&a , 466 , 181 carretta e. , bragaglia a. , gratton r. g. , dorazi v. , lucatello s. , sollima a. , 2014 , a&a , 561 , a87 carretta e. , et al . , 2010 , a&a , 520 , a95 da costa g. s. , armandroff t. e. , 1995 , aj , 109 , 2533 dercole a. , vesperini e. , dantona f. , mcmillan s. l. w. , recchi s. , 2008 , mnras , 391 , 825 decressin t. , meynet g. , charbonnel c. , prantzos n. , ekstrm s. , 2007 , a&a , 464 , 1029 elmegreen b. g. , 2008 , apj , 672 , 1006 dehnen w. , 2000 , apj , 536 , l39 dehnen w. , read j. i. , 2011 , epjp , 126 , 55 dinescu d. i. , majewski s. r. , girard t. m. , cudworth k. m. , 2001 , aj , 122 , 1916 dotter a. , chaboyer b. , jevremovi d. , baron e. , ferguson j. w. , sarajedini a. , anderson j. , 2007 , aj , 134 , 376 duquennoy a. , mayor m. , 1991 , a&a , 248 , 485 edelsohn d. j. , elmegreen b. g. , 1997 , mnras , 290 , 7 giuffrida g. , sbordone l. , zaggia s. , marconi g. , bonifacio p. , izzo c. , szeifert t. , buonanno r. , 2010 , a&a , 513 , a62 gratton r. g. , carretta e. , claudi r. , lucatello s. , barbieri m. , 2003 , a&a , 404 , 187 gunn j. e. , griffin r. f. , 1979 , aj , 84 , 752 harris w. e. , 1996 , aj , 112 , 1487 ibata r. a. , gilmore g. , irwin m. j. , 1994 , natur , 370 , 194 johnston k. v. , spergel d. n. , hernquist l. , 1995 , apj , 451 , 598 king i. r. , 1966 , aj , 71 , 64 kruijssen j. m. d. , 2009 , a&a , 507 , 1409 kpper a. h. w. , kroupa p. , baumgardt h. , heggie d. c. , 2010 , mnras , 407 , 2241 kurucz r. l. , 1993 , phst , 47 , 110 law d. r. , majewski s. r. , 2010 , apj , 718 , 1128 lee h. m. , nelson l. a. , 1988 , apj , 334 , 688 majewski s. r. , et al . , 2004 , aj , 128 , 245 marn - franch a. , et al . , 2009 , apj , 694 , 1498 mcconnachie a. w. , ct p. , 2010 , apj , 722 , l209 milone a. p. , et al . , 2012 , a&a , 540 , a16 montegriffo p. , bellazzini m. , ferraro f. r. , martins d. , sarajedini a. , fusi pecci f. , 1998 , mnras , 294 , 315 mottini m. , wallerstein g. , mcwilliam a. , 2008 , aj , 136 , 614 nipoti c. , londrillo p. , ciotti l. , 2002 , mnras , 332 , 901 ortolani s. , gratton r. , 1990 , a&as , 82 , 71 palma c. , majewski s. r. , johnston k. v. , 2002 , apj , 564 , 736 pasquini l. , et al . , 2002 , msngr , 110 , 1 paust n. e. q. , et al . , 2010 , aj , 139 , 476 pryor c. , meylan g. , 1993 , aspc , 50 , 357 robin a. c. , reyl c. , derrire s. , picaud s. , 2003 , a&a , 409 , 523 salinas r. , jlkov l. , carraro g. , catelan m. , amigo p. , 2012 , mnras , 421 , 960 salpeter e. e. , 1955 , apj , 121 , 161 sarajedini a. , et al . , 2007 , aj , 133 , 1658 sbordone l. , bonifacio p. , buonanno r. , marconi g. , monaco l. , zaggia s. , 2007 , a&a , 465 , 815 schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 siegel m. h. , et al . , 2011 , apj , 743 , 20 sollima a. , bellazzini m. , lee j .- w . , 2012 , apj , 755 , 156 terzan a. , 1968 , crasb , 267 , 1245 venn k. a. , irwin m. , shetrone m. d. , tout c. a. , hill v. , tolstoy e. , 2004 , aj , 128 , 1177 worley c. c. , hill v. , sobeck j. , carretta e. , 2013 , a&a , 553 , a47
we present the results of a spectroscopic survey of red giants in the globular cluster terzan 8 with the aim of studying its kinematics . we derived accurate radial velocities for 82 stars located in the innermost 7@xmath0 from the cluster center identifying 48 bona fide cluster members . the kinematics of the cluster have been compared with a set of dynamical models accounting for the effect of mass segregation and a variable fraction of binaries . the derived velocity dispersion appears to be larger than that predicted for mass - segregated stellar systems without binaries , indicating that either the cluster is dynamically young or it contains a large fraction of binaries ( @xmath130% ) . we detected 7 stars with a radial velocity compatible with the cluster systemic velocity but with chemical patterns which stray from those of both the cluster and the galactic field . these stars are likely members of the sagittarius stream surrounding this stellar system . [ firstpage ] methods : data analysis methods : statistical techniques : spectroscopic stars : kinematics and dynamics stars : population ii globular clusters : invidual : terzan 8
transition metal oxides in perovskite based structures exhibit a wide variety of phases with different electronic , magnetic , and orbital structures , and show rich functionalities such as high-@xmath9 superconductivity , colossal magnetoresistance , and multiferroics . @xcite a recent advance in epitaxial growth techniques has made it even possible to fabricate transition metal oxide heterostructures with sharp and smooth interfaces controlled at the atomic scale . @xcite in these heterostructures , many unique properties , not found in the corresponding alloy compounds made of the same composite elements , have been observed , which include e.g. , two dimensional electron gas with high mobility at the heterostructure interfaces , @xcite indicating the promising potential of oxide heterostructures for future technological applications . @xcite in the case of manganites , @xcite lamno@xmath0 is an a - type antiferromagnetic insulator and srmno@xmath6 is a g - type antiferromagnetic insulator . on one hand , the randomly - cation - doped alloy la@xmath10sr@xmath11mno@xmath6 exhibits a rich magnetic phase diagram , depending on the doping concentration @xmath12 . on the other hand , la / sr cation - ordered analogues forming superlattices behave quite differently from their alloy compounds . @xcite for example , la@xmath13sr@xmath14mno@xmath6 alloy has a mixed valence of mn@xmath15/mn@xmath16 , and the ground state is ferromagnetic half metallic due to the double exchange mechanism . @xcite to the contrary , it is found experimentally that cation - ordered ( lamno@xmath6)@xmath2/(srmno@xmath6)@xmath1 ( 001 ) superlattices are insulating when @xmath17 is larger than 3 . @xcite this change of behavior is easily understood because the number @xmath17 of srmno@xmath6 layers control the quantum confinement potential : when @xmath17 is small , the confinement potential is small and the @xmath18 electrons are distributed uniformly , thus expecting the phases similar to the alloy la@xmath10sr@xmath11mno@xmath6 . when @xmath17 is large , the confinement potential becomes large enough to trap the @xmath18 electrons in lamno@xmath6 layers , and thus the bulk properties of lamno@xmath0 and srmno@xmath6 would be observed . several theoretical studies for ( lamno@xmath6)@xmath2/(srmno@xmath6)@xmath1 superlattices have been reported to understand their electronic and magnetic properties.@xcite more recently , bhattacharya _ et al . _ @xcite have experimentally studied the transport and the magnetic properties of similar superlattices ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 grown on srtio@xmath6 ( 001 ) substrate . they have found that the ground state of these superlattices with @xmath3 are a - type antiferromagnetic metals with nel temperature ( @xmath19 ) which is higher than that observed in any alloy la@xmath10sr@xmath11mno@xmath6 compound . @xcite although the similar physical principles found in ( lamno@xmath6)@xmath2/(srmno@xmath6)@xmath1 superlattices are certainly expected to apply here , the systematic theoretical investigations are required to understand the main ingredients which determine the electronic as well as the magnetic properties of ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 superlattices . here , in this paper , performing first - principles calculations based on the density functional theory , we study the electronic and the magnetic structures of ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 ( 001 ) superlattices with @xmath3 . we show that the magnetic properties are governed not only by the quantum confinement potential caused by periodic alignment of cation ions la@xmath15/sr@xmath20 but also by the strain induced by substrates on which the superlattices are grown . namely , for the case of tensile strain induced by srtio@xmath6 ( sto ) ( 001 ) substrate , our calculations show that the ground state of these superlattices are a - type antiferromagnetic and @xmath7 orbital ordered with higher @xmath19 for @xmath21 than for @xmath22 . this is indeed in excellent agreement with recent experimental observations . @xcite instead , for the case of compressive strain induced by laalo@xmath6 ( lao ) ( 001 ) substrate , we predict c - type antiferromagnetic and @xmath8 orbital orders with higher @xmath19 for @xmath21 than for @xmath22 . the rest of this paper is organized as follows . after describing the computational details in sec . [ method ] , the numerical results for the cases of srtio@xmath6 substrate and laalo@xmath6 substrate are presented in sec . [ results_sto ] and sec . [ results_lao ] , respectively , followed by discussion of the confinement potential in sec . [ confine ] . [ summary ] summaries this paper . we perform the first - principles electronic structure calculations based on the projected augmented wave pseudopotentials using the vienna ab initio simulation package ( _ vasp _ ) . @xcite the valence states include 3p4s3d and 2s2p for mn and o , respectively . the electron interactions are described using the generalized gradient approximation ( gga ) and the rotationally invariant gga+__u _ _ method @xcite with the effective @xmath23 , i.e. , @xmath24 , from 1 ev to 5 ev for _ d _ electron states . compared to the gga , the gga+@xmath25 approach gives an improved description of _ d _ electron localization.@xcite the atomic positions of superlattices are fully optimized iteratively until the hellman - feynman forces are 0.01 ev/@xmath26 or less . the plane - wave cutoff is set to be 500 ev and a @xmath27 monkhorst - pack k - point grid is used in combination with the tetrahedron method . @xcite the supercells considered here consist of 6 mno@xmath28 layers , 2 lao layers , and 4 sro layers for both @xmath21 and 2 , as shown in fig . [ orbital ] ( a ) and fig . [ orbital2 ] ( a ) . we consider 12 and 10 different magnetic moment alignments to search for the ground state magnetic structures for lamno@xmath0/(srmno@xmath0)@xmath28 and ( lamno@xmath0)@xmath28/(srmno@xmath0)@xmath29 superlattices , as shown in fig . [ crystal - la1 ] and fig . [ crystal - la2 ] , respectively . these magnetic structures include not only simple ferromagnetic , a - type , c - type , and g - type antiferromagnetic structures @xcite but also magnetic structures with mixed combinations of these simple magnetic structures . the epitaxial constraint on these superlattices , which is grown on substrates , is to fix the in - plane lattice constants . thus , to simulate the strain effect , we fix the in - plain lattice constants ( @xmath30 ) of the superlattices to the ones of substrates , i.e. , @xmath31 for srtio@xmath6 substrate @xcite and @xmath32 for laalo@xmath6 substrate , @xcite and the lattice constant ( @xmath33 ) perpendicular to mno@xmath28 layers is fully relaxed . atomic positions are also fully optimized . let us first examine ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 ( 001 ) superlattices on srtio@xmath6 ( 001 ) substrate . our systematic gga calculations reveal that the ground states of these superlattices with @xmath21 and 2 are both a - type antiferromagnetic metals . a schematic spin alignment of a - type antiferromagnetic order is shown in fig . [ crystal - la1 ] ( g ) and fig . [ crystal - la2 ] ( a ) . indeed , as shown in fig . [ orbital ] ( d ) and fig . [ orbital2 ] ( d ) , the projected spin density distribution , calculated by integrating spin density of occupied states from fermi level down to @xmath34 ev , clearly indicates the a - type antiferromagnetic spin order . our gga+@xmath25 calculations also find that these a - type antiferromagnetic states are robust against electron correlations , and they are indeed stable up to @xmath35 ev for @xmath21 and @xmath36 ev for @xmath22 ( see fig . [ la - ti ] ) . @xcite ( color online ) . @xmath23 dependence of the relative energies ( calculated using gga+@xmath25 ) for various magnetic structures ( see fig . [ crystal - la1 ] and fig . [ crystal - la2 ] ) compared to a - type antiferromagnetic state for ( lamno@xmath0)@xmath1/(srmno@xmath0)@xmath2 with @xmath21 ( left ) and @xmath22 ( right ) on srtio@xmath6 substrate . , title="fig : " ] ( color online ) . @xmath23 dependence of the relative energies ( calculated using gga+@xmath25 ) for various magnetic structures ( see fig . [ crystal - la1 ] and fig . [ crystal - la2 ] ) compared to a - type antiferromagnetic state for ( lamno@xmath0)@xmath1/(srmno@xmath0)@xmath2 with @xmath21 ( left ) and @xmath22 ( right ) on srtio@xmath6 substrate . , title="fig : " ] since the supercell sizes and the numbers of each type of atoms are the same , we can simply compare the total energy of these two different superlattices . [ table_energy ] summarizes the total energies for the a - type antiferromagnetic states and other magnetic states . since the a - type ( c - type ) magnetic structure is ferromagnetic ( antiferromagnetic ) within the @xmath37 plane and antiferromagnetic ( ferromagnetic ) along the @xmath33 direction , we can approximately estimate an effective magnetic exchange ( @xmath38 ) simply by comparing the total energy of the a - type and the c - type antiferromagnetic states . it is clearly observed in tab . [ table_energy ] that the stabilization energy of the a - type antiferromagnetic state , i.e. , @xmath38 , is larger for @xmath21 than for @xmath22 . this implies that @xmath39 for @xmath21 is higher than that for @xmath22 . these results are in excellent agreement with experimental observations by may _ et al . _ @xcite .[table_energy ] total energies ( in unit of ev ) of ( lamno@xmath0)@xmath1/(srmno@xmath6)@xmath2 superlattices ( @xmath3 ) calculated using gga . fm , a - afm , and c - afm stand for ferromagnatic , a - type antiferromagnetic , and c - type antiferromagnetic states , respectively . [ cols="^,^,^,^,^,^,^ " , ] now , let us study the electronic and the magnetic properties of ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 superlattices ( @xmath3 ) on ( 001 ) laalo@xmath6 substrate . in the alloy manganites la@xmath10sr@xmath11mno@xmath6 , it is known that @xmath40 is a key parameter in determining the magnetic ground states . @xcite here , we demonstrate that even in these superlattices the magnetic structure can be controlled by the substrate strain which varies @xmath40 . because the in - plane lattice constant of laalo@xmath6 is much smaller than that of lamno@xmath6 ( bulk lattice parameter is 3.935@xmath41 ) , it is expected that the laalo@xmath6 substrate induces compressive strain . in fact , we find in tab . [ table_lattice ] that the lattice constant @xmath33 in the superlattices is larger than the in - plain lattice constant @xmath30 . as a result of this tetragonal distortion , mn @xmath18 orbitals are split and @xmath8 orbital is lower in energy than @xmath7 orbital , which thus induces @xmath8 orbital order . a signature of this orbital order can be seen in the projected charge density distributions shown in fig . [ orbital ] ( c ) and fig . [ orbital2 ] ( c ) . because of this orbital order , the magnetic ground state is expected to be c - type antiferromagnetic . considering @xmath42 different candidates for possible magnetic structures as shown in figs . [ crystal - la1 ] and [ crystal - la2 ] , our gga calculations find that the ground states for @xmath21 and 2 are both c - type antiferromagnetic metals [ fig . [ crystal - la1 ] ( b ) and fig . [ crystal - la2 ] ( b ) ] . this magnetic alignment can be indeed clearly seen in the projected spin density distribution as shown in fig . [ orbital ] ( e ) and fig . [ orbital2 ] ( e ) . it is also interesting to note that the lattice distortion along the @xmath33 direction is less pronounced for the case of laalo@xmath6 substrate as compared to the case of srtio@xmath6 substrate . as shown in fig . [ angle ] , mn - o - mn angles between the nearest layers along the @xmath33 direction for the superlattices on laalo@xmath6 substrate is almost @xmath43 , which certainly favors the ferromagnetic double exchange along this direction . we also find that the c - type magnetic structure is robust against electron correlations in mn @xmath44 orbitals up to @xmath45 ev for @xmath21 and @xmath46 ev for @xmath22 ( see fig . [ la - al ] ) . @xcite direction for the relaxed crystal structures ( calculated using gga ) for ( lamno@xmath0)@xmath1/(srmno@xmath0)@xmath2 ( 001 ) superlattices grown on different substrates indicated in the figure . the layer positions in the horizontal axis are indicated in fig . [ orbital ] ( a ) and fig . [ orbital2 ] ( a ) . ] ( color online ) . @xmath23 dependence of the relative energies ( calculated using gga+@xmath25 ) for various magnetic structures ( see fig . [ crystal - la1 ] and fig . [ crystal - la2 ] ) compared to a - type antiferromagnetic state for ( lamno@xmath0)@xmath1/(srmno@xmath0)@xmath2 with @xmath21 ( left ) and @xmath22 ( right ) on laalo@xmath6 substrate . , title="fig : " ] ( color online ) . @xmath23 dependence of the relative energies ( calculated using gga+@xmath25 ) for various magnetic structures ( see fig . [ crystal - la1 ] and fig . [ crystal - la2 ] ) compared to a - type antiferromagnetic state for ( lamno@xmath0)@xmath1/(srmno@xmath0)@xmath2 with @xmath21 ( left ) and @xmath22 ( right ) on laalo@xmath6 substrate . , title="fig : " ] as in the case of srtio@xmath6 substrate , we can discuss the nel temperature @xmath39 for the c - type antiferromagnetic order by calculating the total energy , and the results are summarized in tab . [ table_energy ] . simply by comparing the total energies of the c - type and the a - type antiferromagnetic states , the difference of which gives a rough estimate of an effective magnetic exchange @xmath38 , we find that the stabilization energy of the c - type antiferromagnetic state , i.e. , @xmath38 , is larger for @xmath21 than for @xmath22 . this implies that @xmath39 for @xmath21 is higher than that for @xmath22 . since ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 superlattices ( @xmath3 ) on ( 001 ) laalo@xmath6 substrate have not been studied experimentally , these results provide the theoretical prediction which should be tested experimentally in the future . finally , let us briefly discuss why the magnetic and orbital ground states found here are spatially uniform , in spite of apparent periodic potential modulation caused by different ionic charges , i. e. , la@xmath15 in lamno@xmath6 layers , and sr@xmath20 in srmno@xmath6 . as reported in ref . , one way to estimate the effective potential modulation is to evaluate the oxygen 1@xmath47 core energy level . the results for ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 superlattices with @xmath21 and 2 are shown in fig . [ potential ] . from these figures , we see that ( i ) the potentials are almost the same for both substrates , and ( ii ) as is expected , the confinement potential becomes larger with @xmath17 . the calculated charge density shows that @xmath18 electrons in lamno@xmath6 layers is @xmath480.2 ( 0.1 ) more than that in srmno@xmath6 layers for @xmath22 ( @xmath49 . this suggests that the thickness is still thin enough not to confine @xmath18 electrons in lamno@xmath6 layers . however , we naturally expect that the bulk properties may recover far away from interface when @xmath17 is increased further and a metal - insulator transition should eventually occur . it is also found that the confinement potential can be more easily estimated simply by calculating madelung potential . as shown in fig . [ madelung ] , madelung potential can indeed semi - qualitatively reproduce the values estimated from oxygen 1@xmath47 core energy level . this finding should be very useful in estimating the confinement potential for more complex superlattices in which first - principles electronic structure calculations are computationally expensive . the variations of the relative oxygen _ 1s _ core energy ( calculated using gga ) in each mno@xmath28 layer of ( a ) lamno@xmath0/(srmno@xmath0)@xmath28 and ( b ) ( lamno@xmath0)@xmath28/(srmno@xmath0)@xmath29 superlattices . results for srtio@xmath6 and laalo@xmath6 substrates are indicated by black circles and red squares , respectively . the layer positions are indicated in fig . [ orbital ] ( a ) and fig . [ orbital2 ] ( a ) . ] the variation of madelung potential for mn ions in each mno@xmath28 layer of ( a ) lamno@xmath0/(srmno@xmath0)@xmath28 and ( b ) ( lamno@xmath0)@xmath28/(srmno@xmath0)@xmath29 superlattices . results for srtio@xmath6 and laalo@xmath6 substrates are indicated by black circles and red squares , respectively . here , the ideal crystal structures with no distortion , and the ideal mn valency ( indicated in the figures ) with o@xmath50 , la@xmath15 , and sr@xmath20 are assumed . the layer positions are indicated in fig . [ orbital ] ( a ) and fig . [ orbital2 ] ( a ) . note that the sign convention of madelung potential used here is that electrons prefer to locate in lamno@xmath0 layers . ] using first - principles calculations based on the density functional theory , we have studied the effects of epitaxial strain on the magnetic ground states in ( lamno@xmath6)@xmath1/(srmno@xmath6)@xmath2 ( 001 ) superlattices with @xmath3 . our results clearly demonstrate that as in alloy manganites , even in superlattices , the epitaxial strain induced by substrates enforces tetragonal distortion , which in turn governs the ground state magnetic structure via the inherent orbital ordering . we have found that for the tensile strain induced by srtio@xmath6 ( 001 ) substrate , the ground state is a - type antiferromagnetic metal with @xmath7 orbital order . the approximate estimation of an effective magnetic exchange suggests that the nel temperature @xmath39 of the a - type antiferromagnetic order is higher for @xmath21 than that for @xmath22 . these results are in excellent agreement with experimental observations . @xcite furthermore , we have predicted that for the compressive strain induced by laalo@xmath6 ( 001 ) substrate , the ground state is c - type antiferromagnetic metal with @xmath8 orbital order with higher nel temperature @xmath39 for @xmath21 than that for @xmath22 . these predictions should be confirmed experimentally in the future . the computation is done using the riken integrated cluster of clusters ( ricc ) . q. z. was supported by the natural science foundation of jiangsu province ( bk2012248 ) , and research fund of key laboratory for advanced technology in environmental protection of jiangsu province ( ae201152 ) . s. d. was supported by the 973 projects of china ( 2011cb922101 ) , nsfc ( 11004027 ) , and ncet ( 10 - 0325 ) . b. w. was supported by nsfc ( 11174242 ) . c. h. ahn , a. bhattacharya , m. di ventra , j. n. eckstein , c. d. frisbie , m. e. gershenson , a. m. goldman , i. h. inoue , j. mannhart , a. j. millis , a. f. morpurgo , d. natelson , and j. m. triscone , rev . phys . * 78 * , 1185 ( 2006 ) s. j. may , p. j. ryan , j. l. robertson , j. kim , t. s. santos , e. karapetrova , j. l. zarestky , x. zhai , s. g. e. te velthuis , j. n. eckstein , s. d. bader , and a. bhattacharya , nature mater . * 8 * , 892 ( 2009 ) .
using first - principles calculations based on the density functional theory , we show a strong strain dependence of magnetic order in ( lamno@xmath0)@xmath1/(srmno@xmath0)@xmath2 ( 001 ) superlattices with @xmath3 . the epitaxial strain lifts the degeneracy of mn @xmath4 orbitals , thus inducing an inherent orbital order , which in turn strongly affects the ferromagnetic double exchange of itinerant @xmath4 electrons , competing with the antiferromagnetic superexchange of localized @xmath5 electrons . for the case of tensile strain induced by srtio@xmath6 ( 001 ) substrate , we find that the ground state is a - type antiferromagnetic and @xmath7 orbital ordered , which is in excellent agreement with recent experiments [ s. j. may _ et al . _ , nature materials * 8 * , 892 ( 2009 ) ] . instead , for the case of compressive strain induced by laalo@xmath6 ( 001 ) substrate , we predict that the ground state is c - type antiferromagnetic and @xmath8 orbital ordered .
the lines show the simulated xp spectrum from the band structure results . calculated partial density of states of ( b ) zr 4@xmath7 @xmath1 ( dashed line ) & @xmath8 ( solid line ) symmetries , and ( c ) b 2@xmath9 ( solid line ) and 2@xmath6 ( dashed line ) states . ] valence band spectra obtained using he ii and al @xmath10 excitation energies are shown in fig . 2(a ) exhibiting three distinct features marked by a , b and c beyond 2 ev binding energies . the intensities between 6 - 10 ev binding energies represented by b are prominent in the he ii spectrum , while the relative intensity between 3 - 4 ev binding energies represented by a is enhanced in the @xmath11-ray photoemission ( xp ) spectrum . such change in intensity in the angle integrated spectra may be attributed to the matrix elements associated to different constituent states forming the eigenstates of the valence band , which is a sensitive function of the photon energy @xcite . therefore , the photoemission cross section will vary with the photon energy that can be used to identify the orbital character of the energy bands . considering this feature of the technique , the spectral feature , b can be attributed to dominant b 2@xmath6 orbital character and the feature a to zr 4@xmath7 orbital character . energy band structure of zrb@xmath0 has been calculated within the local density approximations ( lda ) . the calculated partial density of states ( pdos ) corresponding to zr 4@xmath7 & 5@xmath9 , and b 2@xmath9 & 2@xmath6 states are shown in fig . 2(b ) and 2(c ) , respectively . zr 5@xmath9 contribution appears in the energy range of 5 - 7.5 ev above the fermi level with negligible contributions in the energy window studied here as shown by thick solid line in fig . b 2@xmath9 contributions appear beyond 6 ev binding energies . dominant contribution from b 2@xmath6 pdos appears between 1.5 to 10 ev binding energies . the center of mass of the b 2@xmath6 pdos ( considering both occupied and unoccupied parts ) appears around 1.9 ev binding energy . the center of mass of the entire zr 4@xmath7 pdos appear around 2.8 ev above @xmath12 , which is about 4.7 ev above the center of mass of the b 2@xmath6 pdos . thus , an estimate of the charge transfer energy could be about 4.7 ev in this system . the ground state is metallic with flat density of states across @xmath12 arising due to highly dispersive energy bands in this energy range . the density of states at @xmath12 possess close to 2:1 intensity ratio of the b 2@xmath6 and zr 4@xmath7 orbital character . contributions and that on the right panel shows @xmath8 contributions . ] the electronic states with @xmath1 and @xmath8 symmetries , shown in fig . 2(b ) , exhibit almost overlapping center of mass suggesting negligible crystal field splitting ( crystal field splitting @xmath13 100 mev ) . the energy bands are shown in fig . 3 . @xmath8 bands are relatively narrow and exhibit a gap at @xmath12 . the intensity at @xmath12 essentially have highly dispersive @xmath1 symmetry . using the above band structure results , we calculated the xp spectrum considering the photoemission cross section of the constituent partial density of states as discussed later in the method section . the calculated spectrum shown by solid line in fig . 2(a ) exhibits an excellent representation of the experimental xp spectrum . it is clear that the feature , c in the energy range 10 - 14 ev corresponds to photoemission signal from b 2@xmath9 levels . the features , a and b arise due to hybridized b 2@xmath6 - zr 4@xmath7 states . the relative contribution from zr 4@xmath7 states is higher for the feature a. and ( b ) b 2@xmath6 pdos obtained from lda and lda+@xmath14 calculations ( @xmath15 = 6 ev ) . ( c ) zr 4@xmath7 and ( d ) b 2@xmath6 pdos obtained @xmath15 = 6 ev and @xmath16 = 4 ev . ] in order to probe the electron correlation induced effect on the electronic structure , we have carried out electron density of states calculation using finite electron correlation among the zr 4@xmath7 electrons , @xmath15 as well as that among b 2@xmath6 electrons , @xmath16 . in figs . 4(a ) and 4(b ) , we show the calculated results for zr 4@xmath7 and b 2@xmath6 pdos with @xmath15 = 0 ( solid line ) and 6 ev ( dashed line ) for the zr 4@xmath7 electrons . even the large value of @xmath15 of 6 ev does not have significant influence on the occupied part of the electronic structure , while an energy shift is observed in the unoccupied part . such scenario is not unusual considering the large radial extension of 4@xmath7 orbitals and poor occupancy @xcite . the inclusion of an electron correlation , @xmath16 of upto 4 ev for b 2@xmath6 states exhibits insignificant change in the results . considering zr being a heavy element , we carried out similar calculations including spin - orbit interactions and did not find significant modification in the results - we have not shown these later results to maintain clarity in the figure . all these theoretical results suggest that consideration of electron correlation within the lsda+@xmath14 method has insignificant influence in the occupied part of the electronic structure of this system . we now investigate the spectral changes near @xmath12 with high energy resolution in fig . the spectral density of states ( sdos ) are calculated by symmetrizing the experimental spectra ; @xmath17 and shown in figs . 5(c ) and 5(d ) . such an estimation of sdos is sensitive to the definition of the fermi level , which is carefully derived at each temperature by the fermi cutoff in the valence band spectra for silver mounted on the sample holder in electrical connection with the sample and measured using identical experimental conditions . the other effect is the asymmetry across the fermi level . this can be addressed by obtaining the sdos by the division of the resolution broadened fermi - dirac distribution function . the results are shown in figs . 5(e ) and 5(f ) for he ii and he i spectra , respectively . clearly , the sdos obtained following the later procedure exhibits symmetric sdos across @xmath12 . the 30 k and 10 k data exhibit large noise above @xmath12 due to the division of negligibly small intensities by small numbers arising from fermi - dirac function at these temperatures . these results demonstrate the reliability of sdos extraction procedures and also indicate signature of particle - hole symmetry in this system . , ( b ) @xmath18 , and ( c ) @xmath19 . spectral density of states from ( d ) he i and ( e ) he ii spectra . the lines represent the spectral functions simulated to fit the experimental spectra . ] the spectral lineshape near the fermi level often provides important information about the thermodynamic properties of the systems . for example , a lineshape dependence of @xmath20 corresponds to fermi liquid behavior @xcite - absence of linearity of sdos with @xmath20 shown in fig . 6(a ) indicates a deviation from such a behavior in the present case . the decrease in intensity at @xmath12 with the decrease in temperature may arise due to the disorder induced localization of the electronic states . altshuler & aronov showed that the charge disorder in a correlated system leads to @xmath21 dependence of the density of states near fermi level @xcite , which has been proved employing photoemission spectroscopy @xcite . in the present case , a plot of sdos with @xmath22 shown in fig . 6(b ) exhibits deviation from linearity . interestingly , the curvatures at the exponents of 0.5 and 2 are opposite indicating an intermediate exponent for the present system . the simulation of sdos as a function of @xmath23 with @xmath24 exhibits linear dependence with energy for all the temperatures studied ( see fig . 6(c ) ) . such a behavior in a conventional superconductor is curious . most interestingly , in fig . 6(c ) , all the data below 150 k superimpose exactly and correspond to the same energy scale , while the data at 300 k exhibit similar energy dependence with a different slope . these results suggest signature of an incipient phase transition at an intermediate temperature . the sdos at @xmath12 is intense and flat at 300 k with large dispersion of the bands and correspond to a good metallic phase consistent with the band structure results . the decrease in temperature introduces a dip at @xmath12 . the dip gradually becomes more prominent with the decrease in temperature . the same scenario is observed in both , he i and he ii spectra indicating this behavior to be independent of the photon energy used . this dip may be attributed to a pseudogap appearing at low temperatures @xcite . although the probed lowest temperature is slightly above @xmath2 , the dip grows monotonically with the decrease in temperature - a large suppression of the spectral weight at the fermi level is observed at 10 k indicating relation of this dip to the superconducting gap . in order to estimate the energy gap in these spectral functions , we used a phenomenological self - energy , @xmath25 following the literature @xcite . @xmath26\ ] ] where @xmath27 represents the gap size . the first term is the energy independent single particle scattering rate and the second term is the bcs self energy . @xmath28 represents the inverse electron pair lifetime . the spectral functions , the imaginary part of the green s functions , were calculated as follows . @xmath29 $ ] , where @xmath30 and @xmath31 are the real and imaginary part of the self energy . we find that a finite value of @xmath28 is necessary to achieve reasonable representation of the experimental data . the simulated and experimental spectral density of states are shown by superimposing them in figs . 6(d ) and 6(e ) for the he i and he ii spectra , respectively . evidently , the simulated spectra provide a remarkable representation of the experimental spectra . the experimental and calculated band structure results suggest significant hybridization between zr 4@xmath7 and b 2@xmath6 states leading to similar energy distribution of the pdos . here , the eigenstates are primarily constituted by the linear combination of the zr 4@xmath7 and b 2@xmath6 states . the antibonding bands possess large zr 4@xmath7 orbital character and appear in the unoccupied part of the electronic structure @xcite . the energy bands below @xmath12 forming the valence band are the bonding eigenstates consisting of dominant 2@xmath6 orbital character . the covalency between these states could be attributed to the large radial extension of the 4@xmath7 orbitals overlapping strongly with the neighboring states as also observed in other 4@xmath7 systems @xcite . the calculated results within the local density approximation are consistent with the experimental spectra - subtle differences between the experiment and theory could not be captured via inclusion of electron correlation and/or consideration of spin - orbit coupling in these lda calculations . the band structure results exhibit two important features - ( i ) the energy bands with @xmath1 symmetry are highly dispersive compared to @xmath8 bands and ( ii ) the crystal field splitting is negligible . as shown in fig . 1 , the b@xmath0 dodecahedrons form an octahedra around zr sites with the center of mass of the b@xmath0 units at the edge centers of the unit cell . the size of the dodecahedrons are quite large with no boron situating on the unit cell edge . thus , the overlap of b 2@xmath6 orbitals with the @xmath1 orbitals is significantly enhanced due to the proximity of borons along the face diagonals , while that with the @xmath8 orbitals is reduced leading to a comparable crystal field potential on both the orbitals . the larger hybridization with the @xmath1 orbitals are also manifested by the large dispersion of the @xmath1 bands spanning across the fermi level while @xmath8 bands are relatively more localized . from these calculations , the effective valency of zr was found to vary between ( + 2.4 ) to ( + 2.5 ) depending on the electron interaction parameters such as correlation strength , spin orbit coupling _ etc_. bands crossing the fermi level are shown by lines . the symbols represent the first derivative of the same energy bands . the arrows show the fermi level crossings . ] the fermi level is located almost at the inflection points of each of the energy band crossing @xmath12 as shown in fig . this is verified by investigating the first derivative of the energy bands shown by the symbols in the figure . the arrows in the figure indicate fermi level crossing and corresponding point in the first derivative curves . evidently , the band crossing correspond to an extremum in the derivative plot . the inflection point is the point , where the curve changes its curvature that corresponds to a change in the character of the conduction electrons ( the fermi surface ) from hole - like to electron - like behavior . this behavior is termed as lifschitz transition @xcite . thus , this material appear to be lying in the proximity of the lifschitz transition as also observed in various exotic unconventional superconductors such as fe - pnictides @xcite . due to the lack of cleavage plane and hardness of the sample , the fermi surface mapping of this sample could not be carried out to verify this theoretical prediction experimentally . it is well known that the band structure calculations captures the features in the electronic structure well in the weakly correlated systems as also found in the present case suggesting possibility of such interesting phenomena in this system . we hope , future studies would help to enlighten this issue further . , inverse pair lifetime , @xmath28 and single particle scattering rate , @xmath32 as a function of temperature . the open and close symbols correspond to the fitting of he i and he ii data , respectively . ] the high resolution spectra close to the fermi level indeed exhibit deviation from fermi liquid behavior - a suggestive of quantum instability in this system as expected from the above results . r. lortz _ et al . _ @xcite showed that the resistivity of zrb@xmath0 exhibits linear temperature dependence in a large temperature range of 50k - 300k . the resistivity below 50k of the normal phase is complex and does not show @xmath33-dependence . both , the resistivity and specific heat data exhibit signature of anharmonic mode of zr - vibrations . all these observations indicate complexity of the electronic properties , different from a typical fermi - liquid system consistent with the photoemission results . a dip appears at the fermi level at temperatures much higher than the superconducting transition temperature reminiscent of a pseudogap feature . in order to investigate this further , we study the fitting parameters simulating the spectral functions in fig . the energy gap , @xmath27 is found to be about 7.3 mev and remains almost unchanged in the whole temperature range studied . this gap is significantly larger than the value predicted ( 1 - 2 mev ) considering bcs behavior . the magnitude of @xmath34 , which is a measure of single particle scattering rate , is quite large and found to decrease with the decrease in temperature . on the other hand , @xmath35 , the inverse pair lifetime is found to be significant and smaller than 2@xmath27 suggesting proximity to a bcs limit for this system . @xmath28 exhibits temperature dependence quite similar to that of @xmath32 . the decrease in @xmath28 manifests gradual enhancement of the electron pair lifetime with the decrease in temperature that eventually leads to the superconducting phase upon attaining coherence among the pairs . thus , the temperature in the vicinity of 70 k exhibiting the onset of pair formation could be the critical temperature , @xmath36 in this system . the observation of the signature of an energy gap and finite electron pair lifetime above @xmath2 is quite similar to that observed for under - doped cuprates@xcite and is an indication of precursor effect in this system . clearly , the electronic structure of this system is complex , appears to be at the crossover of bcs and unconventional behaviors and further studies are required to understand the origin of pseudogap & unusual spectral lineshape . in summary , we studied the electronic structure of zrb@xmath0 , predicted to be a complex conventional superconductor employing high resolution photoemission spectroscopy and _ ab initio _ band structure calculations . the valence band spectra exhibit multiple features with dominant b 2@xmath6 orbital character close to the fermi level . the experimental results could be captured reasonably well within the local density approximations . the electronic states close to fermi level have @xmath1 symmetry and the filling of the bands is in the proximity of lifschitz transition . the spectral lineshape near the fermi level in the high resolution spectra exhibit deviation from typical fermi liquid behavior . a dip at the fermi level emerges above the superconducting transition temperature and gradually becomes prominent at lower temperatures . the analysis of the spectral functions within the phenomenological descriptions suggests finite electron pair lifetime above @xmath2 as a signature of a precursor effect to the superconducting transition in this system . single crystals of zrb@xmath0 were grown by floating zone technique in a crystal systems incorporated ( csi ) four - mirror infrared image furnace in flowing high purity ar gas at a pressure of 2 bar @xcite . the quality of the crystal was confirmed and orientation was determined from @xmath11-ray laue - diffraction images . the magnetization measurements using a qunatum design mpms magnetometer show a sharp superconducting transition temperature , @xmath37 of 6.1 k @xcite . the photoemission measurements were performed using a gammadata scienta r4000 wal electron analyzer and monochromatic laboratory photon sources . @xmath38-ray photoemission ( xp ) and ultraviolet photoemission ( up ) spectroscopic measurements were carried out using al @xmath10 ( 1486.6 ev ) , he ii ( 40.8 ev ) and he i ( 21.2 ev ) photon energies with the energy resolution set to 400 mev , 4 mev and 2 mev , respectively . the melt grown sample was very hard and possesses no cleavage plane . therefore , the sample surface was cleaned using two independent methods ; top - post fracturing and scraping by a small grain diamond file at a vacuum better than 3@xmath3910@xmath40 torr . both the procedures results to similar spectra with no trace of impurity related signal in the photoemission measurements . the measurements were carried out in analyzer transmission mode at an acceptance angle of 30@xmath41 and reproducibility of all the spectra was ensured after each surface cleaning cycle . the temperature variation down to 10 k was achieved by an open cycle he cryostat from advanced research systems , usa . the energy band structure of zrb@xmath0 was calculated using full potential linearized augmented plane wave method within the local density approximation ( lda ) using wien 2k software @xcite . the energy convergence was achieved using 512 @xmath42-points within the first brillouin zone . the lattice constant of 7.4075 was used considering the unit cell shown in fig . 1 @xcite . in order to introduce electron - electron coulomb repulsion into the calculation , we have employed lda+@xmath14 method with an effective electron interaction strength , @xmath43 ( @xmath44 ; @xmath45 = hund s exchange integral ) setting @xmath45 = 0 following anisimov et al.@xcite . consideration of finite @xmath45 did not have significant influence on the results . we have carried out the calculations for various values of electron correlation upto 6 ev for zr 4@xmath7 electrons and 4 ev for b 2@xmath6 electrons . spin - orbit interactions among the zr 4@xmath7 electrons are also considered for the calculations . to calculate the @xmath11-ray photoemission spectrum , we have multiplied the partial density of states obtained by the band structure calculations by the corresponding photoemission cross sections for al @xmath10 energy . these cross - section weighted pdos is convoluted by two lorentzians with the energy dependent full width at half maxima ( fwhm ) for the hole & electron lifetime broadenings and a gaussian with fwhm representing the resolution broadening . the sum of these spectral functions provides the representation of the experimental valence band spectrum . the authors , k. m. and n. s. acknowledge financial support from the department of science and technology under the swarnajayanti fellowship programme . one of the author , g. b. wishes to acknowledge financial support from epsrc , uk ( ep / i007210/1 ) . vishik , i. m. et al . arpes studies of cuprate fermiology : superconductivity , pseudogap and quasiparticle dynamics . _ new j. phys . _ * 12 * , 105008 ( 2010 ) . timusk , t. & statt , b. the pseudogap in high - temperature superconductors : an experimental survey . phys . _ * 62 * , 61 - 122 ( 1999 ) . hfner , s. , hossain , m. a. , damascelli , a. & sawatzky , g. a. two gaps make a high - temperature superconductor ? phys . _ * 71 * , 062501 ( 2008 ) . littlewood , p. & kos . focus on the fermi surface . _ nature _ * 438 * , 435 ( 2005 ) mannella , n. et . nodal quasiparticle in pseudogapped colossal magnetoresistive manganites . _ nature _ * 438 * , 474 - 478 ( 2005 ) . maiti , k. , medicherla , v. r. r. , patil , s. & singh , r. s. revelation of the role of impurities and conduction electron density in the high resolution photoemission study of ferromagnetic hexaborides . b _ * 99 * , 266401 ( 2007 ) . matthias , b. t. et al superconductivity and antiferromagnetism in boron - rich lattices . _ science _ * 159 * , 530 ( 1968 ) . lortz , r. et al . specific heat , magnetic susceptibility , resistivity and thermal expansion of the superconductor zrb@xmath0 . b _ * 72 * , 024547 ( 2005 ) . wang y. , et al specific heat and magnetization of a zrb@xmath0 single crystal : characterization of a type - ii/1 superconductor . b _ * 72 * , 024548 ( 2005 ) . daghero , d. et al andreev - reflection spectroscopy in zrb@xmath0 single crystals . _ supercond . technol . _ * 17 * , s250-s254 ( 2004 ) . gasparov , v. a. , sidorov , n. s. & zverova , i. i. two - gap superconductivity in zrb@xmath0 : temperature dependence of critical magnetic fields in single crystals . b _ * 73 * , 094510 ( 2006 ) . biswas , p. k. studies of unconventional superconductors . _ d. thesis _ ( university of warwick , warwick , 2012 ) . thakur , s. et al . surface bulk differences in a conventional superconductor , zrb@xmath0 . _ j. appl . phys . _ * 114 * , 053904 ( 2013 ) . tsindlekht , m. i. et al . tunneling and magnetic characteristics of superconducting zrb@xmath0 single crystals . b _ * 69 * , 212508 ( 2004 ) . khasanov , r. et al . anomalous electron - phonon coupling probed on the surface of superconductor zrb@xmath0 . b _ * 72 * , 224509 ( 2005 ) . leviev , g. i. low - frequency response in the surface superconducting state of single - crystal zrb@xmath0 . b _ * 71 * , 064506 ( 2005 ) . yeh , j. j. & lindau , i. atomic subsheell photoionization cross sections and asymmetry parameters : 1@xmath46 103 . data and nucl . data tables _ * 32 * , 1 ( 1985 ) . maiti , k , mahadevan , p. & sarma , d. d. evolution of electronic structure with dimensionality in divalent nickelates . b _ * 59 * , 12457 ( 1999 ) . maiti , k. & sarma , d. d. electronic structure of y@xmath47ca@xmath48banio@xmath49 from photoemission and inverse photoemission . b _ * 58 * , 9746 ( 1998 ) . maiti , k. & singh , r. s. evidence against strong correlation in 4@xmath7 transition - metal oxides caruo@xmath50 and srruo@xmath50 . b _ * 71 * , 161102(r ) ( 2005 ) . efros , a. f. & shklovskii , b. i. coulomb gap and low temperature conductivity of disordered systems . _ j. phys . c : solid state phys . _ * 8 * , l49-l51 ( 1975 ) . massey , j. g. & lee , m. direct observation of the coulomb correlation gap in a nonmetallic semiconductor , si : b. _ phys . lett . _ * 75 * , 4266 - 4269 ( 1995 ) . altshuler , b. l. & aronov , a. g. zero bias anomaly in tunnel resistance and electron - electron interaction . _ solid state commun . _ * 30 * , 115 - 117 ( 1979 ) . sarma d. d. et al . disorder effects in electronic structure of substituted transition metal compounds . _ lett . _ * 80 * , 4004 - 4007 ( 1998 ) . kobayashi , m. , tanaka , k. , fujimori , a. , ray , s. & sarma , d. d. critical test for altshuler - aronov theory : evolution of the density of states singularity in double perovskite sr@xmath51femoo@xmath52 with controlled disorder . lett . _ * 98 * , 246401 ( 2007 ) . medicherla , v. r. r. , patil , s. , singh , r. s. & maiti , k. origin of ground state anomaly in lab@xmath4 at low temperatures . _ * 90 * , 062507 ( 2007 ) . norman , m. r. , randeria , m. , ding , h. & campuzano , j. c. phenomenology of the low - energy spectral function in high - t@xmath53 superconductors . b _ * 57 * , r11093-r11097 ( 1998 ) . shein , i. & ivanovskii , a. band structure of superconducting dodecaborides yb@xmath0 and zrb@xmath0 . . solid state _ , * 45 * 1429 - 1434 , ( 2003 ) . teyssier , j. et al . optical study of electronic structure and electron - phonon coupling in zrb@xmath0 _ phys . b _ , * 75 * 134503 ( 2007 ) . maiti , k. role of covalency in the ground - state properties of perovskite ruthenates : a first - principles study using local spin density approximations . b _ * 73 * , 235110 ( 2006 ) . maiti , k. electronic structure of bairo@xmath50 : a first - principles study using the local spin density approximation . _ phys . rev . b _ * 73 * , 115119 ( 2006 ) . singh , r. s. & maiti , k. manifestation of screening effects and a - o covalency in the core level spectra of a site elements in the abo@xmath50 structure of ca@xmath54sr@xmath48ruo@xmath50 . b _ * 76 * , 085102 ( 2007 ) . lifshitz , i. m. anomalies of electron characteristics of a metal in the high pressure region . . jetp _ * 11 * , 1130 - 1135 ( 1960 ) . liu , c. et al . evidence for a lifshitz transition in electron - doped iron arsenic superconductors at the onset of superconductivity . * 6 * , 419 - 423 ( 2010 ) . balakrishnan , g. , lees , m. r. & paul , d. m. k. growth of large single crystals of rare earth hexaborides . _ j. crystal growth _ * 256 * , 206 - 209 ( 2003 ) . blaha , p. , schwarz , k. , madsen , g. k. h. , kvasnicka , d. & luitz , j. _ wien2k an augmented plane wave + local orbitals program for calculating crystal properties . _ schwarz , k. ( ed . ) universitt , wien , 2001 ) paderno , yu . b. , liashchenko , a. b. , filippov , v. b. & dukhnenko , a. v. _ advantages and challenges , science for materials in the frontier of centuries : _ skorokhod , v. v. ( ed . ) 347 - 348 ( kiev : ipms , 2002 ) . anisimov , v. i. , solovyev , i. v. , korotin , m. a. , czyzyk , m. t. & sawatzky , g. a. density - functional theory and nio photoemission spectra . b _ * 48 * , 16929 - 16934 ( 1993 ) .
we investigate the electronic structure of a complex conventional superconductor , zrb@xmath0 employing high resolution photoemission spectroscopy and _ ab initio _ band structure calculations . the experimental valence band spectra could be described reasonably well within the local density approximation . energy bands close to the fermi level possess @xmath1 symmetry and the fermi level is found to be in the proximity of quantum fluctuation regime . the spectral lineshape in the high resolution spectra is complex exhibiting signature of a deviation from fermi liquid behavior . a dip at the fermi level emerges above the superconducting transition temperature that gradually grows with the decrease in temperature . the spectral simulation of the dip and spectral lineshape based on a phenomenological self energy suggests finite electron pair lifetime and a pseudogap above the superconducting transition temperature . the relationship between pseudogap phase and superconductivity in high temperature superconductors is an issue of discussion for many years now @xcite . while one school believes that the pseudogap arises due to some hidden order and/or effects not associated to superconductivity , the other school attributes pseudogap to the electron pair formation above the transition temperature , @xmath2 as a precursor to the superconducting gap that leads to superconductivity upon establishment of the coherence among the electron pairs at @xmath2 . independent of these differences , it is realized that electron correlation plays a significant role in the occurrence of pseudogap phase in these unconventional superconductors @xcite . several questions are being asked on whether the pseudogap phase is specific to unconventional superconductors , electron correlation or low dimensionality is a necessity for such phase _ etc_. . ( b ) b@xmath0 units are contracted to visualize the isolated b@xmath0 clusters with better clarity . ] apart from these issues , the quest of new superconducting compounds resulted into the discovery of new materials such as cubic hexane , @xmath3b@xmath4 @xcite and dodecaborides , @xmath3b@xmath0 ( @xmath3 = sc . y , zr , la , lu , th ) those attracted much attention due to their interesting electronic properties @xcite . zrb@xmath0 is one such compound exhibiting relatively high superconducting transition temperature of 6 k in the mb@xmath0 family @xcite . it forms in cubic structure as shown in fig . 1 . in fig . 1(a ) , the real lattice positions are shown with the correct bond lengths and atom positions . although , the structure appears to be complex , it is essentially a rock salt structure constituted by two interpenetrated fcc lattices formed by zr and b@xmath0 units as demonstrated in fig . 1(b ) by compressing the b@xmath0 units for clarity . various transport and magnetic measurements suggest a bardeen - cooper - schriefer ( bcs ) type superconductivity in this material , which is termed as conventional superconductivity . since , zr atoms are located within the huge octahedral void space formed by the b@xmath0 units , the superconductivity in these materials could conveniently be explained by the electron - phonon coupling mediated phenomena involving primarily low energy phonon modes associated with the vibration of zr atoms @xcite . the electronic properties of zrb@xmath0 manifest plethora of conflicts involving the mechanism of the transition . for example , there are conflicts on whether it is a type i or type ii superconductor @xcite , whether it possesses single gap @xcite or multiple gaps @xcite , _ etc . _ interestingly , gasparov _ et al_. @xcite found that superfluid density of zrb@xmath0 exhibits unconventional temperature dependence with pronounced shoulder at @xmath5 0.65 . in addition , they find different superconducting gap and transition temperatures for different energy bands , and the values of the order parameter obtained for @xmath6 and @xmath7 bands are 2.81 and 6.44 , respectively . detailed magnetic measurements reveal signature of meissner , mixed and intermediate states at different temperatures and magnetic fields @xcite . some of these variances were attributed to surface - bulk differences in the electronic structure @xcite , the superconductivity at the sample surface @xcite _ etc_. evidently , the properties of zrb@xmath0 is complex despite exhibiting signatures of bcs type superconductivity . here , we probed the evolution of the electronic structure of zrb@xmath0 employing high resolution photoelectron spectroscopy . the experimental results exhibit spectral evolution anomalous to conventional type superconductor and signature of pseudogap prior to the onset of superconductivity .
nowadays it is generally accepted that the non - linearity of non - abelian gauge fields could result in formation of field condensates . the first model of gluon condensate , so called _ `` color ferromagnetic vacuum state '' _ , was proposed thirty years ago by savvidy @xcite . it describes the spontaneous generation of the uniform abelian chromomagnetic field @xmath4 due to a vacuum polarization . unfortunately , this state is unstable because of the tachyonic mode in the gluon spectrum @xcite . this situation is changed at finite temperature when the spectrum stabilization happens due to either a gluon magnetic mass @xcite or a so - called @xmath3-condensate @xcite that is implemented in a stable magnetized vacuum . these are the extensions of the savvidy model to the finite temperature case investigated already by the methods of continuum field theory @xcite . in these ways the possibility of the spontaneous generation of strong temperature - dependent and stable color magnetic fields of order @xmath5 , where @xmath6 is a gauge coupling constant , @xmath7 - temperature , is realized . the field stabilization is ensured by the temperature and field dependent gluon magnetic mass , which serves as a regulator of infrared singularities at @xmath8 . such spontaneously created `` primordial '' color magnetic fields had been generated at a gut scale @xcite . they could serve as seed fields responsible for generation of the large - scale magnetic fields detected in astrophysical observations @xcite . the presence of strong magnetic fields in the early universe is of paramount importance for its evolution . in particular , a present day cosmic magnetic field of order @xmath9 could produce the recently discovered wilkinson microscopic anisotropy probe ( wmap ) anomaly @xcite . in refs . @xcite the spontaneous creation of the chromomagnetic fields was observed in @xmath0- and @xmath10-gluodynamics within the one - loop plus daisy resummations . in ref . @xcite the chromomagnetic condensate of the same order was calculated in a stochastic qcd vacuum model by comparison with some data of lattice simulations . in ref . @xcite the spontaneous generation of the chromomagnetic field at high temperature was investigated in a lattice formulation of @xmath0-gluodynamics . in refs . @xcite the response of the vacuum to the influence of strong external fields at different temperatures was investigated and it has been shown that confinement is restored when the strength of the external field is increased . other condensate , which formation at @xmath11 was investigated in continuum field theory @xcite , is the zero component of the gluon electrostatic potential @xmath12 . it acts to stabilize the qcd vacuum magnetized state , as it was discussed in ref . @xcite . in this paper , however , the magnetic field was considered as external one . whether or not the actual value of the @xmath3 generated in the deconfining phase is sufficient to remove the tachyonic instability was not estimated . the main goal of present paper is to investigate the common generation of the chromomagnetic field and @xmath3-condensate in lattice simulations . to incorporate the chromomagnetic field on a lattice we use the method developed in @xcite . instead of the field strength , which is quantized , the magnetic fluxes are considered as the objects to be investigated . the flux takes continuous values . therefore the minimization of free energy of the flux can be done in a usual known procedure . the values of the strength of spontaneously generated magnetic field and the values of the polyakov loop for corresponding states were obtained from monte carlo ( mc ) simulations . then the value of @xmath3 was derived by using a procedure developed in ref.@xcite . the paper is organized as follows . in section 2 , necessary information about the chromomagnetic fluxes on a lattice is educed and the results of calculations are given . in section 3 , the investigation of the effect of combined generation of chromomagnetic field and @xmath3-condensate is presented . the final section is devoted to discussion . in continuum , to determine the spontaneous generation of a magnetic field one has to minimize the effective potential ( or free energy ) of the background field . the background field is introduced by splitting of the gauge field potential @xmath13 into the quantum @xmath14 and classical @xmath15 parts : @xmath16 . we choice the potential @xmath17 that corresponds to a constant chromomagnetic field directed along the third axis in the euclidean and color space . we relate the free energy density of the flux to the effective action according to the definition , @xmath18 where @xmath19 and @xmath20 are the effective lattice actions with and without chromomagnetic field , correspondingly , @xmath21 is the field flux . to detect the spontaneous creation of the field it is necessary to show that free energy has a global minimum at non - zero flux , @xmath22 . in what follows , we use the hypercubic lattice @xmath23 ( @xmath24 ) with the hypertorus geometry ; @xmath25 and @xmath26 are the temporal and the spatial sizes of the lattice , respectively . in the limit of @xmath27 the temporal size @xmath25 is related to physical temperature . the standard wilson action of the @xmath0 lattice gauge theory can be written as @xmath28;\\\label{plaq } u_{\mu\nu}(x)=u_\mu(x)u_\nu(x+a\hat{\mu})u_\mu^\dagger(x+a\hat{\nu})u_\nu^\dagger(x),\end{aligned}\ ] ] where @xmath29 is the lattice coupling constant , @xmath6 is the bare coupling , @xmath30 is the link variable located on the link leaving the lattice site @xmath31 in the @xmath32 direction , @xmath33 is the ordered product of the link variables . the effective action @xmath34 in ( [ enactions ] ) is the wilson action @xmath35 averaged over the boltzmann configurations produced in the mc simulations . the lattice variable @xmath30 can be decomposed in terms of the unity , @xmath36 , and pauli , @xmath37 matrices in the color space , @xmath38 the chromomagnetic flux @xmath21 through the whole lattice was introduced by applying the twisted boundary conditions @xcite ( fig . 1 ) . in this approach the edge links in all directions are identified as usual periodic boundary conditions except for the links in the second spatial direction , for which the additional phase @xmath21 is added . the magnetic flux @xmath21 is measured in angular units and can take a value from @xmath39 to @xmath40 . the twisted boundary conditions ( t.b.c . ) for the components of ( [ compu ] ) are @xmath41 the relations ( [ tbc2 ] ) and ( [ tbc3 ] ) have been implemented into the kernel of the mc procedure in order to produce the configurations with the chromomagnetic flux @xmath21 . thus , the flux @xmath21 is taken into account while obtaining a boltzmann ensemble at each mc iteration . the mc simulations are carried out by means of the heat bath method . the lattices @xmath42 , @xmath43 and @xmath44 at @xmath45 , @xmath46 are considered . these values of the coupling constant correspond to the deconfinement phase and perturbative regime . to thermalize the system , 200 - 500 iterations are fulfilled . at each working iteration , the plaquette value ( [ plaq ] ) is averaged over the whole lattice leading to the wilson action ( [ wilson ] ) . then the effective action is calculated by averaging over the 1000 - 5000 working iterations . by setting a set of chromomagnetic fluxes @xmath21 in the mc simulations we obtain the corresponding set of values of the effective action . the value of the condensed chromomagnetic flux @xmath47 is obtained as the result of minimization of the free energy density ( [ enactions ] ) in the @xmath21 . the spontaneous generation of chromomagnetic field is the effect of order @xmath48 @xcite . the results of mc simulations show the comparably large dispersion . so , the large amount of the mc data is collected and the standard @xmath1-method for the analysis of data is applied to determine the effect . we consider the results of the mc simulations as observed `` experimental data '' . the effective action depends smoothly on the flux @xmath21 in the region @xmath49 . so , the free energy density can be fitted by a quadratic function of @xmath21 , @xmath50 in eq.([ffit ] ) , there are three unknown parameters , @xmath51 , @xmath52 and @xmath47 . @xmath47 denotes the minimum position of free energy , whereas the @xmath51 and @xmath52 are the free energy density at the minimum and the curvature of the free energy function , correspondingly . the value @xmath47 is obtained as a result of minimization of the @xmath1-function @xmath53 where @xmath54 is the array of the set of fluxes and @xmath55 is the data dispersion , which can be obtained by collecting the data into bins ( as a function of @xmath21 ) ; @xmath56 is a number of points and and @xmath57 is a mean value of free energy density in the considered bin . as it was determined in the data analysis , the dispersion is independent of @xmath21 . the deviation of @xmath47 from zero indicates the presence of the spontaneously generated field . ' '' '' & @xmath42 & @xmath43 & @xmath58 + ' '' '' @xmath45 & @xmath59 & @xmath60 & @xmath61 + ' '' '' @xmath62 & @xmath63 & & + the fit results are given in the table 1 . as one can see , @xmath47 demonstrates the @xmath64-deviation from zero . the 95% c.l . domain of parameters @xmath51 ( @xmath52 for the right figure ) and @xmath47 is represented in fig . the black cross marks the position of the maximum - likelihood values of @xmath51 ( @xmath52 for the right figure ) and @xmath47 . it can be seen that the flux is positively determined . the 95% c.l . area becomes more symmetric with the center at the @xmath51 , @xmath52 and @xmath47 when the statistics is increased . this also confirms the results of the carried out fitting . and @xmath47 , determining the free energy density dependence on the flux @xmath47 on lattice @xmath42 for @xmath45 ( left figure ) . the 95% c.l . area for the parameters @xmath51 and @xmath52 , determining the free energy density dependence on the flux @xmath47 on lattice @xmath43 for @xmath45 ( right figure).,title="fig:",scaledwidth=30.0% ] and @xmath47 , determining the free energy density dependence on the flux @xmath47 on lattice @xmath42 for @xmath45 ( left figure ) . the 95% c.l . area for the parameters @xmath51 and @xmath52 , determining the free energy density dependence on the flux @xmath47 on lattice @xmath43 for @xmath45 ( right figure).,title="fig:",scaledwidth=30.0% ] in this section we investigate an effective potential , taking into consideration the effects of non - trivial @xmath3-condensate and chromomagnetic field @xcite . this is realized by means of calculation of the partition function in a general covariant background field gauge with the background field providing both the chromomagnetic field and the non - trivial polyakov loop . then the data obtained in mc simulations are substituted in this effective potential . if the minimum value of it is negative and the potential is real , the common generation of condensates happens . within the used imaginary time formalism , the temporal direction in the eucleadian space is periodic , with period @xmath65 , and the polyakov loop @xmath66 , defined as the time - ordered exponential , @xmath67,\end{aligned}\ ] ] is a proper order parameter to describe the deconfinement phase transition . it is specified by a constant @xmath3 field , given in the fundamental representation by @xmath68 where @xmath69 , @xmath70 is pauli matrix . the trace of the polyakov loop in the fundamental representation is @xmath71 unless the @xmath72symmetry is spontaneously broken , the variable @xmath73 has to have the value @xmath74 corresponding to @xmath75 . the one - loop contribution to the free energy has the form @xcite @xmath76+\\\nonumber & & \frac12\sum_{m=0}^\infty\sum_{n,\pm}\frac{1}{\beta } \frac{gh}{2\pi}\int\frac{dk_3}{2\pi}\log\left[(\omega_n+\frac{\phi}{\beta})^2 + 2gh(m+\frac12\pm 1)+k_3 ^ 2\right],\end{aligned}\ ] ] where @xmath77 is the matsubara frequency , and the sum over @xmath78 is over all integer values . the sum over @xmath79 is the sum over the landau levels . by applying a standard technique of schwinger to express the logarithms of ( [ totep ] ) as an integral and performing the high - temperature expansion of the effective potential , we obtain @xmath80}{4 \pi ^2}+\frac{g^2 h^2 \log\left[\frac{\sqrt{g h } \beta } { 4 \pi } \right]}{12 \pi ^2}\\\nonumber -\frac{g^2 h^2 \phi ^2 \zeta[3]}{24 \pi ^4}-\frac{g^3 h^3 \beta ^2 \zeta\left[3,-\frac{\phi } { 2 \pi } \right]}{192 \pi ^4}-\frac{g^3 h^3 \beta ^2 \zeta\left[3,\frac{\phi } { 2 \pi } \right]}{192 \pi ^4 } + o(\beta^2(gh)^3).\end{aligned}\ ] ] here @xmath81 , @xmath82 is euller s constant , @xmath83 is the riemann zeta - function , @xmath84 are the bernoulli numbers . we performed the described above lattice calculations and have obtained the values for the spontaneously generated field strength and the polyakov loop of corresponding states . then we considered the expression ( [ totht ] ) as a function of these parameters and studied its minimum . we have determined that for the temperatures @xmath85 , the field strengths @xmath86 , and @xmath87 the one - loop effective potential ( [ totht ] ) is negative , @xmath88 . hence it follows , either gauge field or @xmath3 condensate are spontaneously generated . moreover , in the minimum of the effective potential for used values of the parameters the ratio @xmath89 is of order @xmath90 , that is practically zero for numeric calculations . we conclude that the common generation of the constant abelian chromomagnetic field and the non - trivial polyakov loop results in the stable magnetized vacuum state . of course , the one - loop potential does not exhibit the complete effective potential . however , it gives the main contribution which contains a larger imaginary part . if one adds the second next - to - leading contribution coming from ring diagrams , this part is cancelled @xcite . so , the check of stabilization at one - loop level is important . we have investigated the effect on common spontaneous generation of chromomagnetic field and @xmath3-condensate in @xmath0-gluodynamics at high temperature . the first step was to show the possibility of the spontaneous generation of chromomagnetic field at high temperature @xcite . the obtained results are in a good agreement with that of derived already in the continuum quantum field theory @xcite and in the lattice data analysis @xcite . the actual values of the chrmomomagnetic field strength and the polyakov loop were obtained from the same mc simulations . they allow to conclude that these condensates , chromomagnetic field and @xmath3 , have to be present in the deconfinement phase of qcd . the developed approach joins the calculation of free energy functional @xmath91 and the consequent statistical analysis of its minimum positions for various temperatures and flux values . in this way the spontaneous creation of condensates is realized in lattice simulations . other important result of the present work is that the spontaneously magnetized state is stable . this indicates that a true vacuum at high temperature is formed from these condensates . since the field configuration with constant fields is gauge non - invariant , one could consider this configuration as a domain . a complete structure of whole space can be derived by using the requirement of gauge invariance for the space including domains with different orientation . this problem we left for the future . we would like to conclude that in the deconfinement phase the condensates have to influence various processes that should be taken into consideration to have an adequate concept about this state . savvidy , phys . lett . b * 71 * , 133 ( 1977 ) . skalozub , yadernaya fizika * 28 * , 228 ( 1978 ) . niesen , p. olesen , nucl . b * 144 * , 376 ( 1978 ) . k. enqvist , p. olesen , phys . b * 329 * , 195 ( 1994 ) . a. starinets , a. vshivtsev , v. zhukovsky , phys . b * 322 * , 403 ( 1994 ) . v. skalozub , m. bordag , nucl . phys . b * 576 * , 430 ( 2000 ) . m. pollock , int . j. mod d * 12 * , 1289 ( 2003 ) . d. grasso , h. rubinstein , phys . rept . * 348 * , 163 ( 2001 ) . l. campanelli , p. cea , l. tedesco , phys . letters * 97 * , 131302 ( 2006 ) . v. skalozub , a. strelchenko , eur . phys . j. c * 40 * , 121 ( 2005 ) . n. agasian , phys . b * 562 * , 257 ( 2003 ) . v. demchik , v. skalozub , to be published in yad.fiz . * 1*(2008 ) , hep - lat/0601035 . p. cea , l. cosmai , phys . d * 60 * , 094506 ( 1999 ) ; jhep * 0508 * , 079 ( 2005 ) ; hep - lat/0101017 . k. enqvist , k. kajantie , z. phys . c * 47 * , 291 ( 1990 ) ; + d. ebert , v.ch . zhukovsky , a.s . vshivtsev , int . j. mod . phys . a * 13 * , 1723 ( 1998 ) ; + o. borisenko , j. bohacik , v. skalozub , fortsch . phys . * 43 * , 301 ( 1995 ) . p. meisinger , m. ogilvie , phys . d * 66 * , 105006 ( 2002 ) . g.t hooft , nucl . b * 153 * , 141 ( 1979 ) . v. skalozub , a. strelchenko , eur . j. c * 33 * , 105 ( 2004 ) .
in a lattice formulation of @xmath0-gluodynamics , the spontaneous generation of chromomagnetic fields at high temperature is investigated . a procedure to determine this phenomenon is developed . by means of the @xmath1-analysis of the data set accumulating @xmath2 monte carlo configurations , the spontaneous creation of the abelian color magnetic field is indicated . the common generation of the magnetic field and @xmath3-condensate is also studied . it is discovered that the field configuration consisting of the magnetized vacuum and the @xmath3-condensate is stable .
following cosmological recombination at a redshift @xmath9 , the baryonic gas filling up the universe became predominantly neutral . given that this gas is known to be mostly ionized today , one arrives at one of the major questions in current extragalactic astronomy , namely , when was the cosmic hydrogen re - ionized , and what were the sources responsible ? recent observations have begun to provide preliminary answers to this question . the absorption spectra of sdss quasars at @xmath1 indicate that the neutral fraction of hydrogen increases significantly at @xmath10 ( becker et al . 2001 ; fan et al . 2002 ; cen & mcdonald 2002 ; pentericci et al . 2002 ; white et al . 2003 ; wyithe & loeb 2004 ; mesinger & haiman 2004 ; fan et al . 2005 ) . on the other hand , analysis of the first year data from the _ wmap _ satellite suggests a large optical depth to electron scattering @xmath11 , implying that the universe was significantly reionized at redshifts as high as @xmath12 ( kogut et al . 2003 ; spergel et al . while this result is not surprising in view of earlier calculations of the reionization history by metal - free stars ( e.g. wyithe & loeb 2003a ; cen 2003a ) , it offers an intriguing _ empirical _ path for answering questions regarding the sources of reionization . several authors have pointed out that the combination of these results can only be reconciled with an extended or multi - peaked reionization history ( wyithe & loeb 2003a , c ; cen 2003a , b ; chiu , fan , & ostriker 2003 ; haiman & holder 2003 ; somerville et al . 2003 ; gnedin 2004 ) . a common feature of these models is that the first generation of stars plays a significant role . the first stars in the universe ( pop - iii ) formed out of metal - free gas , relic from the big bang . simulations of metal - free star formation indicate that the first stars may have high masses ( @xmath13 ) , since gas cooling by molecular hydrogen ( h@xmath14 ) can not lower the gas temperature below @xmath15k ( bromm , coppi , & larson 2002 ; abel , bryan & norman 2002 ) . massive , metal - free stars shine at their eddington luminosity , @xmath16 , and have roughly constant effective ( surface ) temperatures of @xmath17k and lifetimes of @xmath18yr , independent of their mass ( bromm , kudritzki , & loeb 2001 ) . this implies that the number of ionizing photons produced per baryon incorporated into these stars ( @xmath19 ) was independent of their mass function , and larger by more than an order of magnitude than that of the observed population - ii ( pop - ii ) metal - rich stars . as a result , the emissivity of ionizing photons may be double peaked , leaving open the possibility of an early reionization by pop - iii stars , with an epoch of recombination following enrichment of the igm above the critical threshold , before a final reionization by pop - ii stars near @xmath20 . indeed , the observed enrichment of the igm is consistent with a substantial contribution from pop - iii stars ( oh , nollett , madau & wasserburg 2001 ) . on the other hand , it has been suggested that the non - monotonic behavior of these models is due to the assumption of a sharp transition in metallicity ( furlanetto & loeb 2005 ) . these authors find that the various feedback mechanisms lead to an extended but monotonic reionization history . in this paper we develop a model which follows the densities of primordial and enriched gas in collapsed systems , and compute co - dependent sf , reionization and metal enrichment histories . while our model is more detailed than many previous semi - analytic approaches in several respects , the primary qualitative difference is that pop - iii sf is allowed to proceed in gas that had already collapsed in minihalos prior to enrichment of the igm , but which has not formed stars , even when the general igm has been enriched to a high metallicity level . we motivate the physics of this conclusion , in addition to modeling its consequences for the reionization history . we show that the additional post enrichment pop - iii sf boosts the total integrated pop - iii sf by an order of magnitude in typical cases . throughout the paper we assume cosmological parameters obtained through fits to _ wmap _ data ( spergel et al . these include density parameter values of @xmath21 in matter , @xmath22 in baryons , @xmath23 in a cosmological constant , and a hubble constant of @xmath24 ( or equivalently @xmath25 ) . for calculations of the press - schechter ( 1974 ) mass function ( with the modification of sheth & tormen 1999 ) we assume a primordial power - spectrum with a power - law index @xmath26 and the fitting formula to the exact transfer function of cold dark matter , given by bardeen et al . ( 1986 ) . we adopt an rms amplitude of @xmath27 for mass density fluctuations in a sphere of radius @xmath28mpc . we make the distinction between the ionizing radiation field due to a possible early population of zero - metallicity stars , and the metal enriched stars observed at lower redshifts . it is thought that the primordial initial mass function favored massive stars ( bromm , copi & larson 1999 , 2001 ; abel , bryan & norman 2000 ; mackey , bromm , hernquist 2002 ) . the possible existence of this population is very important for reionization because the spectrum of these stars would result in an order of magnitude more ionizing photons per baryon incorporated into stars ( bromm , kudritzki , & loeb 2001 ) . the formation of the very massive stars is suppressed as the material out of which stars form is enriched with metals . the fraction of the ionizing photons produced by metal - free stars depends on several unknown parameters , including the mixing efficiency of metals , the environments in which new stars form , and most importantly , the threshold metallicity above which star formation is transformed from being dominated by massive stars to a scalo ( 1998 ) initial mass function ( imf ) . the threshold metallicity is believed to be small ; bromm et al . ( 2001 ) argue for a threshold @xmath29 of the solar metallicity value . the efficiency of mixing of metal enriched outflows from star forming galaxies to the surrounding igm is even more uncertain ; scannapieco , ferrara & madau ( 2002 ) and ricotti & ostriker ( 2004 ) find that the mass weighted mean metallicity can reach values greater than @xmath30 of the solar value at redshifts as high as 20 , and note that the average metallicity scales with star formation efficiency , supernova rate , and the fraction of supernova energy that is channeled into outflows . in a representative region of the igm , the average metallicity increases roughly exponentially with redshift as it is modulated by the exponential growth in the collapse fraction of baryons at high redshifts . we therefore define a transition redshift @xmath31 below which metal - rich stars with a scalo ( 1998 ) imf dominate the production rate of ionizing photons . however , because enrichment will not occur at the same time in all parts of the igm , we allow for a spread in the transition redshift . we assume a gaussian probability function with variance @xmath32 , so that the fraction of igm @xmath33 that is enriched at redshift @xmath34 is given by @xmath35 the effectiveness of a stellar population in ionizing hydrogen in the igm can be parameterized in terms of the number of ionizing photons produced per baryon incorporated into stars ( @xmath36 ) . throughout the paper we assume pop - iii stars to be massive ( @xmath37 ) zero - metallicity stars with the generic spectrum calculated by bromm , kudritzki & loeb ( 2001 ) . the resulting value is @xmath38 . we assume metal enriched stars ( 1/20th solar metallicity ) to have a scalo ( 1998 ) mass - function , and use spectral information from the stellar population model of leitherer et al . this results in a value of @xmath39 . throughout the paper we assume that star - formation is initiated through atomic hydrogen cooling , and so implicitly assume the rapid destruction of molecular hydrogen in the igm by a back - ground of uv photons ; molecular hydrogen had been destroyed long before the universe , on average , became significantly ionized ( an ionized fraction of @xmath40 ) ( haiman , rees & loeb 1997 ; oh & haiman 2003 ) . therefore , at the time when the igm started to become enriched to the critical level for pop - ii sf there was significant gas already collapsed in minihalos ( halos above the cosmological jeans mass , but below the minimum mass for sf ) that had virial temperatures which were too low to initiate sf ( see [ primordev ] ) . if this gas was preserved in its primordial state until sf was initiated through formation of a sufficiently massive system , then the epoch of pop - iii sf could have been increased to redshifts significantly below @xmath31 . it is often assumed that metal enrichment of the intergalactic medium ( igm ) implies metal enrichment of all gas in the universe ( e.g. , wyithe & loeb 2003a , b ; cen 2003a , b furlanetto & loeb 2005 ) . we shall point out that this simple assumption is incorrect , which , as we will show , may lead to qualitatively different conclusions with regard to contributions of pop - iii stars to cosmological reionization . a critical point to note is that metal enrichment process facilitated by shockwaves carrying metals generated in stars is a strong function of the density of the gas that is being enriched . in particular , we will make the distinction between metal enrichment of the average igm and metal enrichment of gas that is inside already formed halos . let us consider metal enrichment of gas in minihalos when a blastwave of velocity @xmath41 sweeps through the igm . this is equivalent to the case of a gas cloud moving through a background medium with velocity @xmath41 . the gas cloud may be subject to a host of hydrodynamical instabilities , including kelvin - helmholtz and richtmeyer - meshkov instabilities , through which metal mixing occurs . several authors ( murray et al . 1993 , m93 hereafter ; klein , mckee , & colella 1994 ; dinge 1997 ; miniati et al . 1997 ) have shown that a non - self - gravitating gas cloud moving at the sound speed of the background medium gets disrupted on a time scale of the dynamic time of the cloud . self - gravity , in our case by the dominant dark matter in a minihalo , could stabilize the gas cloud . m93 studied the case of self - gravitating , non - cooling gas clouds moving through a background medium in the context of a two - phase medium . the gas clouds in our minihalos are physically identical to the case considered by m93 in that there is little cooling ( due to lack of cooling agents at the low temperature ) , and the gas cloud is confined by gravitational potential well of the dark matter minihalo . therefore , we may draw directly upon the calculations by m93 to quantify the whether or not gas clouds in minihalos survive enrichment of the igm . m93 show both analytically and by simulations that the stability condition for a gas cloud moving through a background medium requires the surface gravity of a cloud to be greater than @xmath42 where @xmath43 is the density ratio of the gas cloud to the background gas , @xmath44 is the radius of the gas cloud . if we define @xmath45 as @xmath46 then we find that @xmath47 where we have assumed that the density slope near the virial radius is @xmath48 ( navarro , frenk , & white 1997 ) ; @xmath49 is the mass of the minihalo within its virial radius ; @xmath50 is the mass with radius @xmath51 , @xmath34 is redshift . m93 show at @xmath52 only @xmath53 and @xmath54 of the gas is lost after @xmath55 and @xmath56 times the dynamic time of the cloud , which is about the hubble time . however , they also note that even at @xmath57 , the gas mass loss is still relatively small . while @xmath41 might be large in the immediate vicinity of a shockwave - producing galaxy , one does not expect @xmath41 to be large at large distances . mori , ferrara , & madau ( 2002 ) show , in simulations of propagation of supernova blastwaves from @xmath58 galaxies at @xmath59 , that after more than a hundred million years the relative filling factor for regions being swept by shocks of velocities larger than @xmath60 is roughly @xmath61 . we expect the velocities to be still smaller at the higher redshifts of concern here , due to enhanced cooling and larger hubble velocity . we see from equation ( [ eta ] ) that if @xmath62 @xmath63 of the gas within virialized regions of halos more massive than @xmath64 will have @xmath65 and be relatively un - affected by the shockwaves . this fraction is reduced to @xmath66 and @xmath67 for @xmath68 and @xmath69 , respectively . for lower mass mini - halos ( @xmath70 ) close to the jeans mass ( [ massscales ] ) , we find @xmath71 for a smaller velocity of @xmath72 . however @xmath73 is a high power of the fraction of gas in the halo that is subject to instabilities . thus even for halos of mass @xmath74 , and winds of velocity @xmath75km / s and @xmath76km / s we find values of @xmath77 and @xmath78 yield @xmath79 . combining these results , we expect that most of the gas already virialized with minihalos will be largely unaffected by metal - carrying blastwaves and remain metal - free until its first episode of starformation . in addition to direct enrichment , there is the possibility that virialized primordial gas in mini - halos could be enriched during the process of mergers with other mini - halos containing enriched gas . we do not expect this to be the case for the reasons outlined below . following the transition redshift , if the blastwaves that carry metals sweep through the igm with a velocity of order of @xmath80 km / s , the igm may be collisionally ionized . the collisionally heated igm subsequently cools via the compton cooling process ( by the cosmic microwave background ) , and at the same time recombines in the absence ionizing sources . we find that collisionally ionized gas would be compton - cooled and saturate at a temperature of @xmath81k ( depending on the initial temperature ) after about a hubble time at @xmath82 ; at 1/4 of the hubble time @xmath83 k. hence , igm that is heated at the redshift of interest will not be able to cool below @xmath84 k by radiative processes . subsequent cooling is primarily due to adiabatic expansion , which could achieve a temperature reduction by a factor of a few for the redshift range of interest . for a igm of @xmath85 k at @xmath82 the jeans mass is @xmath86 , which is close to the threshold mass for atomic cooling ( see equation 7 below ) . therefore , metal - enriched and heated igm is likely to be prevented from further accreting onto all minihalos , new and old . thus , gas in collapsed minihalos will not be contaminated by enriched igm gas and will remain metal - free , and newly formed minihalos will be gas - free . to be extremely conservative , let us consider the case where the metal - enriched and heated igm is assumed to have cooled to the temperature of the cosmic microwave background . gas from the igm can enter mini - halos with @xmath87m@xmath88 either through mergers of smaller collapsed objects , or through accretion from the igm . in the former case , the cold gas content is primordial because accretion from the igm at the temperature of the cosmic microwave background is suppressed in low mass mini - halos . in the latter case we argue that newly accreted gas into more massive mini - halos is unlikely to be as dense as the gas already at the virial radius . this may be seen as follows . an estimate of the non - linear overdensity of baryons inside a virialized object is @xmath89 where @xmath90 is the density of baryons inside the virialized object of temperature @xmath91 , and @xmath92 and @xmath93 are the background baryon density and temperature respectively ( barkana & loeb 2001 ) . at redshifts below @xmath94 , and in the absence of a heating source the temperature of an adiabatically cooling igm is @xmath95k ( peebles 1993 ) . this temperature is substantially lower than the temperature of the cosmic microwave background , @xmath96 . thus primordial gas , which has not been subjected to heating accretes into a dark - matter halo with an initial temperature that is substantially lower than does primordial gas which has been heated by the enriching blast - wave before being compton cooled to the temperature of the cosmic microwave background . from equation ( [ baryonod ] ) we therefore find that the ratio of overdensities of primordial to enriched gas following virialization is @xmath97 . this result suggests that in the absence of cooling , enriched gas would be prevented from significantly penetrating the minihalo , so that mixing is expected to be inefficient . as a result primordial gas is not expected to become contaminated during hierarchical growth of mini - halos . upon merging to form a system large enough to initiate sf , does the primordial gas in mini - halos form stars before becoming enriched by other sf in the galaxy ? this is expected to happen to some extent . but other " sf that occurred earlier would have been pop - iii sf . therefore , the question is not whether pop - iii sf will occur or not , but rather what the efficiency of pop - iii formation will be . we do not have a quantitative answer to this at present . it seems plausible to expect that the first sf would occur in the largest mass clump in the galaxy . if the largest mass clump makes up a significant fraction of the total mass of the galaxy , then the overall sf formation efficiency may be largely determined by the sf efficiency within that largest mass clump . we will therefore assume , following conventional wisdom , that pop - iii sf proceeds with efficiency @xmath98 . in summary , the above discussion leads us to explore the effect of the following hypothesis on the pop - iii sf history , and on the reionization history of cosmic hydrogen . we assume that virialized primordial gas within mini - halos is not enriched by super - galactic winds . @xmath100 virialized primordial gas is not enriched following accretion of enriched gas from the igm , or during the mergers of mini - halos . @xmath101 as a result , primordial virialized gas remains in its primordial state until such time as its host halo grows to be massive enough to initiate a burst of pop - iii sf . in order to compute the effect of igm enrichment on the sf history , we need to follow the density of enriched and primordial gas through the hierarchical merging of halos . this section introduces a method for following this evolution . during hierarchical galaxy formation , an average property of the galaxy population , such as the density of galaxies evolves in redshift due to newly collapsing halos , mergers of halos and accretion . we begin by discussing the critical mass - scales in galaxy formation . we then discuss calculation of the rate of newly collapsing halos , before describing how this quantity , along with the merger rates of galaxies can be used as source and sink terms in differential equations that describe the average evolution of the gas content of the galaxy population . there are four critical mass scales affecting galaxy formation . the first is the jeans mass , which corresponds to the smallest mass halo into which gas can accrete from the igm . by considering the response of a baryonic overdensity to the potential well of a dark - matter halo one can find the mass scale at which the baryonic overdensity reaches 100 . we take this mass scale to represent the jeans mass ( barkana & loeb 2001 ) @xmath102 the second mass scale corresponds to the virial temperature above which gas that has accreted into a halo can cool efficiently . we assume this scale to be determined by atomic cooling , and so to correspond to a virial temperature below @xmath103k . the corresponding mass is therefore @xmath104 where @xmath105^{-1}$ ] , @xmath106 and @xmath107 ( see barkana & loeb 2001 for more details ) . the third mass scale becomes important following the reionization of a region of igm at which time it is heated to @xmath108k . the jeans mass is increased by several orders of magnitude , and numerical simulations find that gas infall is suppressed in halos with @xmath109 . there is some disagreement as to the exact value of the halo circular velocity below which gas infall is completely suppressed ( e.g. thoul & weinberg 1996 ; kitayama & ikeuchi 2000 ; quinn , katz & efstathiou 1996 , weinberg , hernquist & katz 1997 ; navarro & steinmetz 1997 ; dijkstra , haiman , rees & weinberg 2004 ) . in what follows we assume a post - reionization minimum temperature of @xmath110 , and note that the assumption of a smaller temperature will not qualitatively effect our conclusions . thus we have @xmath111 finally , the fourth characteristic mass is the non - linear mass - scale ( @xmath112 ) , which corresponds to the co - moving length scale @xmath113 over which typical fluctuations ( i.e. 1-sigma fluctuations ) in the linear density field have a value of @xmath114 . here @xmath115 is the co - moving density of matter , @xmath116 is the linearly extrapolated overdensity at the time of collapse for a dark matter halo , and @xmath117 is the growth factor at redshift @xmath34 . for the cosmology employed in this paper , and at redshifts @xmath118 , a good fit to the dependence of the non - linear mass scale on redshift is given by @xmath119}.\ ] ] from equation ( [ nl ] ) we see that @xmath112 moves through the jeans mass at @xmath120 , and the mass corresponding to the hydrogen cooling threshold ( @xmath121 ) at @xmath122 . the non - linear mass - scale corresponds to the typical mass of newly forming galaxies . equation ( [ nl ] ) therefore encapsulates the reason why reionization is expected to occur at redshifts between @xmath123 and @xmath124 . reionization is a self - limiting process because the non - linear mass - scale does not reach @xmath125 until @xmath126 . prior to the enrichment of the igm with metals produced by the first stars the gas content of collapsed systems below the critical threshold for sf ( minihalos ) was metal - free . systems above the critical threshold had an initial burst of metal - free sf , but then presumably became enriched internally so that subsequent sf was metal enriched . following enrichment of the igm to a metallicity above a critical value , gas that was newly accreted from the igm was enriched , and so resulted in pop - ii sf . however , the metal - free gas in minihalos would not have been enriched and would have moved through mergers into larger systems . when this metal - free gas became part of a galaxy above the sf threshold , further pop - iii sf would have been possible , even significantly after the igm became enriched . indeed , if one calculates the collapsed fraction of gas in galaxies at @xmath82 with masses above the jeans mass , to the collapsed fraction above the mass corresponding to the minimum for sf ( @xmath121 ) one finds that the collapsed fraction differs by a factor of 20 ! figure [ fig1 ] shows the ratio of collapsed fractions above @xmath121 and the jeans mass as a function of redshift . the majority of metal - free gas that has accreted into collapsed systems at the time of enrichment has not yet formed stars . as a result the time of enrichment may not correspond to the time when pop - iii sf ends . indeed the fraction of the pop - iii sf during the history of the universe that has taken place by the transition redshift could be as low as the ratio shown in figure [ fig1 ] . we therefore expect most pop - iii sf to take place at redshifts below @xmath31 . to follow the evolution of primordial gas in a hierarchical cosmology we first need to compute the collapse rate of new halos . let @xmath127 be the press - schechter ( 1974 ) mass function ( number of halos with mass between @xmath128 and @xmath129 per co - moving mpc@xmath130 ) of dark - matter halos at redshift @xmath34 . the rate of change of the density of dark matter halos between @xmath128 and @xmath129 is therefore @xmath131 . let us also define ( lacey & cole 1993 ) @xmath132 the number of mergers per unit time of halos of mass @xmath133 with halos of mass @xmath128 ( forming new halos of mass @xmath134 ) at redshift @xmath34 . we find the component of the change in density with redshift of halos having masses between @xmath128 and @xmath129 that is not due to mergers . this is given by the overall change in the density of halos with mass @xmath128 , minus the change in density due to merger of halos to form a new halo of mass @xmath128 , plus the change in density of halos of mass @xmath128 that merge with other halos . the collapse rate of new halos may be written @xmath135 the value of @xmath136 equals the net change in density due to newly collapsed halos of mass @xmath128 . our next step is to find the evolution of the density of metal - free gas per halo mass @xmath137 . _ note that this density refers to a density of gas that has not been in a galaxy at a time when it underwent a burst of sf_. we assume that if a halo forms with a mass @xmath138 larger than the jeans mass @xmath139 , or if the mass of a halo grows through a merger to become larger than @xmath140 , then enough gas accretes inside the virial radius so that the mass of gas inside the halo becomes @xmath141 , where @xmath142 is the fraction of the mass - density in baryons . therefore , if the halo mass is larger than @xmath139 , then the total density of gas inside halos in a small mass - range between @xmath138 and @xmath143 is @xmath144 our aim is to compute the effect of metal enrichment on the density of cold , metal free gas inside galaxies of mass @xmath138 . prior to enrichment of the igm , the gas in galaxies where sf has not occurred remains metal - free . following enrichment , newly accreted gas is metal enriched . in neutral regions the gas inside halos below @xmath121 is not photo - evaporated , and so , at least initially some halos with masses below @xmath121 will contain metal - free gas . this gas will move into larger halos during subsequent mergers . we can calculate the density of metal - free gas that remains in these low mass halos following enrichment of the igm . if two halos @xmath128 and @xmath133 merge to form a new larger halo @xmath138 with a mass smaller than @xmath121 , then the average mass of metal - free gas in the new halo will equal the sum of the average masses of metal - free gas in each of the initial halos @xmath145 . we can therefore write down the evolution of the density of primordial gas which we denote @xmath146 ( _ recall that this density refers to a density of gas that has not been in a galaxy at a time when it underwent a burst of sf _ ) @xmath147\\ \nonumber & + & f_{\rm tran}(z ) \times \int_{0}^{\frac{m_1}{2}}d\delta m\left.\frac{d^2n_{\rm mrg}}{d\delta mdt}\right|_{m_1-\delta m}\frac{dt}{dz}\frac{dn_{\rm ps}}{d(m_1-\delta m)}f_{\rm b}\left(\frac{\frac{d\rho_{\rm primord}}{dm}}{\frac{dn_{\rm ps}}{dm } } + \frac{\frac{d\rho_{\rm primord}}{d\delta m}}{\frac{dn_{\rm ps}}{d\delta m}}\right)\\ \nonumber & - & \int_{0}^{\infty}d\delta m\left.\frac{d^2n_{\rm mrg}}{d\delta mdt}\right|_{m_1}\frac{dt}{dz}\frac{d\rho_{\rm primord}}{dm_1 } \hspace{23 mm } \mbox{where}\hspace{5mm}m_{\rm j}<m_1<m_{\rm min}\\ \frac{d^2\rho_{\rm primord}}{dm_1dz } & = & 0\hspace{73 mm } \mbox{otherwise . } \end{aligned}\ ] ] the first equation of ( [ gasev ] ) is valid in neutral regions , while once the region become ionized any remaining primordial gas is photo - evaporated out of the minihalos on the halo dynamical timescale ( @xmath148 ) , where it then becomes enriched , so that @xmath149 . the first line contains two terms that relate to regions of igm that have not yet been enriched . the first term corresponds to a source of collapsing halos into which metal - free gas accretes . the second term corresponds to a source of gas accumulation in halos above the jeans mass . this accumulation is the sum of gas that was already in halos if the progenitor was above the jeans mass , and gas newly accreted from the igm if it was not . the third term ( second line ) corresponds to regions with enriched igm and describes the movement of primordial gas from small halos to large halos during mergers . the fourth term describes the loss of gas density in halos of mass @xmath128 that results from the merger of those halos to form larger systems . the primordial gas density in halos smaller than the jeans mass or larger than @xmath150 is zero . the upper left panels of figures [ fig2 ] and [ fig3 ] show the evolution of the primordial gas density with redshift assuming @xmath151 with @xmath152 , and @xmath153 with @xmath154 . the lines show the evolution density at constant halo mass . prior to @xmath31 these follow the expectation from the press - schechter mass function . we compute the density of enriched gas inside halos of mass @xmath138 . in regions of the igm that have not yet been enriched , newly accreted gas is primordial . conversely in enriched regions of the igm , newly accreted gas is metal enriched . prior to the reionization of a region , gas can cool inside a halo whose virial temperature is larger than @xmath155k . however , following reionization of a region , infall is suppressed in halos whose virial temperature is smaller than @xmath156k . in neutral regions the enriched gas inside halos below @xmath121 is not photo - evaporated so that , at least initially , some halos with masses below @xmath150 will contain enriched gas that has not undergone sf . this gas will move into larger halos during subsequent mergers . using a method similar to that of the preceding section , we can calculate the density of enriched gas that remains in these low mass halos following reionization . note here that we assume enriched gas can accrete into halos of all masses in excess of @xmath139 . conversely , in [ haloenrich ] we argued that the igm will remain at a temperature of @xmath157k following enrichment by a blastwave , implying that accretion of enriched gas is suppressed for mini - halos with masses smaller than @xmath158m@xmath88 . however this inconsistency will not affect our estimate of pop - ii sf since @xmath158m@xmath159 . we can therefore write down the evolution of the density of enriched gas which we denote @xmath160 ( _ note that this density refers to a density of gas that has not been in a galaxy at a time when it underwent a burst of sf _ ) . photo - ionization leads to @xmath161 in ionized regions of the igm . in neutral regions we have @xmath162)f_{\rm b}[m_1-\delta m]\}\frac{\frac{}{}}{\frac{}{}}\right]\\ \nonumber & + & \int_{0}^{\frac{m_1}{2}}d\delta m\left.\frac{d^2n_{\rm mrg}}{d\delta mdt}\right|_{m_1-\delta m}\frac{dt}{dz}\frac{dn_{\rm ps}}{d(m_1-\delta m)}f_{\rm b}\left(\frac{\frac{d\rho_{\rm enrich}}{dm}}{\frac{dn_{\rm ps}}{dm } } + \frac{\frac{d\rho_{\rm primord}}{d\delta m}}{\frac{dn_{\rm ps}}{d\delta m}}\right)\\ \nonumber & - & \int_{0}^{\infty}d\delta m\left.\frac{d^2n_{\rm mrg}}{d\delta mdt}\right|_{m_1}\frac{dt}{dz}\frac{d\rho_{\rm enrich}}{dm_1 } \hspace{23 mm } \mbox{where}\hspace{5mm}m_{\rm j}<m_1<m_{\rm min}\\ \frac{d^2\rho_{\rm enrich}}{dm_1dz } & = & 0\hspace{73 mm } \mbox{otherwise , } \end{aligned}\ ] ] where @xmath163 is the heaviside step function . the first two terms ( first line ) only contribute in enriched regions of the igm . the first term corresponds to a source of collapsing halos in which metal enriched gas accretes from the igm . the second term corresponds to a source term of gas accumulation due to progenitor halos which are below the jeans mass . these progenitor halos result in accretion of enriched gas from the igm . the third term describes the movement of enriched gas from small halos to large halos during mergers . the fourth term describes the loss of gas density in halos of mass @xmath128 that results from the merger of those halos to form larger systems . the enriched gas density in halos smaller than the jeans mass or larger than @xmath164 is zero . the upper right panels of figures [ fig2 ] and [ fig3 ] show the evolution of the enriched gas density with redshift assuming @xmath151 with @xmath165 , and @xmath153 with @xmath166 . the fraction of gas in the universe that has collapsed inside galaxies at redshift @xmath34 but not undergone a starburst is @xmath167 for primordial gas , and @xmath168 for enriched gas . these are plotted in the lower left panels of figure [ fig2 ] and [ fig3 ] assuming @xmath151 with @xmath152 , and @xmath153 with @xmath166 respectively . for comparison we also show the collapsed fractions ( @xmath169 ) above @xmath139 , @xmath121 and @xmath170 ( left to right respectively ) . we see that prior to the transition redshift , nearly all gas inside collapsed systems is primordial . after the transition redshift this collapsed primordial gas is replaced with enriched gas . however , the collapsed fraction of primordial gas does not peak until @xmath171 and @xmath172 assuming @xmath151 and @xmath153 respectively . the enriched gas comes to dominate the collapsed gas fraction at around the same time . note that @xmath173 . we model sf to occur following the collapse of baryons into a dark matter halo larger than @xmath121 , or following a major merger with progenitors larger than @xmath121 . halos that merge to form a system greater than @xmath121 for the first time are able to produce pop - ii or pop - iii stars . the evolution of gas densities derived in the previous section may be used to compute the relative quantities of each . barkana & loeb ( 2000 ) suggested a model for the sfr based on the merger history of halos . in their model , gas that has not previously cooled and undergone an episode of sf in a galaxy forms stars with efficiency @xmath45 as it cools following a merger that forms a galaxy larger than @xmath164 . in addition , gas in a galaxy of mass @xmath174 undergoes an additional starburst if the galaxy merges with a second galaxy whose mass is larger than @xmath175 . barkana & loeb ( 2000 ) assume that the cold gas in a galaxy has undergone one previous starburst , so that the remaining gas mass is reduced by a factor @xmath176 . following wyithe & loeb ( 2003a ) we consider two prescriptions for sf which bracket the expected scenario of supernova feedback in low mass galaxies . the scenario described above is referred to as case - a . we also consider a second evolution for the stellar ionizing field , denoted hereafter as case b. analysis of a large sample of local galaxies shows that the ratio @xmath177 ( where @xmath178 and @xmath179 are the total stellar and dark matter halo masses respectively ) scales as @xmath180 for @xmath181 , but is constant for larger stellar masses ( kauffmann et al . since sf is thought to be regulated by supernova feedback ( dekel & silk 1986 ) , the important quantity is the depth of the galactic potential well , or equivalently the halo circular velocity . using the stellar mass tully - fisher relation of bell & de jong ( 2001 ) , we find the threshold circular velocity @xmath182 that at @xmath183 corresponds to a stellar mass of @xmath184 . in this case we define @xmath45 as the sf efficiency in galaxies with circular velocities larger than @xmath185 . the sf efficiency in smaller galaxies is multiplied by a factor @xmath186 , where @xmath187 for @xmath188 and @xmath189 for @xmath190 . the expressions for the sf rate in the following sub - sections have sf efficiencies written as the product @xmath191 . note that while in case - b @xmath186 shows the mass dependence described above , @xmath187 for all masses in case a. we assume that there is a burst of pop - iii sf of mass @xmath192 ( where @xmath191 is the sf efficiency and we assume @xmath193 ) whenever there is a merger of a halo with mass @xmath194 to form a new halo with mass @xmath195 . these starbursts comprise all of the pop - iii sf , since once gas has been involved in a starburst the gas becomes enriched so that a subsequent shock due to a major merger results in pop - ii sf . following reionization of a region of igm , all gas in halos below @xmath121 is photo - evaporated over a timescale @xmath148 where it becomes enriched , and hence subsequent sf there is assumed to be pop - ii . the pop - iii sf rate density in neutral regions is therefore given by @xmath196 + \left[f_{\rm b}\epsilon\eta\frac{\frac{d\rho_{\rm primord}}{d\delta m}}{\frac{dn_{\rm ps}}{d\delta m}}\right]\right)\end{aligned}\ ] ] the pop - iii sf rate in a neutral igm is shown in the lower middle ( case - a ) and lower right ( case - b ) panels of figures [ fig2 ] and [ fig3 ] ( solid lines ) assuming @xmath197 with @xmath152 , and @xmath153 with @xmath154 respectively . the corresponding integrated pop - iii sf densities follow from @xmath198 as noted in figure [ fig1 ] , the collapsed mass of primordial gas at @xmath31 far exceeds the mass of gas involved in sf up until that point . moreover , we see that in the absence of reionization , pop - iii sf can persist long after the enrichment of the igm . indeed , figures [ fig2 ] and [ fig3 ] show that if enrichment occurred at @xmath199 then pop - iii sf would not peak until redshift @xmath200 , and would not cease until @xmath122 . this result is quite insensitive to the exact value @xmath31 and to the mode of sf . in the model of sf employed , gas that has not previously cooled and undergone an episode of sf in a galaxy forms stars with efficiency @xmath191 as it cools following a merger to form a galaxy larger than @xmath164 . in addition , gas in a galaxy of mass @xmath174 undergoes an additional starburst if the galaxy merges with a second galaxy whose mass is larger than @xmath175 . this second star - burst is assumed to be always pop - ii . we may use our calculation of the evolution of @xmath201 to calculate the pop - ii sf rate in neutral regions . @xmath202 + \left[\epsilon\eta\frac{\frac{d\rho_{\rm enrich}}{d\delta m}}{\frac{dn_{\rm ps}}{d\delta m}}\right]\right)\\ \nonumber & & + \int_{m_{\rm min}}^{\infty } dm\int_{m/2}^{m}d\delta m \left.\frac{d^2n_{\rm mrg}}{d\delta mdt}\right|_{m}\frac{dt}{dz}\frac{dn_{\rm ps}}{dm}\\ & & \hspace{10mm}\times\epsilon\eta(1-\epsilon\eta)\left[\theta(m - m_{\rm min})m + \theta(\delta m - m_{\rm min})\delta m\right ] \end{aligned}\ ] ] the first term is for sf where gas crossing @xmath121 undergoes a starburst . the second term corresponds to the secondary starbursts ( in gas that has already undergone a burst of sf ) that follow major mergers . note that while the first term does not contribute in regions of the igm that are not yet enriched , this second term can contribute at all regions of igm that have not been reionized . finally , we calculate the pop - ii sf rate in regions of the igm that have been reionized . @xmath203\right ] + \epsilon\eta\left[f_{\rm b}\delta m\right]\right)\\ \nonumber & & + \int_{m_{\rm min}}^{\infty } dm\int_{m/2}^{m}d\delta m \left.\frac{d^2n_{\rm mrg}}{d\delta mdt}\right|_{m}\frac{dt}{dz}\frac{dn_{\rm ps}}{dm}\\ & & \hspace{10mm}\times\epsilon\eta(1-\epsilon\eta)\left[\theta(m - m_{\rm min})f_{\rm b}m+\theta(\delta m - m_{\rm min})f_{\rm b}\delta m\right].\end{aligned}\ ] ] here the 1st term is for sf where gas in the halo undergoes a starburst when the halo crosses @xmath125 . the 2nd term corresponds to the additional starbursts ( in gas within halos above @xmath121 that has already undergone a burst of sf ) that follow major mergers . the pop - ii sf rates in a neutral ( dark - dashed lines ) and an ionized ( grey - dashed lines ) igm are shown in the lower right panels of figures [ fig2 ] and [ fig3 ] assuming @xmath151 with @xmath152 , and @xmath153 with @xmath166 respectively . as noted by barkana & loeb ( 2000 ) , the sf rate in an ionized universe implied by a sf efficiency of @xmath98 agrees with observational estimates of @xmath204mpc@xmath205yr at @xmath124 . the integrated pop - ii sf density follows from @xmath206 in the previous section we computed the sf rate as a function of time in neutral and ionized regions of the universe for both pop - ii and pop - iii stars . however , the average sf history of the universe can not be computed in isolation from its reionization history , which specifies the fraction of sf that is in neutral and reionized regions as a function of redshift . similarly , the computation of a reionization history requires specification of a sf history . as a result the sf and reionization histories must be computed in a co - dependent way . the simplest estimate of the epoch of reionization is based on the following considerations . given a co - moving density of ionizing photons @xmath207 in a homogeneous but clumpy medium of comoving density @xmath208 ( where the size of the hii region is much larger than the scale length of clumpiness ) , the evolution of the volume filling factor @xmath209 of ionized regions is ( shapiro & giroux 1987 ; haiman & loeb 1997 ; madau et al . 1999 ; barkana & loeb 2001 ) @xmath210 where @xmath211 is the case b recombination coefficient , @xmath212 is the scale factor , @xmath213 is the comoving electron density , and @xmath214 is the clumping factor . this equation describes statistically the transition from a fully neutral universe to a fully ionized one , and yields reionization redshifts for hydrogen of around 7 - 12 for fiducial parameters . however , large uncertainties arise in both the source term and in the value of the clumping factor ( because more rapid recombinations lead to a slower evolution of @xmath209 ) . a more realistic description of reionization in a clumpy medium is provided by the model of miralda - escude et al . ( 2000 ) . in what follows , we draw primarily from their prescription and refer the reader to the original paper for a detailed discussion of its motivations and assumptions . the model assumes that reionization progresses rapidly through islands of lower density prior to the overlap of individual cosmological ionized regions . following overlap , the remaining regions of high density are gradually ionized . it is therefore hypothesized that at any time , regions with gas below some critical overdensity @xmath215 are ionized while regions of higher density are not . the mass fraction @xmath216 ( or equivalently @xmath217 ) therefore evolves according to the equation @xmath218 this equation assumes that all ionizing photons are absorbed shortly after being emitted , so that there is no background ionizing field , and no loss of ionizing photons due to redshift . we therefore implicitly assume that the mean free path of ionizing photons is much smaller than the hubble length . this should be valid at redshifts not too much smaller than the overlap redshift . the integration of equation ( [ postoverlap ] ) requires knowledge of @xmath219 . miralda - escude et al . ( 2000 ) found that a good fit to the volume weighted probability distribution for the density as seen in n - body simulations has the functional form @xmath220}\delta^{-\beta}d\delta,\ ] ] with @xmath221 and @xmath222 , 2.35 and 2.48 , and @xmath223 , 0.599 and 0.611 at @xmath224 , 3 and 4 . at @xmath225 they assume @xmath226 , which corresponds to the distribution of densities of an isothermal sphere , and solve for @xmath227 and @xmath228 by requiring the mass and volume to be normalized to unity . we repeat this procedure to find @xmath219 at higher redshifts . the proportionality of @xmath229 to the scale factor is expected for the growth of structure in an @xmath230 universe or at high redshift otherwise , and its amplitude should depend on the amplitude of the power - spectrum . the simulations on which the distribution in miralda - escude et al . ( 2000 ) was based assumed @xmath231 in matter , @xmath232 in a cosmological constant and @xmath233 , close to the values used in this paper . equation ( [ postoverlap ] ) provides a good description of the evolution of the ionization fraction following the overlap of individual ionized bubbles , because the ionization fronts are exposed to the mean ionizing radiation field . however , prior to overlap , the prescription is inadequate , due to the large fluctuations in the intensity of the ionizing radiation . a more accurate model to describe the evolution of the ionized volume prior to overlap was suggested by miralda - escude et al . ( 2000 ) . in our notation the appropriate equation is @xmath234}{dz } = \frac{1}{n^0}\frac{dn_{\gamma}}{dz } - \alpha_{\rm b}(1+z)^3r(\delta_{\rm c})n_{\rm e}q\frac{dt}{dz}.\ ] ] or @xmath235\frac{q}{f_{\rm m}(\delta_{\rm c})}.\ ] ] in this expression , @xmath209 is redefined to be the volume filling factor within which all matter at densities below @xmath236 has been ionized . within this formalism , the epoch of overlap is precisely defined as the time when @xmath209 reaches unity . however , we have only a single equation to describe the evolution of two independent quantities @xmath209 and @xmath237 . the relative growth of these depends on the luminosity function and spatial distribution of the sources . the appropriate value of @xmath236 is set by the mean free path of the ionizing photons . more numerous sources can attain overlap for smaller values of @xmath236 . assuming @xmath236 to be constant with redshift , we find that results do not vary much ( less than 10% in the optical depth to electron scattering ) for values of @xmath236 ranging from a few to a few tens . at high redshift , these @xmath236 correspond to mean free paths comparable to the typical separations between galaxies or quasars . unless otherwise specified we assume @xmath238 ( which lies between the values for galaxies and quasars ) throughout the remainder of this paper . the presence of minihalos effects the progression of reionization , by delaying the buildup of the ionizing radiation field within ionized regions . many of the ionizing photons will be used to photo - evaporate the minihalos rather than to progress reionization within igm . moreover , the finite time over which this evaporation occurs allows sf to proceed in halos crossing @xmath121 for a short time following the reionization of a region of igm . in the presence of an intense ionizing background radiation field , a large fraction of the gas in halos with virial temperatures below @xmath239k is photo - evaporated ( barkana & loeb 2002 ; iliev , shapiro , & raga , 2005 ; ciardi et al . , 2005 ) on the halo dynamical time @xmath240 . two additions should therefore be made to the above description arising from the effect of minihalos . first , we need to take into account pop - iii sf within ionized regions due to halos crossing the h - cooling threshold which have not yet been photo - evaporated . a simple way to approximate the effect of this additional sf is to reduce the cold gas in minihalos , and therefore the resulting sf rate corresponding to crossing the h - cooling threshold by a factor of @xmath241 at time @xmath242 . here @xmath243 and the pre - factor of 0.5 is approximately the average remaining neutral gas fraction in all minihalos which become exposed to ionizing photons between @xmath244 and zero years prior to a redshift @xmath34 , while the ratio term is just the fraction of minihalos that become exposed to ionizing radiation within the preceding @xmath244 years . second , while the above consideration enhances the sf in ionized igm , the minihalos will delay further reionization for a period lasting for the photo - evaporation timescale through screening of the ionizing sources by minihalos ( barkana & loeb 2002 ) . we therefore need to estimate the fraction of photons that are absorbed by minihalos rather than contributing to the reionization of the igm . this estimate may be made as follows . an ionizing photon will travel a distance @xmath245 through ionized regions of igm before encountering neutral igm . on the other hand , an ionizing photon will travel a different distance @xmath246 before encountering a minihalo that contains cold gas . the fraction of ionizing photons produced in ionized regions that reionize neutral igm rather than intercepting minihalos prior to their evaporation is therefore approximated by @xmath247 [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] the length @xmath245 may be easily computed within the formalism described . an ionizing photon will typically travel a distance @xmath248^{-2/3}$ ] , where @xmath249 is the fraction of the igm at overdensities below @xmath236 ( miralda - escude et al . 2000 ) . on the other hand , to compute the cross - section for interception of a minihalo that contains cold gas we need to find the fraction of gas in minihalos at a redshift @xmath34 that was accreted to within progenitor minihalos within the previous interval @xmath250 . this quantity may be computed from the extended press - schechter formalism ( lacey & cole 1993 ) . for a halo of mass @xmath128 at redshift @xmath34 , the fraction of the halo mass which by some higher redshift @xmath251 had already accumulated in halos is @xmath252}}\right],\ ] ] where @xmath117 is the linear growth factor at redshift @xmath34 , @xmath253 is the variance on mass scale @xmath128 ( defined using the linearly extrapolated power spectrum at @xmath183 ) , and @xmath254 is the jeans mass at @xmath251 . given the virial radius for a halo @xmath255 of mass @xmath128 ( barkana & loeb 2001 ) and the press - schechter ( 1974 ) mass function ( @xmath256 ) for the number density of halos we may then find the length @xmath246 @xmath257\right]^{-1},\ ] ] where @xmath258 . we next describe the co - evolution of the reionization and sf histories . having specified a transition redshift , we computed in the previous section the pop - iii ( @xmath259 ) and pop - ii ( @xmath260 ) sf rates in neutral regions , as well as the pop - ii sf rates in ionized regions ( @xmath261 ) . the ionizing photon production rate then follows from @xmath262 , \end{aligned}\ ] ] where @xmath263 is the proton mass , @xmath264 and @xmath265 are the ionizations per baryon , and @xmath266 and @xmath267 are the escape fractions for pop - iii and pop - ii stars respectively . the quantity @xmath268 is the maximal value that @xmath209 has attained at any time in the reionization history . a region that has been reionized may recombine in the absence of an ionizing radiation field , but the timescale for cooling is longer than the recombination timescale for @xmath269 . the inclusion of @xmath268 ensures that the sf rate in regions that have recombined remains the same as for an ionized region should that region return to a neutral state . the term @xmath270 accounts for the contribution of quasars to the ionizing flux . quasars are included in the model as described in wyithe & loeb ( 2003a ) , with an updated quasar luminosity function model ( wyithe & loeb 2003b ) . if treated as the sole ionizing sources , we find that quasars reionize cosmic hydrogen at around @xmath124 . wyithe & loeb ( 2003c ) found that helium became reionized by quasars at @xmath271 , in agreement with observation ( e.g. theuns et al . . the coupled evolution of equations ( [ postoverlap ] ) , ( [ preoverlap ] ) and ( [ history ] ) yield the sf and reionization histories . we may also estimate the mean level of metal enrichment in the igm . this quantity is important because massive pop - iii stars are expected to form only out of gas that is not enriched to a metallicity above @xmath272z@xmath88 ( bromm et al . we estimate the mean metallicity at a redshift @xmath34 as @xmath273 where @xmath274 is the stellar mass that must be produced per supernova , @xmath275 is the mass of metals produced per supernova , and @xmath276 is the fraction of metals that escape from galaxies into the igm in supernova - driven outflows ( e.g. madau , ferrara & rees 2001 ; thacker , scannapieco & davis 2002 ) . the value of @xmath276 may be large because the early dwarf galaxies had shallow potential wells . for a massive pop - iii imf , furlanetto & loeb ( 2003 ) estimate @xmath277 and @xmath278 . we note that this number is very sensitive to the stellar mass function , because stars with masses @xmath279 are expected to eject their metals while more massive stars collapse to form black holes ( heger & woosley 2002 ) . different reionization histories lead to different column depths of free electrons along a line - of - sight to recombination . an important probe of the reionization history is therefore provided by the observation by the wmap satellite of a large optical depth to electron scattering of cmb photons . the optical depth to electron scattering @xmath280 , depends on the mass filling factor ( @xmath281 ) and is evaluated according to @xmath282 where @xmath283 is the thomson cross - section . figures [ fig4]-[fig9 ] show several examples of the reionization history , which explore a wide range of the various inputs to the model . these inputs are summarized in table [ tab1 ] . in each case we present 4 histories in the upper set of panels . the input values of @xmath266 , together with the value of @xmath284 that leads to a final reionization at @xmath285 are also listed in table [ tab1 ] . the thicker lines correspond to histories with larger values of @xmath266 . note here that the escape fraction is degenerate with the sf efficiency ( for @xmath286 ) . we have chosen @xmath287 in presenting our sf histories , however the values of escape fraction can be adjusted in proportion for different choices of @xmath45 . the total sf rates in pop - ii ( solid lines ) and pop - iii ( dashed lines ) stars are plotted in the upper left panels of figures [ fig4]-[fig9 ] . feedback from reionization quenches sf . however , our histories show that the additional pop - iii sf remains significant at @xmath288 in all cases , and peaks at @xmath289 in many cases . pop - ii stars begin to dominate the global sf rate only around the time of the peak in the pop - iii sf rate . we also find that pop - iii sf at redshifts below @xmath31 forms the dominant contribution to the total integrated pop - iii sf rate in all models . this may be seen through plots of the integrated ionizing photon production as a function of redshift ( upper right panels of figures [ fig4]-[fig9 ] ) . the horizontal dotted lines show the level of ionizing photon production at @xmath290 . the fraction of ionizing photon production by pop - iii at @xmath288 is larger than 90% in nearly all cases , and can be as large as 99.9% . furthermore , pop - iii sf can dominate the total ionizing photon production at @xmath291 for models with large @xmath280 . we note that the comparison of sf in an ionized and in a neutral igm ( see figures [ fig2]-[fig3 ] ) suggests that reionization of the igm should be accompanied by a dip in the sf rate as seen in the calculations of barkana & loeb ( 2000 ) . however , the sf histories that are computed in concert with a reionization history do not show this dip . the reason is that an extended reionization history washes out this feature on average , though it may still be observed in isolated regions . for each of the reionization histories in figures [ fig4]-[fig9 ] we plot @xmath292 ( middle left panels ) . we have chosen @xmath276 ( see values in table [ tab1 ] ) so as to obtain @xmath293 ; close to the threshold deduced by bromm et al . ( 2001a ) at @xmath31 . note that @xmath294 is degenerate with @xmath45 in addition to @xmath276 . however , this simple estimate demonstrates that metal enrichment of the igm by pop - iii stars is consistent with our assumed @xmath31 ( see also mackey , bromm & hernquist 2002 ) . the possible exception is for case - b sf with @xmath295 . in this example the integrated sf rate by @xmath31 is not sufficient to enrich the igm . finally , the parameter @xmath296 is unspecified a - priori . we have chosen @xmath297 , with the exception of figures [ fig8 ] and [ fig9 ] . comparison of figures [ fig8 ] and [ fig9 ] with figures [ fig4 ] and [ fig5 ] demonstrate that while the details of the reionization history depend on the value assumed for @xmath236 , the qualitative nature of the predicted histories and the values of @xmath280 are similar . we therefore find that our conclusions are not sensitive to the choice of @xmath236 . we find that either pop - iii stars substantially reionize the universe , or that the fraction of pop - iii sf which takes place following the transition redshift is large . either way the conclusion is that pop - iii stars played a major role in the reionization history of the universe . in particular if the universe did not become substantially reionized prior to @xmath31 , then the large majority of pop - iii sf takes place at redshifts below @xmath31 , in contrast to previous modeling . each of the histories shown in figures [ fig4]-[fig9 ] was chosen so that overlap occurs at @xmath298 as suggested by the spectra of high redshift sdss quasars . however , different reionization histories result in different densities of electrons as a function of redshift . the values of @xmath280 for the various models are listed in the middle right panels of figures [ fig4]-[fig9 ] and in table [ tab1 ] . wmap suggested @xmath299 . we see that only a small value of @xmath300 for case - a , or @xmath301 for case - b is required to obtain this large optical depth , even where the enrichment of the igm occurred at a redshift as high as @xmath153 . indeed the conclusion that pop - iii stars contributed significantly to reionization is quite insensitive to the transition redshift for @xmath302 , so long as pop - iii stars had a top - heavy imf . the lower three panels of figures [ fig4]-[fig9 ] represent results for a series of reionization histories . for each value of the primordial escape fraction ( @xmath266 ) , we have found the value of @xmath267 that results in a final overlap at @xmath285 ( or as near as possible at @xmath291 ) , and which maintains the reionization until the present day . in the lower left panel we present grey - scale and contours showing the level of overlap as a function of redshift and @xmath266 . the grey - scale shows the level of ionization ( with black representing high ionization ) . the dotted , dashed and solid lines correspond to @xmath303 , 0.5 and .999 respectively . the value of @xmath267 corresponding to these histories is plotted as a function of @xmath266 in the middle panel . in the right panel we plot the optical depth @xmath280 as a function of @xmath266 ( grey line ) . for case - a , the optical depth has a maximum near @xmath304 , but the wmap result is consistent with all values of @xmath266 . for case - b , the optical depth has a maximum near @xmath305 . only larger values of @xmath266 are consistent with the wmap result for case - b . the dark line in the lower - right panel is the ratio of the amount of pop - iii sf already completed at @xmath31 to the total pop - iii sf . this ratio is small ( @xmath306 ) for small values of @xmath280 . in case - a , we find that the ratio can be as small as 0.01 where @xmath153 and 0.1 where @xmath151 . the ratio is only of order unity if @xmath266 is also of order unity . in case - b , we find that the ratio can be even smaller , with values a factor of 3 lower . the ratio does note become of order unity for any value of @xmath266 in case - b . interestingly in case - a a range of histories show two epochs of reionization . however , contrary to earlier works , the tail end of the first reionization has not been set by an abrupt transition redshift @xmath31 . rather the end of the first reionization can come either before , or after @xmath31 depending on the sum of the various astrophysics included in the model . a particularly interesting history is seen where case - a sf is combined with a large escape fraction for pop - iii ionizing photons and a large transition redshift ( see figure [ fig5 ] ) . in this example the reionization history shows three peaks of reionization . thus , the claim made earlier by furlanetto & loeb ( 2005 ) that a temporally smooth metallicity enrichment of the igm gives rise to only monotonic reionization process is not borne out in our more detailed calculations in general . clearly , the information embedded in the reionization history is very rich . it is useful to extract some essential properties with regard to the cmb experiments such as wmap , since the three - year wmap results are expected to be made public shortly . the thin black curves in figures [ fig6 ] and [ fig7 ] show histories with @xmath307 and @xmath308 respectively . these histories follow from a very small pop - iii contribution to reionization , suggesting that without pop - iii contribution we expect @xmath2 . at the other extreme , when we maximize the contribution from pop - iii stars to reionization by using a very high ( perhaps implausible ) escape fraction ( relative to that of pop - ii stars ) , a maximum value of @xmath5 is possible . this suggests that the optical depth reported by the three - year wmap results should not exceed this value ( which is @xmath309 above the central value of the first year wmap result ) , otherwise our understanding of structure formation at high redshift would require a dramatic revision . in a more realistic picture where the product of star formation efficiency and escape fraction for pop - iii stars is comparable to that for pop - ii stars , we find that the range @xmath6 is most likely , consistent with our previous calculations ( cen 2003a , b ) . the three different outcomes described above imply qualitatively different histories . in the small @xmath2 case without significant pop - iii contribution , reionization is expected to be rapid with the neutral fraction quickly rising above @xmath3 at @xmath4 . in the large @xmath310 case with a large pop - iii contribution , the reionization process could be complex and frequently displays multiple peaks . under seemingly more reasonable assumptions regarding the efficiency of pop - iii stars , there appears to be a reionization plateau at @xmath7 where the mean neutral fraction holds in a relatively narrow range of @xmath311 . independent constraints of the reionization history are highly desirable . surveys of ly@xmath312 emitters ( e.g. , malhotra & rhoads 2005 ) and measurements of qso stromgren sphere sizes ( e.g. , mesinger , haiman , & cen 2004 ; wyithe , loeb & carilli 2005 ; fan et al 2005 ) may provide powerful constraints on the evolution of the ionized fraction close to the end of the reionization epoch . in addition future cmb experiments such as planck survey might be able to distinguish ( e.g. , holder et al . 2003 ) between the various histories and if so , it should yield some critical information on sf processes in the high redshift universe . we show that the collapsed fraction of primordial gas that is trapped in virialized minihalos at the redshift ( @xmath31 ) when the igm becomes enriched with metals is larger than the collapsed fraction of gas that has already been involved in sf by a factor of a few to a few tens . we argue that this virialized gas will not be as easily enriched by super - galactic winds as the general igm , because the strong self - gravity created by dark matter and significant overdensity with respect to the background gas help stabilize the collapsed gas against large disruptions by sweeping shockwaves . moreover , enriched gas is expected to have been pre - heated during the process of enrichment , preventing contamination of virialized primordial gas during hierarchical growth of mini - halos . hence , collapsed gas in virialized minihalos may largely remain metal - free , until those halos merge to form larger systems of sufficient mass to initiate sf . we have followed the evolution of virialized primordial gas through hierarchical mergers of their host dark matter halos . this enables the calculation of the pop - iii sf rate arising from the virialized primordial gas when it crosses the hydrogen cooling threshold . our model for the reionization of the igm computes the co - evolution of the reionization and sf histories . in addition to following the gas - content of halos below the minimum mass for sf our model addresses several astrophysical phenomena . these include feedback from reionization in the form of suppression of galaxy formation in a photo - ionized igm , as well as the photo - evaporation of minihalos ( both the screening of ionizing sources and the additional low mass sf in ionized regions prior to photo - evaporation ) . we also allow the igm to be gradually enriched by metals rather than an abrupt transition . we find that the pop - iii sf rate peaks substantially later ( as low as redshift @xmath313 ) in cosmic time than the epoch at which the igm became metal enriched , and may extend to @xmath1 . moreover , we find that the total integrated pop - iii sf is larger , by a factor of a few to a few tens , than the integrated pop - iii sf at the average redshift of igm metal enrichment . this prolonged epoch of pop - iii sf has a pronounced effect on the reionization history of cosmic hydrogen . despite the varied feedback mechanisms included in our model we find that the prolonged epoch of pop - iii sf leads to complex reionization histories which can be multi - peaked under a range of plausible scenarios . observational measurements of the detailed reionization history would provide critical information on the formation process of first stars . specifically , assuming the final overlap redshift to be fixed at @xmath285 ( as implied by observations of high redshift quasars ) , the contribution of pop - iii stars to reionization can be quantified and will be tested by three - year wmap results : ( 1 ) if pop - iii stars do not contribute to reionization , @xmath2 and a rapid reionization at @xmath1 is expected , with the mean neutral fraction quickly exceeding @xmath3 at @xmath4 ; ( 2 ) if the product of star formation efficiency and escape fraction for pop - iii stars is significantly larger than for pop - ii stars , then a maximum @xmath5 is achievable ; ( 3 ) in a ( perhaps ) more plausible scenario where the product of star formation efficiency and escape fraction for pop - iii stars is comparable to that for pop - ii stars , @xmath6 would be observed [ consistent with our previous calculations ( cen 2003a , b ) ] . in this case the histories would be characterized by an extended ionization plateau from @xmath7 where the mean neutral fraction stays in a narrow range of @xmath8 . these results are rather insensitive to the redshift where the igm becomes enriched with metals .
population - iii ( pop - iii ) starformation ( sf ) is thought to be quenched when the metallicity of the star - forming gas has reached a critical level . at high redshift , when the general intergalactic medium ( igm ) was enriched with metals , the fraction of primordial gas already collapsed in minihalos ( above the jeans mass but below the mass corresponding to the efficient atomic cooling threshold for sf ) was significantly larger than the fraction of primordial gas that had already been involved in pop - iii sf . we argue that this reservoir of minihalo gas remained largely in a metal - free state , until these minihalos merged into large systems ( above the hydrogen cooling threshold ) and formed stars . as a result , the era of pop - iii sf was significantly prolonged , leading to a total integrated pop - iii sf that was an order of magnitude larger than expected for an abrupt transition redshift . we find that the contribution of pop - iii sf to the reionization of hydrogen could have been significant until @xmath0 and may have extended to redshifts as low as @xmath1 . our modeling allows for _ gradual _ enrichment of the igm , feedback from photo - ionization and screening of reionization by minihalos . nevertheless , the extended epoch of pop - iii sf may result in complex reionization histories , where multi peaks are possible over some regions of parameter space . the relative contribution of pop - iii stars to reionization can be quantified and will be tested by three - year wmap results : ( 1 ) if pop - iii stars do not contribute to reionization , @xmath2 and a rapid reionization at @xmath1 are expected with the mean neutral fraction quickly exceeding @xmath3 at @xmath4 ; ( 2 ) if the product of star formation efficiency and escape fraction for pop - iii stars is significantly larger than that for pop - ii stars , then a maximum @xmath5 is achievable ; ( 3 ) in a perhaps more plausible scenario where the product of star formation efficiency and escape fraction for pop - iii stars is comparable to that for pop - ii stars , @xmath6 would be observed , with reionization histories characterized by an extended ionization plateau from @xmath7 where the mean neutral fraction stays in a narrow range of @xmath8 . this result holds regardless of the redshift where the igm becomes enriched with metals .
the local bubble ( lb ) is a cavity , elongated towards the galactic north pole , with an average extension of about 100 pc , copiously radiating in soft x - rays . there exist a number of discrepancies between observations and modelling that have all been outlined in a recent panel discussion ( breitschwerdt & cox 2004 ) . most models proceed from a multi - supernova origin of the lb ( cf.berghfer & breitschwerdt 2002 for possible supernova ( sn ) progenitors ) , in which the lb is the result of successive sn explosions , although part of the soft x - ray emission could be of heliospheric origin , generated by charge exchange reactions between solar wind ions and heliospheric plasma . we use a 3d godunov type parallelized amr hydrocode to track small scale structures down to 1.25 pc , where necessary ( for details see avillez 2000 ) . according to berghfer & breitschwerdt ( 2002 ) 20 stars of pleiades subgroup b1 with masses between 10 and 20 @xmath0 were moving through the lb within the last 14 myr , exploding after their main sequence life time , @xmath1 , along a path crossing @xmath2 , @xmath3 pc ( see fig . 1 ) , thus generating the local cavity into which the lb expands . the galactic sn rate has been used for the setup of _ other sne _ in the background disk . first we derive analytic similarity solutions , taking an initial mass function for galactic ob associations with powerlaw index @xmath5 , and using @xmath6)^{-\alpha}$ ] yr ( stothers 1972 ) , with @xmath7 , for stars within the mass range @xmath8 . the mechanical energy input rate ( see berghfer & breitschwerdt 2002 ) then is @xmath9 , where @xmath10 , @xmath11 and @xmath12 yr . modifying suitably the similarity solutions of mccray & kafatos ( 1987 ) , we obtain @xmath13 with @xmath14 . note , that our value of @xmath15 is between the canonical value of @xmath16 for a sedov - type solution and @xmath17 for a wind type solution , due to the declining energy input rate with time . if we now try to match a present day average radius of the lb of even 146 pc at time @xmath18 myr , we need a constant ambient density of @xmath19 , a value way above the average ism density . comparison with our excellently matching numerical results shows , however , that this discrepancy must be due to mass loading , and turbulence in a sn disturbed _ background _ medium . avillez , m.a . : 2000 , _ mnras _ * 315 * , 479 . berghfer , t. and breitschwerdt , d. : 2002 , _ a&a _ * 390 * , 299 . breitschwerdt , d. and cox , d.p . : 2004 , _ apss _ , in press ( astro - ph/0401428 ) . mccray , r. , and kafatos , m. : 1987 , _ apj _ * 317 * , 190 . stothers , r. : 1972 , _ apj _ * 175 * , 431 .
we have performed high resolution 3d simulations of the local bubble ( with 1.25 pc finest resolution ) in a _ realistic background ism _ , jointly with the dynamical evolution of the neighbouring loop i superbubble . we can reproduce ( i ) the size of the bubbles ( in contrast to similarity solutions ) , ( ii ) the interaction shell with loop i , discovered with rosat , ( iii ) predict the merging of the two bubbles in about 3 myr , when the interaction shell starts to fragment , and , ( iv ) the generation of blobs like the local cloud as a consequence of a dynamical instability .
there is a well determined set of cosmological parameters @xmath14 , @xmath15 , @xmath16 , @xmath17 gyr , @xmath18 emerging from a number of observations ( the concordance cosmology , see spergel et al . 2003 ) . also the cosmic density of baryons @xmath19 has been very precisely determined through both the power spectrum of cosmic microwave background anisotropies and measurements of the primordial abundance of light elements ( cyburt et al . 2001 ; olive 2002 ) . an important complementary information is that the density of baryons residing in virialized structures and associated to detectable emissions is much smaller than @xmath20 . in fact , traced - by - light baryons in stars and in cold gaseous disks of galaxies and in hot gas in clusters amount to a @xmath21 ( persic & salucci 1992 , fugukita et al . 1998 , fukugita & peebles 2004 ) . on the other hand , in rich galaxy clusters the ratio between the mass of the dark matter ( dm ) component and the mass of the baryon component , mainly in the hot intergalactic gas , practically matches the cosmic ratio @xmath22 . the circumstance that @xmath20 is a factor of about 10 larger than @xmath23 puts forth both an observational and a theoretical problem . on one side , observations are needed to detect and locate these `` missing '' baryons ( see for a review stocke , penton & shull 2003 ) . on the theoretical side , galaxy formation models have to cope with the small amount of baryons presently in gas and stars inside galaxies . no doubt that feedback from stars and agns played a relevant role in unbinding large amounts of gas and eventually removing them from the host dm halo ( see , e.g. , dekel & silk 1986 ; silk & rees 1998 ; granato et al . 2001 , 2004 ; hopkins et al . 2005 ; lapi , cavaliere & menci 2005 ) , but we need to get a detailed quantitative understanding of these processes , which are crucial to comprehend galaxy formation . the stellar feedback is expected to depend on the star formation history and the total energy released to the gas ultimately depends on the total mass of formed stars and on the present - day galaxy luminosity . correspondingly , the total energy injected by the agn feedback is ultimately related to the final black hole ( bh ) mass . the fraction of the gas removed by feedbacks is expected to depend also on the binding energy of the gas itself , which is determined by the galaxy _ virial _ mass and by its density distribution . therefore the relationships between the galaxy halo mass and the galaxy luminosity , the stellar and baryonic mass , the mass of the central bh , are expected to give extremely useful information on the process of galaxy formation and evolution . an additional relevant piece of information is the link between the galaxy halo mass and the velocity dispersion of the old stellar component . this paper is devoted to a statistical study of these relations . the halo occupation distribution ( see , e.g. , kauffmann , nusser & steinmetz 1997 ; peacock & smith 2000 ; berlind & weinberg 2002 ; magliocchetti & porciani 2003 ) , which specifies the probability @xmath24 that a halo of mass @xmath25 is hosting @xmath26 galaxies , is a helpful statistics to establish the link between the host dm halo mass and the observed galaxy luminosity . an additional tool to explore this link is the formalism of the conditional luminosity function ( yang , mo & van den bosch 2003 ) , which describes how many galaxies of given luminosity reside in a halo of given mass . following this approach , yang et al . ( 2003 ) investigated the relation between halo mass and luminosity . however , particularly for high halo masses , @xmath27 , both methods give information on galaxy systems , more than on large galaxies . marinoni & hudson ( 2002 ) and vale & ostriker ( 2004 ) suggest that a helpful starting point can be the simple hypothesis that there is a one - to - one , monotonic correspondence between halo mass and resident galaxy luminosity . then , by equating the integral number density of galaxies as a function of their luminosity and stellar mass to the number density of galaxy halos , one gets a statistical estimate of the dm halo mass associated to galaxies of fixed luminosity or fixed baryon / stellar mass . however , a major problem of the method is the estimate of the mass function of halos hosting one single galaxy , the galaxy halo mass function ( ghmf ) . to solve the problem , in this paper we use an empirical approach , which takes into account the results of numerical simulations ( see , e.g. , de lucia et al . 2004 and references therein ) on the halo occupation distribution by adding to the halo mass function ( hmf ) the contribution of subhalos ( vale & ostriker 2004 ; van den bosch , tormen & giocoli 2005 ) . at large masses we subtract from the hmf the mass function of galaxy groups ( girardi & giuricin 2000 ; martinez et al . 2002 ; heinmki et al . 2003 ; pisani , ramella & geller 2003 ) . the mass around a galaxy up to a radial distance from its center much larger than the characteristic scale of light distribution can be inferred from detailed x - ray observations ( see , e.g. , osullivan & ponman 2004 ) . also the statistical measurements of the shear induced by weak gravitational lensing around galaxies ( see bartelmann , king , & schneider 2001 ) yield important insights on the halo mass of galaxies ( mckay et al . 2002 , sheldon et al . though these mass estimates have significant uncertainties and their extrapolations to the virial radii are not immediate , they nonetheless provide useful reference values to which we compare the outcomes of our method . the role of stellar and agn feedback has been discussed by several authors ( see , e.g. , dekel & silk 1986 ; silk & rees 1998 ) . granato et al . ( 2001 , 2004 ) have implemented both feedbacks in their model of joint formation of qsos and spheroidal galaxies . more recently the feedbacks have also been introduced into hydrodynamical simulations ( springel , di matteo & hernquist 2005 ) . one of the purposes of this paper is to show how the information coming from the relationships of the galaxy halo mass with measurable galaxy properties ( such as the stellar , baryonic and central bh masses ) can shed light on the role and on relative importance of the two feedbacks . the plan of this work is the following . in @xmath28 we compute the galaxy stellar and baryonic mass functions , exploiting the luminosity function and the mass - to - light ratio of the stellar component inferred from kinematical mass modelling of galaxies . then , in @xmath29 , we derive the mass function of galactic halos , and , exploiting the relevant galaxy statistics ( luminosity function , galaxy star / baryon mass function , velocity dispersion function ) and the galaxy halo mass function , we investigate the relationships of the corresponding galaxy properties with the halo mass . the relation of the halo mass with the mass of the central bh in galactic spheroidal components is deduced in @xmath30 , by comparing the central black hole mass function to the galaxy halo mass function . in @xmath31 we discuss the role of the stellar and agn feedbacks in shaping the relationships between stellar and baryonic mass and halo mass . @xmath32 is devoted to the conclusions . the luminosity function ( lf ) is a fundamental statistics for galaxies . its present form is the result of physical processes involving both baryons and dark matter . in particular , the lfs in the range between about 0.1 to several @xmath33 m probe the stellar component . the mass of stars and baryons associated to galaxies can be derived coupling the lf with estimates of the mass - to - light ratio ( mlr ) of the stellar and gaseous components . as it is well known , the mlr and the fraction of gas depend on galaxy morphology . nakamura et al . ( 2003 ) estimated the lf in the @xmath1-band for early- and late - type galaxies separately . the separation has been done through a light concentration method ( see shankar et al . 2004 for a discussion and a comparison with other lf estimates ) . these early- and late - type galaxy lfs are in reasonable agreement with the lfs of red and blue galaxies , respectively , derived by baldry et al . ( 2004 ) . since the nakamura lf of late - type galaxies is well defined only for luminosities brighter than @xmath34 , we extended it to lower luminosity using the results of zucca et al . ( 1997 ) and loveday ( 1998 ) and translating them from the @xmath35-band to the @xmath1-band using a color @xmath36 , as appropriate for star forming irregular galaxies ( fukugita et al . 1995 ; their table 3 , panels ( j ) and ( m ) with @xmath37 and @xmath38 ) . the conversion to solar luminosities has been done taking @xmath39 ( blanton et al . the resulting lf is well fit , in the range @xmath40 , by @xmath41 where @xmath42 . the mlr of the stellar component can be derived from studies of stellar evolution , with uncertainties associated to the poor knowledge of details of the imf ( see , e.g. , fukugita et 1998 ; bell et al . 2003 ; fukugita & peebles 2004 ; baldry et al . 2004 ; panter , heavens & jimenez 2004 ) . a more direct approach exploits detailed kinematical and photometric studies of galaxies to estimate the amount of mass traced by light and the mass of the dm component , taking advantage of their different distribution inside the galaxies . this method has been used by salucci & persic ( 1999 ) , who estimated the stellar and gaseous mass as a function of the b - band luminosity for late - type galaxies to get the baryon mass @xmath43 . we have approximated their results for the stellar and the gas component , respectively as @xmath44 for @xmath45 , and @xmath46 in the range @xmath47 . combining these relations with the lf of eq . ( [ eq|lf ] ) , assuming a gaussian distribution around the mean relations with a dispersion of about 20% , we derived the stellar mass function and the baryonic mass function of late - type galaxies . to do that we have taken @xmath48 ( binney & merrifield 1998 ) and @xmath49 ( see table 3 , panel ( m ) of fukugita et al . 1995 ) . a similar approach can be followed for early - type galaxies . for about 9000 such galaxies extracted from the sloan digital sky survey ( sdss ) bernardi et al . ( 2003 ) estimated the mlr ( within the characteristic half luminosity ratio @xmath50 , in solar units ) @xmath51 @xmath52 in the @xmath1-band ( @xmath53 ) . these authors derived the mass inside @xmath50 using the relation @xmath54 , where @xmath4 is the central velocity dispersion , and assuming @xmath55 . however , the value of @xmath56 depends on the light profile ( @xmath57 for a de vaucouleurs profile , see prugniel and simien 1997 ) and on the dm distribution ( borriello et al . 2003 ) . using @xmath58 and rescaling the zero point of the mlr by bernardi et al . ( 2003 ) , we obtain : @xmath59 as discussed by bell et al . ( 2003 ) and by baldry et al . ( 2004 ) , the uncertainties related to the imf and to the star formation history imply an uncertainty of about @xmath60 in the mean value of the mlr ; the dispersion around the mean relation , eq . ( [ eq|mslr ] ) , is of about @xmath61 . by convolving the @xmath1-band luminosity function of early - type galaxies of nakamura et al . ( 2003 ) with the distribution of mlrs , assuming a gaussian distribution around the mean relation of eq . ( [ eq|mslr ] ) , with a dispersion of 20% , we estimate the galaxy stellar mass function ( gsmf ) of @xmath62 galaxies , which essentially coincides with the galaxy baryonic mass function ( gbmf ) , since in early - type galaxies the gas gives a negligible contribution to the baryon mass . the total gsmf , holding over the mass range @xmath63 , is shown by the solid line in fig . [ fig|gsmf ] , where the shaded area corresponds to the 30% uncertainty in the mean mlr . this is a safe range to determine the gsmf as at lower masses the uncertainty in the lf grows while the increase in the total stellar mass density including @xmath64 is rather small , @xmath65 . the upturn at @xmath66 corresponds to the appearance of the dwarf galaxy population , represented by the second term at r.h.s . ( [ eq|lf ] ) . however the contribution in stellar mass density in the range @xmath67 is just @xmath68 3% of the total . in the inset of fig . [ fig|gsmf ] we have displayed the gas fraction as a function of the stellar mass . a very accurate analytical representation ( actually indistinguishable from the solid line showing the numerical results ) is provided by a schechter function plus a power law term : @xmath69 where @xmath70 . the gbmf is easily computed by adding the appropriate gas contribution . recent estimates of the gsmf and of the gbmf have been produced by cole et al . ( 2001 ) , bell et al . ( 2003 ) and baldry et al . ( 2004 ) , exploiting 2df , sdss and two micron all sky survey data . the estimate by bell et al . ( 2003 ) is shown by the dashed line in fig . [ fig|gsmf ] . the difference with our estimate is mainly due to the difference in adopted mlr . bell et al . ( 2003 ) have derived their mlr by fitting the broad band sed with stellar population models . the mean mlr adopted here , based on kinematic determinations , is a factor of about @xmath71 higher at high luminosities , while at very low luminosities ( @xmath72 ) it is about a factor of @xmath73 lower . however , the flatness of the gsmf at small masses conceals the difference in the mlr , while at large masses the almost exponential decline of the lf amplifies it . it is worth noticing that the determination of the mlr of low luminosity objects is hampered by many effects related to the episodic star formation history , to the presence of dust , to the irregularity of their shapes , and to the dm predominance . cole et al . ( 2001 ) presented estimates of the gsmf for two choices of the initial mass function ( imf ) : salpeter s ( 1955 ) and kennicutt s ( 1983 ) . for a salpeter imf their gsmf is consistent with ours , within our estimated uncertainties : the main difference is a @xmath74 excess for @xmath75 . for a kennicutt imf , their gsmf drops for @xmath76 about 0.2 dex lower than that by bell et al . the estimate by baldry et al . ( 2004 ) is close to that by bell et al . ( 2003 ) , as expected since they exploit very similar lfs and mlrs . all in all , methods based on kinematic measurements and on stellar population synthesis yield gsmfs and gbmfs in reasonable agreement , and establish a sound confidence interval . integrating the gsmf for @xmath77 , the mass density parameter of baryons condensed in stars associated to late - type galaxies is found to be @xmath78 where the label `` kin '' indicates that the stellar mass of galaxies has been estimated using kinematic data . the corresponding neutral gas density amounts to @xmath61 of this value and it is concentrated in late - type , low mass systems with @xmath79 . here and in the following eqs . ( [ eq|omegaearly ] ) and ( [ eq|omegagal ] ) , the errors reflect the uncertainties on the gsmf . the star density parameter associated to early - type galaxies amounts to @xmath80 it is well known that in early - types the amount of cold gas is negligible . therefore , the overall local stellar mass density in galaxies with stellar masses in the range @xmath81 is @xmath82 this value is in good agreement with the recent estimates obtained through spectro - photometric galaxy models ( bell et al . 2003 ; fukugita 2004 ; fukugita & peebles 2004 ) . the contribution of the cold gas to the baryon density in galaxies is only @xmath83 , and thus @xmath84 . this result confirms the well known conclusion that only a small fraction , @xmath85 , of the cosmic baryons is today in stars and cold gas within galaxies . it is worth noticing that the star formation rate integrated over the cosmic history matches the overall local mass density in stars ( see nagamine et al . this mass density has been accumulated at high redshifts , @xmath86 , for early - type galaxies and later on for late - types , as indicated by their respective stellar populations . by subtracting from the cosmic matter density the contribution of baryons ( @xmath87 ) and that of dark matter ( dm ) in groups and clusters of galaxies ( @xmath88 ; reiprich & bohringer 2002 ) , we obtain the dm mass density associated with galaxies @xmath89 , in excellent agreement with the determination by fukugita & peebles ( 2004 ) . the average dm - to - baryon ( essentially stars ) mass fraction in galaxies turns out to be around @xmath90 . this value must be compared with the cosmological ratio @xmath91 . in fact , in rich galaxy clusters the baryon mass , mostly in form of diffuse gas , and the dm halo mass have a ratio consistent with the `` cosmic '' dm to baryon ratio ( see , e.g. , ettori , tozzi & rosati 2003 ) . this evidences that the baryon fraction in galaxies decreases on average by a factor of about @xmath92 relative to the initial value , due to a number of astrophysical processes associated to the formation of these objects . in the following we will use the cosmic fraction @xmath93 . in order to investigate the relationships between the stellar and baryonic mass and the halo mass in galaxies in a one - to - one correspondence , the statistics of halos containing _ one _ single galaxy , the galaxy halo mass function ( ghmf ) , has to be estimated . the overall hmf as found by numerical simulations ( see , e.g. , jenkins et al . 2001 ; springel et al . 2005 ) is well described by the press & schechter ( 1974 ) formula as modified by sheth & tormen ( 1999 ) . however , in order to compute the ghmf , we have to deal with the problem of the halo occupation distribution ( hod ; peacock & smith 2000 ; berlind et al . 2003 ; magliocchetti & porciani 2003 ; kravtsov et al . 2004 ; abazajian et al . two effects need to be taken into account : ( i ) the number of galaxies is actually larger than the number of dm halos , since a halo may contain a number of sub - halos , each hosting a galaxy , and ( ii ) the probability that very massive halos ( @xmath95@xmath96 ) host a single giant galaxy drops rapidly with increasing halo mass . to account for the effect ( i ) we use the results by vale & ostriker [ 2004 ; see their eqs . ( 1 ) and ( 3 ) ] and we _ add _ to the hmf the subhalo mass function they have derived ; we have checked that this procedure does not alter substantially the overall mass density in the galactic range . the subhalo mf by van den bosch et al . ( 2005 ) is extremely close to the vale & ostriker one over the mass range of interest here , and gives essentially indistinguishable results . to account for effect ( ii ) we _ subtract _ from the hmf the halo mass function of galaxy groups and clusters . estimates of the latter obtained by different groups ( girardi & giuricin 2000 ; martinez et al . 2002 ; heinmki et al . 2003 ; pisani et al . 2003 ; eke et al . 2006 ) are in reasonable agreement for @xmath97 . at lower masses the galaxy group mass function is very uncertain , and the recent study by eke et al . ( 2006 ) finds a larger abundance of low luminosity groups than previously reported from smaller samples . on the other hand , from fig . 8 ( plus fig . 15 ) of eke et al . ( 2006 ) , it looks plausible that galaxies dominate the mf for @xmath98 . we stress , however , that , as discussed in the following , in the mass range considered here our results are only weakly sensitive to whether or not the galaxy group mass function is subtracted from the hmf . the ghmf obtained subtracting from the hmf the group and cluster mass function estimated by martinez et al . ( 2002 ) is shown in fig . [ fig|ghmf ] and is well fit , in the range @xmath99 , by a schechter function @xmath100 with @xmath101 , @xmath102 and @xmath103 . the fall off at high masses ( where early - type galaxies dominate ) mirrors the increasing probability of multiple occupation of mass halos found by magliocchetti & porciani ( 2003 ) for @xmath104 ( see also zehavi et al . weak lensing measurements also suggest an upper galaxy mass limit @xmath105 ( kochanek & white 2001 ) . if two galaxy properties , @xmath106 and @xmath107 , obey a monotonic relationship we can write @xmath108 where @xmath109 is the number density of galaxies with measured property between @xmath106 and @xmath110 and @xmath111 is the corresponding number density for the variable @xmath107 . the solution is based on a numerical scheme that imposes that the number of galaxies with @xmath106 above a certain value @xmath112 must be equal to the number of galaxy halos with @xmath107 above @xmath113 ( see , e.g. , marinoni & hudson 2002 ; vale & ostriker 2004 ) , i.e. , @xmath114 in the following @xmath115 and @xmath116 , while the variable @xmath106 will be , in turn , the luminosity , the stellar mass , the velocity dispersion , and the central bh mass . it is worth noticing that in this way we establish a direct link between the specific galaxy property and the halo mass without any assumption or extrapolation of the dm density profile . the monotonicity assumption is obviously critical for the applicability of the present approach . however direct evidence of monotonic relationships between several pairs of these quantities has been reported ( hring & rix 2004 ; ferrarese 2002 ; baes et al . 2003 ; tremaine et al . 2002 ) , and additional data supporting the relationships derived here are discussed in the following . we also implicitly assume that all galactic - size halos contain a visible galaxy . this assumption underlies all major semi - analytic models for galaxy formation , including the one by granato et al . ( 2004 ) , that we adopt as our reference model . the successes of this model in reproducing the redshift - dependent galaxy luminosity function in different wavebands provide strong support to this view . the result for the stellar mass , obtained setting @xmath117 [ eq . ( [ eq|gsmf ] ) ] is plotted in fig . [ fig|mslrmh]@xmath118 . we find that its relationship with the host halo mass is well approximated by @xmath119 the calculations for the baryonic mass is strictly analogous . setting @xmath120 [ eq . ( [ eq|lf ] ) ] we also derived the approximated behavior of the luminosity as a function of the halo mass @xmath121 and of the halo mass as a function of luminosity @xmath122~. { m_h\over 3\times 10^{11}\ , m_{\odot } } = \left[\left(\frac { l_{r}}{1.3\times 10^{10}\ , l_{\odot}}\right)^{0.35}+\left(\frac { l_{r}}{1.3\times 10^{10}\ , l_{\odot}}\right)^{1.65}\right]~.\ ] ] both stellar mass and luminosity exhibit a double power law dependence on halo mass with a break around @xmath123 , corresponding to a luminosity @xmath124 . the derivation of the @xmath2@xmath0 relation is quite sensitive to uncertainties in the lf and in the ghmf . in the range @xmath125 the lf is rather precisely known ( see the comparison of different lf estimates in fig . 1 of shankar et al . 2004 ) , while the effect of uncertainties in the ghmf becomes significant at large masses . on the whole , the @xmath2@xmath0 relation is quite accurately determined in the range @xmath126 ( fig . [ fig|mslrmh]@xmath127 ) . at low luminosities the errors on the lf rapidly increase , becoming a factor of about @xmath73 for @xmath128 . in order to illustrate the consequences on the @xmath2@xmath0 relation , we have computed it using the @xmath129 upper and lower boundaries of the lf by nakamura et al . ( 2003 ) . in the former case , the low - luminosity portion of the @xmath2@xmath0 relation flattens from a slope of @xmath130 to @xmath131 ; in the latter case , it decreases almost exponentially . therefore , the extrapolation of the above relationships to @xmath132 and , correspondingly , to @xmath133 must be taken cautiously . of course , the same conclusion holds for relationships involving other statistics of galaxies and of their host halos . this emphasizes the need of a precise determination of the low luminosity end of the lf . in fig . [ fig|mslrmh]@xmath127 our estimate of the @xmath2 vs. @xmath0 relation is compared with observational results for galaxies whose @xmath0 could be derived based on two different methods : ( i ) an x - ray - based mass model of the isolated elliptical galaxy ngc 4555 ( osullivan & ponman 2004 ) , in which the gravitational potential is known up to about @xmath134 of the virial radius ( the mass within this radius is shown as a lower limit in fig . [ fig|mslrmh]@xmath127 ) ; ( ii ) weak - lensing observations that provide the shear field around a number of galaxies of average luminosity @xmath94 , from which it is possible to infer the projected mass density and eventually to extrapolate the virial mass by assuming a dm profile ( guzik & seljak 2002 ; kleinheinrich et al . 2004 ; hoekstra et al . 2004 ) . in fig . [ fig|mslrmh]@xmath127 we also show for comparison the @xmath2 vs. @xmath0 relation obtained from eq . ( [ eq|jacobian ] ) setting @xmath135 . it is worth noticing that if the group and cluster mf is not subtracted from the hmf , our results vary only at high masses , and the changes do not exceed 0.2 dex up to @xmath136 . as a further check , our estimate of the ratio @xmath137 as a function of @xmath3 , is compared in fig . [ fig|ratio ] with estimates derived by extending to the virial radius the inner mass models of a number of giant ellipticals ( gerhard et al . 2001 ) and spirals ( persic , salucci & stel 1996 ; salucci & burkert 2000 ) . we stress that these results require extrapolations to the virial radius of the density profile , assumed to have the navarro , frenk & white ( 1996 ) shape , while our estimate does not need any assumption on dm density profile . it is apparent that these independent results are in nice agreement with our findings . the dependence of the luminosity on the halo mass has been investigated also by vale and ostriker ( 2004 ) ; their result is also shown in fig . [ fig|mslrmh]@xmath127 . they exploited the 2df galaxy luminosity function in the @xmath138-band ( norberg et al . 2002 ) , extrapolating it beyond the range of magnitudes where it was defined . the difference in the @xmath2-@xmath0 relation between our estimate and theirs is due to the steeper slope of the lf adopted by them . there is also a small difference in the normalization of the lf , but this is of minor importance . at high masses the direct comparison of the galaxy lf to the halo plus subhalo number density [ cf . their eq . ( 9 ) ] results in a slight flattening of the relation ( dot - dashed curve for vale & ostriker , dashed curve for our calculations ) . as vale and ostriker ( 2004 ) pointed out , in this way the mass term refers to the entire halo hosting the group or the cluster and not to just the galaxy halo . guzik & seljak ( 2002 ) modelled the galaxy - galaxy lensing trying to separate the central galactic contributions from contributions of the surrounding groups and clusters . their model applied to the sdss data on galaxy lensing yields @xmath139 at the characteristic luminosity @xmath140 for early - type galaxies , in keeping with our results . they also found a luminosity dependence @xmath141 , compatible with the high - luminosity slope of eq . ( [ eq|mhlr ] ) . however , we find that , at low luminosities , the slope significantly flattens toward a dependence @xmath142 , while guzik & seljak ( 2002 ) assume a single power law relation . van den bosch et al . ( 2003 , 2005 ) computed the conditional luminosity function of early- and late - type galaxies , a statistics linking the distribution of galaxies to that of the dm . they concluded that the mlr has a minimum @xmath143 at @xmath144 , consistent with eq . ( [ eq|mhlr ] ) . marinoni & hudson ( 2002 ) investigated the mlrs of the virialized systems , which include galaxies , groups and clusters . by comparing their luminosity function to the @xmath145cdm halo mass function , they concluded that the mlr has a broad minimum around @xmath146 . the slopes at low and high luminosity are @xmath147 and @xmath148 , respectively . our slope is similar at low masses , where practically all virialized systems are galaxies and thus the comparison is meaningful . the studies by eke et al . ( 2004 , 2006 ) of the variation of the mlr with size of galaxy groups is fully consistent with a minimum at approximately the same luminosity . by comparing the hmf and the lf , as we have done for local galaxies , it is possible to infer the @xmath3@xmath149 relation even at substantial redshifts . for the gsmf we use a simple linear fit [ @xmath150 , holding for @xmath151 to the data by fontana et al . ( 2004 ) at @xmath152 , and we approximate the ghmf with the hmf computed at the same redshift . the result , shown by the shaded area in fig . [ fig|mslrmh]@xmath118 , has to be taken as an upper limit since we have neglected the contribution of galaxy groups to the hmf . clearly , our estimate becomes increasingly uncertain as we approach the upper limit of the interval where gsmf is observationally estimated ; this is reflected in the increased width of the shaded area . nevertheless , the @xmath3@xmath0 relation at @xmath152 turns out to be quite close to the local one , indicating that for large galaxies the @xmath3@xmath149 relation was already in place at redshift @xmath153 in line with the anti - hierarchical baryon collapse scenario developed by granato et al . ( 2001 ; 2004 ) . sheth et al . ( 2003 ) estimated the velocity dispersion function ( vdf ) of spheroidal galaxies using a sample drawn from the sdss survey and have built a simple model for the contribution to the vdf of the bulges of spirals , which dominate at low velocity dispersions . their estimate covers the range @xmath154 km s@xmath155 km s@xmath12 . comparing the global vdf ( including both early and late type galaxies , as shown in fig . 6 of sheth et al . ) , and the ghmf with the same technique presented above [ cf . ( [ eq|jacobian ] ) and ( [ eq|criterion ] ) ] , we can derive the @xmath4@xmath0 relationship shown in panel ( c ) of fig . [ fig|mslrmh ] . for @xmath156 , the relationship is accurately represented by a simple power - law : @xmath157 while it steepens for lower halo masses . the uncertainties strongly increase for @xmath158 km s@xmath12 , corresponding to @xmath159 . note that these relationships must break down in the low-@xmath4 ( and low-@xmath0 ) regime . in fact , the close match found by cirasuolo et al . ( 2005 ) between the vdf and the virial velocity function ( derived from the halo mass distribution function integrated over redshift ) indicates that halos more massive than @xmath160 are generally associated to spheroidal galaxies or to later type galaxies with bulge velocity dispersions @xmath161 km s@xmath12 . on the other hand , the fraction of galactic halos associated to essentially bulgeless late - type galaxies must increase with decreasing @xmath0 , so that the integral of the vdf deviates from that of the ghmf and eq . ( [ eq|criterion ] ) no longer applies . figure [ fig|efficiency ] shows the ratio of the mass in stars to the initial baryon mass associated with each halo , assumed to be @xmath162 , as a function of @xmath0 . it illustrates the `` inefficiency '' of galaxies , especially of those of low halo mass , in retaining baryons . as discussed in [ sect : feed ] , the shape of the @xmath163 can be understood in terms of feedbacks : at low masses the sn feedback is very efficient in removing the gas , thus quenching the star formation ; moving toward higher masses ( for @xmath164 , corresponding to @xmath165 ) the agn feedback becomes more and more powerful , to the point of sweeping out most of the initial baryons . our result is at odds with the claim by guzik & seljak ( 2002 ) of a high efficiency , up to @xmath166 , in turning primordial gas into stars . however , the claim is based on a mlr @xmath167 in the @xmath168 band for late - type galaxies , a factor of 3 lower than the value found for the early - type ones . as the authors themselves point out , the statistical significance of this result is marginal , due to the weak lensing signal for the fainter late - type galaxy sample . the gsmfs by cole et al . ( 2001 ) and bell et al . ( 2003 ) yield similar efficiencies , which are very close to our estimates for relatively low halo masses , but lower for large masses , yet within the estimated uncertainties . the relation between the central supermassive black hole and the halo mass @xmath149 is relevant in the framework of theories for the origin and evolution of both galaxies and agns ( silk & rees 1998 ; monaco , salucci & danese 2000 ; granato et al . 2001 ; ferrarese 2002 ; granato et al . 2004 ) . to constrain such relation we adopted the procedure presented in the previous section [ eqs . ( [ eq|jacobian ] ) and ( [ eq|criterion ] ) ] , replacing the function @xmath109 with the local bh mass function ( shankar et al . we assume that each galactic halo hosts just one supermassive bh . our result is shown in fig . [ fig|mbhmh ] , where the barred area illustrates the errors due to the observational uncertainties on the bh mass function , as estimated by shankar et al . we find good agreement , within the estimated uncertainties , with the predictions of the granato et al . ( 2004 ) model . the relationship can be approximated by @xmath169 again a double power law with a break at @xmath170 is a very good representation of our results . at the high mass end , the bh mass is nearly proportional to the halo mass ( @xmath171 ) , while at low masses the relation steepens substantially ( @xmath172 ) . in fig . [ fig|mbhmh ] we also compare our estimate of the @xmath5@xmath0 relation with that of ferrarese ( 2002 ) , who first investigated this issue from an observational point of view . she derived a power - law relationship between the bulge velocity dispersion and the maximum circular velocity , @xmath173 , for a sample of spiral and elliptical galaxies spanning the range @xmath174 km s@xmath12 , and combined it with the relationship between @xmath173 and the virial velocity , @xmath175 based on the numerical simulations by bullock et al . ( 2001 ) and with the @xmath0@xmath175 relationship given by the @xmath145cdm model of the latter authors for a virialization redshift @xmath176 , to obtain a @xmath0@xmath4 relation . coupling it with one version of the observed bh mass vs. stellar velocity dispersion relationship ( @xmath177 ) she obtained @xmath178 , with @xmath1791.82 . baes et al . ( 2003 ) with the same method , but assuming @xmath180 and with new velocity dispersion measurements of spiral galaxies with extended rotation curves , yielding a slightly different @xmath173-@xmath4 relation , found @xmath181 . it is apparent from fig . [ fig|mbhmh ] that our result differs substantially in normalization , while the high - mass slope is remarkably close to that obtained by baes et al . it should be noted that the @xmath0@xmath182 relation depends on the virialization redshift . for @xmath183 [ the median virialization redshift for galaxies with velocity dispersions in the range probed by ferrarese and baes et al . , according to the analysis by cirasuolo et al . ( 2005 ; see their fig . 1 ) ] , its coefficient would be a factor of @xmath184 lower than that used by ferrarese ( 2002 ) and baes et al . ( 2003 ) and the coefficients of the @xmath185 relations would be larger by a factor of @xmath186 in the case of eq . ( 6 ) of ferrarese(2002 ) or of @xmath187 in the case of baes et al . ( 2003 ) , bringing them much closer to ours . in fact , as suggested by loeb & peebles ( 2003 ) and shown in details by cirasuolo et al . ( 2005 ) , the velocity dispersion of the old stellar population ( whose mass is related to the central bh mass ) is closely linked to halo mass and characteristic velocity at the virialization redshift . as a consistency test , we have combined the @xmath5@xmath0 relation , shown in the upper panel of fig . [ fig|mbhmh ] , with the @xmath4@xmath0 relation , shown in panel ( c ) of fig . [ fig|mslrmh ] , to obtain the @xmath5@xmath4 relation ( see ferrarese & ford 2005 for a review ) , which turns out to be consistent with the data , as shown by the lower panel in fig . [ fig|mbhmh ] . the dependence of the star and bh masses on the halo mass found in the previous section , suggests that different physical mechanisms are controlling the star formation and the bh growth above and below @xmath188 , corresponding to @xmath189 after eq . ( [ eq|msmh ] ) , and to @xmath190 ( or @xmath191 ) after eq . ( [ eq|lrmh ] ) . it is worth noticing that the analysis of a huge sample of galaxies drawn from the sdss shows that around @xmath192@xmath193 and @xmath194 there is a sort of transition in the structure and stellar ages of galaxies ( kauffmann et al . 2003 ; baldry et al . 2004 ) . the efficiency of star formation within galactic halos of different masses is the result of several processes . the most important are : ( i ) the cooling of the primordial gas within the virialized halos ( white & rees 1978 ) ; ( ii ) the injection of large amounts of energy into the ism by supernova explosions ( dekel & silk 1986 ; white & frenk 1991 ; granato et al . 2001 ; romano et al . 2002 ) and by the central quasar ( silk & rees 1998 ; granato et al . 2001 , 2004 ; lapi et al . all these processes have been implemented in the model of granato et al . a simplified , analytical rendition of this model is presented in the appendix a. the impact of stellar and agn feedback is illustrated by fig . [ fig|feedback ] . the binding energy of baryons in the dm potential well per unit baryonic mass as a function of the halo mass [ cf . ( [ eq|binding ] ) ] for @xmath195 is shown by the shaded area . to compute the overall energy per unit baryonic mass injected in the gas by supernovae ( @xmath196 ) and by agns ( @xmath197 ) , we have exploited eqs . ( [ eq|stellarfeed ] ) and ( [ eq|agnfeed ] ) , respectively , where @xmath3 and @xmath5 as functions of the halo mass are given by eqs . ( [ eq|msmh ] ) and ( [ eq|mbhmh ] ) , respectively . then we divided the overall energies by the initial baryon mass @xmath198 . figure [ fig|feedback ] shows that for large masses the gas can be efficiently removed by the agn feedback which overwhelms the binding energy . for small masses the supernova feedback dominates but appears to be insufficient to remove the gas associated to the host halo , due to the above mentioned problem that the observed @xmath3@xmath0 relation inferred from the data exhibits a too steep low - mass slope . the flattest slope allowed by the data , discussed in @xmath199[sect : ghmf ] , would largely alleviate , but not completely overcome , this problem . the relative importance of the two feedbacks depends on their efficiency in transferring the available energy to the gas . it is interesting that with the efficiencies used in fig . [ fig|feedback ] the crossing point is quite close to @xmath200 . as discussed by granato et al . ( 2004 ) and cirasuolo et al . ( 2005 ) , a more accurate evaluation of the efficiencies can be obtained by fitting statistics of galaxies and of agns , such as lfs at low and high redshift , the faber and jackson relation , and the local bh mass function . a more quantitative insight into the role of the key ingredients of the model is provided by the analytic calculations presented in the appendix a. so long as the star formation rate obeys eq . ( [ eq|dotmc ] ) , the mass in stars at a the present time @xmath201 , assumed to be @xmath202 , is given , after eq . ( [ eq|ms ] ) , by : @xmath203 where @xmath204 is the fraction of stars that survive up to now . for large halo masses , where the stellar feedback is less efficient ( @xmath205 ) , the quantity @xmath206 is a slowly decreasing function of the halo mass , so that @xmath3 is approximately proportional to @xmath0 . however , in this case , the fraction of gas turned into stars is controlled by the agn feedback , which , as shown by the full treatment by granato et al . ( 2004 ) , for @xmath207 expels an approximately constant fraction of the initial gas , thus preserving the approximate proportionality between @xmath3 and @xmath0 , in agreement with eq . ( [ eq|msmh ] ) . the effective optical depth , which rules the flow of the cold gas into the reservoir around the bh [ cf . ( [ eq|dotmres ] ) ] , is large ( @xmath208 ) for large galaxies , implying , after eq . ( [ eq|mbh ] ) , @xmath209 . as a consequence , in the high mass limit , the bh mass must have a dependence on the halo mass very similar to that of the stellar mass , in agreement with eq . ( [ eq|mbhmh ] ) . for @xmath210 the dominant term in the denominator in the right - hand side of eq . ( [ eq|msmhfeed ] ) is the effective efficiency of the sn feedback , @xmath211 [ cf . ( [ eq|sneff ] ) ] . therefore we get @xmath212 this limiting dependence has been derived theoretically also by dekel & woo ( 2003 ) with similar assumptions . on the other hand , such a relation is significantly flatter than that inferred from the data [ cf . ( [ eq|msmh ] ) and fig . [ fig|feedback ] ] . possible explanations may be that in less massive halos the initial baryon fraction is lower by effect of reheating of the intergalactic medium , hindering the infall of baryons into the shallower potential wells , or that the sn efficiency in removing the gas is higher . however , the difference must not be overemphasized , in view of the large uncertainties on the shape of the @xmath3@xmath0 and @xmath0-@xmath2 relations at low masses / luminosities , induced by our poor knowledge of the low - luminosity portion of the lf . as discussed in [ sect : ghmf ] , the data are consistent also with a flatter relation ( @xmath213 ) . since in the mass range @xmath214 the optical depth is small ( @xmath215 ) , from eqs . ( [ eq|mbh ] ) and ( [ eq|optdepth ] ) we obtain : @xmath216 thus this simple model predicts that the low mass slope of the @xmath5@xmath0 relation is steeper than that of the @xmath217 relation , because of the decrease of the optical depth with mass @xmath218 , entailing a lower capability of feeding the reservoir around the bh . interestingly , a steepening by approximately this amount is also found from our analysis of observational data [ cf . ( [ eq|msmh ] ) and ( [ eq|mbhmh ] ) ] . we computed the stellar and baryon mass function in galaxies exploiting @xmath219 ratios for stars and gas derived from galaxy kinematics . the results turn out to be in agreement with previous analyses based on stellar population models . the total baryonic mass density in galactic structures amounts to @xmath220 , of which @xmath221 resides in late - type galaxies . this result confirms the well - known conclusion that only a fraction @xmath222 of the cosmic baryons are presently in stars and cold gas within galaxies . the present - day galaxy halo mass function , i.e. , the number density of halos of mass @xmath0 containing a single baryonic core , has been estimated by adding the subhalos to the halo mass function and by subtracting the contributions from galaxy groups and clusters . such subtraction , which is required to single out galactic halos , has however a minor effect over the mass range of interest here . approximated analytic relationships between the halo mass , @xmath0 , the mass in stars , @xmath3 , and the @xmath1-band luminosity , @xmath2 , have been obtained from the functional equations @xmath223= \mathrm{\psi}(>q)$ ] , where @xmath224 is the number density of galactic halos larger that @xmath0 and @xmath225 is the number density of galaxies with either stellar mass greater than @xmath3 or luminosity greater than @xmath2 . the results are in good agreement with @xmath226 ratios inferred through x - ray mapping of the gravitational potential and through gravitational lensing . both relations exhibit a double - power law shape with a break around @xmath227 , corresponding to @xmath228 and to an absolute magnitude @xmath229 . a transition at about the same magnitude in the galaxy properties has been evidenced by kauffmann et al . ( 2003 ) and baldry et al . ( 2004 ) . an additional interesting outcome of our analysis is that the @xmath3@xmath0 relation is already established at redshift @xmath230 , in line with the theoretical expectation of the anti - hierarchical baryon collapse scenario ( granato et al . 2004 ) . applying the same technique to the local velocity dispersion function of galaxies and to the black hole mass function , we have also computed the @xmath4@xmath0 and @xmath5@xmath0 relationships . the former is quite close to a single power law @xmath231 . the latter is again a double power law breaking approximately at @xmath232 , corresponding to @xmath233 . the associated velocity dispersion , @xmath234 km s@xmath12 , is very close to the first estimate of the critical velocity dispersion for the gas removal by sn explosions given by dekel & silk ( 1986 ) , who found a critical halo velocity @xmath235 km s@xmath12 , corresponding to a critical velocity dispersion @xmath236 . as a test of our results , we combined the @xmath5@xmath0 relation [ eq . ( [ eq|mbhmh ] ) ] with the @xmath4@xmath0 relation [ eq . ( [ eq|sigmamh ] ) ] ; as shown by the lower panel of fig . [ fig|mbhmh ] , the resulting @xmath5@xmath4 relation is consistent with the observational data . the relationships we have obtained are model - independent and can be interpreted in terms of feedback effects by supernovae and agns in galactic structures . we also presented a simple feedback model , which nicely reproduces the approximate proportionalities @xmath237 observed in the high mass range , and the break of these relationships at @xmath238 . at low masses , the @xmath3@xmath0 relation derived here ( @xmath239 ) is steeper than that yielded by the model ( @xmath240 ) , and would imply that only a tiny fraction of the baryons initially associated with the halo remains within it in the form of stars ( and we know that the gas does not add much to the baryon content of galaxies ) . on the other hand , if the amount of stars formed is so low , for a standard salpeter imf the energy injected by sn explosions is insufficient to expel the residual gas if the baryon fraction is close to the cosmic value . thus , if the slope of the @xmath3@xmath0 relation really is as steep as its face value suggests , we must conclude that either the initial baryon fraction in low - mass galaxies was substantially lower than the cosmic value ( due , e.g. , to a pre - heating of the intergalactic medium hampering the infall of baryons into shallow potential wells ) or that the sn feedback in these objects was substantially stronger than in more massive galaxies . as discussed in @xmath199[sect : ghmf ] , however , the uncertainties on the low luminosity portion of the lf are large enough to allow for a flatter slope , closer to the model prediction and almost sufficient to grant the gas removal by sn feedback . the errors shown in the figures mostly reflect uncertainties in the @xmath219 ratio . we must not forget however , other error sources . for example , fig . [ fig|mslrmh]@xmath127 shows that different choices for the ghmf yield a systematic difference in the results , that , at the high - mass end , become comparable to the scatter considered in the same figure . further uncertainties come from estimates of the gsmf ; these are illustrated , in figs . [ fig|mslrmh ] , [ fig|ratio ] , [ fig|efficiency ] , and [ fig|feedback ] , by comparisons with results obtained using the gsmf by bell et al . nevertheless , our approach provides results consistent with observations , and have comparable uncertainties . moreover , since our approach bypasses any assumption on the dm profile , it could provide a valuable tool to discriminate among the different models of dm mass distribution in galaxies . our analysis has shown that the relationships presented above bear the imprint of the processes ruling the galaxy formation and evolution . models should eventually comply with them . we thank s. borgani , m. girardi , and g.l . granato for helpful discussions , and the referee for a very careful reading of the manuscript and many constructive comments that helped substantially improving this paper . this work is supported by asi and miur grants . abazajian , k. , et al . 2005 , apj , 625 , 613 baes , m. , buyle , p. , hau , g.k.t . , & dejonghe , h. 2003 , mnras , 341 , l44 baldry , i.k . , glazebrook , k. , brinkmann , j. , ivezic , z. , lupton , r.h . , nichol , r.c . , & szalay , a.s . 2004 , apj , 600 , 681 bartelmann , m. , king , l.j . , & schneider , p. 2001 , a&a , 378 , 361 bell , e.f . , mcintosh , d.h . , katz , n. , & weinberg , m.d . 2003 , apjs , 149 , 289 berlind , a.a . , & weinberg , d.h . 2002 , apj , 575 , 587 berlind , a.a . , 2003 , apj , 593 , 1 bernardi , m. , et al . 2003 , aj , 125 , 1849 binney , j. , & merrifield , m. 1998 , galactic astronomy , princeton university press blanton , m.r . , et al . 2001 , aj , 121 , 2358 borriello , a. , salucci , p. & danese , l. 2003 , mnras , 341 , 1109 bullock , j.s . , kolatt , t.s . , sigad , y. , somerville , r.s . , kravtsov , a.v . , klypin , a.a . , primack , j.r . , & dekel , a.2001 , mnras , 321 , 559 chartas , g. , brandt , w.n . , gallagher , s.c . , & garmire , g.p . 2002 , apj , 579 , 169 chartas , g. , brandt , w.n . , & gallagher , s.c . 2003 , apj , 595 , 85 cirasuolo , m. , shankar , f. , granato , g.l . , de zotti , g. , & danese , l. 2005 , apj , 629 , 816 cole , s. , et al . 2001 , mnras , 326 , 255 cyburt , r.h . , fields , b.d . , & olive , k.a . 2001 , newa , 6 , 215 de lucia , g. , et al . 2004 , mnras , 348 , 333 dekel , a. , & silk , j. 1986 , apj , 303 , 39 dekel , a. , & woo , j. 2003 , mnras , 344 , 1131 eke , v. r. , et al . 2004 , mnras , 355 , 769 eke , v. r. , baugh , c.m . , cole , s. , frenk , c.s . , & navarro , j.f . , 2006 , mnras , submitted ( astro - ph/0510643 ) ettori , s. , tozzi , p. & rosati , p. 2003 , a&a , 398 , 879 ferrarese , l. 2002 , apj , 578 , 90 ferrarese , l. , & ford , h. 2005 , sp . sc . , 116 , 523 fontana , a. , et al . 2004 , a&a , 424 , 23 fukugita , m. 2004 , iau symp . ryder , d.j . pisano , m.a . walker , and k.c . freeman ( san francisco : asp ) , 227 fukugita , m. , hogan , c.j . , & peebles , p.j.e . 1998 , apj , 503 , 518 fukugita , m. , & peebles , p.j.e . 2004 , apj , 616 , 643 fukugita , m. , shimasaku , k. , & ichikawa , t. 1995 , pasp , 107 , 945 gerhard , o. , kronawitter , a. , saglia , r.p . , & bender , r. 2001 , aj , 121 , 1936 girardi , m. & giuricin , g. 2000 , apj , 540 , 45 granato , g.l . , de zotti , g. , silva , l. , bressan , a. , danese , l. 2004 , apj , 600 , 580 granato , g.l . , silva , l. , monaco , p. , panuzzo , p. , salucci , p. , de zotti , g. , danese , l. 2001 , mnras , 324 , 757 guzik , j. , & seljak , u. 2002 , mnras , 335 , 311 hring , n. , & rix , h .- w . 2004 , apj , 604 , l89 heinmki , p. , einasto , j. , einasto , m. , saar , e. , tucker , d.l . , & mller , v. 2003 , a&a , 397 , 63 hoekstra , h. , yee , h.k.c . , gladders , m.d . 2004 , apj , 606 , 67 hopkins , p.f . , hernquist , l. , cox , t.j . , di matteo , t. , robertson , b. , & springel , v. 2005 , apj , submitted ( preprint astro - ph/0506398 ) inoue , s. , & sasaki , s. 2001 , apj , 562 , 618 jenkins , a. et al . 2001 , mnras , 321 , 372 kauffmann , g. , nusser , a. , & steinmetz , m. 1997 , mnras , 286 , 795 kauffmann , g. , et al . 2003 , mnras , 341 , 33 kawakatu , n. , & umemura , m. 2002 , mnras , 329 , 572 kennicutt , r.c . 1983 , apj , 272 , 54 kleinheinrich , m. , et al . 2004 , a&a , submitted ( preprint astro - ph/0412615 ) kochanek , c.s . , & white , m. 2001 , apj , 559 , 531 kravtsov , a.v . , et al . 2004 , apj , 609 , 35 lapi , a. , cavaliere , a. , & menci , n. 2005 , apj , 619 , 60 loeb , a. , & peebles , p.j.e . 2003 , apj , 589 , 29 loveday , j. 1998 , in dwarf galaxies and cosmology , proc . xviii moriond astrophys . meeting , ed . thuan et al . ( paris : editions frontieres ) , astro - ph/9805255 magliocchetti , m. , & porciani , c. 2003 , mnras , 346 , 186 marinoni , c. , & hudson , m. 2002 , apj , 569 , 101 martinez , h.j . , et al . 2002 , mnras , 337 , 1441 mckay , t.a . , et al . 2002 , apj , 571 , l85 mo , h.j . , mao , s. 2004 , mnras , 353 , 829 monaco , p. , salucci , p. , & danese , l. 2000 , mnras , 311 , 279 nagamine , k. , cen , r. , hernquist , l. , ostriker , j.p . & springel , v. 2004 , apj , 610 , 45 nakamura , o. , fukugita , m. , yasuda , n. , loveday , j. , brinkmann , j. , schneider , d.p . , shimasaku , k. , & subbarao , m. 2003 , aj , 125 , 1682 nath , b.b . , & roychowdhury , s. 2002 , mnras , 333 , 145 navarro , j.f . , frenk , c.s . , & white , s.d.m . 1997 , apj , 490 , 493 norberg , p. , et al . 2002 , mnras , 336 , 907 olive , k.a . 2002 , in astroparticle physics , ed . h. athar , g .- lin , and k .- w . ng ( singapore : world scientific ) , 23 osullivan , e. , & ponman , t.j . 2004 , mnras , 354 , 935 panter , b. , heavens , a.f . , & jimenez r. 2004 , mnras , 355 , 764 peacock , j.a . , & smith , r.e . 2000 , mnras , 318 , 1144 persic , m. , & salucci , p. 1992 , mnras , 258 , 14 persic , m. , salucci , p. & stel , f. 1996 , mnras , 281 , 27 pisani , a. , ramella , m. , & geller , m.j . 2003 , aj , 126 , 1677 press , w.h . , & schechter , p. 1974 , apj , 187 , 425 prugniel , p. , & simien , f. 1997 , a&a , 321 , 111 reiprich , t.h . , & bhringer , h. 2002 , apj , 567 , 716 romano , d. , silva , l. , matteucci , f. , & danese , l. 2002 , mnras , 334 , 444 salpeter , e.e . 1955 , apj , 121 , 161 salucci , p. , & burkert , a. 2000 , apj , 537 , 9 salucci , p. , & persic , m. 1999 , mnras , 309 , 923 shankar , f. , salucci , p. , granato , g.l . , de zotti , g. , & danese , l. 2004 , mnras , 354 , 1020 sheldon , e. , et al . 2004 , aj , 127 , 2544 sheth , r.k . , & tormen , g. 1999 , mnras , 308 , 119 sheth , r.k . , 2003 , apj , 594 , 225 silk , j. , & rees , m.j . 1998 , a&a , 331 , l1 spergel , d.n . , et al . 2003 , apjs , 148 , 175 springel , v. , di matteo , t. , & hernquist , l. 2005 , mnras , 361 , 776 stocke , j.t . , penton , s.v . , & shull , j.m . 2003 , in the igm / galaxy connection : the distribution of baryons at @xmath241 , ast . j.l . rosenberg and m.e . putman ( dordrecht : kluwer academic publishers ) , 57 tremaine , s. , et al . 2002 , apj , 574 , 740 vale , a. , & ostriker , j.p . 2004 , mnras , 353 , 189 van den bosch , f.c . , yang , x. , & mo , h.j . 2003 , mnras , 340 , 771 van den bosch , f.c . , tormen , g. , & giocoli , c. 2005 , mnras , 359 , 1029 white , s.d.m . , & frenk , c.s . 1991 , apj , 379 , 52 white , s.d.m . , & rees , m.j . 1978 , mnras , 183 , 341 yang , x. , mo , h.j . , & van den bosch , f.c . 2003 , mnras , 339 , 1057 zehavi , i. , et al . 2005 , apj , 621 , 22 zhao , d.h . , mo , h.j . , jing , y.p . , & brner g. 2003 , mnras , 339 , 12 zucca , e. , et al . 1997 , a&a , 326 , 477 the rate at which the gas mass @xmath242 falls toward the central star forming regions can be written as @xmath243 the infalling gas mass thus declines exponentially @xmath244 , where @xmath245 $ ] is the maximum between the cooling and the dynamical time at the virial radius , while @xmath246 is the initial gas mass . the time derivative of the cold , star forming gas mass is given by @xmath247 where @xmath248 is the star formation rate ( sfr ) , @xmath249 is the fraction of mass restituted by evolved stars ( @xmath250 for a salpeter imf ) , and @xmath251 is effective efficiency for the removal of cold gas by the supernova feedback . here @xmath252 is the number of sne per unit solar mass of condensed stars , @xmath253 the energy per sn used to remove the cold gas , and @xmath254 the binding energy of the gas within the dm halo , per unit gas mass . following zhao et al . ( 2003 ) and mo & mao ( 2004 ) , the latter quantity can be written as @xmath255 where @xmath256 is the circular velocity at the virial radius for a halo mass @xmath0 , @xmath257 is a weak function of the concentration @xmath56 , and we have assumed that , initially , the gas fraction is equal to the cosmic baryon to dark - matter mass density ratio , @xmath258 . taking into account the dependence of @xmath256 on the halo mass and redshift , we get , for @xmath86 : @xmath259 the effective efficiency is well approximated by @xmath260 further setting @xmath261 @xmath262 being the star - formation timescale , eq . ( [ eq|dotmc ] ) is easily solved for @xmath263 . the mass @xmath3 cycled through stars is then straightforwardly obtained , using eq . ( [ eq|sfr ] ) : @xmath264~,\ ] ] where @xmath265 . in the above formula , @xmath266 , since we expect that in the central , clumpy regions the cooling and dynamical times are shorter than @xmath267 , which is estimated at the virial radius . the mass in stars at the present time only includes the fraction @xmath204 of stars still surviving : @xmath268 . the survived fraction depends on the imf and on the history of star formation ; as a reference , for a salpeter imf after about 10 gyr from a burst we have @xmath269 . if we assume that most of the stellar feedback comes from sn explosions , then the total energy injected into the gas is given by @xmath270 as long as a significant amount of cool gas is present in the central regions , we can imagine that there are mechanisms able to remove angular momentum from gas clouds bringing them into a reservoir around the central bh . one of these mechanisms is the radiation drag ( kawakatu & umemura 2002 ) according to which , as shown by granato et al . ( 2004 ) , the reservoir is fuelled at a rate @xmath271 where @xmath272 is the effective optical depth of the central star forming regions [ cf . ( 14 ) , ( 15 ) and ( 16 ) of granato et al . if most of the mass in the reservoir is ultimately accreted onto the central bh , we expect @xmath273 granato et al . ( 2004 ) assumed that the effective optical depth depends on the cold gas metallicity and mass @xmath274 . the outcome of their numerical code yields , on average , @xmath275 in the mass range @xmath276 [ cf . their figs . ( 5 ) and ( 8) ] . since @xmath277 , one gets @xmath278 as for the agn feedback , we use the prescription of granato et al . ( 2004 ) ; we rewrite their eq . ( 28 ) for the kinetic luminosity that can be extracted from agn - driven winds , as @xmath279 where @xmath280 is the covering factor of the wind and @xmath281 is the gas column density in units of @xmath282 @xmath283 . if we assume that the bh mass is growing at around the eddington rate , the total kinetic energy in winds emitted by a bh of mass @xmath5 is @xmath284 . this shows that the action of the agn occurs on a short timescale , the @xmath285-folding timescale @xmath286 ( where @xmath287 is the eddington time and @xmath288 is the bh mass to energy conversion efficiency ; for @xmath289 , @xmath290yr ) . if a fraction @xmath291 of the agn kinetic energy is transferred to the gas , its total energy input is @xmath292 since studies of bal qsos suggest that @xmath293 , @xmath294 ( see , e.g. , chartas et al . 2002 ; chartas , brandt & gallagher 2003 ) and @xmath295 ( see , e.g. , inoue & sasaki 2001 ; nath & roychowdhury 2002 ) , we can take @xmath296 . it is interesting to compare this energy input to the total energy released by accretion , @xmath297 erg : @xmath298
we present new relationships between halo masses ( @xmath0 ) and several galaxy properties , including @xmath1-band luminosities ( @xmath2 ) , stellar ( @xmath3 ) and baryonic masses , stellar velocity dispersions ( @xmath4 ) , and black hole masses ( @xmath5 ) . approximate analytic expressions are given . in the galaxy halo mass range @xmath6 the @xmath0@xmath2 , @xmath3@xmath0 , and @xmath7@xmath0 are well represented by a double power law , with a break at @xmath8 , corresponding to a mass in stars @xmath9 , to a @xmath1-band luminosity @xmath10 , to a stellar velocity dispersion @xmath11 km s@xmath12 , and to a black hole mass @xmath13 . the @xmath4@xmath0 relation can be approximated by a single power law , though a double power law is a better representation . although there are significant systematic errors associated to our method , the derived relationships are in good agreement with the available observational data and have comparable uncertainties . we interpret these relations in terms of the effect of feedback from supernovae and from the active nucleus on the interstellar medium . we argue that the break of the power laws occurs at a mass which marks the transition between the dominance of the stellar and the agn feedback .
separation of the cosmic microwave background ( cmb ) signal from extragalactic and galactic foregrounds ( gf ) is one of the most challenging problems for all the cmb experiments , including the ongoing nasa and the upcoming esa mission . the gf produces the major ( in amplitude ) signal in the raw maps , which is localized at a rather small latitude band @xmath2 . to avoid any contribution of the gf to the derived cmb map , starting from to experiments , a set of masks and disjoint regions of the map are in use for extraction of the cmb anisotropy power spectrum @xcite . the question is , what kind of assumption about the properties of the foregrounds should we apply for the data processing and what criteria determines the shape and area of the mask and the model of the foregrounds ? to answer these questions we need to know the statistical properties of the gf to determine the strategy of the cmb signal extraction from the observational data sets . these questions are even more pressing for the cmb polarization . unlike temperature anisotropies , our knowledge about the polarized foregrounds is still considerably poor . additionally , we have yet to obtain a reasonable truly _ whole - sky _ cmb anisotropy maps for statistical analysis , while obtaining a whole - sky polarization map seems to be a more ambitious task . modeling the properties of the foregrounds thus needs to be done for achieving the main goals of the mission : to the cmb anisotropy and polarization signals for the whole sky with unprecedented angular resolution and sensitivity . apart from modeling the foregrounds , @xcite ( hereafter toh ) propose the `` blind '' method for separation of the cmb anisotropy from the foreground signal . their method ( see also @xcite ) is based on minimizing the variance of the cmb plus foreground signal with multipole - dependent weighting coefficients @xmath3 on k to w bands , using 12 disjoint regions of the sky . it leads to their foreground cleaned map ( fcm ) , which seems to be clean from most foreground contamination , and the wiener - filtered map ( wfm ) , in which the instrumental noise is reduced by wiener filtration . it also provides an opportunity to derive the maps for combined foregrounds ( synchrotron , free - free and dust emissions etc . ) . both fcm and wfm show certain levels of non - gaussianity @xcite , which can be related to the residuals of the gf @xcite . therefore , we believe that it is imperative to develop and refine the `` blind '' methods for the mission , not only for better foreground separation in the anisotropy maps , but also to pave the way for separating cmb polarization from the foregrounds . the development of `` blind '' methods for foreground cleaning can be performed in two ways : one is to clarify the multipole and frequency dependency of various foreground components , including possible spinning dust , for high multipole range and at the high frequency instrument ( hfi ) frequency range . the other requires additional information about morphology of the angular distribution of the foregrounds , including the knowledge about their statistical properties in order to construct realistic high - resolution model of the observable foregrounds . since the morphology of the cmb and foregrounds is closely related to the phases @xcite of @xmath4 coefficients from spherical harmonic expansion @xmath5 , this problem can be re - formulated in terms of analysis of phases of the cmb and foregrounds , including their statistical properties @xcite . in @xcite , it is reported that a major part of the gf produces a specific correlation in spherical harmonic multipole domain at @xmath6 : between modes @xmath7 and @xmath8 . the series of @xmath9-correlation from the gf requires more investigation . this paper is thus devoted to further analysis of the statistical properties of the phases of the foregrounds for such correlation . we concentrate on the question as to what the reason is for the @xmath9 correlation in the data , and can such correlation help us to determine the properties of the foregrounds , in order to separate them from the cmb anisotropies . in this paper we develop the idea proposed by @xcite and demonstrate the pronounced symmetry of the gf ( in galactic system of coordinates ) is the main cause of the @xmath9 correlation . the estimator designed in @xcite to illustrate and tackle such correlation can help us understand gf manifestation in the harmonic domain , leading to the development of `` blind '' method for foreground cleaning . in combination with multi - frequency technique proposed in @xcite , the removal of @xmath9 correlation of phases can be easily used as an effective method of determination of the cmb power spectrum without galactic mask and disjoint regions for the data . it can serve as a complementary method to the internal linear combination method @xcite and to the toh method as well , in order to decrease the contamination of the gf in the derived maps . such kind of correlation should be observed by the mission and will help us to understand the properties of the gf in details , as it can play a role as an additional test for the foreground models for the mission . this paper is organized as follows . in section 2 we describe the @xmath10 estimator for @xmath9-correlation in the coefficients @xmath4 and its manifestation in the observed signals . in section 3 we apply the estimator on 3 toy models which mimics galactic foregrounds to investigate the cause of such correlation . in section 4 we discuss the connection between the @xmath9 correlation and the foreground symmetry . we also examine the power spectrum of the estimator and the correlations of @xmath10 estimator in section 5 . the conclusion is in section 6 . as is shown in @xcite to illustrate the @xmath9-correlation , we recap the estimator taken from the combination of the spherical harmonic coefficients @xmath4 , @xmath11 where @xmath12 , and the coefficients @xmath13 are defined by the standard way : @xmath14 @xmath15 is the whole - sky anisotropies at each frequency band , @xmath16 are the polar and azimuthal angles of the polar coordinate system , @xmath17 are the spherical harmonics , @xmath18 and @xmath19 are the amplitudes ( moduli ) and phases of @xmath20 harmonics . the superscript @xmath21 in @xmath10 characterizes the shift of the @xmath22-mode in @xmath23 and , following @xcite , we concentrate on the series of correlation for @xmath24 , @xmath25 . note that the singal of galaxy mostly lies close to @xmath26-plane . the estimator @xmath10 in form eq.([eq1 ] ) is closely related with phases of the multipoles of the @xmath15 signal . taking eq.([eq2 ] ) into account , we get @xmath27 . \label{eq3}\ ] ] from eq.([eq3 ] ) one can see that , if the phase difference @xmath28 , then @xmath29 if @xmath30 , the map synthesized from the @xmath23 estimator is simply a map from the @xmath4 with phases rotated by an angle @xmath31 and the amplitudes lessened by a factor @xmath32 , while for non - correlated phases @xmath33 we have specific ( but known ) modulation of the @xmath4 coefficients ( see the appendix ) . a non - trivial aspect of @xmath23 estimator is that it significantly decrease the brightest part of the galaxy image in the k - w maps . in the following analysis we use a particular case @xmath34 so that @xmath35 , although it can be demonstrated that for @xmath36 the results of analysis do not change significantly as long as @xmath37 , where @xmath38 is the multipole number in the spectrum where the instrumental noise starts dominating over the gf signal . in this section we show how the @xmath39 estimator transforms the gf image in the k - w maps , taking from the nasa lambda archive @xcite . in fig.[fig1 ] we plot the maps synthesized from the @xmath10 estimator for k - w band ( for @xmath35 and @xmath40 ) @xmath41 note that the amplitudes are significantly reduced in each map . it should be emphasized that the @xmath42 map is not temperature anisotropy map , as the phases are altered . let us discuss some of the properties of the @xmath39 estimator , which determine the morphology of the @xmath42 maps . first of all , from eq.([eq1 ] ) one can find that for all @xmath43 modes , the estimator is equivalent to zero if @xmath44 and it is non - zero ( and doubled ) if @xmath45 . in terms of phase difference in eq.([eq4 ] ) this means that for @xmath43 modes @xmath39 estimator removes those which have the same phases , while doubles the amplitudes of others whose phases differing by @xmath46 in the @xmath42 maps . however , such specific case of @xmath23 estimator for @xmath43 modes is not unique for @xmath24 only . it seems typical for any values of @xmath21 parameter . what is unique in the data is that for @xmath47 the order of sign for @xmath43 modes leads to the @xmath42 image without strong signal from the galactic plane . we present in fig.[fig2 ] the images synthesized of the even and odd @xmath48 modes from the w band signal . the even and odd @xmath48 modes reflect different symmetry of the signal , related to the properties of the spherical harmonics and the corresponding symmetries of the foregrounds . for even @xmath48 the brightest part of the signal is mainly localized in the galactic plane area ( the top panel ) , while for odd @xmath48 modes the signal has less dominated central part from the gf , but it has well presented periodic structure in @xmath49 direction ( horizontal stripes ) , from the north to south pole caps crossing the galactic plane . in fig.[fig3 ] we present the symmetry of the gf for the w band signal for even and odd @xmath22 harmonics , including all corresponding @xmath48 modes . as one can see from fig.[fig2]-[fig3 ] , the even @xmath48 and the even @xmath22 maps ( the top of fig.[fig2 ] and fig.[fig3 ] ) have a common symmetrical central part , which looks like a thin belt covered in @xmath50 area and all @xmath51 range . for odd modes the brightest gf mainly concentrate locally in @xmath49 and @xmath52 rectangular areas . additionally , for the maps of even and odd @xmath22 harmonics in fig.[fig3 ] we have periodic structure of the signal in @xmath49 direction , which is determined by the properties of the spherical harmonics , and , more importantly , by the properties of @xmath7 coefficients of decomposition , which reflect directly corresponding symmetry of the gf . in this section we want to examine why the @xmath9 correlation appears . in order to answer this question we introduce 3 toy models for the galaxy emissivity , which reflect directly different symmetries of the galactic signal . in appendix we analyze more general situation . these 3 toy models are the belt , the rectangular and the spots models . all 3 models are the simple geometrical shapes added with the ilc map . 1 . the belt model : + we add on top of the ilc map with @xmath53 if @xmath54,and @xmath55 $ ] , where @xmath56 is the halfwidth of the belt . we set @xmath57 mk and @xmath58 . 2 . the rectangular model : + we add on top of the ilc map with @xmath59 for @xmath60 , and @xmath61 . we further use two variants * @xmath58 and @xmath62 , * @xmath58 and @xmath63 . 3 . the spots model : + this model is to mimic the properties of symmetric bright point - like sources convolved with a gaussian beam @xmath64 with fwhm=@xmath65 . so on ilc map we add + @xmath66 + where @xmath67 is the dirac-@xmath56 function , and the amplitudes of the point sources @xmath68 are in order of 10 mk . below we examine these toy models to see how they can illustrate the @xmath9 correlation . ideologically this model best illustrates that the symmetry of the gf signal can help remove the gf itself without any additional assumption about properties of the foregrounds and of galactic mask as well . the theoretical basis is that the properties of the phases for the belt signal are related to @xmath43 modes ( see below ) . thus , any method that can remove all @xmath43 modes from the map is automatically able to remove the belt - like galactic signal and reproduce the ilc signal for all @xmath69 modes without any errors . however , these @xmath43 modes contribute significantly to the low multipole part of the power spectrum @xmath70 for the reconstructed cmb signal ( which is in our model the ilc signal for @xmath71 modes ) . using @xmath23 estimator we can avoid the problem of reconstruction of the @xmath43 modes and the corresponding power spectrum of the @xmath42 map ( see section [ power ] ) . in fig.[sim ] we plot the resultant @xmath42 maps for the belt - like model . as one can see , the particular case seems peculiar for the @xmath72 estimator , which does not remove the galactic signal properly . according to the private communication with eriksen et al . , this simple model of the galactic signal should reflect the properties of spherical harmonics , namely the @xmath9 correlation mentioned in @xcite . the symmetry of this model is related to @xmath73 correlation of even harmonics , while for odd harmonics the galactic signal must vanish . however , for some of the @xmath43 harmonics the @xmath74 estimator differs from zero , and the corresponding stripes determines the morphology of the image ( see fig.[sim ] , the second from the top ) . to show that effect clearly , in fig.[sim ] ( the third ) we plot the map , which contains the non - zero @xmath74 modes . remarkably , the morphology of defects in the second and third one are practically the same . to demonstrate that , in the @xmath42 map ( the second in fig.[sim ] ) we simply remove all non - zero @xmath74 modes and get the map shown in fig.[sim ] as the 4th . so , the second and the 4th maps are the same , excluding all non - zero modes @xmath74 from the @xmath42 map . this result is not surprising , if one takes into account the symmetry of the belt signal , all the phases of @xmath75 harmonics for @xmath69 are exactly those of @xmath71 harmonics in the ilc map , while all @xmath76 in the first from the top in fig.[sim ] map are mainly determined by the belt signal ( galaxy ) . to show that in fig.[sim ] we simply remove all @xmath76 modes from the ilc plus the belt map and the resultant is shown in fig.[sim ] the bottom map . one can see , that this is the ilc map for all @xmath77 harmonics . what properties of this toy model are crucial in producing the @xmath9 multipole correlation of the phases ? taking the definition of the belt signal into account , one can obtain that for the belt signal @xmath78 , \label{bel}\ ] ] where @xmath79 , @xmath80 is the kronecker symbol , @xmath81 is the legendre polynomials and @xmath82 . for @xmath83 the asymptotic of @xmath84 is @xcite @xmath85\delta_{\l,2n}. \label{bel1}\ ] ] then for @xmath86 from eq.([eq2 ] ) we obtain @xmath87 \right . \nonumber \\ & & \left . -\frac{|\sin\left[(\l+\frac{1}{2})\delta\right]| } { |\sin\left[(\l+\delta+\frac{1}{2})\delta\right]| } \sin\left[(\l+\delta+\frac{1}{2})\delta\right]\right\}. \label{bel2}\end{aligned}\ ] ] as is seen from eq.([bel2 ] ) , @xmath86 depends on the sign of the first and second terms in the brackets . let us assume that for some @xmath88 the first term is positive and the second term is negative . in this case @xmath89 \sin\left[(\l+\delta+\frac{1}{2})\delta\right ] , \label{bel3}\end{aligned}\ ] ] and unlike compensation of the modes @xmath90 , we have the signal in order of magnitude close to @xmath91 . moreover , taking into account that the amplitude of that signal decreases very rapidly in comparison with amplitudes of the cmb signal , starting from some multipoles @xmath92 the order of sign in eq.([eq2 ] ) is determined by the cmb signal and the number of horizontal stripes increase rapidly , as it is seen from the fig.[sim ] ( the second map from the top ) . thus , for a given `` belt '' model of the galaxy emissivity for the range of multipoles @xmath93 the @xmath9 multipole correlation does not exist . however , in the opposite approximation @xmath94 the properties of the multipole coefficients are determined by the asymptotic of legendre polynomials in eq.([bel ] ) @xmath95^{\frac{1}{2 } } , \label{bbel}\end{aligned}\ ] ] and formally @xmath96 for all @xmath97 , @xmath98,but @xmath99 . as follows from eq.([bbel ] ) , the symmetry of the belt model does not require @xmath9 correlation at all . if @xmath21 parameter is even @xmath100 , the sum @xmath101 is even too , and the corresponding @xmath84 coefficient vanishes . the case for @xmath24 is a special one for the more general correlation @xmath100 , which will be broken at the limit @xmath83 . however , the belt model as discussed above provides a useful estimation of the properties of the @xmath10 estimator for more general cases , when the symmetry of the gf is not so high as in the belt toy model . particularly , to prevent any contribution to the @xmath42 map from the highly symmetrical part of the gf , we will use further generalization of the @xmath10 estimator as follows : @xmath102 we will use this generalization for @xmath10 in all the subsequent discussions ( excluding descriptions of the toy models in 3.2 and 3.3 ) . let us discuss the properties of the model which mimics the galaxy image in the map as a rectangular area , characterized by halfwidth @xmath56 in @xmath49 direction and @xmath103 in @xmath52 direction and with a constant amplitude of the signal @xmath104 in the area @xmath105 , @xmath106 . for that model the corresponding @xmath7 coefficients are @xmath107^{\frac{1}{2}}\frac{\sin(m\phi)}{m } \nonumber\\ & \times & \int_{\pi/2-\delta}^{\pi/2+\delta}d\theta\sin\theta p_\l^m(\cos\theta ) . \label{rec}\end{aligned}\ ] ] the properties of the integral in eq.([rec ] ) depend on the parameter @xmath108 . if @xmath99 , then @xmath109 , where @xmath110^{-\frac{1}{2 } } , \label{rec1}\ ] ] while for @xmath83 we get @xcite @xmath111 . \label{rec2}\end{aligned}\ ] ] thus , for these asymptotics we have @xmath112^{\frac{1}{2 } } \label{rec3}\end{aligned}\ ] ] if @xmath99 , and @xmath113^{\frac{1}{2 } } \left[\frac{\gamma(\l+m+1)}{\gamma(\l+\frac{3}{2})}\right ] \nonumber\\ & & \times \left\ { \cos(\frac{\pi(\l+m)}{2})\sin \left[(\l+\frac{1}{2})\delta \right]- \right . \nonumber\\ & & \left.\frac{(2\l+1)\delta}{\pi(\l+m)}\sin\left[\frac{\pi(\l+m)}{2}+(\l+\frac{1}{2})\delta\right ] \right \ } \nonumber\\ \label{rec4}\end{aligned}\ ] ] if @xmath83 . for odd @xmath114 , as seen from eq.([rec3 ] ) , @xmath115 if @xmath99 is in order of magnitude @xmath116 . that means that the main term in eq.([rec ] ) , proportional to @xmath117 is related with the even harmonics @xmath118 . the leading term , which determines the sign of @xmath119 is @xmath120 in the denominator of eq.([rec4 ] ) . for that term we have @xmath121 and for @xmath122 the sign of @xmath119 is the same , as for @xmath123 . taking into account that for @xmath124 estimator the order of signs for @xmath75 and @xmath125 is crucial , we can conclude that the compensation of the central part of the signal requires @xmath122 . however , it does not guarantee that for @xmath83 modes @xmath126 criteria leads the compensation of the signal . to show this let us describe the asymptotic @xmath83 , when the symmetry of the @xmath127 coefficients is determined by eq.([rec2 ] ) . as one can see from this equation , if @xmath128 the sign of @xmath119 is now determined by the combination @xmath129 $ ] and does not show any @xmath130-correlation at all . moreover , according to the properties of the sine mode the shift of the argument @xmath22 by the factor @xmath126 just transforms it to the combination @xmath131&=&\sin\left[(\l+\frac{1}{2})\delta\right]\cos(4k\delta)\nonumber\\ & + & \cos\left[(\l+\frac{1}{2})\delta\right]\sin(4k\delta ) . \label{sin}\end{aligned}\ ] ] thus , one can see that @xmath130-correlation requires some restriction on the @xmath56-parameter @xmath132 thus , for @xmath133 and @xmath134 the halfwidth of the rectangular area must be close to @xmath135 . if , for example , @xmath136 then we will have correlation for @xmath137 , but not for @xmath133 . practically speaking , for @xmath138 we can have some particular symmetry , but not general symmetry @xmath24 . conclusions concerning this rectangular model of gf are clearly seen in fig.[fign ] . to understand how each sort of defects is related with corresponding @xmath9-correlation , we introduce the model of defects , which can be describe as a sum of peaks with amplitudes @xmath68 and coordinates @xmath139 . for analytical description of the model we neglect the beam convolution of the image of point sources ( ps ) , but we include it in the numerical simulation . for the model of defects @xmath140 for the @xmath4 coefficients of the spherical harmonics expansion from eq.([ps ] ) we get @xmath141 as for the rectangular model , we will assume that all @xmath142 and simply we will have @xmath143 for @xmath83 , as well as @xmath99 . as it was shown in previous section , @xmath144 clearly demonstrate @xmath9 correlation . we would like to point out that for the spots model this correlation now is strong , unlike the belt and rectangular models .- correlation . ] moreover , implementation of the gaussian shape of the ps which come from beam convolution does not change that symmetry at all . to show that , in fig.[ps ] we plot the model of two ps with amplitudes in order to 10 mk , combined with the ilc map . the reason for such effect is quite obvious . the beam convolution does not change the symmetry of the model , but rescale the amplitudes of the ps by a factor @xmath145/2\sigma^2 $ ] , if we assume the gaussian shape of the beam . an important question is that is the symmetry of the galaxy image in @xmath52 direction important for extraction of the brightest part of the signal , or is the effect simply determined by the symmetry of the galaxy image in @xmath49 direction ? to answer this question , in fig.[ps5 ] we plot the result from the @xmath23 estimation in the model with 5 spots located at @xmath142 with different amplitudes and different @xmath146 . as one can see no symmetry in @xmath52 direction was assumed . the result of reconstruction clearly shows that location of the sources in the galactic plane in @xmath52 direction is crucial . unlike the model with symmetric location of the spots in fig.[ps ] , now the residuals of the extraction of the spots dominate over the rest of the signal in the galactic plane . however , as is seen from fig.[ps5 ] , the @xmath9-correlation of image exists . the properties of the @xmath7 coefficients in the spots model are related with the sum ( see eq.([ps1 ] ) ) @xmath147 actually , eq.([aa ] ) determines the phases of the @xmath7 coefficients . let us discuss the model of two symmetrically situated ps with the same amplitudes @xmath148 and @xmath149 , @xmath150 . for this particular case @xmath151=0 $ ] , while @xmath152=2a\cos(\pi m/2 ) $ ] . for @xmath153 we get @xmath152={\rm i m } [ s(m)]=0 $ ] and the contribution of the strong signal to the map ilc + two ps vanishes . this means that for amplitudes @xmath154 , where @xmath155 and @xmath156 correspond to the ilc and ps signals respectively ) and phases @xmath157 of @xmath7 coefficients , we have @xmath158 where @xmath159 are the ilc phases . as one can see , this is a particular example , when strong , but symmetric in @xmath52 direction signal do not contribute to the set of @xmath127 coefficients at least for the defined range of multipoles . let us discuss the other opposite model , in which the number of spots in the galactic plane is no fewer than 2 , and their @xmath146 coordinates are random in some range @xmath160 . no specific assumptions about the amplitudes are needed . in this model the sum @xmath161 in eq.([aa ] ) mostly is represented by @xmath43 modes , @xmath162 , while all @xmath69 modes @xmath163 because of randomness of the phases @xmath164 . this model , actually is close to the rectangular model , in which the width of rectangular side in the @xmath52 direction now is @xmath160 . at the end of this section we would like to demonstrate , how the symmetry of the galaxy image in @xmath52 direction can determine the properties of the @xmath165 map . for that we rotate the w - band map by @xmath166 along the pole axis and produce the same estimation of @xmath23 , as is done for the galactic reference system . the result of estimation is shown in fig.[rot ] . for comparison , in this figure we plot the difference and sum between @xmath42 maps before and after rotation . from these figures , these new symmetry of the w band map after rotation simply increase the amplitude of signal in galactic plane zone , especially in the the central part of it . at the end of this section , we summarize the main results of investigation of the given models of the gf signal . * for highly symmetrical signal , like the belt model , all @xmath127 coefficients vanish for the multipole numbers @xmath167 , but @xmath168 . modification of the @xmath39 estimator in form of eq.([eq1b ] ) is crucial to prevent any contribution to the @xmath42 map from the gf . * less symmetric model , like the rectangular model , requires @xmath9 transition for the multipole numbers for @xmath23 estimator , which appears for the range of multipoles @xmath168 . if the resolution of the map we are dealing with is low , @xmath169 that @xmath9 correlation appears for all @xmath170 and the corresponding @xmath127 coefficients for gf are in order of magnitude @xmath171 . for @xmath172 the @xmath9 correlation of phases does not exist at all . * the amplitude of the gf signal , @xmath104 , and its dependency on @xmath22 ( like @xmath173 , @xmath174 is the minimal multipole number for which @xmath175 achieve the maxima , and @xmath176 is the power index ) are crucial for establishing of the @xmath9 correlation of phases . taking asymptotic @xmath168 into account , and defining the critical multipole number @xmath177 , we can estimate the corresponding amplitudes @xmath178 . if at that range of @xmath22 we get @xmath179 , where @xmath180 is the cmb power spectrum , the @xmath9 correlation would be established for all range of multipoles @xmath170 , even if it vanishes for the gf signal for @xmath172 . starting from @xmath181 and for @xmath182 the corresponding @xmath7 for gf play a small role in amplitude noise , in comparison to the amplitudes of the cmb signal . * the @xmath23 estimator effectively decreases the amplitudes of the point - like sources located in the galactic plane , if they have nearly the same amplitudes and are symmetrically distributed in @xmath52 direction around galactic center . non - symmetrical and different in amplitudes point - like sources after implementation of @xmath23 estimator produce significant residues . in this section we apply the proposed @xmath23 estimator to the maps for q , v and w band foregrounds ( which are sum of synchrotron , free - free and dust emission ) . we then transform them by the @xmath23 estimator . these foreground maps do not contain the cmb signal and instrumental noise , therefore they allow us to estimate the properties of the gf in details . in fig.[fgd ] we plot the @xmath42 maps for q , v and w band foregrounds ( @xmath35 ) for the multipole range @xmath183 . this range is determined by the resolution of the foregrounds maps ( @xmath184 ) . as one can see from these maps , the gf perfectly follows to @xmath9 multipole correlation , which remove the brightest part of the signal down to the level @xmath185 50 mk for the q band , @xmath186 mk for the v band , @xmath187 mk for the w band and @xmath188 mk for the @xmath42 map , the difference between v and w foregrounds . note that these limits are related with the brightest positive and negative spots ( point sources ) in the maps , while diffuse components have significantly smaller amplitudes . to show the high resolution @xmath42 map which characterizes the properties of the foregrounds in v and w band , in fig.[pow ] we plot the map of difference @xmath189 bands , and the corresponding @xmath42 map for @xmath190 . note that @xmath191 map does not contain the cmb signal , but for high @xmath22 the properties of the signal are determined by the instrumental noise . to characterize the power spectrum of the @xmath42 maps we introduce the definition @xmath193 if the derived @xmath23 signal is gaussian , that power represents all the statistical properties of the signal . for non - gaussian signal , @xmath194 power characterizes the diagonal elements of the correlation matrix . from fig.[psim ] it can be clearly seen that for foregrounds , especially for v and w bands , the power spectra of @xmath42 are significantly smaller than the power of the cmb , for estimation of which we simply use the power of toh fcm map , transformed by @xmath23 estimator as @xmath195 assuming that fcm map is fairly clean from the foreground signal . an important point of analysis of the foregrounds is that for v and w bands @xmath23 estimator decreases significantly the amplitude of gf , practically by 1 to 2 order of magnitude below the cmb level . the most intriguing question related to @xmath9-correlation of the derived map from the v and w band signals is what is reproduced by the @xmath23 estimator ? the next question , which we would like to discuss is why the power spectrum of @xmath23 estimation of the v and w bands shown in fig.[fig22 ] are practically the same at the range of multipoles @xmath196 , when we can neglect the contribution from instrumental noise to both channels and differences of the antenna beams . the equivalence of the powers for these two signals , shown in fig.[fig22 ] , clearly tell us that these derived maps are related with pure cmb signal ( which we assume to be frequency independent ) . in this section we present some analytical calculations which clearly demonstrate what kind of combinations between amplitudes and phases of the cmb signal in the v , w bands and phases of foregrounds are represented in the @xmath23 estimator . as was mentioned in section 1 , this estimator is designed as a linear estimator of the phase difference @xmath197 , if the phase difference is small . let us introduce the model of the signal at each band @xmath198 , where @xmath155 is frequency independent cmb signal and @xmath199 is the sum over all kinds of foregrounds for each band @xmath200 ( synchrotron , free - free , dust emission etc . ) . according to the investigation above on the foreground models , it is realized that without the ilc signal the @xmath23 estimation of the foregrounds , especially for v and w bands , corresponds to the signal to simplify the formulas ] @xmath201 the power of which is significantly smaller then that of the cmb @xmath202 in terms of moduli and phases of the foregrounds at each frequency band @xmath203 where @xmath204 and @xmath159 are the phases of foreground and the cmb , respectively . and from eq.([dd0 ] ) we get @xmath205 and practically speaking , we have @xmath206 . thus , taking the @xmath9 correlation into account , we can conclude that it reflects directly the high correlation of the phases of the foregrounds , determined by the gf . moreover , if any foreground cleaned cmb maps derived from different methods display the @xmath9 correlation of phases , it would be evident that foreground residuals still determine the statistical properties of the derived signal . one of the basic ideas for comparison of phases of two signals is to define the following trigonometric moments for the phases @xmath207 and @xmath157 as : @xmath208 where @xmath209 . we apply these trigonometric moments to investigate the phase correlations for toh fcm and wfm . for that we simply substitute @xmath210 in eq.([def2 ] ) , and define @xmath159 as the phase of fcm and @xmath157 as that of wfm . the result of the calculations is presented in fig.[comp ] . from fig.[comp ] it can be clearly seen that the fcm has strong @xmath211 correlations starting from @xmath212 which rapidly increase for @xmath213 , while for wfm these correlations are significantly damped , especially at low multipole range @xmath214 . however , the @xmath23 estimator allow us to clarify the properties of phase correlations for low multipole range . the idea is to apply @xmath23 estimator to fcm and wfm , and to compare the power spectra of the signals obtained before and after that . according to the definition of @xmath23 estimator , the power spectrum of the signal is given by eq.([pp1 ] ) , which now has the form @xmath215 . \label{dd}\ ] ] the last term in eq.([dd ] ) corresponds to the cross - correlation between @xmath75 and @xmath216 modes , which should vanish for gaussian random signals after averaging over the realization . for a single realization of the random gaussian process this term is non - zero because of the same reason , as well known `` cosmic variance '' , implemented for estimation of the errors of the power spectrum estimation ( see naselsky et al . thus @xmath217 and error of @xmath194 is in order to @xmath218 to evaluate qualitatively the range of possible non - gaussianity of the fcm and wfm , in fig.[fpow ] we plot the function @xmath219/[d(\l)+c(\l)]$ ] for fcm and wfm , in which we mark the limits @xmath220 . as one can see , potentially dangerous range of low multipoles is @xmath221 , @xmath222 , @xmath223 for the wfm . non - randomness on some of the multipole modes is mentioned in @xcite . at the end of this section we would like to demonstrate that application of @xmath23 estimator to maps with foregrounds residuals , such as the fcm , provides additional `` cleaning '' . in fig.[clean ] we present the @xmath224 and @xmath225 trigonometric moments for the fcm with shift of the multipoles @xmath226 . one can see that the @xmath227 correlation of phases is strong ( practically , they are at the same level as @xmath228 correlations ) . however , after @xmath23 filtration these correlations are significantly decreased . the implementation of the @xmath229 estimator to the non - gaussian signal significantly decreases these correlations . the properties of the @xmath23 estimator described can manifest themselves more clearly in terms of images of the cmb signal . in fig.[clean1 ] we plot the results of the maps with @xmath23 implemented on fcm and wfm , in order to demonstrate how the estimator works on the non - gaussian tails of the derived cmb maps . in fig.[clean1 ] we can clearly see that the morphology of the @xmath42 maps are the same and difference between @xmath230 and @xmath231 is related to point sources residuals localizes outside the galactic plane ( see the 3rd panel ) . a direct substraction of the wfm from the fcm reveals significant contamination of the gf residuals and non -galactic point sources ( the third from the bottom and bottom maps ) . the second from the bottom map corresponds to difference between @xmath230 and @xmath231 for which the amplitudes of the signal represented in colorbar limit @xmath232 mk . one can see that the gf is removed down to the noise level . in combination of the phase analysis we can conclude that the implementation of the @xmath23 estimator looks promising as an additional cleaning of the gf residuals and can help investigate the statistical properties of derived cmb signals in more detailed . in this paper we examine a specific group of correlations between @xmath22 , which is used as an estimation of the statistical properties of the foregrounds in the maps . these correlations , in particular , among phases are closely related to symmetry of the gf ( in galactic coordinate system ) . an important point of analysis is that for the foregrounds the correlations of phases for the total foregrounds at v and w bands have specific shape when @xmath233 . these correlations can be clearly seen in the w band of the data sets down to @xmath40 and must be taken into account for modeling of the foreground properties for the upcoming mission . we apply the @xmath23 estimator to the toh fcm , which contains strong residuals from the gf and show that these residuals are removed from the @xmath231 map . moreover , in that map the statistics of the phases display the gaussian statistics closer than the original fcm ( no correlation of phases between different @xmath20 modes except between @xmath125 and @xmath75 , which is chosen as a basic one , defined by the form of @xmath23 estimator . ) in this paper we do not describe in details the properties of the signal derived by @xmath23 estimator from the v and w bands . further developments of the method , including multi - frequency combination of the maps and cmb extraction by the estimator will be in a separate paper . to avoid misunderstanding and confusion , here we stress again that any @xmath42 maps synthesized from the @xmath234 are _ by no means _ the cmb signals ( since the phases of the these signals are not the phases of true cmb ) and the true cmb can be obtained after multi - frequency analysis , which is the subject of our forthcomming paper . we thank h.k . eriksen , f.k . hansen , a.j . banday , c. lawrence , k.m . gorski and p.b . lilje for their comments and critical remarks . we acknowledge the use of nasa legacy archive for microwave background data analysis ( lambda ) and the maps . we also acknowledge the use of the @xcite and the package @xcite to produce @xmath4 from the data sets . in this appendix we would like to describe general properties of the @xmath9 periodicity of the galactic signal , taking into account its symmetry . we adopt the following model of the signal , which seems to be general . let define some area around galactic plane @xmath235 , where @xmath236 is the pixel area and index @xmath200 mark the location of the pixel . we assume for simplicity that all the pixels in the map are have the same area . in polar system of coordinates corresponding angles @xmath237 and @xmath146 mark the position of @xmath200-th pixel in the map . let us define the amplitude of the signal per each pixel as @xmath238 . thus the map which corresponds to the galactic signal is now @xmath239 let assumes that galaxy image is localized in @xmath49-direction as @xmath240 and it could be or could not be localized in @xmath52-direction . additionally we will assume that signal per each pixel @xmath241 is the sum of galactic foreground signal @xmath242 and cmb plus instrumental noise signal @xmath243 . important to note that statistical properties of these two components are different as in terms of amplitudes , as in terms of pixel - pixel correlations @xmath244 . particularly , in the area @xmath245 we have @xmath246 , while outside @xmath245 we assume that @xmath247 . using proposed model of the signal in the map we can obtain corresponding @xmath7 coefficients of the spherical harmonic expansion which can be represented as a sum of foreground @xmath249 coefficients and the cmb plus noise coefficients @xmath155 . in order to understand the nature of @xmath9-periodicity of the galactic foreground , let discuss the model when @xmath250.then , from eq.([eq1 ] ) the subject of interest would be the phases of foreground @xmath204 related to the @xmath249 coefficients as follows @xmath251 where sum over @xmath200 corresponds to the pixel in the area @xmath245 . let s define the difference of phases , using their tangents . @xmath252 where @xmath253 as one can see from eq.([a4 ] ) , if @xmath254 , then @xmath255 , which determine the properties of @xmath23 estimator for correlated phases ( see eq.([eq1 ] ) . below the object of out investigation is function @xmath256 from eq.([a5 ] ) . particularly we are interesting in asymptotic @xmath257 , which should reflect directly the symmetry of the foreground signal . simple algebra allows us to represent @xmath256 function in the following form @xmath258 taking into account that area @xmath245 is located close to the @xmath26 , let us discuss the properties of n - function at the limit @xmath259 , using asymptotic of the legendre polynomials . after simple algebra we obtain @xmath260 where @xmath261 + \cos[(\l+\frac{1}{2})(\theta_j-\theta_k)]\}(\cos\theta_k\delta-\cos\theta_j\delta ) \nonumber \\ -\{\sin[(\l+\frac{1}{2})(\theta_j+\theta_k)+m\pi-\frac{\pi}{2 } ] -\sin[(\l+\frac{1}{2})(\theta_j-\theta_k)]\ } ( \sin\theta_k\delta-\sin\theta_j\delta ) . \label{a8}\end{aligned}\ ] ] from eq.([a8 ] ) one can see the symmetry of the legendre polynomials which manifest themselfs trough @xmath262 and @xmath263 modes . if @xmath142 , then depending on @xmath21 we will have @xmath264 thus , choosing @xmath265 mode we take the corresponding properties of the legendre polynomials into consideration . however , as one can see from eq.([a9 ] ) @xmath265 periodicity of the galactic image is not exact . in reality we have pixels which containt galaxy signal having @xmath266 coordinate close to @xmath267 , but not exactly equivalent to @xmath267 . let us introduce a new variable @xmath268 characterized the deviation of the @xmath200-th pixel location from the @xmath269 plane . from eq.([a8 ] ) one can find that for @xmath265 @xmath270 . thus , if pixel @xmath200 containt the signal from the galactic foreground , the deviation from the center of the galactic plane should be small enough:@xmath271 . it is clear that this condition does not necessarily correspond to the properties of the galactic image , which is clearly seen from the k , ka and q band signals . taking the above - mentioned properties of @xmath272 function , we represent the asymptotic of this function at the limit @xmath271 , which is applicable for analysis of the galactic signal at v and w bands . @xmath273+(-1)^{\l+m } \cos[(\l+\frac{1}{2})(\delta_k+\delta_j)]\right\ } \nonumber \\ + \delta(\delta_j-\delta_k ) \left\{\sin[(\l+\frac{1}{2})(\delta_k-\delta_j)]+(-1)^{\l+m } \sin[(\l+\frac{1}{2})(\delta_k+\delta_j)]\right\}. \label{a10}\end{aligned}\ ] ] thus , combining eq.([a7 ] ) and eq.([a10 ] ) , we obtain @xmath274 one may think that choose of described above @xmath9-mode of legendre polynomials automatically guarantee cancellation of the brightest part of the signal from the map without any restriction on symmetry and amplitude of the foreground . to show that symmetry of the galactic signal is important , let us discuss a few particular cases , which illuminate this problem more clearly . firstly , let take a look at galactic center ( gc ) , which is one of the brightest sources of the signal . for the gc corresponding amplitudes @xmath241 are localized per pixels , for which @xmath275 in the galactic system of coordinate . from eq.([a11 ] ) one can see that for gc the function @xmath256 is equivalent to zero . more accurately , taking into account that image of the gc has characteristic sizes @xmath276 , where fwhm is the full width of half maximum of the beam , in eq.([a11 ] ) additionally to @xmath277 parameter we get small parameter @xmath278 . secondly , let discuss the model of two bright point like sources , located symmetrically relatively to the gc . let assumes that for that point sources @xmath279 , but @xmath280 . once again , from eq.([a11 ] ) we get @xmath281 for all @xmath48 and these point sources will be automatically removed by the @xmath282 estimator even if they have @xmath283 . another possibility related to the symmetry of the galactic image in @xmath49 direction . we would like to remind , that eq.([a11 ] ) was obtained under approximation @xmath284 , where @xmath285 . this means , that @xmath286 can be as big enough ( @xmath287 ) , as small ( @xmath288 ) as well . for @xmath289 from eq.([a10 ] ) we obtain @xmath290 + \delta ( \l+\frac{1}{2})[-(\delta_j-\delta_k)^2 + ( -1)^{\l+m}(\delta^2_j-\delta^2_k ) ] . \label{a12}\ ] ] as one can see from eq.([a12 ] ) the bright sources located on the same @xmath49 coordinates ( @xmath291 ) does not contribute to @xmath282 estimator . grski , k. m. , hivon , e. , wandelt , b. d. , 1999 , in a. j. banday , r. s. sheth and l. da costa , proceedings of the mpa / eso cosmology conference `` evolution of large - scale structure '' , printpartners ipskamp , nl
we study a specific correlation in spherical harmonic multipole domain for cosmic microwave background ( cmb ) analysis . this group of correlation between @xmath0 , @xmath1 is caused by symmetric signal in the galactic coordinate system . an estimator targeting such correlation therefore helps remove the localized bright point - like sources in the galactic plane and the strong diffused component down to the cmb level . we use 3 toy models to illustrate the significance of these correlations and apply this estimator on some derived cmb maps with foreground residuals . in addition , we show that our proposed estimator significantly damp the phase correlations caused by galactic foregrounds . this investigation provides the understanding of mode correlations caused by galactic foregrounds , which is useful for paving the way for foreground cleaning methods for the cmb .
circuit quantum electrodynamics ( qed ) is a novel field combining atomic physics and quantum optical cavity qed concepts with superconducting circuits.@xcite its fundamental dynamics is understood within the jaynes - cummings model , describing the interaction between a two - level system and a single field mode.@xcite in circuit qed , superconducting qubits are considered as artificial atoms @xcite interacting with on - chip one - dimensional resonators playing the role of cavities.@xcite these mesoscopic devices are candidates for implementations in quantum information processing @xcite due to their inherent tunability and scalability properties , at least as good as trapped ions @xcite or quantum dots.@xcite to implement quantum algorithms in circuit qed@xcite within the standard quantum computing approach , the sequential realization of fast high - fidelity quantum gates is required.@xcite several implementations of quantum operations in coupled superconducting qubits have been proposed@xcite and the implementation of a controlled - not gate has been achieved.@xcite furthermore , coherent coupling@xcite and quantum information exchange@xcite of two superconducting qubits , through a cavity bus and cavity state synthesis@xcite , have been realized . scalable architectures in circuit qed may require efficient designs of two - dimensional cavity arrays.@xcite these allow scalable standard quantum computation , and also open the possible realization of two - dimensional cluster states for one - way quantum computing.@xcite another crucial challenge in quantum information and also in circuit qed is to speed up operations and , thus , beat decoherence . for this purpose it would be desirable to employ resonant gates . indeed , based on first - order couplings between qubits and cavities , resonant gates are much faster than the commonly used dispersive gates , based on second - order couplings . in cavity qed , a controlled - phase ( cphase ) gate has been already implemented in the resonant regime @xcite , but these ideas are difficult to scale up and build , for example , cluster states in multiqubit systems . in this work , we focus on the field of circuit qed and propose the implementation of a toolbox of resonant quantum gates improving on speed and fidelity when compared to slower dispersive gates in circuit qed . to this end , a key concept from quantum optics will be the use of an auxiliary excited state for the qubits and the use of the cavity as a resonant mediator of the qubit interactions . the paper is organized as follows : in section ii , we will explain how to realize , resonantly and efficiently , the paradigmatic two - qubit cphase gate . the theoretical protocol is inspired by cavity qed experimental works@xcite . to prove its feasibility in circuit qed , we test our results with full numerical simulations involving decoherence mechanisms , reaching a fidelity of @xmath4 . then , in section iii , we extend the ideas of previous section and propose the implementation of other two - qubit gates : the iswap gate and the bogoliubov gate . @xcite some of these two - qubit gates , together with suitable one - qubit gates , can form universal sets for quantum computation . in section iv , as an example of the resonant cphase gate presented before , we propose the realization in circuit qed of a cluster state for one - way quantum computing@xcite in a two - dimensional array of cavities . the cphase gate is one of the paradigmatic two - qubit gates for quantum information . indeed , by combining it with one - qubit gates , it forms a set of universal gates . this cphase gate produces a phase - shift @xmath5 only when both qubits are excited . in the computational basis of two qubits , @xmath6 , the cphase matrix reads @xmath7 where @xmath8 and @xmath9 represent the ground and excited states of the qubits , labeled with subindices @xmath10 and @xmath11 . in a recent experiment @xcite , where a quantum algorithm was implemented in circuit qed , a dispersive interaction using the third level of a superconducting qubit was used . to implement a cphase gate in the resonant regime , in which the detuning is zero , we borrow from quantum optics the use of the cavity as a mediator for the operation and the use of auxiliary qubit levels.@xcite we develop below a specific modelisation in the context of circuit qed and present numerical simulations , including realistic losses , of the involved superconducting qubits and microwave resonators . we consider two three - level superconducting qubits , 1 and 2 , with @xmath12 , @xmath13 , and @xmath14 being their first three lower energy levels , respectively . the usefulness of auxiliary qubit levels in the superconducting qubit context has been recognized only recently , for phase qubit operations,@xcite for circuit qed quantum optics applications,@xcite as well as for a recent demonstration of gates in circuit qed.@xcite the qubits are coupled to a coplanar waveguide cavity and their dynamics is described , after a rotating - wave approximation , by the hamiltonian @xmath15 here , @xmath16 , @xmath17 , while @xmath18 are the bosonic annihilation ( creation ) operators of the resonator field mode , @xmath19 is the energy for level @xmath20 of qubit @xmath21 , @xmath22 is the cavity mode frequency , and the @xmath23 s denote the vacuum rabi coupling strengths . we assume that the qubit levels are anharmonic and that each transition can be tuned to match the cavity frequency , e.g. by using ac - stark - shift fields,@xcite thereby effectively switching on the qubit - cavity couplings for the respective transition . conversely , we could make use of a tunable cavity.@xcite we assume that at a certain point of a quantum computation the tripartite system ( qubits+cavity ) is in the following state @xmath24 where @xmath25 and @xmath26 are arbitrary complex amplitudes and @xmath27 is the cavity vacuum . a cphase gate is implemented on qubits 1 and 2 after the protocol displayed in fig . 1 is applied . in step ( i ) , the state of qubit 2 , encoded in its two lowest energy levels , @xmath28 and @xmath29 , is mapped onto a photonic cavity qubit through a resonant coupling between qubit 2 and the cavity mode , while the first qubit remains off - resonant . it is possible to achieve this mapping without introducing extra phase factors , such that the state at that stage is @xmath30 . in step ( ii ) @xcite , a 2@xmath5 resonant pulse between the two upper energy levels of qubit 1 , @xmath31 and @xmath32 , and the cavity mode realizes a cphase gate among them , yielding @xmath33 . in the last step ( iii ) , the cavity qubit is mapped back to qubit 2 , thus implementing an overall cphase gate between qubits 1 and 2 , leaving the cavity qubit decoupled in the vacuum state @xmath34 [ cols="<,^,^,^ " , ] as application of these resonant toolbox of two - qubit gates , we propose in the next section a protocol to implement a cluster state of @xmath35 superconducting qubits . in this cluster state , each qubit is entangled with its nearest neighbor with the help of a resonant cphase gate . one - way quantum computing@xcite is an alternative model to the standard approach based on quantum gates @xcite . it requires the initial generation of a two - dimensional cluster state , containing all necessary entanglement for the local and sequential implementation of a quantum algorithm . a cluster state can be obtained by the action of a cphase gate between neighbouring qubits , assuming all of them initially in the state @xmath36 . its experimental realization is one of the main difficulties when trying to implement one - way quantum computing in different physical systems . we present here a resonant protocol for realizing a cluster state of @xmath35 superconducting qubits . to achieve this goal , we consider a two - dimensional cavity grid@xcite as the most suitable architecture in circuit qed . indeed this architecture consists of a 2d square array of one - dimensional resonators where the superconducting qubits are placed at the crossings of the cavities . we assume that the qubits are initially in their ground states and the cavity - resonators contain zero photons . to generate this initial state , one can send an external field resonant to the transition frequency of all qubits to implement a y - rotation for a @xmath37 pulse . these local qubit rotations can be done simultaneously so that the corresponding operation time is @xmath38 , with @xmath39 the frequency of the external field . the qubits are then entangled by performing a cphase gate between each qubit and their nearest neighbours . although all cphase gates mathematically commute and could be implemented simultaneously,@xcite the proposed architecture of the cavity grid imposes critical constraints . in particular , one cphase gate can be implemented at most in one cavity at the same time . a solution to optimize the number of steps is shown in fig . [ simultgates ] . it consists in implementing simultaneously all cphase gates on the first row using the vertical cavities , plus the ones we can do on the first column with the unused horizontal cavities . we then repeat this step n times by shifting the row to the top and the column to the right . as a result , we will have implemented all required cphase gates between neighbouring qubits except the ones at the crossing points between the horizontal and vertical cavities during the process . last , we implement the missing gates on the two diagonals in two steps . the total time used to realize the 2d cluster state is @xmath40 where the second and third term on the r.h.s . of the first line correspond to the simultaneously done and the last missing cphase gates , respectively . the total time scales with @xmath1 , whereas the number of qubits is @xmath35 , producing a resonant and efficient technique to build 2d cluster states in the cavity grid architecture . cavities and contains @xmath35 qubits , one at each crossing point . the performed cphase gates for the lines are displayed in red and for the columns in blue . the row is shifted to the top and the column to the right . the missing gates at the crossings are indicated as dashed black lines.,title="fig : " ] cavities and contains @xmath35 qubits , one at each crossing point . the performed cphase gates for the lines are displayed in red and for the columns in blue . the row is shifted to the top and the column to the right . the missing gates at the crossings are indicated as dashed black lines.,title="fig : " ] cavities and contains @xmath35 qubits , one at each crossing point . the performed cphase gates for the lines are displayed in red and for the columns in blue . the row is shifted to the top and the column to the right . the missing gates at the crossings are indicated as dashed black lines.,title="fig : " ] cavities and contains @xmath35 qubits , one at each crossing point . the performed cphase gates for the lines are displayed in red and for the columns in blue . the row is shifted to the top and the column to the right . the missing gates at the crossings are indicated as dashed black lines.,title="fig : " ] we consider now how errors would accumulate during this process in the case of the generation of a cluster state in a cavity grid with an array of @xmath41 superconducting qubits . this example implies operating on 9 qubits , a rather advanced platform when compared with the state - of - the - art case of 3 qubits @xcite . for each one - qubit gate , we consider a fidelity of @xmath42 , which is in agreement with recent experimental results.@xcite . although a fidelity of @xmath4 for the cphase gates was achieved in our numerical simulation of section ii , we consider a more conservative value of @xmath43 in our present error estimations . following the steps presented before , we have to implement @xmath44 one - qubit operations and 12 cphase gates , reaching a global fidelity of @xmath45 . given the number of operations , this fidelity can be considered as good for the sake of demonstrating the scheme with a small number of qubits . clearly , full studies of decoherence sources ! and popu with this work , we extended ideas of quantum optics and cavity qed to the field of circuit qed . we considered the two - qubit cphase gate to introduce the tools used to implement this unitary two - qubit operation in the resonant regime and to prove its feasibility in circuit qed . indeed , we have shown numerically that one can obtain fast resonant cphase gates of high fidelity , above @xmath43 by using an auxiliary qubit level and the cavity photon to establish qubit communication . then we have applied these tools to the implementation of other two - qubit gates , the iswap and bogoliubov gates . all these resonant gates , together with high - fidelity one - qubit gates , constitute a circuit qed toolbox of resonant gates for the sake of implementing standard quantum computing . as possible application of this resonant toolbox , we have introduced a method of generating efficiently a cluster state in circuit qed in a two - dimensional architecture for implementations of one - way quantum computing . we expect the introduced toolbox of resonant and efficient gates will enhance the theoretical and experimental research in quantum information applications in circuit qed . the authors acknowledge valuable discussions with a. wallraff . this work was funded by the dfg through sfb 631 and the nanosystems initiative munich ( nim ) as well as the emmy - noether program . e.s . thanks support from upv / ehu grant giu07/40 , spanish micinn project fis2009 - 12773-c02 - 01 , eurosqip and solid european projects . g.h . acknowledges the support of the swiss nsf . o. buisson and f. hekking , in macroscopic quantum coherence and quantum computing , edited by d. v. averin , b. ruggiero and p. silvestrini ( kluwer , new york ) 2001 ; f. marquardt and c. bruder , phys . b * 63 * , 054514 ( 2001 ) ; j. q. you and f. nori , phys . b * 68 * , 064509 ( 2003 ) . i. chiorescu , p. bertet , k. semba , y. nakamura , c. j. p. m. harmans , and j. e. mooij , nature * 431 * , 159 ( 2004 ) ; a. wallraff , d. i. schuster , a. blais , l. frunzio , r .- s . huang , j. majer , s. kumar , s. m. girvin , and r. j. schoelkopf , nature * 431 * , 162 ( 2004 ) . 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 ) .
we propose the implementation of fast resonant gates in circuit quantum electrodynamics for quantum information processing . we show how a suitable utilization of three - level superconducting qubits inside a resonator constitutes a key tool to perform diverse two - qubit resonant gates , improving the operation speed when compared to slower dispersive techniques . to illustrate the benefit of resonant two - qubit gates in circuit qed , we consider the implementation of a two - dimensional cluster state in an array of @xmath0 superconducting qubits by using resonant controlled - phase ( cphase ) and one - qubit gates , where the generation time grows linearly with @xmath1 . for @xmath2 , and taking into account decoherence mechanisms , a fidelity over @xmath3 for the generation of this cluster state is obtained .
the smallness of the neutrino mass compared with the other quarks and leptons are one of the mysteries of nature . recently , a new mechanism generating small dirac mass terms for neutrino has been proposed @xcite . the similar mechanism generating the small neutrino dirac mass term for the tev seesaw mechanism is also proposed in @xcite and phenomenology is studied in @xcite and @xcite . there are also models with radiatively generated dirac mass term in @xcite . the interesting feature of the model proposed in @xcite is the tiny vacuum expectation value for an extra higgs su(2 ) doublet @xcite . the small neutrino mass is realized without introducing tiny yukawa coupling for neutrinos . a softly broken global u(1 ) symmetry guarantees the tiny vacuum expectation value for the extra doublet . in addition to the small softly breaking mass parameter , the mass squared parameter for the extra higgs is chosen to be positive so that the light pseudo nambu goldstone bosons due to the softly broken global symmetry do not appear . this is a contrast to the mass squared parameter for the standard model like higgs boson . in the present paper , we study the global minimum of the tree level higgs potential by explicitly solving the stationary conditions . there are many studies of the tree level higgs potential of general two higgs doublet model @xcite . ( see also @xcite for recent review of two higgs doublet model . ) it has been shown that the charge neutral vacuum is lower than the charge breaking vacuum @xcite . also the vacuum energy difference of two neutral minima was derived @xcite . we make use of the results and identify the vacuum of the present model . when the u(1 ) symmetry breaking term is turned off , the tree level higgs potential and the phase structure of the present model is rather similar to the model with @xmath0 discrete symmetry @xcite . in contrast to @xmath0 symmetric case , it is essential to keep the soft breaking term when finding the true vacuum . if we set the symmetry breaking term at zero , then the order parameter corresponding to the softly broken u(1 ) symmetry becomes redundant parameter and can not be determined . we treat the soft breaking term as small expansion parameter and obtain the vacuum expectation values and the vacuum energies in terms of the parameters of the higgs potential . the constraints on the parameters of the model for which the desired vacuum can be realized , are derived and they are rewritten in terms of higgs masses and a few coupling constants which can not be directly related to the higgs masses . these constraints are fully used when we study the radiative corrections to the vacuum expectation values numerically . beyond the tree level , we study the radiative correction to the higgs potential and the vacuum expectation values of higgs . since the neutrino masses are proportional to the vacuum expectation value of one of higgs , one can also compute the radiative corrections to neutrino masses . as already noted in @xcite the radiative correction to the softly breaking mass parameter is logarithmically divergent and it is renormalized multiplicatively . we derive the formulae for the one loop corrected vacuum expectation values for two higgs doublets by studying one loop corrected effective potential . the corrections are evaluated numerically by exploring the parameter regions allowed from the global minimum condition for the vacuum . we show how the radiative corrections change depending on the extra higgs spectrum . the radiative corrections are also evaluated for the case that a relation among the coupling constants is satisfied . the paper is organized as follows . in section ( ii ) , we derive the condition for the desired vacuum being global minimum . in section ( iii ) one loop effective potential is derived and one loop corrections to the vacuum expectation values are obtained in section ( iv ) . in section ( v ) , the corrections are evaluated numerically for various choices of parameters of the higgs potential . section ( vi ) is devoted to summary and discussion . the model of the dirac neutrino is proposed in @xcite . in @xcite , two higgs su(2 ) doublets are introduced , _ 1= _ 1 ^ 1 + i _ 1 ^ 2 + _ 1 ^ 3 + i _ 1 ^ 4 , _ 2= _ 2 ^ 1 + i _ 2 ^ 2 + _ 2 ^ 3 + i _ 2 ^ 4 , where @xmath1 s vacuum expectation value is nearly equal to the electroweak breaking scale and the second higgs @xmath2 has a small vacuum expectation value which gives rise to neutrino mass . the higgs potential in @xcite is , v_tree=_i=1,2 ( m_ii^2 _ i^_i+ ( _ i^_i)^2 ) -(m_12 ^ 2 _ 1^_2 + h.c.)+ _ 3 ( _ 1^_1)(_2^_2 ) + _ 4 second higgs . the u(1)@xmath3 global symmetry is broken softly with the term @xmath4 . in this paper , we introduce the following real o(4 ) representation for each doublet , because this parametrization is convenient when computing the one loop corrected effective potential . _ 1^a= _ 1 ^ 1 + _ 1 ^ 2 + _ 1 ^ 3 + _ 1 ^ 4 , _ 2^a= _ 2 ^ 1 + _ 2 ^ 2 + _ 2 ^ 3 + _ 2 ^ 4 , _ 1^a= -_1 ^ 2 + _ 1 ^ 1 + -_1 ^ 4 + _ 1 ^ 3 . [ eq : phi ] using the notation above , the tree level effective potential introduced in eq.([eq : tree ] ) can be written as , v_tree&=&m_11 ^ 2 _ a=1 ^ 4 ( _ 1^a)^2 + m_22 ^ 2 _ a=1 ^ 4 ( _ 2^a)^2 -m_12 ^ 2 _ a=1 ^ 4 _ 1^a _ 2^a + & + & ( _ a=1 ^ 4 _ 1^a^2)^2 + ( _ a=1 ^ 4 _ 2^a^2)^2 + ( _ a=1 ^ 4 _ 1^a^2 ) ( _ a=1 ^ 4 _ 2^a^2 ) + & + & ( ( _ a=1 ^ 4 _ 1^a _ 2^a)^2 + ( _ a=1 ^ 4 _ 1^a _ 2^a)^2 ) , where one can choose @xmath4 real and positive . with the notation of eq.([eq : phi ] ) , the softly broken global symmetry u(1)@xmath3 corresponds to the following transformation on @xmath5 , _ 2^=o_u(1)^ _ 2 & = & ( cccc & - & 0 & 0 + & & 0 & 0 + 0 & 0 & & - + 0 & 0 & & ) _ 2 . [ eq : u(1 ) ] @xmath6 does not transform under u(1)@xmath3 . therefore u(1)@xmath3 is broken softly when @xmath4 does not vanish . without loss of generality , one can choose the vacuum expectation values of higgs with the form given as , _ 1 = 0 + 0 + v + 0 + , _ 2 = v ^ + -v ^ + v ^ + -v ^ , [ eq : v2gene ] where the range for @xmath7 is @xmath8 and the range for @xmath9 and @xmath10 is @xmath11 $ ] . we call the four order parameters as @xmath12 , @xmath13 . when @xmath14 vanishes , by taking @xmath15 in eq.([eq : u(1 ) ] ) , one can rotate @xmath7 away in eq.([eq : v2gene ] ) . for the most general case , in total , there are four independent order parameters when u(1)@xmath3 symmetry is broken . for completeness of our discussion , we give the constraints on the quartic couplings from condition that the tree level potential is the bounded below , @xcite , @xcite , @xcite . _ 1 > 0 , _ 2 > 0 , [ eq : gcon0 ] + - _ 3 , [ eq : gcon1 ] + - _ 3+_4 . [ eq : gcon2 ] in addition to the conditions on the quartic terms , one can constrain the parameters including the quadratic terms so that the desired vacuum satisfies the global minimum conditions of the potential . about the global minimum of the tree potential , it was shown that the energy of charge neutral vacuum is lower than that of the charge breaking vacuum @xcite . we therefore set @xmath10 zero . we also require the vacuum expectation value of the second higgs is much smaller than that of the first higgs , which implies that @xmath16 is small . in terms of the parametrization in eq.([eq : v2gene ] ) with @xmath17 , the potential can be written as , v_tree(v , , ^ ) = a ( ) v^4 + b ( , ^ ) v^2 , where , a ( ) & = & ^4 + ^4 + ( + ) ^2 ^2 , + b ( , ^ ) & = & ^2 + ^2 -m^2_12 ^. we first find the global minimum of @xmath18 . the stationary conditions @xmath19 @xmath20 , are written as , & & v(2 a v^2+b)=0 , [ eq : v2 ] + & & 2 r_4=2 , [ eq : beta ] + & & m_12 ^ 2 ^ 2= 0 , [ eq : thetap ] where @xmath21 are defined as , r_1 & = & , + r_2 & = & , + r_3 & = & , + r_4 & = & . the stationary conditions eq.([eq : v2 ] ) and eq.([eq : beta ] ) correspond to eq.(36 ) of @xcite . here we solve them explicitly by treating the soft breaking term @xmath14 as perturbation . the non - zero solution for @xmath22 in eq.([eq : v2 ] ) is written as , v^2&=&-=-4 , where @xmath23 . substituting it into @xmath18 , one obtains , v_tree & & v_min.=- . for non - zero @xmath4 and @xmath24 , the solution of eq.([eq : thetap ] ) is @xmath25 . one still needs to find @xmath9 among the solutions of eq.([eq : beta ] ) , which leads to the minimum of @xmath26 . we solve eq.([eq : beta ] ) and determine @xmath9 by treating @xmath27 as a small expansion parameter . one can easily find the approximate solutions as , \(1 ) = , ^= sign(m_22 ^ 2 _ 1-m_11 ^ 2 _ 34 ) , + ( 2 ) = , ^= sign(m_11 ^ 2 _ 2-m_22 ^ 2 _ 34 ) , + ( 3 ) 2 = + o(r_4 ) , [ eq : solution ] corresponding to each solution , ( 1)@xmath28(3 ) of eq.([eq : solution ] ) , the vacuum expectation value @xmath22 and the minimum of the potential are obtained . & & ( v^2,v_min)= + & & \(1 ) ( -+ 2 _ 1(m_22 ^ 2-m_11 ^ 2 ) ( ) ^2 , - + ) , + ( 2 ) ( -+ 2 _ 2(m_11 ^ 2-m_22 ^ 2 ) ( ) ^2 , - + ) , + ( 3)(2+o(r_4 ) , -+o(r_4 ) ) . + [ eq : v2vmin ] the leading terms of the vacuum expectation values agree with those obtained in @xmath0 symmetric model @xcite . if @xmath29 , then @xmath30 must be vanishing and @xmath31 from eq.([eq : beta ] ) and eq.([eq : thetap ] ) . .classification of the solutions with non zero @xmath24 of the stationary conditions of higgs potential . for ( 3 ) , o(@xmath30 ) correction is not shown . [ cols="^,^",options="header " , ] in figures 1,2 , and 3 , we also show the correction @xmath32 as functions of @xmath33 . @xmath34 is independent on @xmath35 and does not necessarily vanish at the same points where @xmath36 vanishes . with @xmath37 and @xmath38 ( gev ) , when the pseudoscalar higgs mass is much larger than that of charged higgs mass , we find very large correction to @xmath39 . in fig.4 , we show that the two dimensional surface which corresponds to @xmath40 . we find that the interior of the surface corresponds to the region of the positive correction ; @xmath41 while the exterior region of the surface corresponds to the negative correction ; @xmath42 . + in fig.5 , we have shown the regions of ( @xmath43 ) which correspond to that the corrections of @xmath44 and @xmath45 have the definite values ( @xmath46 ) . the dark gray shaded area corresponds to the region where both @xmath34 and @xmath36 can vanish with taking account of the conditions ; eq.([eq : gcon0 ] ) , eq([eq : gcon1 ] ) and eq.([eq : gcon2 ] ) . we note that for @xmath47 ( gev ) , the quantum corrections vanish around the region where the charged higgs degenerates with the pseudoscalar higgs . when the corrections become larger , the larger mass splitting of the pseudoscalar higgs and charged higgs is allowed . however as the average mass of the charged higgs and pseudoscalar higgs increases , the allowed mass splitting becomes smaller . in this paper , the dirac neutrino mass model of davidson and logan is studied . in the model , one of the vacuum expectation values of two higgs doublets is very small and it becomes the origin of the mass of neutrinos . the ratio of the small vacuum expectation value @xmath48 and that of the standard like higgs @xmath49 is @xmath50 . therefore @xmath16 is very small and typically it is @xmath51 . the smallness of @xmath16 is guaranteed by the smallness of the soft breaking term of @xmath52 . we have treated the soft breaking term as perturbation and calculated , in particular , the vacuum expectation of higgs in the leading order of the perturbation precisely . as summarized in table i , only by including the soft breaking terms , one can argue which of the local minima minimizes the potential and becomes the global minimum . we have studied the global minimum of the tree level higgs potential including the effect of the soft breaking term as perturbation . beyond the tree level , we study the quantum correction to the vacuum expectation values and @xmath16 in a quantitative way . in one loop level , we confirmed that tree level vacuum is stable , i.e. , the order parameters which vanish at tree level do not have the vacuum expectation value as quantum correction . in one loop level , we derived the exact formulae for the quantum correction to @xmath9 in the leading order of expansion of the soft breaking parameter @xmath4 . we have confirmed not only that the loop correction to @xmath16 is proportional to the soft breaking term but also found that the correction depends on the higgs mass spectrum and some combination of the quartic coupling constants of the higgs potential . technically , we carried out the calculation of the one loop effective potential by employing o(4 ) real representation for su(2 ) higgs doublets . dependence of the corrections on the higgs spectrum is studied numerically . we first derive a relation of the coupling constants which corresponds to the condition that the correction to @xmath9 vanishes for degenerate extra higgs masses . next , we study the effect of non - degeneracy of the charged higgs and pseudoscalar higgs on the correction . if the charged higgs mass is as light as 100 ( gev ) @xmath28 200 ( gev ) , allowing the mass difference of charged higgs and pseudoscalar higgs is about @xmath53(gev ) , the quantum corrections to both @xmath9 and @xmath39 are within a few @xmath54 for @xmath55 . if the charged higgs is heavy @xmath56 ( gev ) , a slight increase of the pseudoscalar higgs mass from the degenerate point leads to very large corrections to @xmath9 and @xmath39 . one can argue the size of the quantum corrections to the neutrino mass of the model , because the ratio of the tree level neutrino mass and one loop correction can be written as , = + , [ eq : neutrino ] where we take account of the corrections only due to higgs vacuum expectation values . the formulae eq.([eq : neutrino ] ) implies that radiative correction to neutrino mass is related to the higgs mass spectrum . therefore once higgs mass spectrum is measured in lhc , one can compute the radiative correction to the mass of neutrinos using the formulae eq.([eq : neutrino ] ) . we would like to thank d. kimura for reading the manuscript and y. kitadono with discussion . the work of k. tamai is supported by hiroshima university research assistant fellowship . t. m. would like to thank the aspen center for physics and nsf grant no . 1066293 where a part of the work completed . the work of t. m. is supported by kakenhi , grant - in - aid for scientific research(c ) no.22440283 from jsps , japan . + * note added * + after submitting the paper , we are aware that the stability of the model studied in this paper was also discussed in @xcite . compared to their analysis , we derived the 1 loop effective potential taking into account all the interactions of higgs sector while they consider a part of the interactions and study the stability in a qualitative way . using the effective potential , we carried out the quantitative analysis of the quantum corrections . in this appendix , we give the details of the derivation of the one - loop effective potential and the counterterm in eq.([eq : v1loop ] ) . one can split @xmath57 in eq.([eq : mtmphi ] ) into the diagonal part and the off diagonal part as , @xmath58 the divergent part of @xmath59 can be easily computed by expanding it up to the second order of @xmath60 , v_1loop&=&v^(1)+v_c , + v^(1)&= & \ { ( d_ii^0 -1+m_ii^2 ( ) ) _ ij+m^2_ij -_1 m_12 ^ 2 } + & = & _ i=1 ^ 8 \{d_ii^0 -1+m_ii^2 ( ) } + & - & _ i , j=1 ^ 8 d_ii ( m^2-_1 m_12 ^ 2)_ij d_jj ( m^2-_1 m_12 ^ 2)_ji+ ... , where , d_ii^-1&=&d_ii^0 -1+m_ii^2 ( ) , + & = & c m_ii^2+m_11 ^ 2-k^2 ( 1 i 4 ) , + m_ii^2+m_22 ^ 2-k^2 ( 5 i 8) . the diagonal parts of the propagators are given as , d_ii= c ( 1 i 4 ) , + ( 5 i 8) . in the modified minimal subtraction scheme , feynman integration is carried out with help of the well known formulae of dimensional regularization , ^4-d ( m^2-k^2 ) = - m^4 + ( - ) , [ eq : integ ] and , ^4-d _ div . = , with @xmath61 and @xmath62 . the divergent part of @xmath63 is , v^(1)_div.=&- & \ { _ i=1 ^ 4 ( m^2_ii+m_11 ^ 2)^2 + _ i=5 ^ 8 ( m^2_ii+m_22 ^ 2)^2 } + & - & _ i j=1 ^ 8 ( m^2-m_12 ^ 2 _ 1)_ij ( m^2-m_12 ^ 2 _ 1 ) _ ji , + = & - & ( m_11 ^ 2 _ i=1 ^ 4 m^2_ii ( ) + m_22 ^ 2 _ i=5 ^ 8 m^2_ii ( ) + 2(m_11 ^ 4 + m_22 ^ 4 ) ) + & - & ( m^2()-m_12 ^ 2 _ 1)(m^2()-m_12 ^ 2 _ 1 ) , + = & -&[m_t^4 ] . [ eq : div ] the trace of eq.([eq : div ] ) is calculated in eq.([eq : decom0 ] ) and eq.([eq : mmmm ] ) of appendix b and the result is , v^(1)_div . = & - & m_11 ^ 2 \ { 6_1 ( _ 1^_1 ) + 2(2_3+_4)(_2^_2 ) } + & + & m_22 ^ 2 \ { 2(2_3+_4)(_1^_1 ) + 6_2 ( _ 2^_2 ) } + & + & ( 2_3 + 4_4 ) ( _ 1^_2 + _ 2^_1 ) + & - & + & - & ( 12_1 ^ 2 + 4 _ 3_4 + 4_3 ^ 2 + 2_4 ^ 2)(_1^_1)^2 + & + & ( 12_2 ^ 2 + 4 _ 3_4 + 4_3 ^ 2 + 2_4 ^ 2)(_2^_2)^2 + & + & ( 12_1 _ 3 + 4_1 _ 4 + 8_3 ^ 2 + 4_4 ^ 2 + 12 _ 2 _ 3 + 4_2 _ 4)(_1^_1)(_2^_2 ) + & + & ( 4_1_4 + 16_3_4 + 8_4 ^ 2 + 4_2_4)_1^_2 ^2 . [ eq : veffdiv ] now the counterterms for the one loop effective potential are simply given by changing the sign of the divergent part of eq.([eq : veffdiv ] ) , v_c&=&-v^(1)_div . + & = & [ m_t^4 ] . [ eq : counterterms ] using eq.([eq : counterterms ] ) and eq.([eq : integ ] ) , one can derive the finite part of the 1 loop effective potential given in eq.([eq : v1loop ] ) . in this section , we present the derivation of eq.([eq : veffdiv ] ) . we start with the quartic interaction terms of the higgs potential , v^(4)&= & ( _ i=1 ^ 4 _ i^2)^2 + ( _ i=5 ^ 8 _ i^2)^2 + ( _ i=1 ^ 4 _ i^2 ) ( _ j=5 ^ 8 _ j^2 ) + & + & ( ( _ 1 _ 5 + _ 2 _ 6 + _ 3 _ 7 + _ 4 _ 8)^2 + ( _ 1 _ 6 + _ 3 _ 8 -_2 _ 5 -_4 _ 7)^2 ) . by taking the derivatives of @xmath64 , one can obtain the mass squared matrix @xmath65 . one first computes the first derivative of @xmath64 with respect to @xmath66 , = 2(_j=1 ^ 4 _ j^2 ) 2 _ i + _ i _ j=5 ^ 8 _ j^2 + \ { ( _ 1 _ 5 + _ 2 _ 6 + _ 3 _ 7 + _ 4 _ 8) _ i+4 + + ( _ 1 _ 6 + _ 3 _ 8 -_2 _ 5 -_4 _ 7 ) ( _ 1i _ 6 -_2i _ 5 + _ 3i _ 8 -_4i _ 7 ) } , ( 1 i 4 ) + 2(_j=5 ^ 8 _ j^2 ) 2 _ i + _ i _ j=1 ^ 4 _ j^2 + \ { ( _ 1 _ 5 + _ 2 _ 6 + _ 3 _ 7 + _ 4 _ 8) _ i-4 + + ( _ 1 _ 6 + _ 3 _ 8 -_2 _ 5 -_4 _ 7 ) ( -_5i _ 2 + _ 6i _ 1 -_7i _ 4 + _ 8i _ 3 ) } . ( 5 i 8) . + the second derivatives are given as , & & = + & & ( _ ij _ k=1 ^ 4 _ k^2 + 2 _ j _ i ) + _ ij ( _ k=5 ^ 8 _ k^2 ) + \ { _ j+4 _ i+4 + . ( _ 1j _ 6 -_2j _ 5 + _ 3j _ 8 -_4j _ 7 ) ( _ 1i _ 6 -_2i _ 5 + _ 3i _ 8 -_4i _ 7 ) } , ( 1 i , j 4 ) , + _ 3 _ i _ j + \ { _ i+4 _ j-4 + _ k=1 ^ 4 _ i+4,j_k _ k+4 + ( -_5j _ 2 + _ 6j _ 1 -_7j _ 4 + _ 8j _ 3 ) + ( _ 1i _ 6 -_2i _ 5 + _ 3i _ 8 -_4i _ 7 ) + ( _ 1 _ 6 + _ 3 _ 8 - _ 2 _ 5 - _ 4 _ 7 ) + ( _ 1i_6j + _ 3i_8j-_2i_5j-_4i_7j ) } , ( 1i 4 , 5 j 8) , + _ 3 _ i _ j + \ { _ i-4 _ j+4 + _ k=1 ^ 4 _ i-4,j_k _ k+4 + ( _ 1j _ 6 -_2j _ 5 + _ 3j _ 8 -_4j _ 7 ) + ( -_5i _ 2 + _ 6i _ 1 -_7i _ 4 + _ 8i _ 3 ) + ( _ 1 _ 6 + _ 3 _ 8 - _ 2 _ 5 - _ 4 _ 7 ) + ( _ 1i_6j + _ 3i_8j-_2i_5j-_4i_7j ) } , ( 5 i 8 , 1 j 4 ) , + ( _ ij _ k=5 ^ 8 _ k^2 + 2 _ j _ i ) + _ ij ( _ k=1 ^ 4 _ k^2 ) + \ { _ j-4 _ i-4 + + ( -_5j _ 2 + _ 6j _ 1 - _ 7j _ 4 + _ 8j _ 3 ) ( -_5i _ 2 + _ 6i _ 1 - _ 7i _ 4 + _ 8i _ 3 ) } , ( 5 i , j 8) . + [ eq : second ] with eq.([eq : second ] ) , the diagonal sums of @xmath67 are given as , _ i=1 ^ 4m^2_ii&= & 3 _ 1 _ i=1 ^ 4 _ i^2 + 2_3 _ i=5 ^ 8 _ i^2 + _ 4 _ i=5 ^ 8 _ i^2 = 6_1 _ 1^_1 + ( 4_3 + 2_4)_2^_2 , ( 1i 4 ) , + _ i=5 ^ 8m^2_ii&= & 3 _ 2 _ i=5 ^ 8 _ i^2 + 2_3 _ i=1 ^ 4 _ i^2 + _ 4 _ i^2 = 6_2 _ 2^_2 + ( 4_3 + 2_4)_1^_1 , ( 5i 8) . + [ eq : b4 ] the counterterm in eq.([eq : counterterms ] ) includes the following contribution , [ ( m^2()-m_12 ^ 2 _ 1 ) ( m^2()-m_12 ^ 2 _ 1 ) ] = tr[m^2 ( ) m^2 ( ) -2m_12 ^ 2 _ 1 m^2 ] + 8 m_12 ^ 4 . [ eq : ct ] the second term of eq.([eq : ct ] ) is proportional to , [ m_12 ^ 2 _ 1 m^2 ] & = & ( 2_3 + 4_4 ) ( _ 1 _ 5 + _ 2 _ 6 + _ 3 _ 7 + _ 4 _ 8 ) m_12 ^ 2 + & = & ( 2_3 + 4_4 ) ( _ 1^_2 + _ 2^_1 ) m_12 ^ 2 . [ eq : decom0 ] the first term of eq.([eq : ct ] ) can be decomposed as , [ m^2()m^2()]&=&_i , j=1 ^ 4 m^2()_ijm^2()_ji+ 2 _ i=1 ^ 4 _ j=5 ^ 8 m^2()_ijm^2()_ji + & + & _ i , j=5 ^ 8 m^2()_ijm^2()_ji . [ eq : decom ] each term of eq.([eq : decom ] ) is given as , & & _ i , j=1 ^ 4m^2()_ij m^2()_ji = 3 _ 1 ^ 2 ( _ i=1 ^ 4 _ i^2 ) ^2 + 3 _ 1_3 _ i=1 ^ 4 _ i^2 _ j=5 ^ 8 _ j^2 + _ 1_4 \ { _ i=5 ^ 8 _ i^2 _ j=1 ^ 4 _ j^2 . + & & .+ ( _ 1 _ 5 + _ 2 _ 6 + _ 3 _ 7 + _ 4 _ 8 ) ^2 + ( _ 1 _ 6 + _ 3 _ 8 -_2 _ 5 -_4 _ 7)^2 } + & & + _ 3_4 ( _ i=5 ^ 8 _ i^2 ) ^2 + _ 3 ^ 2 ( _ i=5 ^ 8 _ i^2 ) ^2 + ( _ i=5 ^ 8 _ i^2 ) ^2 + & & = 12 _ 1 ^ 2 ( _ 1^_1 ) ^2 + ( 12_1 _ 3 + 4 _ 1_4)(_1^_1)(_2^_2 ) + & & + 4_1 _ 4_1^_2 ^2 + ( 4_3 _ 4 + 4_3 ^ 2 + 2_4 ^ 2)(_2^_2)^2 , [ eq : decom3 - 1 ] + & & _ i=1 ^ 4 _ j=5 ^ 8 m^2()_ij m^2()_ji= _ 3 ^ 2 _ i=5 ^ 8 _ i^2 _ j=1 ^ 4 _ j^2 + 2_3_4 \ { _ i=1 ^ 4 _ i_i+4 _ j=1 ^ 4 _ j _ j+4 . + & & + . ( _ 1_6 - _ 2_5 + _ 3 _ 8 - _ 4_7)^2 } + & & + \ { _ i=1 ^ 4 _ i^2 _ j=5 ^ 8_j^2 + 2 ( _ i=1 ^ 4_i _ i+4 ) ^2 + 2(_1_6-_2_5 + _ 3_8-_4_7)^2 } + & & = ( 4_3 ^ 2 + 2_4 ^ 2 ) ( _ 1^_1)(_2^_2 ) + ( 8 _ 3_4 + 4_4 ^ 2 ) _ 1^_2 ^2 , [ eq : decom3 - 2 ] + & & _ i , j=5 ^ 8 m^2()_ij m^2()_ji = 3 _ 2 ^ 2 ( _ i=5 ^ 8 _ i^2 ) ^2 + 3 _ 2_3 _ i=5 ^ 8 _ i^2 _ j^2 + _ 2_4 \ { _ i=1 ^ 4 _ i^2 _ j=5 ^ 8 _ j^2 + . + & & . ( _ 1 _ 5 + _ 2 _ 6 + _ 3 _ 7 + _ 4 _ 8 ) ^2 + ( _ 1 _ 6 + _ 3 _ 8 -_2 _ 5 -_4 _ 7)^2 } + & & + _ 3_4 ( _ i=1 ^ 4 _ i^2 ) ^2 + _ 3 ^ 2 ( _ i=1 ^ 4 _ i^2 ) ^2 + ( _ i=1 ^ 4 _ i^2 ) ^2 + & & = 12 _ 2 ^ 2 ( _ 2^_2 ) ^2 + ( 12_2 _ 3 + 4 _ 2_4)(_1^_1)(_2^_2 ) + & & + 4_2 _ 4_1^_2 ^2 + ( 4_3 _ 4 + 4_3 ^ 2 + 2_4 ^ 2)(_1^_1)^2 . [ eq : decom3 - 3 ] from eq.([eq : decom3 - 1 ] ) , eq.([eq : decom3 - 2]),and eq.([eq : decom3 - 3 ] ) , one obtains , [ m^2 ( ) m^2 ( ) ] & = & ( 12_1 ^ 2 + 4 _ 3_4 + 4_3 ^ 2 + 2_4 ^ 2)(_1^_1)^2 + & + & ( 12_2 ^ 2 + 4 _ 3_4 + 4_3 ^ 2 + 2_4 ^ 2)(_2^_2)^2 + & + & ( 12_1 _ 3 + 4_1 _ 4 + 8_3 ^ 2 + 4_4 ^ 2 + 12 _ 2 _ 3 + 4_2 _ 4)(_1^_1)(_2^_2 ) + & + & ( 4_1_4 + 16_3_4 + 8_4 ^ 2 + 4_2_4)_1^_2 ^2 . [ eq : mmmm ] using eq.([eq : b4 ] ) , eq.([eq : ct ] ) , eq.([eq : decom0 ] ) , and eq.([eq : mmmm ] ) , one can derive eq.([eq : veffdiv ] ) . in this appendix , we show @xmath70_{jj}$ ] and @xmath69 which are needed to calculate one loop corrections to the order parameters @xmath71 in eq.([eq : ordersimp ] ) . @xmath70_{jj}$ ] ( @xmath72 ) are given as , & & [ o^t o]_jj=0,_jj=0 . [ eq : althp ] & & [ o^t o]_jj=2 v [ o^t o]_jj = + & & ( _ 1+_2 + 6 _ 3 - 2 _ 4-(4 ) ( _ 1+_2 - 2 ( _ 3+_4 ) ) ) + ( _ 1+_2 + 6 _ 3 - 2 _ 4-(4 ) ( _ 1+_2 - 2 ( _ 3+_4 ) ) ) + ( _ 1+_2 + 6 _ 3 + 6 _ 4-(4 ) ( _ 1+_2 - 2 ( _ 3+_4 ) ) ) + 12\{_2 ^2 ^ 2 + ^2 ^2 _ 1}+(3 2 ( -)-2 ( + ) + 2 ) ( _ 3+_4 ) + 12\{_1 ^2 ^2 + ^2 ^2 _ 2}+(-3 2 ( -)+2 ( + ) + 2 ) ( _ 3+_4 ) , + [ eq : v ] and , & & [ o^t o]_jj = v^2 + & & _ 2 ^2()- ^2 ( ) _ 1- ( 2 ) ( _ 3+_4 ) + _ 2 ^2()- ^2 ( ) _ 1- ( 2 ) ( _ 3+_4 ) + _ 2 ^2()- ^2 ( ) _ 1- ( 2 ) ( _ 3+_4 ) + 3 _ 2 ^2()-3 ^2 _ 1 + ( ( 2 ( + ) ) -3 ( 2 ( - ) ) ) ( _ 3+_4 ) + -3 _ 1 ^2()+3 ^2 ( ) _ 2 - ( ( 2 ( + ) ) -3 ( 2 ( - ) ) ) ( _ 3+_4 ) , + [ eq : bet ] next we show @xmath69 in eq.([eq : lij ] ) . note that @xmath69 is symmetric @xmath73 and its non - zero elements are , l_11&= & ^2m_11 ^ 2+^2m_22 ^ 2 -2 ( ) ( ) m_12 ^ 2 + & + & 3 v^2 \ { _ 1 ^4 ( ) + ^2 ( ) ( 2 ( _ 3+_4 ) ^2()+ ^2 ( ) _ 2 ) } , + l_22&=&v^2 \{- ( _ 1+_2 - 2 ( _ 3+_4 ) ) v^2 + ( _ 2-_1 ) v^2 + & + & 2 m_12 ^ 2 2- 2(m_11 ^ 2-m_22 ^ 2 ) } , + l_12&=&l_21=v \{- ( _ 1+_2 - 2 ( _ 3+_4 ) ) v^2 + 2(_2-_1 ) v^2 , + & -&2 m_12 ^ 22 - 2(m_11 ^ 2-m_22 ^ 2 ) } + l_33&=&- v^2 ( 2 ) ( v^2 ( 2 ) _ 4 - 4 m_12 ^ 2 ) , + l_44&=&v^2 ( ) ( ) m_12 ^ 2 . [ eq : elementofl ] here we show the orthogonal matrix @xmath74 in eq.([eq : higgsmass ] ) . o= 0 & - & 0 & 0 & 0 & 0 & & 0 + - & 0 & 0 & 0 & 0 & & 0 & 0 + 0&0 & 0 & & & 0 & 0 & 0 + 0&0 & - & 0 & 0 & 0 & 0 & + 0 & & 0 & 0 & 0 & 0 & & 0 + & 0 & 0 & 0 & 0 & & 0 & 0 + 0 & 0 & 0 & & - & 0 & 0 & 0 + 0 & 0 & & 0 & 0 & 0 & 0 &
we study a dirac neutrino mass model of davidson and logan . in the model , the smallness of the neutrino mass is originated from the small vacuum expectation value of the second higgs of two higgs doublets . we study the one loop effective potential of the higgs sector and examine how the small vacuum expectation is stable under the radiative correction . by deriving formulae of the radiative correction , we numerically study how large the one loop correction is and show how it depends on the quadratic mass terms and quartic couplings of the higgs potential . the correction changes depending on the various scenarios for extra higgs mass spectrum . s /
in this paper we study linear functional differential equations with infinite delay . consider @xmath0 x(\theta ) & = \phi(\theta),\quad \theta { \in}(-\infty,0]\end{aligned}\ ] ] where @xmath1 is an arbitrary sequence of real numbers , @xmath2 is a strictly increasing sequence of strictly positive reals such that @xmath3 and @xmath4\longrightarrow\mathbb{r}$ ] is continuous . for the special case @xmath5 , ( 1.1 ) can be uniquely solved for any given @xmath6 bc @xmath7 $ ] , the space of all bounded real - valued continuous functions . the proof of this is indicated in example1.2 . denote this solution by @xmath8 . consider the family of operators @xmath9 on bc @xmath7 $ ] defined as @xmath10(\theta ) & = x_{\phi } ( t + \theta ) , \quad\hbox{if}\quad t + \theta > 0\nonumber\\[.2pc ] & = \phi(t + \theta ) , \quad\hbox{if}\quad t + \theta \leq 0.\end{aligned}\ ] ] it is elementary to see that @xmath11 is _ not _ a strongly continuous semigroup on bc @xmath12 $ ] . we prove this as follows if @xmath13 is a semigroup , we must have @xmath14 given @xmath15 , we can find @xmath16 such that @xmath17 for @xmath18 . now , let @xmath19 . consider @xmath20 $ ] with @xmath21 let @xmath22 . note that @xmath23 and @xmath24 consider @xmath25 & \leq |(s_{t}\phi ) ( \theta_{1 } ) - \phi(\theta_{1})|\\[.2pc ] & \leq \|s_{t}\phi -\phi\|_\infty\\[.2pc ] & \leq \epsilon.\end{aligned}\ ] ] thus , we have shown that @xmath26 is uniformly continuous on @xmath27 $ ] . by the uniform continuity of @xmath26 on @xmath28 $ ] , uniform continuity of @xmath26 on @xmath12 $ ] follows . but this is a contradiction as there are bounded continuous functions which are not uniformly continuous . on the space buc @xmath12 $ ] , the space of bounded uniformly continuous functions , the family of operators @xmath13 defined by ( 1.2 ) do form a semigroup but this space is properly contained in bc@xmath12 $ ] . in the literature , certain banach spaces , which contain bc@xmath12 $ ] , which are classes of functions satisfying certain growth conditions , are used in the context of infinite delay equations . refer @xcite . our approach is to find a frechet space that contains bc@xmath12 $ ] on which @xmath29 form a semigroup . instead of constructing a weight function which is related to @xmath30 and @xmath31 , we obtain a family of semi - norms for the initial function @xmath26 that enable us to get estimates for the solution and also to capture a frechet space @xmath32 such that ( 1.2 ) defines a strongly continuous semigroup on @xmath32 . we do not make any explicit summability assumption on the sequence @xmath33 , but the space @xmath32 heavily depends on the properties of @xmath33 . if @xmath34 , then bc@xmath7 \subset f$ ] but if @xmath35 , bc@xmath7 $ ] is not contained in @xmath32 . the basic theory of finite delay differential equations is covered in @xcite . @xcite and @xcite are some basic references for banach phase spaces related to infinite delay equations . consider the following examples let @xmath36 = bc@xmath37 and @xmath38 be defined as @xmath39(x ) = { \varphi}'(x)$ ] . define @xmath40 , as @xmath41 . it is easy to see that @xmath42 satisfies the following conditions 1 . the resolvents @xmath43 exist for @xmath44 where @xmath45 is the identity operator on @xmath46 2 . @xmath47 and that @xmath13 satisfy the following conditions + .4pc + \(i ) @xmath48 + \(ii ) @xmath49 . + despite the above observations @xmath13 is not strongly semigroup on @xmath36 since the condition + \(iii ) @xmath50 + does _ not _ hold . the proof that ( iii ) does not hold is similar to the proof of the analogous assertion in the case of the infinite delay equations . the hille yosida theorem is not applicable precisely because @xmath42 is not densely defined . one way to overcome this difficulty is to construct a _ smaller _ banach space called the hille yosida space for the operator @xmath42 @xcite on which the restriction of @xmath42 generates a semigroup . in this construction , the condition ( a ) plays a crucial role . but consider the operator @xmath51 defined as @xmath52(x ) = { \varphi}'(x)$ ] . @xmath53 is an ` extension ' of @xmath42 that does generate the semigroup @xmath54 defined as @xmath55\varphi(x)=\varphi(t + x)$ ] on the _ frechet space _ @xmath56 . thus@xmath57 it is clear that by considering the hille yosida space@xmath57 a lot of useful information is lost . we now indicate the proof of assertion that for @xmath58 , ( 1.1 ) can be uniquely solved for any given @xmath59 $ ] . for @xmath60 , t -\tau_{i } \in ( -\infty , 0]$ ] and hence for @xmath61 , \phi(t -\tau_{i})$ ] is meaningful . define @xmath62 $ ] as @xmath63 . now @xmath64 } |\phi(t -\tau_{i})| \leq @xmath65 is in @xmath66 , the series @xmath67 converges . hence @xmath68 converges in @xmath69 $ ] . thus , @xmath70 $ ] . now , consider the ordinary differential equation @xmath71,\\[.4pc ] x(0 ) & = \phi(0)\end{aligned}\ ] ] whose solution exists and is unique . now , assume the existence of a unique function @xmath72 \longrightarrow\mathbb{r}$ ] such that @xmath73 for @xmath74 $ ] and whose restriction to @xmath75 $ ] is a solution to @xmath76,\\[.4pc ] x(0 ) & = \phi(0).\end{aligned}\ ] ] as before , the expression @xmath77 defines a continuous function on @xmath78 $ ] . now , consider @xmath79,\\ x(k \tau_{1 } ) & = y_{k}(k\tau_{1})\end{aligned}\ ] ] which has a unique solution @xmath80 . define @xmath81 as @xmath82 $ ] and @xmath83 $ ] . thus , we obtain a solution to ( 1.1 ) in @xmath84 $ ] . by induction and patching up of solutions , we get a unique solution to ( 1.1 ) on the whole of @xmath85 . let @xmath86 $ ] and @xmath5 . define @xmath87 \hbox{\rm : } \ { \varphi}'(0 ) = \sum\limits^{\infty}_{i=1 } b_{i}\varphi(-\tau_{i})\right\}\longrightarrow { \hbox{bc}}(-\infty,0]\end{aligned}\ ] ] as @xmath88(x ) = { \varphi}'(x).\ ] ] @xmath89 is not dense in @xmath36 and hence @xmath42 does not generate a semigroup . motivated by example 1.1 , we look for a frechet space @xmath32 that contains bc@xmath12 $ ] and an ` extension ' @xmath53 to @xmath42 such that @xmath53 generates a semigroup on @xmath32 . the general theory of semigroups on frechet spaces is very complicated . for example@xmath57 even a bounded linear operator on a frechet space need not generate a semigroup @xcite . in @xcite@xmath57 a generalisation of the hille yosida theorem for a closed and unbounded operator in a locally convex space is proved . but the hypotheses of this theorem are not easily verified in many concrete cases . by proving various estimates for the solution @xmath90 of ( 1.1)@xmath57 we are able to capture a frechet space @xmath32 on which the solution of ( 1.1 ) gives rise to a semigroup . refer to @xcite , @xcite and @xcite for applications of semigroups on locally convex spaces to pde s . we need the following definitions and results in the next section . @xmath91 1 . a topological vector space @xmath36 is said to be a frechet space if its topology is generated by a family of countable semi - norms @xmath92 and @xmath36 is complete with respect to the family @xmath93 . 2 . a linear map @xmath94 is said to be bounded if for every @xmath95 , there are finitely many indices @xmath96 and a constant @xmath97 such that for all @xmath98 , @xmath99 the basic theory of frechet spaces and the proof of the following proposition can be found in @xcite . @xmath91 a linear map @xmath100 is continuous if and only if it is bounded . @xmath91 a family of bounded linear operators @xmath101 on @xmath36 is said to be a strongly continuous semigroup if the properties ( i ) , ( ii ) and ( iii ) of example 1.1 hold . @xmath91 let @xmath102 denote the banach space of all sequences @xmath103 of elements of a banach @xmath36 such that @xmath104 . let @xmath105 ; l^{1})$ ] be the banach space of all continuous functions @xmath106\longrightarrow l^{1}$ ] . the banach space @xmath107))$ ] is isometrically embedded in @xmath108;l^{1})$ ] . let @xmath109))$ ] . define @xmath110 \longrightarrow l^{1 } ( \mathbb{r})$ ] as @xmath111^{(i ) } = g^{(i)}(s)$ ] . it is given that @xmath112\ } < \infty.\ ] ] it is easy to check that @xmath113 is a bounded function and that @xmath114^{(i)}|\hbox{\rm : } \ s \in [ 0,t]\right\ } = \sum\limits_{i = 1}^{\infty } \hbox{sup } \{|g^{(i)}(s)|\hbox{\rm : } \ s \in [ 0,t]\}.\ ] ] it remains to be shown that @xmath115^{(i ) } - [ g^{*}(t_{0})]^{(i)}| = 0.\ ] ] by the hypothesis for each @xmath116 , @xmath117^{(i ) } - g^{*}(t_{0})]^{(i)}| = 0 $ ] and for a given @xmath15 , there exists @xmath118 such that @xmath119\ } < \epsilon/3.\ ] ] further , there exists @xmath16 such that @xmath120 implies that @xmath121^{(i ) } - [ g^{*}(t_{0})]^{(i)}| < \epsilon /3.\ ] ] hence for all @xmath122 in @xmath123 $ ] with @xmath124 , @xmath125^{[(i ) } \!-\ ! [ g^{*}(t_{0})]^{[(i)}| & = \sum\limits^{k}_{i = 1}|[g^{*}(t)]^{[(i ) } - [ g^{*}(t_{0})]^{[(i)}|\\[.5pc ] & \quad\,+ \sum\limits^{\infty}_{i = k + 1}|[g^{*}(t)]^{[(i ) } - [ g^{*}(t_{0})]^{[(i)}| \leq { \epsilon}/3\\[.5pc ] & \quad\,+ \sum\limits^{\infty}_{i = k + 1}|[g^{*}(t)]^{[(i)}| + \sum\limits^{\infty}_{i = k + 1}|[g^{*}(t_{0})]^{[(i)}|\\[.5pc ] & \leq { \epsilon}/3 + \epsilon/3 + \epsilon/3 = \epsilon.\end{aligned}\ ] ] the result is proved . the following definitions and statements on frechet space valued riemann integral are found in @xcite . let @xmath36 be a frechet space and let @xmath126 \longrightarrow x$ ] be continuous . the integral @xmath127 can be defined uniquely which has the following properties 1 . for every continuous linear functional @xmath128 , @xmath129 2 . for every seminorm @xmath130 , @xmath131 3 . @xmath132 4 . @xmath133\hbox{\rm d}t = \int^{b}_{a}u(t)\hbox{\rm d}t + \int^{b}_{a}v(t)\hbox{\rm d}t$ ] 5 . @xmath134 . @xmath91 let @xmath135 be a sub interval of @xmath136 . a function @xmath137 is said to be differentiable at @xmath138 if there exists @xmath139 with the following property for every @xmath15 and @xmath140 , there exists @xmath16 such that for all @xmath141 with @xmath142 and @xmath143 , we have @xmath144 further , the element @xmath145 is denoted by @xmath146 . @xmath91 let @xmath147 be a sub interval of @xmath136 . a function @xmath148 is said to be continuously differentiable on @xmath147 , if @xmath149 is differentiable at every point @xmath138 and the function mapping @xmath150 to @xmath151 is continuous on @xmath147 . the class of all such functions is denoted by @xmath152 . if @xmath126\longrightarrow x$ ] is continuously differentiable , then @xmath153 for a given @xmath154)$ ] define the family of seminorms @xmath155 as @xmath156\}.\ ] ] let @xmath157 and @xmath158 be as in the beginning of this section . given @xmath159 , define @xmath160 as the smallest positive integer such that @xmath161 for all @xmath162 . define @xmath32 as @xmath163\hbox{\rm : } \ p_{k}(\varphi ) < \infty \quad\hbox{for all } \ \ k \in \mathbb{n}\},\ ] ] where the seminorms @xmath164 are defined as follows@xmath165\}.\ ] ] @xmath91 the space @xmath32 equipped with the topology generated by the family of seminorms @xmath166 is a frechet space . let @xmath167 be a cauchy sequence in @xmath32 . clearly , there exists @xmath168 $ ] such that @xmath167 converges to @xmath26 uniformly on every compact set of @xmath12 $ ] . consider the banach space @xmath169))$ ] . for every @xmath170 , define @xmath171))$ ] as @xmath172 by the hypothesis , @xmath173 is a cauchy sequence in @xmath174))$ ] and hence converges to some @xmath175))$ ] . for fixed @xmath116 and @xmath176 , @xmath177 and hence @xmath178 . this implies that @xmath179 and @xmath167 converges to @xmath26 in @xmath32 . the proof is complete . now we state the definition of a mild solution of an abstract cauchy problem in frechet spaces from @xcite . let @xmath36 be a frechet space whose topology is given by the family of seminorms @xmath180 and @xmath89 be a subset of @xmath36 . for a closed operator @xmath181 and @xmath182 , consider the abstract cauchy problem @xmath183 u(0 ) & = \phi.\end{aligned}\ ] ] @xmath91 @xmath91 a function @xmath184 that satisfies ( 1.3 ) is said to be a _ solution _ of problem ( 1.3 ) . a function @xmath185 is said to be a _ mild solution _ of ( 1.3 ) , if @xmath186 , for all @xmath187 , and @xmath188 let @xmath32 be as in the previous section . define @xmath189 and @xmath190 and @xmath191 as @xmath192 where @xmath193 @xmath91 we say that a function @xmath194 is a solution to ( 1.1 ) if the following hold 1 . @xmath90 is continuous and @xmath195 for all @xmath196 $ ] . the restriction of @xmath90 to @xmath37 is continuously differentiable . @xmath197 for all @xmath187 . note that our definition of the solution does not imply that @xmath90 is differentiable from the left at @xmath198 . let @xmath199 @xmath200 and @xmath42 be as in the beginning of this section . then @xmath42 generates a strongly continuous semigroup @xmath201 of bounded linear operators on @xmath32 . further@xmath57 for a given @xmath202 the map @xmath203 defined as @xmath204,\\[.2pc ] x(t ) & = [ s_{t}\phi](0),\quad t\in ( 0,\infty)\end{aligned}\ ] ] is a unique solution to @xmath205 . besides , fixing @xmath206 and defining @xmath207 as @xmath208 is a mild solution to the abstract cauchy problem ( 1.3 ) . we need the following lemmas to prove theorem 2.3 and we actually define the semi - group via the solution to ( 1.1 ) . let @xmath206 . the problem @xmath205 has a unique solution @xmath209 . further@xmath57 for each @xmath159 there exists a finite subset @xmath210 of @xmath211 and a constant @xmath212 such that @xmath213\}\leq c_{k } \max \{q ( \phi)\hbox{\rm : } \ q\in \lambda\}.\ ] ] consider @xmath214 $ ] . clearly , @xmath215 for all @xmath216 . thus for @xmath217 , t - t_{i}\leq 0 $ ] . define @xmath218\rightarrow \mathbb{r}$ ] as the unique solution of the initial value problem . @xmath219 x(0 ) & = \phi ( 0).\end{aligned}\ ] ] note that as @xmath220 defines a continuous function on @xmath221 $ ] . we have @xmath222 define @xmath223\rightarrow \mathbb{r}$ ] as @xmath224\\[.2pc ] & = y_{1 } ( s),\quad s \in [ 0 , \tau_{1}].\end{aligned}\ ] ] in the remaining part of the proof we shall assume that @xmath225 . the estimates for @xmath226 are easier to obtain . clearly , @xmath227\}\leq \left(\sup\limits_{s\in [ 0,\tau_{1 } ] } \hbox{e}^{as}\right ) \|\phi \|_{1 } + \left(\sup\limits_{s\in [ 0,\tau_{1 } ] } \frac{\hbox{e}^{as}-1}{a}\right)p_{1 } ( \phi).\ ] ] here , note that for @xmath228 for all @xmath229 . now we claim that for each @xmath159 , there exists a function @xmath230\rightarrow \mathbb{r}$ ] with the following properties 1 . for each @xmath231 $ ] , @xmath232 converges and this summation defines a continuous function on @xmath233 $ ] . @xmath234 is the unique solution to @xmath235\nonumber\\ \hskip -1.25pc x(\theta ) & = \phi ( \theta),\quad\hbox{for } \ \ t \in ( -\infty , 0].\ ] ] 3 . there exists a finite subset @xmath210 of @xmath236 and a constant @xmath237 such that @xmath238\}\leq c_{k } \max \{q ( \phi)\hbox{\rm : } \ q\in \lambda\}.\ ] ] we prove this by induction on @xmath239 . the case @xmath240 is already proved . assuming that our claim is true for arbitrary @xmath159 , we show that the claim is true for @xmath241 . define @xmath242 $ ] and @xmath243 as @xmath244\}.\ ] ] for @xmath245 and @xmath246 $ ] and so we can consider the summation @xmath247 the first summation involves only finitely many terms . for @xmath248 $ ] and @xmath249 and hence @xmath250 so , for @xmath251 , the summation @xmath252 defines a continuous function on @xmath253 . for a given @xmath254 , define @xmath255 as the smallest positive integer such that @xmath256 the following estimates follow from the definition of the seminorms and the integers @xmath255@xmath257\right\ } \leq p_{k + 1}(\phi).\ ] ] moreover , @xmath258\right\}\nonumber\\[.5pc ] & \quad\,\leq \sum_{i=1}^{n(k+1)-1 } |b_{i}|\times \max \{\|\phi\|_{m(k+1 ) } , \|x_{k}\|_{i_{k}}\}.\end{aligned}\ ] ] define @xmath259\rightarrow \mathbb{r}$ ] as @xmath260 from ( 2.5 ) and ( 2.6 ) , we get the estimate @xmath261\right\}\nonumber\\[.4pc ] & \quad\,\leq \left[\sum_{i=1}^{n(k+1)-1 } |b_{i}|\times \max \{\|\phi\|_{m(k+1 ) } , \|x_{k}\|_{i_{k}}\}+p_{k+1}(\phi)\right].\ ] ] defining @xmath262\rightarrow \mathbb{r}$ ] as @xmath263\\[.2pc ] & = y_{k+1}(s ) , s \in ( k\tau_{1},(k+1)\tau_{1}],\end{aligned}\ ] ] our claims ( i ) and ( ii ) are proved . now , we proceed to prove ( iii ) for @xmath264 . from the definition of @xmath265 and the estimate ( 2.7 ) , we get , for @xmath159 , @xmath266\}=\|x_{k+1}\|_{i_{k+1}}\nonumber\\[.4pc ] & \quad\,= \sup\{|y_{k+1}(s)|\hbox{\rm : } \ s \in [ k\tau_{1 } , ( k+1 ) \tau_{1}]\}\nonumber\\[.4pc ] & \quad\,\leq \left(\sup\limits_{s \in[0,k\tau_{1 } ] } \hbox{e}^{as}\right)\|x_{k}\|_{i_{k } } + \left(\sup\limits_{s \in[0,\tau_{1}]}\frac{\hbox{e}^{a(k+1)s } - 1}{{a}}\right)\nonumber\\[.4pc ] & \qquad\ \times \left[\sum_{i=1}^{n(k+1)-1 } |b_{i}| \times \max \{\|\phi\|_{m(k+1 ) } , \|x_{k}\|_{i_{k}}\}+p_{k+1}(\phi)\right].\ ] ] since @xmath267\}\ ] ] the assertion ( iii ) for @xmath264 follows from the estimate ( 2.8 ) and hence the assertion ( iii ) for @xmath239 . the solution to ( 1.1 ) is obtained by patching the functions @xmath234 . uniqueness of @xmath90 now follows . let @xmath268 and @xmath269 be a continuous function such that @xmath270 for all @xmath271 $ ] . define @xmath272)$ ] as @xmath273(\theta)=x(t+\theta)$ ] . then @xmath274 . fix @xmath275 $ ] . it is trivial to check that @xmath276)$ ] . we also have the estimate @xmath277\hbox{\rm : } \ |[u(t)](\theta)|\}\nonumber\\[.2pc ] & \leq \max ( \|\varphi \|_{k } , \sup \{|x(t)|\hbox{\rm : } \ t \in [ 0 , j\tau_{1}\}).\ ] ] now we show that @xmath278 for each @xmath279 . let @xmath280 $ ] and @xmath281 $ ] . it is clear that @xmath282 and we have the following assertions : @xmath283 and @xmath284 thus , for every @xmath285 , we have @xmath286(s-\tau_{i } ) & = \sum_{i = n(k)}^{n(k+j)-1 } b_{i}[u(t)](s-\tau_{i})\\[.5pc ] & \quad\,+ \sum_{i = n(k+j)}^{k } b_{i}[u(t)](s-\tau_{i})\\[.5pc ] & = \sum_{i = n(k)}^{n(k+j)-1 } b_{i}x(t+s-\tau_{i})\\&\quad\,+ \sum_{i = n(k+j)}^{k } b_{i}\phi(t+s-\tau_{i}).\end{aligned}\ ] ] now choosing a positive integer @xmath287 , we have the estimate @xmath288 equations ( 2.9 ) and ( 2.10 ) show that @xmath289 . next , we show the continuity of @xmath149 at an arbitrary @xmath290 . let @xmath15 be given . by the uniform continuity of @xmath90 on @xmath291 $ ] , there exists @xmath16 such that for @xmath292 $ ] with @xmath293 . take @xmath294 and consider @xmath176 with @xmath295 . clearly , for any @xmath296 , t_{0}+ \theta$ ] and @xmath297 both belong to @xmath298 $ ] and hence @xmath299\ } < \epsilon.\ ] ] that is , given @xmath15 , there is a @xmath300 such that @xmath301 whenever @xmath302 . next , we claim that for a given @xmath15 and @xmath303 $ ] , there exists @xmath16 such that @xmath304 whenever @xmath303 $ ] satisfies @xmath305 . by proposition 1.6 , and the definition of @xmath32 , for a given @xmath306\rightarrow l^{1}(\mathbb{r})$ ] defined as @xmath307^{(i)}=b_{n(k)+(i-1)}\varphi ( s-\tau_{n(k)+(i-1)})$ ] is a continuous function . consider @xmath308(s-\tau_{i})&= \sum_{i = n(k)}^{n(k+j)-1 } b_{i}[u(t_{0})-u(t)](s-\tau_{i})\\[.5pc ] & \quad\,+ \sum_{i = n(k+j)}^{k } b_{i}[u(t_{0})-u(t)](s-\tau_{i})\\[.5pc ] & = \sum_{i = n(k)}^{n(k+j)-1 } b_{i}[x(t_{0}+s-\tau_{i})-x(t+s-\tau_{i})]\\[.5pc ] & = \sum_{i = n(k+j)}^{k } b_{i}[\phi(t_{0}+s-\tau_{i})-\phi ( t+s\!-\tau_{i})]\end{aligned}\ ] ] and hence @xmath309(s-\tau_{i } ) & = \sum_{i = n(k)}^{n(k+j)-1 } b_{i}[x(t_{0}+s-\tau_{i})-x(t+s-\tau_{i})]\\[.5pc ] & + \sum_{i = n(k+j)}^{\infty } b_{i}[\phi(t_{0}+ s-\tau_{i})-\phi(t+s\!-\tau_{i})].\end{aligned}\ ] ] by the uniform continuity of @xmath310 on @xmath311 $ ] , given @xmath15 , there exists a @xmath312 such that @xmath313 implies that @xmath314| < \epsilon/2.\ ] ] so , for @xmath315 and @xmath281 $ ] with @xmath316 , @xmath317\right\ } < \epsilon/2.\ ] ] since the expression @xmath318 $ ] involves evaluation of @xmath90 over a compact set , given @xmath319 , there exists @xmath320 such that @xmath321 $ ] and @xmath322 . our claim is now proved by taking @xmath323 . by lemma 2.4 , for a given @xmath206 , we have a solution @xmath324 to ( 1.1 ) . define @xmath325 as @xmath10(\theta ) & = x_{\phi}(t+\theta),\quad t+\theta>0\\[.2pc ] & = \phi ( t+\theta),\quad t+\theta\leq 0.\end{aligned}\ ] ] lemma 2.5 shows that @xmath326 for all @xmath327 and the map @xmath328 is a continuous function from @xmath329 into @xmath32 . from the definition of @xmath13 one can verify linearity of each @xmath13 and the property @xmath330 . we need only to check that each @xmath13 is a bounded linear map on @xmath32 . this follows from the estimates ( 2.9 ) , 2.10 ) and ( 2.1 ) next , we have @xmath331(\theta)&=\phi ( t+\theta),\ \ t+\theta \leq 0\nonumber\\[.4pc ] [ u(t)](\theta)&=\phi ( 0)+ \int_{0}^{t + \theta } l(u(s))\hbox{d}s,\ \ t + \theta > 0 \ ] ] and using ( 2.11 ) , it is easy to see that @xmath149 is indeed a mild solution to ( 1.3 ) . the banach space @xmath332 defined below was the phase space used in [ 4 ] for the study of infinite delay equations of which ( 1.1 ) is a special case . an interesting observation is that @xmath32 contains @xmath332 . let @xmath333\rightarrow [ 1 , \infty)$ ] be a continuous non - increasing function such that @xmath334 consider the space @xmath332 of all @xmath335\rightarrow \mathbb{r}$ ] such that @xmath336\right\ } < \infty.\ ] ] then @xmath337 . consider @xmath338 . for @xmath339 $ ] and @xmath340 and hence @xmath341\}\leq ( \|\phi\|_{g } ) |b_{i}|g(-\tau_{i}).\ ] ] now , the assertion follows from ( 2.12 ) . major part of this work was done while the author was with the department of mathematics , indian institute of science , bangalore . the author acknowledges the referee whose comments greatly improved the presentation of the paper and who pointed out many misprints .
in this paper , we show existence and uniqueness of a solution to a functional differential equation with infinite delay . we choose an appropriate frechet space so as to cover a large class of functions to be used as initial functions to obtain existence and uniqueness of solutions . [ theore]_lemma _ [ theore]proposition [ theore]definition [ theore]_remark _ [ theore]_proof of theorem _ [ theore]_example _
dark matter mini - halos in the mass range @xmath5 solar mass virializing at high redshift are believed to collect enough primordial gas within them to host the formation of the first generation of stars ( population iii ) . mediated by molecular hydrogen cooling , gas in the centers of these halos condenses and becomes unstable to gravitational fragmentation with a typical mass scale of a few hundred solar masses . pioneering high resolution 3d hydrodynamic cosmological simulations by abel et al . ( 2002 ) and bromm et al . ( 2002 ) , and more recently by yoshida et al . ( 2006 ) and oshea & norman ( 2007a ) have shown that gravitational fragmentation produces only one massive fragment per halo . if this gas does not fragment further as it approaches stellar density , pop iii stars with this typical mass would be produced . the fates and feedback effects of such stars has many interesting consequences for the early structure formation as discussed elsewhere in this volume ( see review by ciardi ) . while this picture is now firmly established , based as it is on rather simple physics elucidated by the simulations , it is important to realize the remaining uncertainties . first , as of 2003 when the first reviews of primordial star formation were being written ( barkana & loeb 2001 , bromm & larson 2004 , glover 2005 ) , a rather small number of such simulations had been done , raising the question of how ubiquitous this mode of star formation was in the early universe . second , despite the large range of scales covered by the simulations , they were necessarily terminated well before stellar density was reached due to missing physics . the possibility of sub - fragmentation could not be ruled out , with consequent uncertainty on the mass scale of the first stars . third , as emphasized by barkana & loeb ( 2004 ) , for technical reasons the simulations were carried out in quite small cosmological volumes ( < 1 mpc comoving ) , the result being that the redshift of formation of the first stars found ( @xmath6 ) was underestimated . larger volumes would contain rarer peaks in the density field , and these would presumably form pop iii stars earlier ( white & springel 2000 ) . could the higher densities and temperatures of this earlier epoch alter the mass scale of the truly first stars ? and finally , there was a host of complicated feedback effects radiative , kinetic , and chemical which were not included in these first simulations which could mess up " the simple picture by altering the initial conditions and the cooling properties of the collapsing cloud cores . the most concerning of these feedback effects was the build - up of an fuv background by emissions from the first stars that could photo - dissociate the hydrogen molecules which mediate pop iii star formation in the first place ( haiman , rees & loeb 1997 , ciardi , ferrara & abel 2000 , haiman , abel & rees 2001 ) . the first simulations examining this negative feedback effect " ( machacek , bryan & abel 2001 , hereafter mba ; and yoshida et al 2003 , hereafter y03 ) concluded that pop iii star formation would not be suppressed , but merely delayed . the concept of a critical halo mass capable of forming a pop iii star was introduced by mba . they found that the critical halo mass is an increasing function of the fuv background mean intensity in the lyman - werner bands ( 11.2 - 13.6 ev ) suggesting that pop iii star formation would become self - limiting . y03 found that gas cooling becomes inefficient above @xmath7 due to the photo - destruction of 2 . here @xmath8 is the mean intensity of the fuv background in units of @xmath9 ergs/@xmath10/sec / ster / hz . using this information y03 constructed a semi - analytic model of the cosmic history of pop iii star formation . they found a rapidly rising population of pop iii stars over the redshift interval 35 > z > 20 the rise of pop iii " , which begins to become self - regulated at @xmath11 when the fuv background attains @xmath7 . they did not consider the lower redshift evolution as this was out of the range of their simulations , but rather speculated that pop ii star formation dominated the cosmic star formation history below @xmath12 . it is important to point out that neither the mba simulations nor the y03 simulations were evolved below redshifts of @xmath12 or had the spatial resolution to follow the chemo - thermal - hydrodynamics of cloud collapse on scales below @xmath13 pc , making the fall of pop iii " by radiative feedbacks alone quite uncertain and understudied . indeed , y03 found the pop iii global sfr was still rising at the end of their simulations , implying that we do nt even know when the pop iii epoch peaked . ricotti , gnedin & shull ( 2002 ) addressed this question using hydrodynamic cosmological simulations that included the feedback of both photo - dissociating and photo - ionizing radiation from primeval galaxies . although their simulations were of even lower resolution than mba and y03 and parameterized star formation with ad hoc prescriptions , they found a cosmic star formation history quite similar to that predicted by y03 despite small box size effects . they too could not simulate sufficiently large boxes to obtain converged results in the interesting redshift regime 20 > z > 6 , but rather found that rare luminous objects strongly affected the average sfr within the simulation volume . it is fair to say that we have a much better idea when the first pop iii star formed in the universe than when the last one formed . indeed , some models have been presented which predict pop iii stars forming as low as redshift 3 ( schneider et al . 2006 ) . this review updates progress on direct numerical simulations of pop iii star formation by several groups since the reviews of bromm and larson ( 2004 ) , glover ( 2005 ) , and ciardi and ferrara ( 2005 ) . new high resolution simulations have been carried out by yoshida et al . ( 2006 ) , oshea & norman ( 2007a ) and gao et al . ( 2007 ) which test the robustness of the basic picture described above by pushing the collapse to higher central densities , although not yet stellar densities , and by considering ensembles of simulations . the effects of a fuv background on pop iii star formation has recently been revisited by oshea & norman ( 2007b ) and by wise and abel ( 2007 ) . it is found that 2 cooling remains important in halos with masses approaching @xmath14 and fuv backgrounds as strong as @xmath15 . this suggests that the pop iii epoch may be more extended than previously thought and therefore occurring in rather different environments than originally simulated . pop iii star formation has been studied in new and different environments as well . pop iii star formation in warm dark matter models has been studied by oshea & norman ( 2006 ) and by gao & theuns ( 2007 ) . finally , pop iii star formation in gas that has been pre - processed by earlier generation of pop iii stars has been simulated by oshea et al . ( 2005 ) , mesinger , bryan & haiman ( 2006 ) , and by yoshida et al . ( 2007a , b ) . the surprising result is that despite the diversity of the environments studied , it is found that pop iii.2 stars form in basically the same way as pop iii.1 stars and that they are massive , although they appear to be somewhat less massive than pop iii.1 based on lower cloud core temperatures and accretion rates . however , the new simulations show that environment and formation redshift / history can have a substantial effect on protostellar accretions rates , suggesting some spread in the primordial imf . no 3d simulation has yet been done that follows the entire accretion history of the star , and therefore final masses are still uncertain . however , approximate integrations have been done assuming spherical symmetry which suggest that the entire protostellar envelope can be accreted ( omukai & palla 2003 , yoshida et al . 2006 , 2007a , b ) . verifying this result with fully 3d simulations represents a grand challenge for the field . hereafter we will refer to pop iii stars forming from undisturbed cosmological initial conditions pop iii.1 , in keeping with the taxonomy introduced by mckee ( these proceedings ) . our understanding of the formation of pop iii.1 stems primarily from high resolution 3d hydrodynamic cosmological simulations supplemented with primordial gas chemistry including the all - important reaction kinetics of molecular hydrogen . the simulations of abel , bryan & norman ( 2000 , 2002 ) and bromm , coppi & larson ( 2002 ) showed that sufficient 2 is formed in the cores of mini - halos of mass @xmath17 at @xmath12 so that ro - vibrational line cooling of 2 will precipitate jeans instability . typical densities and temperatures in this gas prior to collapse are @xmath18 and t @xmath19 200 k , implying a jeans mass of @xmath20 . of gravitationally unstable primordial gas cooling by molecular hydrogen line radiation . the dark patch at lower left is a 0.01 @xmath21 fully molecular hydrogen , optically thick cloud core cooling by collisionally induced emission with central density @xmath22 . accretion of the massive envelope onto this seed will form a pop iii star at @xmath23 . from yoshida al . ( 2006 ) . ] the collapse of this gas initially proceeds quasi - statically , and then dynamically when the density threshold for 3-body 2 formation is reached at @xmath24 . the collapse proceeds from the inside out " , analogous to protostellar cloud collapse models that have been studied for decades in connection with current day star formation as reviewed by larson ( 2003 ) . the infalling envelope assumes a density distribution that is well approximated by a powerlaw : @xmath25 although the temperature distribution is distinctly not isothermal . both groups found that the collapsing core did not fragment into smaller units , implying a massive star would eventually be formed . both the abel and bromm simulations were forced to stop well before stellar density was reached due to limits of physics and resolution . the simulation by abel , bryan and norman ( 2002 ) achieved a central density of @xmath26 on scales @xmath27 au where a fully molecular core of @xmath28 was found accreting its envelope at a rate of @xmath29/yr . at this rate > 100 @xmath21 of material could accrete within a kelvin time , suggesting that pop iii.1 stars were very massive . new simulations have confirmed and extended this basic picture . table 1 summarizes the simulation details for the adaptive mesh refinement ( amr ) simulation of oshea & norman ( 2007a ; hereafter on07a ) and the smoothed particle hydrodynamics ( sph ) simulation of yoshida et al . ( 2006 ; hereafter y06 ) . both groups assumed wmap 1 cosmological parameters , had similar mass and spatial resolution , and ran their simulations until a central density of @xmath30 particles / cc was reached . the sph simulation included more physics and is thus more reliable in the range of densities where radiation transport effects , collisionally - induced - emission ( ripamonti & abel 2004 ) , and equation of state effects become important ( @xmath31 ) . y06 ran a single simulation in a box 0.3 mpc on a side , while on07a ran a dozen simulations in three different box sizes : 0.3 , 0.45 and 0.6 @xmath32 . we will therefore use the on07a simulations to discuss the effect of formation redshift and environment , the y06 simulation to discuss high density evolution , and compare the two of them over the range of densities where they can be compared . .simulation parameters [ cols="<,<,<",options="header " , ] fig . 1 shows the radial density and temperature profiles when the central density has reached @xmath33 . although the on07a simulation is not valid at these high densities , the density profiles agree extremely well over a large range of radii and enclosed masses considering the very different numerical techniques employed . the density profile is well approximated as power law @xmath34 over many decades in radius . the temperature distributions also agree extremely well for n < @xmath1 where opacity effects become important . the temperature distribution reflects four regions . proceeding from larger to smaller radii we have first the region of cosmological infall , where the temperature increases to its maximum value ( @xmath35k ) at the virialization shock located at @xmath36 pc . the second is a cooling zone where the gas temperature drops to the minimum allowed by 2 cooling ( @xmath37 k ) at a radius of a few pc . the third sees the gas temperature rise again to @xmath35k due to compressional heating in the quasi - statically collapsing cloud core at @xmath38 pc . the fourth and innermost ( @xmath39 pc ) is a cooling zone accompanying the formation of the fully molecular core due to the 3-body 2 reaction . this innermost cooling zone is absent in the y06 simulation because the cooling radiation is trapped at densities @xmath31 , whereas it is allowed to escape unphysically in the on07a simulation . when opacity effects are included , the central temperature rises to @xmath40 k. the difference in thermal structure seems to have little effect on the density profile , however . pc ) where the gas is optically thin to 2 cooling radiation . ] fraction and radial velocity from the amr and sph simulations referred to in fig . 2 . ] fig . 3 shows a comparison of profiles for molecular hydrogen fraction and radial velocity , analogous to fig . three body 2 formation creates a fully molecular core of size @xmath41 pc and mass @xmath28 . the massive envelope accretes onto this seed , driving its thermal and chemical evolution . a maximum infall velocity of @xmath42 km / s is achieved in both simulations at a radius of @xmath43 pc . the collapse of the cloud core is relatively unimpeded by angular momentum effects . 4 shows the specific angular momentum profiles from both simulations . again , the agreement is excellent both qualitatively and quantitatively . the star forming cloud has a sharply increasing specific angular momentum distribution as a function of enclosed mass , which is well - approximated by @xmath44 . this implies the first stars form from very low angular momentum material . a plot of the ratio of the circular velocity to the local keplerian velocity on mass shells ( dashed line of fig . 4b ) indicates that the infalling envelope is never rotationally supported . how is this angular momentum profile established ? the multiple lines plotted in fig . 4a indicate different output times corresponding to the density profiles shown in fig . the change with time of the angular momentum profile implies redistribution of angular momentum within the innermost 1000 @xmath21 of the cloud . this is seen in both amr and sph simulations , and is understood to be the result of angular momentum segregation , wherein lower angular momentum fluid elements in the turbulent cloud preferentially sink to the center ( oshea 2005 , y06 ) . at densities above @xmath45 3-body formation of 2 becomes efficient at converting the mostly atomic gas into molecular hydrogen . at densities above @xmath1 it is necessary to include the trapping of the 2 line radiation , collisionally induced emission ( cie ) , and chemical heating of the gas as the core becomes fully molecular . these effects have been included in the simulation of y06 , and therefore we refer to their results in the following . the effect of radiation trapping and chemical heating is to raise the temperature of the core to @xmath40 k as n reaches @xmath33 . prior to reaching these densities and temperatures the core passes through a chemo - thermally ( ct ) unstable regime which silk ( 1983 ) calculated would lead to fragmentation . y06 evaluated the growth time of the ct instability using the conditions in the center of their cloud , and found that it briefly becomes comparable to the dynamical time at @xmath46 ( fig . 5 ) , but that there is insufficient material satisfying these conditions to form multiple fragments . cie becomes the dominant coolant above @xmath47 , however the dynamical timescale in the core remains short compared to the timescale for isobaric modes of the thermal instability , and therefore the core does not fragment into multiple objects . as the central density and temperature increases , the jeans mass continues to drop , suggesting low mass stars could be produced if a fragmentation mechanism existed that operated quickly enough . however the simulations and the linear analysis reveals that multiple fragments are not produced , but rather a single fragment which is the accretion center for the rest of the cloud . abel refers to this process as russian doll fragmentation . " this result needs to be kept in mind when interpreting the claims that very low mass stars can results from dust - induced fragmentation ( ciardi , these proceedings ) . 3d simulations have not yet been able to evolve the cloud core to stellar densities , although valiant attempts are being made as reported by yoshida and turk at this meeting . additional physical processes must be included , such as collisional dissociation of 2 and the trapping of continuum radiation due to the cie opacity . the further evolution of the core depends critically on the role of angular momentum in halting core collapse and building a centrifugally supported disk through which most of the envelope mass would presumably be accreted ( tan & mckee 2004 ) . in the absence of rotation , we have the 1d spherically symmetric models of omukai and nishi ( 1999 ) and omukai & palla ( 2003 ) , which show that the entire envelope can be accreted provided the late time accretion rate does not exceed the eddington rate of @xmath48/yr . a plot of the instantaneous accretion rate at the end of the y06 simulation is shown in fig . the accretion rate drops below the eddington - limited rate after 10 @xmath21 has been accreted , just as the protostar begins its kelvin - helmholtz contraction to the zams ( fig . 6-bottom ) . thus most of the protostar s pre - main sequence evolution is sub - eddington , implying that rotation , not radiation , sets its zams mass . the three lines in fig . 6b show protostellar evolution tracks for different assumptions about how much centrifugal forces reduces the accretion rate onto the protostar . for even quite significant reductions in the accretion rate by rotation , the pop iii star is predicted to enter the main sequence with > 60 @xmath21 . on07a and gao et al . ( 2007 ) have studied the dependence of the central protostellar cloud s properties on formation redshift and cosmological environment . as pointed out by white and springel ( 2000 ) , dark matter mini - halos in the mass range capable of forming pop iii.1 would be expected to form in protocluster " environments at redshifts considerably higher than 20 . a pure dark matter n - body study of such an environment by gao et al . ( 2005 ) found that formation redshifts as high as 50 are likely ( assuming @xmath49 ) . in this case , the higher densities and temperatures in the collapsing cloud would shorten formation timescales and increase accretion rates . on07a simulated 12 cases by varying box size and random seed assuming a wmap1 cosmological parameters . their sample of halos formed stars over the redshift interval 33 > z > 19 . gao et al . ( 2007 ) simulated 8 cases using a multimass technique whereby a select region from a much larger cosmological volume is re - simulated at much higher mass and spatial resolution . they considered wmap1 and wmap3 cosmological parameters ( differing principally in the amount of small - scale power ) , and also different baryon fractions . they were thus able to study halos forming stars over the larger redshift interval 50 > z > 10 . both simulations terminated their evolutions when central densities of @xmath1 were reached and ignored the effects of a photo - dissociating uv background and opacity effects in the collapsing cloud core . . density profile is insensitive for formation redshift and environment , while temperature profile shows considerable variation in the central 1 pc . from oshea & norman ( 2007a ) . ] 7 shows the density and temperature distributions at the end of the 12 simulations in on07a . the density profiles are very similar to one another despite differences in formation redshift and environment , while the temperature profiles show a considerable amount of variation in the core region . this is reflected in the instantaneous accretion rate plots ( fig . while all 12 simulations show the generic feature of an early rapid rise to a maximum followed by a power - law decline , there is a two order of magnitude scatter in peak accretion rates . on07a found a correlation between redshift of collapse ( which correlates with box size ) and peak accretion rate , with higher redshift minihalos exhibiting lower accretion rates . this was understood to be the result of the early rapid formation of 2 in the denser , hotter high redshift mini - halos which results in lower final core temperatures and hence accretion rates , since @xmath50 . gao et al . ( 2007 ) also found considerable scatter in accretion rates , but no redshift correlation . rather , they found their cores had a variety of shapes and amount of centrifugal support , and that their centrifugally - supported disk - like cores had the lowest accretion rates . clearly , both of these effects could be operating and bear further study . given the importance of the late - time accretion rates on setting the stellar mass , a careful study of angular momentum evolution in the cloud cores is needed . ) . from oshea & norman ( 2007a ) . ] oshea & norman ( 2006 ; hereafter on06 ) and gao & theuns ( 2007 ; hereafter gt07 ) have considered the formation of pop iii star in a universe whose dominant mass component is warm dark matter ( wdm ) . wdm has been advanced as a solution to the many problems of cdm on sub - mpc scales ( e.g. , bode , ostriker & turok 2001 ) . operationally , wdm simulations are similar to cdm simulations except that the cdm power spectrum is assumed to by exponentially suppressed below some scale set by the free streaming scale of the wdm particle , which depends upon its mass for thermal relics . fig . 9 shows the suppression mass scale for gravitino wdm , given by @xmath51 as can be seen , the mini - halos which form pop iii.1 stars in become suppressed for @xmath52<20 kev , raising the interesting question of what is the nature of the first cosmological objects in ? < 20 kev . from oshea & norman ( 2006 ) . ] on06 carried out 8 high resolution amr simulations with resolutions similar to table 1 however varying the wdm particle mass over the range @xmath53 kev . they followed the evolution until the baryonic core had collapsed to baryon densities of @xmath54 . 10 shows the collapse redshift as a function of @xmath52 . on06 found that the @xmath52=40 kev evolution is virtually indistinguishable from the case computed with the same random seed , which collapsed at z=17.8 . the effects of decreasing the wdm particle mass is to modify the dark matter distribution at virialization from spheroidal halos to elongated filaments ( bode , ostriker & turok 2001 ) and to reduce the core collapse redshift . the @xmath52=12.5 kev case produces a collapsing primordial cloud core at z=12 in the center of a cosmological filament ( fig . 11 ) which is considerably more massive ( @xmath55 ) than its cdm counterpart . despite these differences , it is found that the high density ( @xmath56 ) evolution of the central @xmath20 of the collapsing cloud core is very similar to the cdm case ( fig . 12 ) , implying that similarly massive stars would be produced . = 0.01 a different conclusion was reached by gt07 , who simulated the more extreme case @xmath52=3 kev . in this case , the suppression mass scale is @xmath57 . objects at this mass scale virialize into long ( @xmath27 kpc comoving ) slender filaments with no dark matter substructure around which the baryons coalesce . their simulation was terminated when the filament achieved a central density of @xmath58 and temperature @xmath59 k , but before it fragments . presumably gravitational instability would fragment the filament into lumps of order the central jeans mass which is a few hundred solar masses for the values above . yet they state otherwise . they claim the filament will continue to collapse to higher densities in analogy to cloud cores , and thereby fragment into lower mass protostellar seeds . they suggest the seeds will then accrete from the cylindrical gaseous envelope while they merge along the filament axis , producing a broad distribution of pop iii stellar masses . this scenario needs to be investigated with high resolution numerical simulations because as we have seen , whether the cloud fragments or not depends not only on the local jeans mass , but on a competition of timescales which can only be determined from a self - consistent multidimensional calculation . = 12.5 kev simulation shown in fig . 11 . for @xmath60 the @xmath20 core evolves very similarly to the case despite the delay in achieving collapse conditions . from oshea & norman ( 2006 ) . ] in this section we return to simulations of the formation of pop iii stars in a universe which examine the influence of radiation backgrounds on their formation history . we discuss two cases in particular , the formation in the presence of a concurrent photo - dissociating lyman - werner background , and formation from gas that has previously been ionized by stellar sources ( relic hii regions ) . machacek , bryan & abel ( 2001 ) showed using moderate resolution amr survey simulations that a lyman - werner background radiation field delays but does not extinguish pop iii star formation by raising the minimum halo mass capable of cooling by 2 line transitions . they carried out simulations for fixed lw backgrounds corresponding to @xmath61 . the formula for a fit to their minimum halo mass is given by ciardi in these proceedings , but generally rises from @xmath62 at @xmath8=0 to @xmath63 at @xmath64 . wise and abel ( 2005 ) extrapolated the fit to lw backgrounds as high as @xmath8=1 and used a press - schechter - type analysis to estimate the evolution of the lw background for three values of the assumed mass of the the pop iii stars produced . they found that pop iii star formation becomes self limiting by z=25 and that the lw background reaches @xmath65 in the redshift range 20 > z > 10 . y03 carried out sph survey simulations similar in resolution to those of mba and found that 2 cooling becomes inefficient in primordial mini - halos for @xmath8>0.01 , making the extrapolation of the mba minimum halo mass fitting formula to @xmath8=1 somewhat suspect . neither the mba nor the y03 simulations had the mass or spatial resolution to simulate baryonic core collapse . motivated by this , oshea & norman ( 2007b ) carried out such simulations . 13 shows the time evolution of central values for gas density , temperature , entropy , and 2 fraction for four values of @xmath8 . the figures show the time delay effect very clearly . the physics is very simple . for @xmath8>0.01 the photodissocation time is shorter than the 2-body formation time , consequently the 2 fraction assumes its equilibrium value . the cooling time is then proportional to @xmath8 , and reaches @xmath66 yr for @xmath8=1 . the surprising result is that collapse eventually occurs even in the extreme case @xmath8=1 despite the greatly reduced 2 fraction in the core . the explanation for this result is that the time delay allows the virial mass and temperature of the halo to increase through mergers to @xmath67 and @xmath68 k where the per molecule cooling rate in the core is greatly elevated . in the most extreme case , central temperatures reach 5000 k which accelerates production and cooling of 2 to the point of baryonic collapse . 14-top shows the increase of the mass - weighted temperature in the central @xmath69 of gas at the collapse redshift for increasing values of @xmath8 . this translates into larger accretion rates onto the protostar at early times ( fig . 14-bottom ) exceeding @xmath70/yr for @xmath8>0.01 . taken at face value , this suggests that pop iii.2 stars forming in high lw backgrounds may be more massive than pop iii.1 stars , although we must bear in mind that it is the late - time accretion rate that governs the final mass of the star . this suggestion bears further study . cloud core as a function of lw mean intensity . top : mean temperature , bottom : accretion rate . because a lw background delays baryonic collapse , virial temperatures , core temperatures and hence accretion rates are elevated in late - forming primordial mini - halos . from oshea & norman ( 2007b ) . ] wise & abel ( 2007 ) independently carried out a study similar to oshea & norman ( 2007b ) , but considered even more massive halos ( @xmath73 ) where 2 cooling gives way the lyman alpha cooling . they find a smooth transition in collapse redshift and halo mass between those that cool by 2 lines and those that cool by atomic hydrogen lines ( fig . they find that collisional ionization of h in the warmest halos produces enough electrons to catalyze the formation of 2 even in the presence of strong lw backgrounds . the conclusion from both of these works is that 2 cooling remains the dominant mediator of early star formation even for strong uvbs , and that by restricting ones attention to only those halos for which @xmath74 k one misses a large fraction of star forming halos . finally we consider the case of a mini - halo that has previously been ionized by a strong uv source , pop iii or otherwise . as reviewed by ciardi ( these proceedings ) , pop iii stars photo - evaporate most of the baryons from the halos that formed them , and create large hii regions of 1 - 2 kpc in diameter . after a few myr the pop iii star dies , leaving behind a relic hii region . gas cools and recombines out of equilibrium , and the large residual electron fraction quickly catalyzes the formation of 2 via the h- reaction ( oshea et al . 2005 ) . the question is under what conditions does a neighbor halo form a primordial star , and in those that do , what are their properties ? the survival of neighbor halos depends on a number of factors including halo mass , evolutionary stage , proximity to the ionizing source , and its ionizing luminosity ( ahn & shapiro 2007 , whalen et al . the most important factor appears to be the central density of the neighboring halo at the time of the time it is illuminated by the ionizing source . densities of @xmath18 , corresponding to the loitering phase " of the evolution of primordial cloud cores , are high enough to self - shield the cloud against the lw radiation from the first star . cores of lower density are photo - dissociated , but crushed to higher density by the cometary i - front . the residual ionization in these cores rapidly produces 2 after the first star dies . only the lowest density neighbor halos have their baryons stripped , and are therefore unable to form stars . overall , the effect on star formation appears to be neutral ( ahn & shapiro 2007 ) . oshea et al . ( 2005 ) carried out a high resolution 3d amr cosmological simulation of the evolution of a halo forming inside the relic hii region of a pop iii.1 star . resolutions similar to that given in table 1 were used to follow the collapse of the second halo , which began 23 myr after the first star died . the collapse proceeded in a way very similar to that of a pop iii.1 star once central densities of @xmath75 were reached , indicating the universality of how pop iii stars are formed . however , the instantaneous accretion rate when central densities of @xmath1 were reached was quite low ( @xmath76/yr ) due to the fact that the second halo had a large amount of angular momentum and produced a centrifugally - supported disk in its center . so far we have considered only pop iii stars forming from gas that has remained neutral since recombination . stars forming in chemically pristine gas that has previously been ionized start from initial conditions that differ thermally , chemically , structurally , and kinematically from pop iii.1 . photoionization will raise the entropy of the gas and smooth out the clumping due to photoheating and photo - evaporation ( oh & haiman 2003 ) . in addition , the high electron fraction will catalyze rapid formation of 2 molecules via the h- channel when the hii region begins to recombine after the ionizing source switches off ( oshea et al . 2005 ) . in the absence of a high lw background , 2 fractions are elevated by 2 orders of magnitude at the cosmic mean density relative to the pop iii.1 case . finally , these conditions are ideal for the formation of hd , which becomes an important coolant at moderate densities ( @xmath77 ) and low temperatures ( t<150k ) due to chemical fractionation and hd s large permanent dipole moment ( glover , these proceedings ) . the effect of these processes on the evolution of pop iii star - forming halos has been examined by mesinger , bryan & haiman ( 2006 ) ( see also bryan et al . , these proceedings ) . they carried out high resolution amr simulations using enzo of a region that forms a pop iii.1 star at z=25 in the absence of radiative feedback . without hd cooling , they found that the negative and positive feedback effects roughly cancel , with negative feedback ( density smoothing ) slightly outweighing positive feedback ( accelerated 2 formation ) . pop iii.2 stars formed in such regions are found to have very similar properties as pop iii.1 stars as estimated by mass accretion rates . when hd cooling is included , however , the simulations predict a reduced mass accretion rate because of the lower tempersture and hence soundspeed in the accreting gas . pop iii.2 stars with typical masses of @xmath78 are claimed . a similar conclusion was reached by yoshida et al . ( 2007a , b ) , who recently have simulated the second primordial star to form in a given primordial mini - halo . they self - consistently simulate the formation of a pop iii.1 star as in y06 , and follow the expansion of the hii / heiii region it produces . as in earlier studies ( whalen et al . 2004 , kitiyama et al . 2004 ) , gas in the minihalo core is swept out to the virial radius by a strong photo - evaporative flow . after the pop iii.1 star dies , this gas recombines and cools , forming hd . they continue the calculation until this gas recollects into the dark matter mini - halo core and forms a pop iii.2 star . they find that it takes about 100 myr for the second primordial star to form , and that as a consequence of hd cooling , the mass accretion rate is reduced , in agreement with the results of mesinger , bryan & haiman ( 2006 ) . in yoshida , omukai & hernquist ( 2007 ) they calculate the accretion history of the pop iii.2 protostar using a 1d protostellar accretion code . the result is shown in fig . it enters the main sequence with a mass of @xmath4 , significantly less than the pop iii.1 star that pre - processed its birth cloud . mln wishes to thank the conference organizers for their financial support for attending the meeting in santa fe , and their patience while this manuscript was being prepared . mln also wishes to thank tom abel , brian oshea , and naoki yoshida for permission to reproduce their figures . the research of mln is supported by the national science foundation through grant ast-0307690 . abel , t. , bryan , g. & norman , m. l. , 2000 , apj , 540 , 39 abel , t. , bryan , g. & norman , m. l. , 2002 , science , 295 , 93 ahn k. , shapiro p. r. , 2007 , mnras , 375 , 881 barkana , r. , & loeb , a. , 2001 , ara&a , 39 , 19 barkana , r. , & loeb , a. , 2004 , apj , 609 , 474 bode , p. , ostriker , j. p. & turok , n. , 2001 , apj , 556 , 93 bromm , v. , coppi , p. , & larson , r. , 1999 , apj , 527 , 5 bromm , v. , coppi , p. , & larson , r. , 2002 , apj , 564,23 bromm , v. , & larson , r. , 2004 , ara&a , 42 , 79 ciardi b. , ferrara a. , abel t. , 2000 , apj , 533 , 594 ciardi b. , ferrara a. , 2005 , space science reviews , 116 , 625 . gao , l. , white , s. , jenkins , a. , frenk , c. , & springel , v. , 2005 , mnras , 363 , 379 gao , l. , yoshida , n. , abel , t. , frenk , c. , jenkins , a. , & springel , v. , 2007 , mnras , 378 , 449 gao , l. , & theuns , t. , 2007 , science , 317 , 1527 glover , s. , 2005 , space sci . , 117 , 445 haiman z. , abel t. , rees m. j. , 2000 , apj , 534 , 11 haiman z. , rees m. j. , loeb a. , 1997 , apj , 476 , 458 larson , r. b. , 2003 , reports on progress in physics , 66 , 1651 machacek m. m. , bryan g. l. , abel t. , 2001 , apj , 548 , 509 mesinger , a. , bryan , g. , & haiman , z. , 2006 , apj , 648,835 oh s. p. , haiman z. , 2003 , mnras , 346 , 456 omukai k. , nishi r. , 1999 , apj , 518 , 64 omukai , k. & palla , f. , 2003 , apj , 589 , 677 oshea , b. w. , 2005 , phd thesis , university of illinois oshea b. , abel . t. , whalen , d. , & norman m. l. , apjl , 628 , l5 oshea b. , norman m. l. , 2006 , apj , 648 , 31 oshea b. , norman m. l. , 2007 , apj , 654 , 66 oshea b. , norman m. l. , 2007 , astro - ph/0706.4416 ricotti m. , gnedin n. y. , shull j. m. , 2002 , apj , 575 , 49 ripamonti , e. , & abel . t. , 2004 , mnras , 348 , 1019 schneider r. , omukai k. , inoue a. k. , ferrara a. , 2006 , mnras , 369 , 1437 silk , j. , 1983 , mnras , 205 , 705 tan , j. , & mckee , c. f. , 2004 , apj , 603 , 383 whalen d. , abel t. , norman m. l. , 2004 , apj , 610 , 14 whalen , d. , oshea , b. , smidt , j. , & norman , m. , 2007 , astro - ph/0708.1603 white , s. d. m. , & springel , v. , 2000 , in the first stars , eds . a. weiss et al . , springer - verlag , berlin , 327 wise , j. h. , & abel , t. , 2005 , apj , 629 , 615 wise , j. h. , & abel , t. , 2007 , apj , 671 , 1559 yoshida n. , abel t. , hernquist l. , sugiyama n. , 2003 , apj , 592 , 645 yoshida , n. , omukai , k. , hernquist , l. , & abel , t. , 2006 , apj , 652 , 6 yoshida n. , oh s. p. , kitayama t. , hernquist l. , 2007 , apj , 663 , 687 yoshida n. , omukai , k. , hernquist l. , 2007 , apj , 663 , 687
we review recent 3d cosmological hydrodynamic simulations of primordial star formation from cosmological initial conditions ( pop iii.1 ) and from initial conditions that have been altered by radiative feedback from stellar sources ( pop iii.2 ) . we concentrate on simulations that resolve the formation of the gravitationally unstable cloud cores in mini - halos over the mass range @xmath0 and follow their evolution to densities of at least @xmath1 and length scales of @xmath2 pc such that accretion rates can be estimated . the advent of ensembles of such simulations exploring a variety of conditions permits us to assess the robustness of the standard model for pop iii.1 star formation and investigate scatter in their formation redshifts and accretion rates , thereby providing much needed information about the pop iii imf . the simulations confirm the prediction that pop iii.1 stars were massive ( @xmath3 ) , and form in isolation in primordial mini - halos . simulations of pop iii.2 star forming in relic hii regions suggest somewhat lower masses ( @xmath4 ) which may help explain the chemical abundances of extremely metal poor stars . we note that no 3d simulation at present has achieved stellar density let alone followed the entire accretion history of the star in any scenario , and thus the imf of pop iii stars remains poorly determined theoretically . address = center for astrophysics and space sciences , university of california , san diego , la jolla , california usa ; [email protected]
the dynamical evolution of gravitationally bound stellar clusters has been extensively studied for decades , and the basic theory is thought to be secure . populous systems evolve , over many crossing times , through the processes known collectively as dynamical relaxation ( * ? ? ? * chapter 7 ) . the inner core of the cluster contracts , effectively transferring energy to the outer halo , which expands as a result . concurrently , stars of relatively high mass sink toward the cluster center . theory predicts further that the interior contraction leads eventually to core collapse , a catastrophic rise in central density @xcite . as first suggested by @xcite , the runaway is halted when hard binaries form near the center and release energy through three - body encounters . observations of globular clusters , which can be significantly older than their relaxation times , have confirmed these expectations beautifully . the surface brightness profiles of milky way globular clusters indicate that some 20 percent harbor collapsed cores @xcite . near the centers of many systems are x - ray binaries and blue stragglers @xcite , both created at high stellar density , perhaps during the collapse phase . globular clusters have impressive populations ( ) , but are relatively rare and distant groups . open clusters are sparser ( ) , but much more common and closer at hand , with over a thousand catalogued @xcite ; the sample is thought to be complete out to 2 kpc @xcite . ironically , their evolutionary status is much less clear . half of open clusters disintegrate within after birth @xcite , a span corresponding to at most a few initial relaxation times . not surprisingly , there is little observational signature that relaxation has occurred , aside possibly from mass segregation , first found by @xcite . a small fraction of open clusters , located at the outskirts of the galaxy , have survived for over 1 gyr @xcite . even these ancient systems show no sign of core collapse . a prototypical example , m67 , has a smooth surface density profile that , unlike post - collapse systems , is well fit by a king model @xcite ; the system appears to be in the last stages of tidal disruption @xcite . @xcite performed @xmath0-body simulations to follow the evolution of m67 from infancy to its inferred age of 4 gyr . their preferred model began with 12,000 single stars and an equal number of binaries ; only about 10 percent of these stars survived to the end . even over the protracted time of the simulation , the cluster never exhibited classical dynamical relaxation . instead , the central mass density rose slightly and then declined . @xcite attributed this behavior to the binary - rich initial population . hard binaries undergo superelastic encounters with other stars and effectively heat the cluster @xcite , counteracting the outward energy transfer driving dynamical relaxation . indeed , it has long been appreciated that the presence of even a few binaries can radically alter the evolution of the sparsest groups @xcite . these findings prompt us to ask a more general question : under what conditions does binary heating prevent significant core collapse ? in this paper , we begin to address this larger issue , utilizing our own suite of @xmath0-body simulations . the basic answer to our question is that _ the presence of massive stars _ is essential . these massive stars couple with others to form pairs that , through three - body interactions , frustrate core contraction relatively early , so that there is little or no rise of the central density . the system thereafter undergoes _ global expansion_. here , the stellar density falls everywhere . however , there a mitigating factor in this scenerio - stellar evolution . in more populous clusters with longer relaxation times , the most massive stars die out . binary heating is tamed , and does not effectively oppose core contraction until later in the cluster s evolution . as a result , this contraction proceeds to yield a higher density contrast . such is the case in the globular clusters that have undergone true core collapse . we elucidate these processes in a step - by - step fashion , beginning with very simple , highly idealized systems , and progressively adding more realistic features . in section 2 below , we introduce an energy analysis that quantitatively distinguishes classical dynamical relaxation from global expansion . we apply this analysis to both single - mass systems and those with a more realistic stellar mass distribution ; only in the latter does binary heating come into play . section 3 describes in more detail the discontinuous manner in which binary heating actually operates , while section 4 shows how stellar evolution lessens the effect . finally , in section 5 , we discuss the implications of our results for both open clusters , which we have simulated more or less accurately , and globular clusters , which we can not model directly . we have run a suite of n - body simulations , all employing the publicly available code starlab ( * ? ? ? * appendix b ) . significantly for our purposes , the code uses no softening in the gravitational potential , so that the formation and dynamical interactions of binaries are followed accurately . as we did not have access to any special - purpose hardware , our simulations were limited by time constraints to . let us first adopt the simplified , and assuredly unrealistic , assumption that all stars have identical mass . many of the classic theoretical papers in stellar dynamics , as well as textbook accounts , have utilized such single - mass models . we assume the cluster starts out in virial equilibrium , with a mass density profile corresponding to an polytrope . we recently found that this particular configuration best describes the very early state of the pleiades , just after gas removal ( * ? ? ? * hereafter paper i ) . we stress , however , that the precise initial state is of little consequence ; the system loses memory of this state well within one relaxation time , as in the case of the pleiades . in this and our other simulations , we model only isolated systems , with no tidal gravitational field either from the galaxy or from passing molecular clouds . more complete models should include such an external field , which eventually destroys all clusters . however , the presence of the field does not qualitatively alter our main conclusions regarding cluster evolution up to the point of dissolution . figure 1 shows the evolution of lagrangian mass shells in a cluster with the representative population of . here we display the temporal change of the shell radii , expressed as fractions of the cluster s initial virial radius , @xmath1 . each shell has the indicated value of @xmath2 , where @xmath3 is the interior mass and @xmath4 the initial mass of the whole cluster . the time itself is normalized to the initial relaxation time @xmath5 , for which we utilize equation ( 1.37 ) in @xcite : @xmath6 where . following standard practice ( e.g. , * ? ? ? * section 2.4 ) , the crossing time @xmath7 is given by @xmath8 we need not choose values for @xmath1 or @xmath7 as long as we compare only nondimensional versions of all the relevant quantities @xcite . as we see in the figure , interior shells contract , while those closer to the cluster boundary expand . this behavior is the hallmark of dynamical relaxation . in this plot , the steeply accelerating contraction that signifies core collapse is not present , simply because of the limited time range covered . @xcite , who investigated single - mass models using a special - purpose ( grape-4 ) computer , found core collapse to occur after about 6 initial relaxation times . ( see his figure 1 plotting the central density for the 4k run , after noting that @xmath5 corresponds to 62 of his scaled n - body time units . ) our results , over a more restricted interval , are fully consistent with makino s . other researchers , utilizing a variety of techniques , have verified through simulations that dynamical relaxation occurs in single - mass systems ( e.g. , * ? ? ? * ; * ? ? ? nevertheless , it is worth revisiting the basic energetics of the process . the cluster s total energy is conserved , so its dual contraction and expansion reflects energy _ transfer _ from the inside out . according to figure 1 , the shell with grows only slowly . thus , this shell lies just outside the core - halo boundary . we arrive at the same conclusion by calculating directly the mean rate of energy transfer . let @xmath9 be the total kinetic energy within mass @xmath3 , and @xmath10 the time derivative of this quantity . after finding the best - fit straight line to @xmath11 over the full time span of the simulation ( ) , we then calculate @xmath10 as the slope of this line . figure 2 shows @xmath10 as a function of the mass fraction @xmath2 . what we actually display is the nondimensional quantity @xmath12 , where @xmath13 and where @xmath14 is the cluster s total initial kinetic energy . the curve in figure 2 has a small , central dip , a numerical artifact of the large scatter in @xmath11 over this region containing relatively few stars . thereafter , @xmath10 rises , attains a maximum , and then monotonically declines . within the rising portion of the curve , the kinetic energy in a shell of thickness @xmath15 _ increases _ with time that is , @xmath16 a certain , interior region of the cluster is thus gaining kinetic energy . self - gravitating systems have negative heat capacity . thus , the increasing kinetic energy ( and therefore temperature ) of the inner core signifies a _ decreasing _ total energy . analogous reasoning shows that the region corresponding to the descending portion of the curve is _ gaining _ total energy , and therefore comprises the halo , which receives its energy from the core . it is natural , therefore , to locate the core - halo boundary at the peak of the curve , i.e. , where . according to figure 2 , this boundary is at , in agreement with the analysis of figure 1 . we next eliminate the most egregious simplification in the model , the assumption of a uniform stellar mass . as has long been appreciated ( e.g. , * ? ? ? * ; * ? ? ? * ) , relaxing this assumption has a profound effect on cluster evolution . we turn again to our recent study of the pleiades ( paper i ) , and use , as our stellar mass distribution , the one characterizing the cluster in its infancy , soon after gas dispersal . this distribution was a lognormal , joining smoothly onto a power law at higher masses . the full distribution is given in equation ( 17 ) of paper i , with the parameter values listed in table i of the paper . following that study , we take the minimum and maximum stellar masses to be @xmath17 and @xmath18 , respectively . ( see section 3.3 for reconsideration of the maximum mass . ) once again , we assume that the mass density profile of the cluster is that of an polytrope . cluster members are all single stars , whose masses are drawn randomly from the assumed stellar distribution . we set , and ignore both mass loss during stellar evolution and any tidal gravitational field . the left panel of figure 3 shows the evolution of lagrangian mass shells , in a manner analogous to figure 1 . in this case , we note first that radii tend to exhibit more jitter in their evolution . this characteristic stems from the redistribution of stellar mass over the crossing time . even after averaging over the jitter , interior mass shells do not monotonically contract , as they did before . these radii initially shrink . however , at some relatively early time , they reach a minimum and begin to expand . ( we will later identify this time with binary formation ; see below . ) this expansion continues , with ups and downs , for the remainder of the simulation . radii corresponding to expand from the start . for the single - mass model , the shrinking of any interior radius unambiguously signifies that the average distance between stars is also diminishing in that region . in the present case , the interpretation of early contraction is complicated by the phenomenon of mass segregation . the mass of any star is drawn from the same distribution , regardless of that object s initial location in the cluster . thus , there is no mass segregation initially . however , relatively massive stars quickly drift toward the center , under the influence of dynamical friction . as these stars accumulate , the radius of any region _ of fixed mass _ may shrink , even if the average interstellar spacing does not . to illustrate this point graphically , the right panel of figure 3 shows , for the same simulation , the evolution of radii containing a fixed _ number fraction _ , @xmath19 , of the cluster . here , @xmath20 is the interior number of stars . since no mass segregation was imposed at the start , the lagrangian mass shell with initially has the same radius as the `` number shell '' with . the radius of the former contracts at early times , but figure 3b shows that the radius of the latter stays constant and later grows . the average interstellar separation within the volume is not shrinking . tracking the radii of lagrangian mass shells is a widely employed technique for visualizing cluster evolution . other authors who have studied relatively low-@xmath0 systems under similar assumptions have documented the early contraction of interior shells . this development is said to demonstrate core collapse ( e.g. , * ? ? ? * ; * ? ? ? the putative collapse occurs within one initial relaxation time , much earlier than in single - mass models , and ends before the central mass density has risen dramatically . however , contraction of inner mass shells may be due to mass segregation and thus it alone does not definitively show the occurrence core collapse . in any cluster containing a range of stellar masses , any temporal increase in mass segregation obscures the physically distinct phenomenon of core contraction . the latter may still be occurring , and is in this case to a limited degree . consider again figure 3b , where we have added a deeply embedded , number shell corresponding to . this shell _ does _ contract initially , although only by a relatively small amount before the turnaround at @xmath21 . true dynamical relaxation occurs at the start , but the accompanying interior contraction is weak , and is soon aborted . as in the single - mass model , calculation of the mean energy transfer rate adds physical insight . figure 4 is an energy transfer profile before turnaround , i.e. , over the interval . the curve is similar to that in figure 2 . the rate @xmath22 eventually rises smoothly , peaks a bit beyond , and thereafter declines . such a profile is again indicative of dynamical relaxation . identifying the core - halo boundary with the peak of the curve is consistent with the pattern of mass shell curves in figure 3b . thus , the radius corresponding to initially contracts slightly , while that with expands . figure 5 displays the energy transfer profile after the turnaround . here , the mean rate @xmath10 is determined over the interval . the profile is now qualitatively different , and illustrates a distinct mode of cluster evolution . the kinetic energy in every mass shell falls with time . so does , therefore , the kinetic energy of the entire cluster . from the virial theorem , the cluster as a whole is gaining in _ total _ energy . this injection of energy accounts for the system s global expansion , as seen in all the radii of figure 3 for . the central engine driving the expansion is binary heating , as we verify shortly . we have run analogous simulations for cluster populations ranging from to 16,384 . the upper limit was a practical one ; the last case required three months on a desktop computer . all simulations gave qualitatively the same result . the cluster experiences an early , transient phase of dynamical relaxation . during this epoch , the central number density rises by only a modest amount , typically a factor of 2 . the central mass density rises by about a factor of 10 , with the larger increase reflecting the onset of mass segregation . in all cases the end of this early period coincides with the formation of the first long - lived binary system , with @xmath23 for all @xmath0 , consistent with the findings of @xcite . from this early epoch until the end of the simulation , the cluster undergoes global expansion . as we shall next see , even the brief , transient period of dynamical relaxation was itself an artifact that vanishes under more realistic initial conditions . one feature of our simplified cluster models is that they consist initially of only single stars . it is well known that most field stars of solar - type mass have at least one binary companion @xcite . the observational assessment of binarity in even the nearest open clusters is challenging , but the indication so far is that the fraction is comparable to the field - star value ( e.g. , * ? ? ? * ; * ? ? ? since , as we will see , binary heating plays a key role in dynamical evolution , we should try to gauge the influence of primordial pairs . for this purpose , we may utilize our own simulated history of the pleiades ( paper i ) . our initial state , another polytrope , was that which evolved , over the 125 myr age of the cluster , to a configuration most closely resembling the current one . in a previous investigation @xcite , we found that 76 percent of the stellar systems today are binary . the best - fit initial state in paper i consisted essentially of _ all _ binaries , with the corresponding fraction being 95 percent . we endowed these binaries with a lognormal period distribution and a thermal distribution of eccentricities , reflecting both conditions in the field population and in the pleiades itself @xcite . in addition , the masses of the primary and secondary stars were correlated ( see section 2.1.2 of paper i ) . finally , the initial state had a finite degree of mass segregation , i.e. , the masses and energies of stellar systems were also correlated . the reader is again referred to paper i for the detailed prescription . mass segregation may be characterized quantitatively through the gini coefficient ( * ? ? ? * section 4.2 ) . this quantity measures how fast the cumulative mass increases outward relative to the cumulative number of systems . the initial state of the pleiades had . the initial number of stellar systems , both binary and single , was . figure 6 shows the evolution of mass- and number - shell radii . note that the current age of the pleiades corresponds to about 0.5 initial relaxation times . thus , these plots span a significantly briefer interval than those in figures 1 and 3 . bearing this fact in mind , we see that the curves are generally similar to those in figure 3 . after a brief initial plunge , the radii of mass shells with expand , while the shell contracts , at least over this time . number shells undergo an analogous , early contraction , and then either remain static ( ) or expand . if binaries are energetically significant , why is the cluster evolution not radically altered ? this is an important question , to which we shall return presently ( see section 3.2 ) . the early dips seen in all the curves of figure 6 signify that the cluster as a whole initially contracts . this behavior is an artifact of our specific method for implementing mass segregation . as explained in section 3 of paper i , the configuration starts out in precise virial equilibrium . however , the redistribution of higher stellar masses toward the center alters slightly the gravitational potential from that associated with an polytrope . over a period lasting about two crossing times ( ) , the cluster `` bounces , '' and then settles into a configuration that evolves smoothly thereafter . the bounce does not occur if we impose no mass segregation initially . in that case , both the mass density of stars and the gravitational potential correspond exactly to an polytrope . figure 7 shows results from such a simulation . in this `` pleiades - like '' cluster , the initial state is identical to that in figure 6 , but without mass segregation . over the time span covered ( @xmath24 ) , number shells either remain static or expand ( figure 7b ) . the early contraction of the innermost shells seen in figure 3 never occurs , due to the heating by primordial binaries . in summary , there is no evidence of classical dynamical relaxation ; the cluster evolves purely through expansion . the radii of interior mass shells do contract ( figure 7a ) as a result of increasing mass segregation ; the gini coefficient grows from 0 to 0.15 over this time ( see figure 8 of paper i ) . returning to the more realistic pleiades simulation , it is again instructive to visualize the internal transport of energy . from our description thus far , there should be no core - halo boundary , identified by the peaks of the energy transfer profiles in figures 2 and 4 . figure 8 , which plots @xmath10 as a function of @xmath3 , bears out this expectation . here , we have computed @xmath10 by a linear fit over the full time span of the simulation . we see that @xmath10 monotonically falls from zero to increasingly negative values . ( compare figure 5 and the accompanying discussion . ) the cluster as a whole is cooling down , and is therefore gaining in total energy . the pleiades evolved to its present state through global expansion , not dynamical relaxation . let us reconsider the highly idealized clusters with which we began - one with stars of identical mass , and the other with a continuous range of stellar masses . in both cases , the initial systems contained neither binaries nor higher - order multiple systems . the evolving , single - mass cluster spawned no new binaries over the duration of our simulation . however , @xcite found , in his more extensive investigation of the single - mass model , that binaries do eventually form in the contracting interior , and that their heating reverses core collapse at . the core subsequently undergoes the gravothermal oscillations predicted by @xcite and @xcite using fluid models with an internal energy source . in our cluster with a realistic stellar mass distribution , the interior contraction ends much sooner , within a single initial relaxation time . is this prompt reversal also due to binary heating ? the answer is yes . we have confirmed that the turnaround at coincides with the appearance of the first hard binary . here , we remind the reader that a `` hard '' binary is one whose gravitational binding energy exceeds the average , center - of - mass kinetic energy of all other stellar systems . it is only such pairs that donate energy to neighboring stars during a close flyby , and thereby become even harder . this is the essence of binary heating @xcite . why do binaries form so much earlier in this cluster than in the single - mass model ? closer inspection reveals that these new systems are comprised of stars that are appreciably more massive than the average cluster member . this fact is readily understood in a qualitative sense . in clusters with no initial mass segregation , the relatively massive stars promptly sink to the center . once in close proximity , these objects have a stronger mutual attraction than other cluster members , and are thus more prone to forming binaries . in more detail , a gravitationally bound pair of such stars can only form by giving energy to a third star . binary formation is thus a three - body process . in the traditional analysis of cluster evolution based on single - mass models , three - body encounters throughout the bulk of the system are considered too rare to be of significance . @xcite show that @xmath25 , the time for the first binary to form via this route , is much longer than @xmath5 . specifically , they estimate that @xmath26 our superscript on @xmath25 emphasizes that this time pertains to the highly specialized case of equal - mass stars . the derivation of equation ( 5 ) assumes that the binary - forming stars reside in a region of average density . this assumption breaks down if the interactions occur in a deeply collapsing core . three - body interactions _ can _ proceed here efficiently ( e.g. , * ? ? ? * ) . however , the process is too slow in regions where the density is not greatly enhanced . the situation changes dramatically once the cluster is endowed with a distribution of stellar masses . here , binaries can form even where the density is close to the average . we may demonstrate this fact through a slight alteration of the heuristic derivation for @xmath25 given in @xcite . suppose that stars require a minimum mass @xmath27 to be part of a binary . the time @xmath28 for a given one of these objects to come within distance @xmath29 of another with comparable mass is @xmath30 here @xmath31 is the number fraction of such stars , @xmath32 is the average cluster number density , and @xmath33 the velocity dispersion . during this encounter , there is a probability that a third star will also be within the interaction distance @xmath29 . this star can have the average mass . thus , the time for the original star to suffer a binary - forming triple encounter is about . there are @xmath34 such stars in the cluster . the time for _ any _ such star to form a binary is @xmath35 in order for a hard binary to form , the gravitational potential energy of the binary must be equal to or greater than the average kinetic energy in the cluster : @xmath36 thus , @xmath37 from the virial theorem , , where @xmath1 is the cluster s virial radius . using this expression along with the approximation that we find @xmath38 now the relaxation time from equation ( 1 ) may be approximated as @xmath39 dividing equation ( 10 ) by equation ( 11 ) yields @xmath40 which is a simple modification of the analogous equation ( 5 ) . on the righthand side of equation ( 12 ) , the factor @xmath41 is necessarily greater than unity . on the other hand , is , in practice , so small that . consider , for example , the models described in section 2.2 , which had a stellar mass distribution appropriate for the infant pleiades . here , . in our cluster , we find empirically that the minimum mass in any newly formed binary is ; the corresponding @xmath31-value is @xmath42 . equation ( 12 ) then predicts that , in good agreement with our numerical results . this derivation is , of course , highly simplified , and the quantitative result above should not be given too much weight . the relative velocity of an encounter in the core will typically be larger than in the rest of the cluster , and the core density will be higher than the average . a more complete derivation of @xmath21 would also consider the physical basis for the minimum mass @xmath27 . presumably , this limit is set by the rate at which dynamical friction allows stars of various mass to drift inward . we will not embellish the argument along these lines , but simply note that equation ( 12 ) adds justification for our main points : ( 1 ) the rate of binary formation is very sensitive to the stellar mass distribution , and ( 2 ) even in hypothetical clusters composed entirely of single stars , binaries form relatively quickly . it is only by adopting the extreme assumption that these single stars have identical mass that binary formation can be delayed to the point of core collapse . a hard binary that resides within a cluster , no matter how it formed , adds energy to the whole system . the process , like the creation of new pairs , is a three - body interaction . as a result of the encounter , the binary usually tightens and releases energy . this heating accounts for the expansion of both the mass and number shells in figure 3 , for . expansion driven by binaries is global , and differs qualitatively from the dual contraction and expansion seen in the single - mass model ( figure 1 ) . this difference is also apparent when we view the evolution of the cluster s aggregate energy . first , we need to distinguish _ internal _ and _ top - level _ energies . in the first category is the gravitational binding energy of each binary , and the kinetic energy of both component stars with respect to their center of mass . in the top - level category are the center - of - mass kinetic energies of all bound stellar systems , whether single or multiple , and the gravitational potential energy of this array . thus , the kinetic energy @xmath9 considered previously was actually a top - level quantity . the cluster s total energy @xmath43 is the sum of the two contributions : @xmath44 here , we are ignoring the relatively small amount of energy carried off by escaping stars . in the absence of an external tidal field , @xmath43 remains strictly constant . binary heating , whether by creation of a new pair or interaction of an existing pair with single stars , lowers @xmath45 and transfers the same amount of energy to @xmath46 . the solid curve in figure 9 shows the evolution of the top - level energy in the model cluster with a realistic stellar mass distribution ( ) . here @xmath46 is normalized to @xmath47 , its initial value . since this cluster begins with all single stars , @xmath46 and @xmath43 are identical at the start . upon the formation of the first hard binary at , @xmath46 takes a substantial , upward jump . subsequent jumps occur whenever new hard binaries form , or when existing ones impulsively heat the cluster . as an instructive comparison , the dashed curve in figure 9 shows @xmath46 for the single - mass model described in section 2.1 . the curve is very nearly flat . despite some weak and transient interactions , no stable , hard binaries form over the span of the simulation . one interesting feature of figure 9 is that the jumps tend to diminish with time . indeed , @xmath48 scales roughly with @xmath43 , where the latter approaches zero as the cluster inflates . to see the origin of this scaling , consider in more detail the energetics of the three - body interaction . the energy released as the binary tightens is shared by that pair and the passing star . both recoil from the site of the original encounter . in our simulations , the binary is lifted to a much higher orbit , but usually does not become unbound . the pair then drifts back down , via dynamical friction , and gives its energy to surrounding stars . if denotes this contribution to the total energy change , then , where @xmath49 is the binary mass . the recoiling single star , of mass @xmath50 , rockets away at high speed , much larger than @xmath33 , and is lost to the cluster . on its way out , the star does work on the system . here , @xmath51 is the depth of the top - level gravitational potential well at the interaction site , which is close to the cluster center . now both @xmath52 and @xmath51 are proportional to @xmath46 itself . thus , the total energy change , , is also proportional to @xmath46 . the time when the first hard binary appears , , is also when the cluster begins to expand ( figure 3 ) . thus , the binary immediately begins to heat the system through interactions with its neighbors . eventually , the binary itself is ejected as a result of such an encounter , to be replaced later by another . over the course of the simulation , a total of 4 hard binaries arise . but a snapshot of the cluster at any time shows it to contain either a single binary or none at all . for example , the flat portion of the energy curve between and represents such a barren period . it is indeed remarkable , as many investigators have noted , how a handful of binaries control the fate of a populous cluster . in more detail , there is variation of the heating rate with @xmath0 . smaller systems experience fewer binary interactions . on the other hand , each interaction creates a larger @xmath48 relative to @xmath43 . larger systems have more frequent interactions , with each contributing less relative energy . in the end , the rate of energy input actually varies little , when averaged over a sufficiently long period . what if the cluster is seeded with many binaries initially ? figure 10 shows the evolution of the top - level energy for the pleiades simulation . there are now many binaries even at the start , and thus no initial period of constant @xmath46 . remarkably , however , the evolution is quite similar to the case of no primordial binaries . the top - level energy is changed in a few discrete jumps these few major interactions always involve binary ( or triple ) systems composed of the few most massive stars ( see also @xcite ) . the important lesson is that only a special subset of binaries strongly influences a cluster s evolution . these are systems which are relatively massive , wide enough to have a significant interaction cross section with other stars , and yet tight enough to be hard . to be sure , the primordial binaries in the pleiades - like simulation shown in figure 7 do halt the initial contraction . relatively little energy input is required to do so . virtually all primordial binaries are either of too low a mass , or are so tight that they effectively interact as a single system . it is the subsequent coupling of relatively few massive stars that inject much greater energy and principally drive the cluster s expansion . we have seen how binary heating can dominate a cluster s evolution . for a realistic stellar mass spectrum , the effect begins very quickly , in less than a single relaxation time . under these circumstances , the cluster is still very far from the point of true core collapse . following our pleiades study ( paper i ) , we have set the maximum mass at @xmath53 . the reasoning here was that more massive objects would have ionized the parent cloud , allowing the stars to disperse before they could form a bound cluster . in any event , it is instructive when elucidating basic physical principles , to relax this assumption and gauge the effect . we now allow stars in our @xmath54 cluster to be drawn from the same mass function as before , but with a nominal upper limit of @xmath55 . in practice , no star ever realizes this mass ; the largest generated is about @xmath56 . again , there are no primordial binaries . based on our earlier arguments , we would expect binary formation to begin even sooner . indeed this is the case . the first stable , hard binary forms at , a factor of 2 earlier in time . figure 11 shows the evolution of both lagrangian mass and number shells . by either measure , the cluster undergoes global expansion at all radii . not surprisingly , there are detailed differences from the @xmath53 case . the apparent early contraction of the innermost mass shells is now _ entirely _ due to mass segregation . even the number shell with @xmath57 expands from the start . a closer analysis shows that there is again never more than a single binary in the cluster at any instant , although the specific pair changes identity in time . both components of this pair are always among the top 5 stars by mass . these binaries generate especially strong heating . figure 12 shows the evolution of the top - level energy . for comparison , we also reproduce the analogous plot from figure 9 for the simulation with @xmath53 . binary heating now begins much sooner , and the individual three - body encounters inject larger amounts of energy . this result corroborates our earlier conclusion that @xmath58 is proportional to the mass of the binary system . in the presence of very massive stars , the cluster energy , @xmath59 , diminishes to only 7 percent of its initial value over the time considered . in some of our simulations , the heating was so severe as to effectively dissolve the cluster , inflating it to thousands of times its initial size ( in the absence of a tidal field ) . globular clusters may have been born in parent clouds so massive that even multiple stars producing hii regions do not disrupt them @xcite . why , then , are young globular clusters not dispersed by binary heating ? how do they evolve to the point of core collapse ? to answer these questions , we now include the last key ingredient - stellar evolution . it has long been appreciated that the mass loss associated with stellar evolution can have a dramatic effect on the early life of a cluster @xcite . as its largest stars die out , the cluster s total mass can decrease significantly . the loss of gravitational binding causes the cluster to expand . this initial phase of expansion , which is ubiquitous in simulations , is quickly stifled because lower mass stars survive much longer . the loss of the cluster s most massive stars has another effect , more relevant here , that is not as widely appreciated ( see however * ? ? ? * ) . as we have seen , it is these same stars that reverse core contraction and drive global expansion through binary formation and heating . because of mass loss , however , the objects die out before they can pair with others . stellar evolution thus tamps down binary heating and postpones the global expansion that this heating drives . the code starlab is able to track stellar evolution , including mass loss , by applying analytic fitting formulae . once we switch on this module , however , we need to give an explicit size scale for our cluster , in order to set the relation between dynamical and stellar evolutionary times . we select a virial radius of @xmath60 pc as a representative value . in our stellar mass function , we continue to set @xmath61 m@xmath62 . as before , we focus on the case @xmath54 . our cluster has an initial crossing time of 8 myr , and an initial relaxation time of @xmath63 myr . we follow the evolution of the cluster for 8.5 gyr , which is about 15 relaxation times . figure 13 shows the evolution of lagrangian radii . at first glance , the pattern looks similar to figure 11 , which shows the same cluster , but without stellar evolution . closer inspection reveals important differences . the cluster now undergoes a rapid expansion . during the first 800 myr , corresponding to 1.5 initial relaxation times , the virial radius increases by a factor of 2 once the maximum mass of the stars falls below about 2 m@xmath62 , the expansion slows . the next , slower phase of expansion lasts until about 3 gyr , or 5.5 @xmath5 . here , heating is provided by an 18 m@xmath62 black hole left behind by a formerly 67 m@xmath62 star . due to its mass , it readily forms a binary system with another star , and this system is the source of the heating . the quantitative details of this phase are as uncertain as our knowledge of the late stages of massive stellar evolution . for example , @xcite find that the same 67 m@xmath62 star leaves behind a 4 m@xmath62 black hole , which would create much less heating and expansion . neither of these phases are in the previous figure 11 , which omitted stellar evolution . instead there is an initial brief contraction of the innermost mass shells . as noted earlier , the innermost number shells do not contract , so we are actually witnessing the effects of mass segregation . in the present case , significant contraction of both the mass and number shells occurs . the 18 m@xmath62 black hole and its companions have been ejected , and no new binaries form . hence the system is undergoing true dynamical relaxation . compared to the system with no stellar evolution , this phase is quite protracted , lasting 6 @xmath5 . again , the most massive binaries , that would have halted contraction earlier have died off . eventually , however , new binaries do form . as before , it is the highest mass stars present that interact and cause heating . the cluster thereafter enters a prolonged phase of global expansion . this lasts through the end of the simulation . in summary , stellar mass loss has delayed binary formation , and therefore cluster expansion , but not prevented their occurrence . the simulations we performed without stellar evolution all found that @xmath21 , the epoch marking the onset of binary formation , was a fixed fraction of @xmath5 . we have just seen , in the specific case of @xmath54 that stellar mass loss modifies this result , increasing @xmath64 . another path to the same conclusion comes from equation 12 . stellar evolution lowers the minimum mass @xmath27 of stars that are around to form a hard binary that can heat the cluster . the ratio @xmath65 thus increases , and @xmath64 rises accordingly . as we consider clusters of higher @xmath0 , a basic point to note is that @xmath5 itself increases . if the average mass @xmath66 is unchanged , then equation 11 shows that @xmath5 scales as @xmath67 , ignoring the logarithmic factor . thus , for similar virial radii @xmath1 , clusters of higher population take longer to relax . the binary formation time @xmath21 in these systems is longer , and , because of stellar evolution , is a higher fraction of @xmath5 itself . by a given age , therefore , a cluster of higher population has experienced more dynamical relaxation . that is , its core has contracted further . in our view , this trend represents the critical difference between open and globular clusters . the former undergo , at best , a brief , tepid period of core contraction . in our pleiades simulation , even this mild contraction is stifled by heating from primordial binaries . globular clusters , on the other hand , undergo much longer periods of dynamical relaxation . in some cases , this prolonged epoch results in true core collapse . throughout this study , we have carried out our simulations to arbitrary times , just long enough to illustrate the main evolutionary phases . the smaller-@xmath0 groups on which we focus , are eventually destroyed tidally , either by the general galactic field or by the close passage of giant molecular clouds . * equation ( 8.57 ) ) give the cloud disruption time scale as @xmath68 where @xmath69 is the half - mass radius . consider again our simulation of an @xmath54 cluster , with stellar evolution included . here , @xmath70 m@xmath62 and @xmath71 pc . according to equation ( 14 ) @xmath72 myr , or @xmath73 . figure 13 shows that the cluster is torn apart very early , during the initial phase of rapid expansion accompanying the death of its most massive stars . tidal disruption by passing clouds has long been considered the dominant cluster disruption mechanism @xcite . as noted , however , even the galactic tidal field will eventually do the job . in our pleiades simulation of paper i the cluster was largely destroyed in this way by 700 myr . an isolated cluster of this size would be just entering its phase of weak contraction . because of this external tidal field , however , the simulated pleiades never even began to contract , but globally expanded until it totally dissolved . such tidal disruption may account for the mass - independent pattern of cluster death observed in the antennae galaxies @xcite as well as the the magellanic clouds @xcite . a few open clusters do survive for ages much longer than the ones just mentioned @xcite . these lie in the outer reaches of the galaxy , where the encounter rate with giant molecular clouds is relatively low , and the general tidal field is also weaker . our isolated @xmath54 cluster eventually begins core contraction at 3 gyr , corresponding to @xmath74 . even the weakened galactic tidal field will begin to disrupt the system by this age , as seen in the simulation of the even richer system m67 @xcite . the lesson here is that , while open clusters can in principle enter a phase of tepid core contraction , none reach this point in reality . the classical theory of dynamical relaxation is relatively simple , an elegant illustration of how systems with negative heat capacity evolve @xcite . however the theory does not describe accurately real clusters , at least those of modest population on which we have focused . the two main factors changing the picture are binary heating and stellar evolution . both processes are , of course , well understood , but their combined effect has not been appreciated . all clusters are born with a large fraction of binaries , but these do not provide the largest effect . it is the system s most massive stars coupling together that generate most of the heating through three - body interactions . this heating easily reverses incipient core contraction , so that the central density climbs only slightly before the new phase of global expansion begins . this phase resembles , at least qualitatively , the post - collapse evolution described by @xcite . however , the reversal from contraction occurs at much lower density than in earlier accounts . mass loss accompanying stellar evolution modifies the picture , but does not change it qualitatively . since the most massive stars die out before they can couple with others , the degree of binary heating , and therefore the vigor of global expansion , is less . in addition , the earlier phase of core contraction lasts longer and leads to a higher central density before reversal . both modifications increase with the cluster population @xmath0 . we thus see why some globular clusters indeed reach the point of true core collapse , which can be reversed only by the tightest of binaries . the new picture of cluster evolution presented here is more complex than the classical one , but it is motivated by the basic physical effects that are incorporated in modern numerical simulations . with the benefit of hindsight , it is easy to see why earlier , simplified methods reinforced the impression that dynamical relaxation is ubiquitous . in single mass models , binary formation is so delayed that it becomes irrelevant . statistical models , based on solving the fokker - planck equation , neglect three - body effects entirely . finally , the contraction of lagrangian mass shells is not a reliable sign of core contraction , but may reflect a different phenomenon , mass segregation . our new picture is itself far from complete . future simulations carried out at higher @xmath0 will reveal in detail how the transition is made to a more vigorously contracting central core . we are grateful to douglas heggie and simon portegies - zwart for helping us navigate the literature of dynamical relaxation . onsi fakhouri also provided useful suggestions on the visualization of energy transfer . this research was supported by nsf grant ast 0908573 . angeletti , l. & giannone , p. 1977 , , 58 , 363 applegate , j.h . 1986 , , 301 , 132 bailyn , c. d. 1995 , , 33 , 133 bastian , n. , gieles , m. , goodwin , s. p. , trancho , g. , smith , l.j . , konstantopoulos , i. , & efremov , yu . 2008 , , 389 , 223 baumgardt , h. , heggie , d. c. , hut , p. , & makino , j. 2003 , , 341 , 247 bettwieser , e. & sugimoto , d. 1984 , , 208 , 493 binney , j. & tremaine , s. 2008 , galactic dynamics , princeton : princeton u. press bica , e. & bonatto , c. 2005 , , 437 , 483 bouvier , j. , rigaut , f. , & nadeau , d. 1997 , , 323 , 139 brown , a. g. a. 2001 , , 11 , 89 chandar , r. , fall , s. m. , & whitmore , b.c . 2010 , , 711 , 1263 chernoff , d. & djorgovski , s. g. 1989 , , 339 , l904 converse , j. m. & stahler , s. w. 2008 , , 678 , 431 converse , j. m. & stahler , s. w. 2010 , , 405 , 666 ( paper i ) davenport , j. r. a. & sandquist , e. l. 2010 , , 711 , 559 dawson , s. a. , & schrder , k. p. 2010 , , 404 , 917 dias , w. s. , alessi , b. s. , moitinho , a. , & lepine , j. r. d. 2002 , , 416 , 125 djorgovski , s. g. & king , i. r. 1986 , , 305 , l61 duquennoy , a. & mayor , m. 1991 , , 88 , 241 fabian , a. c. , pringle , j. e. , & rees , m. j. 1975 , , 172 , 15 fall , s.m . , chandar , r. , & whitmore , b.c . 2009 , , 704 , 453 friel , e. 1995 , , 33 , 381 de la fuente marcos , r. 1995 , , 301 , 407 de la fuente marcos , r. 1996 , , 308 , 141 de la fuente marcos , r. 1996 , , 314 , 453 giersz , m. & heggie , d. c. 1997 , , 286 , 709 goodman , j. 1987 , , 313 , 576 heggie , d. c. 1975 , , 173 , 729 heggie , d. c. 1984 , , 206 , 179 heggie , d. c. , hut , p. , & mcmillan , s. l. w. 1996 , , 467 , 359 heggie , d. c. & mathieu , r. d. 1986 , in lecture notes in physics , vol . 267 , the use of supercomputers in stellar dynamics , ed . p. hut & s. l. w. mcmillan ( berlin : springer - verlag ) , 233 hnon , m. 1972 , in the gravitational n - body problem , ed : m. lecar , ( dodrecht : reidel ) , p. 406 hills , j. g. 1975 , , 80 , 1075 hurley , j. r. , tout , c. a. , aasreth , s. j. , & pols , o. r. 2004 , , 355 , 1207 hurley , j. r. , pols , o. r. , aarseth , s. j. & tout , c. a. 2005 , , 363 , 293 hurley , j. r. , pols , o. r. & tout , c. a. 2000 , , 315 , 543 inagaki , s. & wiyanto , p. 1984 , , 36 , 391 kroupa , p. & boily , c. m. 2002 , , 336 , 1188 lynden - bell , d. & wood , r. 1968 , , 138 , 495 lynden - bell , d. 1999 , physica a , 263 , 293 makino , j. 1996 , , 471 , 796 portegies - zwart , s. f. , hut , p. , makino , j. , & mcmillan , s. l. w. 1998 , , 337 , 363 portegies - zwart & s. f. , mcmillan 2002 , , 576 , 899 portegies - zwart , s. f. , mcmillan , s. l. w. , hut , p. , & makino , j. 2001 , , 321 , 199 spitzer , l. 1958 , , 127 , 17 takahashi , k. 1995 , , 47 , 561 terlevich , e. 1987 , , 224 , 193 trager , s. c. , king , i. r. , & djorgovski , s. g. 1993 , , 109 , 218 van den bergh , s. & sher , d. 1960 , publ . david dunlop obs . , 2 , 7 wielen , r. 1974 , in stars and the milky way system , ed . l. n. mavridis ( berlin : springer - verlag ) , p. 326
utilizing a series of @xmath0-body simulations , we argue that gravitationally bound stellar clusters of modest population evolve very differently from the picture presented by classical dynamical relaxation theory . the system s most massive stars rapidly sink towards the center and form binary systems . these binaries efficiently heat the cluster , reversing any incipient core contraction and driving a subsequent phase of global expansion . most previous theoretical studies demonstrating deep and persistent dynamical relaxation have either conflated the process with mass segregation , ignored three - body interactions , or else adopted the artificial assumption that all cluster members are single stars of identical mass . in such a uniform - mass cluster , binary formation is greatly delayed , as we confirm here both numerically and analytically . the relative duration of core contraction and global expansion is effected by stellar evolution , which causes the most massive stars to die out before they form binaries . in clusters of higher @xmath0 , the epoch of dynamical relaxation lasts for progressively longer periods . by extrapolating our results to much larger populations , we can understand , at least qualitatively , why some globular clusters reach the point of true core collapse .
super - kamiokande has given strong evidence for a maximal oscillation between and @xmath2 @xcite , and several projects with accelerators have been designed to check this result . the first results of the k2k experiment @xcite confirm the oscillation , and future projects ( minos in the usa , opera and icarus at gran sasso ) should refine the oscillation parameters by 2010 . more recently , after the results from sno @xcite and kamland @xcite , a solid proof for solar neutrino flavour oscillations governed by the so - called lma solution has been established . we can no longer escape the fact that neutrinos have indeed a mass , although the absolute scale is not yet known . furthermore , the large mixing angles of the two above - mentioned oscillations and their relative frequencies open the possibility to test cp violation in the neutrino sector if the third mixing angle , @xmath0 , is not vanishingly small ( we presently have only an upper limit set at 10 degrees on @xmath0 , provided by the chooz experiment @xcite ) . such a violation could have far reaching consequences , since it is a crucial ingredient of leptogenesis , one of the presently preferred explanations for the matter dominance in our universe . the ideal tool for these studies is thought to be the so - called neutrino factory , which would produce through muon decay intense neutrino beams aimed at magnetic detectors placed several thousand kilometers away from the neutrino source . however , such projects would not be launched unless one is sure that the mixing angle @xmath0 , governing the oscillation between and at the higher frequency , is such that this oscillation is indeed observable . this is why physicists have considered the possibility of producing new conventional neutrino beams of unprecedented intensity , made possible by recent progress on the conception of proton drivers with a factor 10 increase in power ( 4 mw compared to the present 0.4 mw of the fnal beam ) the present limit on @xmath0 is 10 degrees , these new neutrino `` superbeams '' would explore @xmath0 down to 1 degree ( i.e a factor 100 improvement on the - oscillation amplitude ) . european working groups have studied a neutrino factory at cern for some years , based on a new proton driver of 4 mw , the spl . along the lines described above , a subgroup on neutrino oscillations has studied the potentialities of a neutrino superbeam produced by the spl . the energy of produced neutrinos is around 270 mev , so that the ideal distance to study to oscillations happens to be 130 km , that is exactly the distance between cern and the existing frjus laboratory . the present laboratory can not house a detector of the size needed to study neutrino oscillations , which is around 1 million cubic meters . but the recent decision to dig a second gallery , parallel to the present tunnel , offers a unique opportunity to complete the needed extension in 2012 for a reasonable price . due to the schedule of the new gallery , a european project would be competitive only if the detector at frjus reaches a sensitivity on @xmath0 around 1 degree , since other projects in japan ( jhf phase 1 ) and usa ( numi off - axis ) will have reached 2.5 degrees by 2013 . the working group has then decided to study directly a water erenkov detector with a mass around 1 megaton , necessary to reach the needed sensitivity . it has benefited from a similar study by our american colleagues , the so - called uno detector @xcite with a total mass of 660 kilotons . simulations have shown that the sensitivity on @xmath0 at a level of 1 degree could indeed be fulfilled . however , the study of cp violation requires the spl to be run sequentially with neutrinos and antineutrinos , and due to the fact that less antineutrinos are produced ( less @xmath3 than @xmath4 are produced ) and that the cross section is 5 times lower at the considered energies , 10 years of running should be shared roughly in 2 years with @xmath5 and 8 years with . this would be a strong limitation on cp sensitivity . this is where the beta beam concept , initially proposed by piero zucchelli @xcite , comes into play . the idea is to produce well collimated and intense ( ) beams by producing , collecting , accelerating to energies with @xmath6 factor around 100 and storing in a final decay ring radioactive ions chosen for their ability to be copiously produced and with a lifetime around 1 second . the best candidates happen to be @xmath7 for and @xmath8 for . a baseline study for such a betabeam complex has been produced at cern @xcite , where there is a strong expertise on ion beams , both for nuclear physics through isolde and for high energy experiments . the initial goal was to produce a beam which could be run simultaneously with the spl superbeam , so that 10 years of data could be accumulated for the each of the 2 time reversed oscillations , @xmath9and @xmath10 . this project was already presented at nufact02 @xcite . a workshop took place in march 2003 at les arcs @xcite , where nuclear physicists ( mainly those concerned with the eurisol project ) , neutrino physicists and machine scientists have met to discuss betabeam issues . the aim of this workshop was to get updated on recent progress on the betabeam project , explore the synergies between beta beams and eurisol @xcite , and identify common studies which could benefit to both communities . apart from new ideas which simplify the overall betabeam design , the major innovation was the proposal to run simultaneously with both types of ions ( @xmath11 and @xmath12 emitters ) stored in the ring . this opens up the exciting possibility of performing efficiently cp violation studies with beta beams alone , and get very useful redundancies by comparing superbeam and betabeam data . the section 2 describes the machine aspects of beta beams , the section 3 gives the expected performances on the measurement of the mixing angle @xmath0 and the cp violating @xmath1 phase . the beta - beam complex is shown schematically on figure [ fig : sketch ] . technical details and recent progress on this project can be found at http://beta-beam.web.cern.ch/beta-beam/ the protons are delivered by the super proton linac ( spl ) @xcite , which is being studied at cern in the framework of the neutrino factory @xcite . such an intense proton driver would deliver 2ma of 2.2 gev ( kinetic energy ) protons hopefully by 2012 . an isol target would need only 100 @xmath13a , that is 5 % only of the total proton intensity . the targets are similar to the ones envisioned by eurisol @xcite : for @xmath14he , it consists either of a water cooled tungsten core or of a liquid lead core which works as a proton to neutron converter surrounded by beryllium oxide @xcite , aiming for 10@xmath15 fissions per second . @xmath16ne can be produced by spallation reactions , in this case protons will directly hit a magnesium oxide target . the collection and ionization of the ions is performed using the ecr technique . the pulsed `` ecr - duoplasmatron '' under development at grenoble , using very dense plasmas ( 10 @xmath17/@xmath18 ) and high magnetic fields ( 2 to 3 teslas ) submitted to high frequencies ( 60 to 90 ghz ) is aimed to produce 10@xmath19 to 10@xmath20 ions in very short bunches ( 20 to 100 @xmath13s ) at 100 kev with repetition rates reaching 16 hz @xcite . the advantage over the standard ecr technique is that the downstream complex can be simplified , due to the achieved bunching and hopefully to ions which are totally ionized . then the first acceleration process can be achieved using a linac rather than a cyclotron or a ffag as initially considered . ions would be accelerated to 20 - 100 mev / u in 16 batches per second . then comes the first storage ring , a rapid cycling synchrotron using multiturn injection ( 40 turns ) , delivering a single 150 ns bunch at 300 mev / u . 16 bunches ( consisting of 2.5 10@xmath19 ions each in the case of he ) are then accumulated into the ps , and reduced to 8 bunches during their acceleration to intermediate energies . due to the fact that the ps is a slow machine , this is the place where radiation levels due to ion decays is the most severe ( the replacement of the ps by a more rapid machine would ease this problem and many others in the cern complex of accelerators ) . furthermore , the space charge bottleneck at sps injection will require a transverse emittance blow - up . the sps will finally accelerate the 8 bunches to the desired energy ( @xmath21 100 ) using a new 40 mhz rf system and the existing 200 mhz rf system , before ejecting them in batches of four 10 ns bunches into the decay ring . this ring has the shape of an hippodrome , with a total length of 6880 m ( matching the sps ) and straight sections of 2500 m each ( 36% ) . due to the relativistic time dilatation , the ion lifetimes reach several minutes , so that stacking the ions in the decay ring is mandatory to get enough decays and hence high neutrino fluxes . the challenge is then to inject ions in the decay ring and merge them with existing high density bunches . as conventional techniques with fast elements are excluded , a new scheme ( asymmetric merging ) was specifically conceived for this task @xcite . it schematically consists in injecting an off - momentum ion bunch on a matched dispersion trajectory , then rotate this fresh bunch in longitudinal phase space by a 1/4 turn into a starting configuration for bunch merging . this technique had been proven to work on computer simulations @xcite , but it received very recently a first experimental confirmation @xcite . one of the ideas presented at the moriond meeting was that it should be possible to run together neon and helium ions in the decay ring ( of course , in different bunches ) . due to their different rigidities , these ions would have relativistic @xmath6 factors in the 5 to 3 ratio , which is quite acceptable for the physics program . this will impose constraints on the lattice design for the decay ring , but no impossibility has been identified . an ecr source coupled to an eurisol target would produce @xmath22 @xmath14he ions per second . taking into account all decay losses along the accelerator complex , and estimating an overall transfer efficiency of 50% , one estimates that @xmath23 ions would permanently reside in the final decay ring for @xmath6 = 60 . that would give an antineutrino flux aimed at the frjus underground laboratory of @xmath24 per standard year ( 10@xmath25 s ) . for @xmath16ne , the yield is expected to be only @xmath26 ions per second . due to this smaller yield , which could be certainly improved with some r&d , it was then proposed to use 3 eurisol targets in sequence connected to the same ecr source . again taking into account decay losses plus a 50% efficiency , this means that @xmath27 such ions would reside in the decay ring for @xmath6 = 100 , giving rise to a neutrino flux of @xmath28 per standard year . all these numbers are preliminary and need to be refined . they are however based on the present state of the art for the technology , and suppose using the present ps , while the sps cycle is set at 16 s ; a shorter cycle for the sps would improve the accumulation factor substantially , while a faster ps would increase the intensity of ions making it to the decay ring . in the following study , it was supposed that the neutrino flux from @xmath29 could be increased by a factor 3 over the present conservative estimate , having room for improvements both in the cycle duration of ps and sps and in the @xmath29production at the targets with a dedicated r&d , while only a 40 % improvement was put on antineutrino fluxes . the main losses are due to decays of he ions , and reach 1.2 w / m in the ps and 9 w / m in the decay ring . this seems manageable , although the use of superconducting bending magnets in the decay ring requires further studies . activation issues have been recently addressed @xcite , and show that the dose rate on magnets in the arcs is limited to 2.5 msv / h at contact after 30 days operation and 1 day cooling . furthermore , the induced radioactivity on ground water will have no impact on public safety . the following study is based on the hypothesis that a uno - like water cerenkov detector ( 440 kt fiducial mass ) will be installed in the underground frjus laboratory and receive neutrino beams produced at cern , 130 km away . the neutrino beam energy depends on the @xmath6 of the parent ions in the decay ring . as discussed in ref . @xcite , the optimization of this energy , is a compromise between the advantages of the higher @xmath6 , as a better focusing , higher cross sections and higher signal efficiency ; and the advantages of the lower @xmath6 values as the reduced background rates ( see the following ) and the better match with the probability functions . given the decay ring constraint ( see sect . [ sec : decay ring ] ) : @xmath30 the optimal @xmath6 values result to be @xmath31 and @xmath32 . a flux of @xmath33 @xmath34 decays / year and @xmath35 @xmath29 decays / year , as discussed in sect . [ sec : fluxes ] , will be assumed . [ fig : fluxes ] shows the betabeam neutrino fluxes computed at the 130 km baseline , together with the spl super beam ( spl - sb ) . the mean neutrino energies of the , beams are 0.24 gev and 0.36 gev respectively . they are well matched with the cern - frejus 130 km baseline . on the other hand energy resolution is very poor at these energies , given the influence of fermi motion and other nuclear effects and in the following all the sensitivities are computed for a counting experiment with no energy cuts . [ cols="^,^,^ " , ] a facility where the neutrino fluxes are known with great precision is the ideal place where to measure neutrino cross sections . in the beta beam the neutrino fluxes are completely defined by the parent ions beta decay properties and by the number of ions in the decay ring . a close detector of @xmath36kton placed at a distance of about 1 km from the decay ring could then measure the relevant neutrino cross sections . furthermore the @xmath6 factor of the accelerated ions can be varied . in particular a scan can be initiated below the background production threshold , allowing a precise measurement of the cross sections for resonant processes . it is estimated that a residual systematic error of 2% will be the final precision with which both the signal and the backgrounds can be evaluated . the @xmath0and @xmath1 sensitivities are computed taking into account a 10% error on the solar @xmath37 and @xmath38 , already reached after the recent sno - salt results @xcite and a 5% and 1% error on @xmath39 and @xmath40 respectively , as expected from the j - parc neutrino experiment @xcite . only the diagonal contributions of these errors are considered . in the following the default values for the oscillation parameters will be , , , , @xmath41=+1 . correlations between @xmath0and @xmath1 are fully accounted for , and indeed they are negligible as can be seen in the fits to @xmath0and @xmath1 shown in fig . [ fig : many_plots ] . the net effect of the @xmath41ambiguity is to make undetectable sign(@xmath42 ) . this derives by the negligible matter effects at the 130 km , so that a change of @xmath41is equivalent to a change of the sign of @xmath1 . the performances of the betabeam to the two opposite values of sign(@xmath42 ) are different because the neutrino and antineutrino runs have different statistics and backgrounds . this effect will be illustrated in fig . [ fig : beta : sens ] and fig . [ fig : cp : delta ] . finally the @xmath43 ambiguity is formally taken into account , but no effect is found because the betabeam performances are computed for the central value of superkamiokande : @xmath44 . a study of the performances of the betabeam for different values of @xmath45 is beyond the purpose of this article . we stress the fact that an experiment working at very short baselines has the smallest possible parameter degeneracies and ambiguities and it is the cleanest possible environment where to look for genuine leptonic cp violation effects . the @xmath0angle can be independently explored both with and disappearance measurements . we note that the comparison of the and disappearance experiments could set limits to cpt violation effects . sensitivities to @xmath0 , computed for a 5 yr run and for systematic errors equal to 2% , 1% and 0.5% are shown fig . [ fig : th13]left ) . for comparison sake , shown in the same plot are the sensitivities reachable with the appearance channels , computed for @xmath47 . indeed @xmath0and @xmath1 are so tightly coupled in the appearance channels that the sensitivity expressed for @xmath47 is purely indicative . a better understanding of the sensitivity of the betabeam is expressed in the @xmath48 plane , having fixed all the other parameters ( @xmath49 ev@xmath50 ) , as shown in fig . [ fig : th13]right ) . in the same plot the sensitivity of the spl - sb computed for a 5 yrs run is displayed . it can be noted the very large variation of the spl - sb sensitivity for the different values of @xmath1 , characteristic of the single flavour run . the betabeam , having both cp neutrino states in the same run , exhibits a much more favourable dependence to the cp phase @xmath1 . a search for leptonic cp violation can be performed running the beta beam with @xmath16ne and @xmath14he , and fitting the number of muon - like events to the @xmath51and to the p(@xmath52 ) probabilities . the fit can provide the simultaneous determination of @xmath0 and @xmath1 , see fig . [ fig : many_plots ] . event rates are summarized in table [ tab : beta : rates ] . the region of 99% cl sensitivity to maximal cp violation ( @xmath53 ) in the @xmath54and @xmath0parameter space , following the convention of @xcite , is plotted in fig . [ fig : beta : sens ] . the @xmath55 sensitivity to @xmath1 , having fixed @xmath56 ev@xmath50 , is shown in fig . [ fig : cp : delta ] . the beta beam needs the spl as injector , but consumes at most @xmath57 of the spl protons . the fact that the average neutrino energies of both the superbeam and the beta beam are below 0.5 gev ( cfr . fig . [ fig : fluxes ] ) , with the beta beam being tunable , offers the fascinating possibility of exposing the same detector to @xmath58 beams ( and @xmath59 and ) having access to cp , t and cpt searches in the same run . it is evident that the combination of the two beams would not result only in an increase in the statistics of the experiment , but it would also offer clear advantages in the reduction of the systematic errors , and it would offer the necessary redundancy to firmly establish any effect of violation of cp within the reach of the experiment . the cp violation sensitivities of the combined betabeam and spl - sb experiments are shown in fig . [ fig : beta : sens ] and fig . [ fig : cp : delta ] . betabeams are a novel concept which can give very precise insight on the problem of neutrino mixing . their design has many common features with the eurisol project aiming at producing high intensity radioactive beams for nuclear physics studies with astrophysical applications . this synergy has been outlined at the moriond workshop , and common technical studies are already going on . recently , the potentialities of low energy betabeams ( @xmath6 factors below 10 ) has also been emphasized @xcite . on the other hand , there have been for some time projects of megaton detectors to study proton decay and detect supernova explosions , but they never got financial support . the fact that they would also be perfect targets for low energy neutrino superbeams or betabeams considerably increases the physical interest of these detectors . fortunately enough , a possible site exists near cern at the right distance and could house as soon as 2015 such a detector . cern has a long lasting expertise on ion production and acceleration , and has announced officially to be site candidate for the eurisol project . this offers europe and cern a unique opportunity to contribute significantly to megaton physics , neutrino physics , and eurisol physics by joining efforts of several communities on an ambitious and multidisciplinary project . the super - kamiokande collaboration , s. fukuda _ et al . _ , phys.rev.lett . 85 ( 2000 ) 3999 - 4003 the k2k collaboration , m.h . et al . _ , ( 2003 ) 041801 s. n. ahmed _ et al . _ [ sno collaboration ] , arxiv : nucl - ex/0309004 . kamland collaboration , k. eguchi _ et al . _ , phys.rev.lett . 90 ( 2003 ) 021802 m. apollonio _ et al . b 466 ( 1999 ) 415 uno collaboration , hep - ex/0005046 p. zucchelli , phys . b * 532 * ( 2002 ) 166 . b. autin _ et al . _ , `` the acceleration and storage of radioactive ions for a neutrino factory , '' arxiv : physics/0306106 . m. mezzetto , j.phys.g29:1771-1776 , 2003 ; hep - ex/0302005 . m. mezzetto , j.phys.g29:1781-1784 , 2003 ; hep - ex/0302007 . moriond workshop on `` radioactive beams for nuclear physics and neutrino physics '' , les arcs ( france ) march 17 - 22 , 2003 http://moriond.in2p3.fr/radio/index.html b. autin _ et al . _ , `` conceptual design of the spl , a high - power superconducting h- linac at cern , '' cern-2000 - 012 m. apollonio _ et al . _ , `` oscillation physics with a neutrino factory . '' arxiv : hep - ph/0210192 . j. nolen , npa 701 ( 2002 ) 312c p. sortais , presentations at the moriond workshop on radioactive beams , les arcs ( france ) 2003 `` ecr technology '' , http://moriond.in2p3.fr/radio/moriond-sortais_1.ppt http://beta-beam.web.cern.ch/beta-beam/presentations/asymmerging.gif m. benedikt , s. hancock and j - l . vallet , `` a proof of principle of asymmetric bunch pair merging '' , cern note ab - note-2003 - 080 md `` parameters of radiological interest for a beta - beam decay ring '' , m. magistris and m. silari , note cern - tis-2003 - 017-rp - tn d. casper , nucl . * 112 * ( 2002 ) 161 [ arxiv : hep - ph/0208030 ] . y. itow _ et al . _ , `` the jhf - kamioka neutrino project , '' , hep - ex/0106019 . p. migliozzi and f. terranova , phys . b * 563 * ( 2003 ) 73 [ arxiv : hep - ph/0302274 ] . m. v. diwan _ et al . _ , phys . d * 68 * , 012002 ( 2003 ) [ arxiv : hep - ph/0303081 ] . g. tzanakos , proceedings of this conference . j. burguet - castell _ et al . _ , nucl . b * 608 * , 301 ( 2001 ) [ arxiv : hep - ph/0103258 ] . c. volpe , `` what about a beta beam facility for low energy neutrinos ? , '' arxiv : hep - ph/0303222 .
we give the present status of the beta - beam study , which aims at producing intense and beams from the decay of relativistic radioactive ions . the emphasis is put on recent technical progress and new ideas . the expected performances in terms of neutrino mixing parameters ( @xmath0 and cp violating phase @xmath1 ) using a megaton water cerenkov detector installed in the frjus underground laboratory are shown to be excellent , and the synergy with a a companion superbeam is underlined . address = dapnia , cea saclay , france address = cern , geneva , switzerland address = istituto nazionale fisica nucleare , sezione di padova , italy . , [email protected]
the study of the intersections of stable and unstable manifolds of maps and flows has a strong influence on dynamical systems . in particular , the existence of a transverse intersection is associated with the onset of chaos , and gave rise to the famous horseshoe construction of smale . the poincar - melnikov method @xcite is a widely used technique for detecting such intersections . given a system with a pair of saddles and a degenerate heteroclinic or saddle connection between them , the classical melnikov function computes the rate at which the distance between the manifolds changes with a perturbation . there have been many formulations of the melnikov method for two - dimensional maps or flows @xcite and for higher - dimensional symplectic mappings @xcite . recently , the geometric content of melnikov s method was exploited in order to detect heteroclinic intersections of lagrangian manifolds for the case of perturbed hamiltonian flows @xcite . here it was shown that the heteroclinic orbits are in correspondence with the zeros of a geometric object , the so - called melnikov one - form . for maps , the melnikov function is an infinite sum whose domain is a saddle connection between two hyperbolic invariant sets . as usual , a simple zero of this function corresponds to a transverse intersection of stable and unstable manifolds of a perturbation of the original map . melnikov s method can also be used to compute transport fluxes . in particular , a _ resonance zone _ for a two - dimensional mapping is a region bounded by alternating segments of stable and unstable manifolds that are joined at primary intersection points @xcite . because the intersection points are primary , a resonance zone is bounded by a jordan curve and has exit and entry sets @xcite . the images of these sets completely define the transport properties of the resonance zone . moreover , the integral of the melnikov function between two neighbouring primary intersection points is the first order approximation to the geometric flux escaping from the resonance zone @xcite . the method has also been applied to the case of periodically time - dependent , volume - preserving flows @xcite and more generally to volume - preserving maps with fixed points @xcite and invariant circles @xcite . volume - preserving maps provide perhaps the simplest , natural generalization of the class of area - preserving maps to higher dimensions . moreover , they naturally arise in applications as the time - one poincar map of incompressible flows even when the vector field of the flow is nonautonomous . thus the study of the dynamics of volume - preserving maps has application both to fluids and magnetic fields . our goal in this paper is to develop , based on the theory of deformations , a general , geometrical description of the melnikov displacement and to compare our theory to classical results . deformation theory was first introduced in the theory of singularities @xcite , but was soon used in the contexts of volume and symplectic geometry . its application to dynamical systems in @xcite provide results that are close to our goals . let @xmath0 be a smooth family of diffeomorphisms such that the unperturbed map @xmath1 has a saddle connection @xmath2 between a pair of compact @xmath3-normally hyperbolic invariant manifolds . let @xmath4 be the algebraic normal bundle of the saddle connection . we show that there exists a canonical @xmath5 section @xmath6 , called the _ melnikov displacement _ , that measures the splitting of the saddle connection in first - order . we will prove that the melnikov displacement is given by the absolutely convergent series @xmath7 where @xmath8 is the vector field defined by @xmath9 . these sums do not converge in the tangent space @xmath10 , but only in the algebraic normal bundle @xmath11 . the use of the algebraic normal bundle in the study of normally hyperbolic manifolds goes back to @xcite . the fact that ( [ mainformula ] ) always converges also addresses the question of how to deal with melnikov s method when the original pair of compact @xmath3-normally hyperbolic invariant manifolds ( that are connected with the saddle connection ) are not fixed with the perturbation . in addition , we will also show that the melnikov displacement has a number of geometric properties . the main result in this direction is that any change of coordinates acts on the displacement by its pullback . this result will be used to obtain the natural action of any symmetries , reversing symmetries or integrals of the dynamical system on the displacement . similarly , if the map preserves a symplectic or volume form , this gives additional structure to the displacement . for example , if @xmath0 is a family of exact symplectic maps and the normally hyperbolic invariant manifolds are fixed points , then we will show that there exists a function @xmath12 , the _ melnikov potential _ , such that @xmath13 , where @xmath14 is the symplectic two - form . this relation is reminiscent of the definition of globally hamiltonian vector fields . when the normally hyperbolic invariant manifolds are not fixed points ( or isolated periodic points ) , its stable and unstable manifolds are coisotropic , but not isotropic , and so the relation @xmath13 makes no sense . we complete this introduction with a note on the organization of this paper . the general theory is developed in [ sec : melnikovdisplacement ] . in [ sec : classical ] , we show how our theory reproduces the classical methods of poincar and melnikov . the study of exact symplectic maps and volume - preserving maps is contained in [ sec : symp ] and [ sec : volume ] , respectively . in this subsection we present the deformation calculus for families of diffeomorphisms and submanifolds . we shall begin with diffeomorphisms by defining a vector field associated with a deformation . next , we construct a vector field for the deformation of ( immersed ) submanifolds . finally , we will combine these results to define and compute the melnikov displacement . in this paper , we consider smooth families of diffeomorphisms @xmath15 , where @xmath16 is an @xmath17-dimensional smooth manifold . here , the term smooth family means that @xmath18 is @xmath19 in both variables . the map @xmath1 will be called unperturbed " . the generating vector field of a smooth family of diffeomorphisms @xmath0 is the unique vector field @xmath8 such that @xmath20 if we regard @xmath8 as a nonautonomous vector field with time @xmath21 , then the function @xmath22 represents its flow ( * ? ? ? consequently , if @xmath0 is volume preserving , then @xmath8 has zero divergence ( * ? ? ? 2.2.24 ) , and if @xmath0 is ( exact ) symplectic , then @xmath8 is ( globally ) hamiltonian . these geometric equivalences form the basis of the deformation calculus . the generating vector field @xmath8 was also called the _ perturbation vector field _ in @xcite ; however , we adopt the older terminology here . sometimes , we will also refer to @xmath8 as the _ generator _ of the family @xmath0 . we will always use the convention that , given a family of diffeomorphisms denoted by italic letters @xmath0 , its generator is denoted by the same letter in calligraphic capitals . we have collected in the proposition below some well - known relations among generators , in which appear push - forwards and pullbacks . recall that the pullback @xmath23 and push - forward @xmath24 of a diffeomorphism @xmath25 act on a vector field @xmath26 as follows : @xmath27 we note that @xmath28 . next , our goal is to develop a first - order deformation calculus for invariant manifolds of smooth diffeomorphisms . we are mainly interested in stable and unstable invariant manifolds of @xmath3-normally hyperbolic manifolds for some @xmath29 , which unfortunately are just immersed submanifolds ( not embedded submanifolds ) , and @xmath30 ( not @xmath19 ) . this gives rise to a few technicalities . recall that a map @xmath31 is an _ immersion _ when its differential has maximal rank everywhere . if @xmath32 is one - to - one onto its image , then @xmath33 is an immersed submanifold of the same dimension as @xmath34 . we will denote an immersion by @xmath35 or simply @xmath36 when the immersion @xmath32 does not matter . for brevity , we will sometimes omit the term immersed " . if @xmath37 is the stable ( resp . , unstable ) invariant manifold of a fixed point of a diffeomorphism , then @xmath38 ( resp . , @xmath39 ) , where @xmath40 and @xmath41 are the number of stable and unstable directions at the hyperbolic point . when the fixed point is hyperbolic , @xmath42 . stable and unstable manifolds are typically not embedded because they can have points of accumulation . in this case the immersion is not a proper map . we consider families of submanifolds of the form @xmath43 , where @xmath44 is @xmath30 in both variables , for some @xmath45 . all the elements of such a family are diffeomorphic ( as immersed submanifolds ) , because they are diffeomorphic to the same base " manifold @xmath34 . just as for @xmath1 , the unperturbed submanifold is denoted by @xmath46 . [ def : adapted ] if @xmath47 is a @xmath30 family of immersed submanifolds , a family of diffeomorphisms @xmath48 is an adapted deformation when @xmath49 and @xmath50 is @xmath30 in both variables . adapted deformations exist since it suffices to take @xmath51 . while there is quite a bit of freedom in the choice of @xmath52 , only its normal component is relevant , since this measures the actual motion of @xmath53 with @xmath21 . the normal component will be defined using the algebraic normal bundle . for an immersed submanifold @xmath54 , this is defined as the set of equivalence classes @xmath55 when @xmath16 is riemannian , this normal bundle is isomorphic to the more familiar geometric normal bundle , @xmath56 ( cf . @xcite ) . in general @xmath57 is a manifold of dimension @xmath17 and is defined independently of any inner product structure on @xmath58 . the displacement vector field of a @xmath30 family of immersed submanifolds @xmath47 is the @xmath5 section @xmath59 where @xmath60 is any adapted deformation . the displacement vector field is well - defined ; that is , its definition is independent of the choice of the adapted deformation as is shown in the following lemma . [ lem : adapted ] let @xmath61 be two adapted deformations . then @xmath62_{\epsilon=0 } \in t_\xi w_0 \;,\qquad \forall \xi \in w_0 \;.\ ] ] for each fixed @xmath63 , the map @xmath64 describes a @xmath30 curve in @xmath46 such that @xmath65 and @xmath66 . using the fact that @xmath67 , we then have @xmath62_{\epsilon=0 } = { \rm d } \phi_0(\xi ) c'(0 ) = c'(0 ) \in t_\xi w_0 \;,\ ] ] because @xmath68 . when the submanifold is invariant under a diffeomorphism , its deformations are related by means of a fundamental iterative relationship between the generating vector field of the family of diffeomorphisms and the displacement vector field of the family of submanifolds . let @xmath0 be a smooth family of diffeomorphisms , and @xmath47 be a @xmath30 family of immersed submanifolds that are invariant under @xmath0 . then @xmath69 on the unperturbed submanifold @xmath46 , where @xmath70 is the displacement vector field . the tangent space @xmath71 is invariant under the pullback @xmath72 , so the term @xmath73 is well - defined as a section of the normal bundle @xmath74 . if @xmath48 is any adapted deformation , then @xmath75 is as well . differentiating @xmath76 with respect to @xmath21 yields @xmath77 where we used . however , by lemma [ lem : adapted ] the displacement is independent of the adapted deformation , so @xmath78 applying @xmath79 to both sides finishes the proof . the identity is equivalent to @xmath80 . thus , we can work either with push - forwards or pullbacks . to obtain the melnikov displacement we will iterate these identities on the stable and unstable manifolds of a family of diffeomorphisms . the melnikov displacement will be defined for a saddle connection between a pair of normally hyperbolic invariant manifolds . in this section we recall the definitions of these objects . there are many slightly different definitions of normally hyperbolic manifolds , see @xcite . in this paper , we adopt the following . let @xmath81 be a submanifold invariant under a smooth diffeomorphism @xmath82 . we say that @xmath83 is @xmath3-normally hyperbolic when there exist a riemann structure on @xmath58 , a constant @xmath84 , and a continuous invariant splitting @xmath85 such that if @xmath86 and @xmath87 are the canonical restrictions of the linear map @xmath88 associated to the splitting , then 1 . @xmath89 , and 2 . @xmath90 for all @xmath91 , and for all @xmath92 . as is usual in the literature , the term _ normally hyperbolic _ will be taken to mean @xmath93-_normally hyperbolic_. note that setting @xmath94 in the previous definition implies that the linearizations @xmath95 and @xmath96 of @xmath97 restricted to the stable and unstable spaces of @xmath83 have the uniform bounds @xmath98 in this paper we will assume that each @xmath3-normally hyperbolic invariant manifold @xmath83 is compact , although it would be sufficient that our diffeomorphisms be uniformly @xmath30 in some neighbourhood of @xmath83 . we will also assume , without loss of generality , that @xmath83 is connected . one consequence is that the sets @xmath99 and @xmath100 are @xmath30 immersed submanifolds of @xmath16 that are tangent at @xmath83 to @xmath101 , see @xcite . in particular , @xmath102 . moreover , @xmath83 and its stable and unstable invariant manifolds are persistent : given any smooth family of diffeomorphisms @xmath0 such that @xmath103 , then for each small enough @xmath21 there exists a nearby @xmath3-normally hyperbolic invariant manifold @xmath104 with @xmath30 families of immersed submanifolds @xmath105 . to compute the melnikov displacement , we will need to show that certain series are geometrically convergent ; the following lemma is a key component in this proof . [ lem : lnbound ] let @xmath106 be a diffeomorphism with a compact normally hyperbolic invariant manifold @xmath83 , and fix a point @xmath107 . then given any splitting @xmath108 , there exists a constant @xmath109 and an integer @xmath110 such that @xmath111 where @xmath112 are the restrictions of @xmath113 to the subspaces @xmath114 $ ] , @xmath115 . a similar bound holds for @xmath116 . there exists a unique point @xmath92 such that @xmath117 , where @xmath118 . according to the @xmath119-lemma for normally hyperbolic manifolds @xcite , the complementary subspaces @xmath120 tend to the unstable subspaces @xmath121 , as @xmath122 . thus , the maps @xmath123 tend to the unstable restrictions @xmath124 as @xmath122 , and the lemma follows from . it suffices to take any @xmath125 . the melnikov displacement will be defined as a function on the normal bundle of a _ saddle connection _ , which is defined as follows . let @xmath106 be a diffeomorphism with a pair of compact normally hyperbolic invariant manifolds @xmath83 and @xmath126 . a saddle connection between @xmath83 and @xmath126 is an invariant submanifold @xmath127 such that @xmath128 for all @xmath129 . [ rem : normalbundles ] the coincidence of the tangent spaces is needed in order that the manifolds have the same algebraic normal bundles on the saddle connection : @xmath130 since @xmath131 , for all @xmath129 . by definition , @xmath132 , and the manifolds @xmath83 and @xmath126 are not part of the saddle connection . the simplest ( and most common ) saddle connections are of the form @xmath133 . in this case , we say that the unperturbed invariant manifolds are _ completely doubled_. many melnikov problems studied in the literature fall into this category . nevertheless , in some problems there may exist points @xmath134 , @xmath135 , such that @xmath136 , see @xcite . in that case the saddle connection is strictly contained in the intersection of the stable and unstable invariant manifolds : @xmath137 . in this subsection we use the fundamental iterative equation to obtain infinite series for the displacements of the stable and unstable manifolds . these series are absolutely convergent , but only , as we must stress , when they are evaluated on their corresponding normal bundles . indeed , the tangential components of these series can be unbounded . consequently , in order to compute these sums , each term must be projected onto the normal bundle . an example will be given in [ ssec : suris ] . the proof of the following proposition is inspired by a proof given in @xcite , the main difference is that our setting is more geometric . [ pro : duds ] let @xmath0 be a smooth family of diffeomorphisms such that the unperturbed map @xmath1 has a compact normally hyperbolic invariant manifold @xmath138 with stable and unstable invariant manifolds @xmath139 . then the displacement vector fields @xmath140 and @xmath141 of the families of perturbed stable and unstable invariant manifolds are given by the absolutely convergent series @xmath142 we prove the claim about the stable displacement @xmath143 ; the unstable result is obtained analogously . repeatedly applying the iterative formula , yields @xmath144 for any integer @xmath145 . therefore , it suffices to check that the term @xmath146 tends geometrically to zero on the normal bundle of the unperturbed stable manifold . for any point @xmath147 , let @xmath148 , and @xmath149 . then @xmath150 where @xmath151 . we must show that @xmath152 tends geometrically to zero as an element of the quotient space @xmath153 . given any splitting @xmath154 , let @xmath155 be the projections onto the linear subspaces @xmath156 $ ] , and let @xmath112 be the corresponding restrictions of the linear maps @xmath157 . then , @xmath158 because @xmath159 and @xmath160 . let @xmath161 be the integer referred to in lemma [ lem : lnbound ] , and define @xmath162 . lemma [ lem : lnbound ] gives the bound @xmath163 for some @xmath109 . moreover , the sequence @xmath164 is bounded due to the compactness of the normally hyperbolic manifold @xmath138 and the continuity of the displacement vector field @xmath165 . therefore , @xmath166 tends geometrically to zero as @xmath167 , and so , @xmath152 tends geometrically to zero in the quotient space @xmath153 . we now will use the displacement vector fields to study the splitting of a saddle connection upon perturbation . as usual , @xmath0 denotes a smooth family of diffeomorphisms such that the unperturbed map @xmath1 has a saddle connection @xmath168 between a pair of compact @xmath3-normally hyperbolic invariant manifolds @xmath138 and @xmath169 . these manifolds persist and remain @xmath3-normally hyperbolic for small @xmath21 . we want to study the distance between the perturbed manifolds @xmath170 and @xmath171 the growth rate of this distance with @xmath21 is obtained simply by taking the difference between the displacement vector fields of both families . [ def : melnikov ] under the previous assumptions , the melnikov displacement is the canonical @xmath5 section of the normal bundle @xmath11 defined by @xmath172 where @xmath141 and @xmath140 are the displacement vector fields of @xmath173 and @xmath174 , respectively . the melnikov displacement @xmath70 makes sense only on the saddle connection , where the tangent spaces of the unperturbed invariant manifolds @xmath175 and @xmath176 coincide . away from @xmath2 , the difference @xmath177 is undefined because each term is a section of a different ( algebraic ) normal bundle , see remark [ rem : normalbundles ] . let @xmath15 be a smooth family of diffeomorphisms verifying the assumptions of definition [ def : melnikov ] . then its melnikov displacement @xmath178 is given by the absolutely convergent sums @xmath179 from proposition [ pro : duds ] , we get that @xmath180 , where the last equality follows from the identity @xmath181 . the melnikov displacement has been defined in a canonical way as a section of the normal bundle , as such it has strong geometric properties . we will show next that any change of variables acts as a pullback on it . this will imply a number of geometrical properties , for example that the displacement is invariant under the pullback and the push - forward of the unperturbed map . in addition , we will see that if there exist symmetries , reversors , first integrals , or saddle connections , then these have natural implications on the melnikov displacement . these claims are the subject of proposition [ pro : geometric ] . we recall that a diffeomorphism @xmath82 is _ symmetric _ when there exists a diffeomorphism @xmath182 such that @xmath183 , and then @xmath40 is called a _ symmetry _ of the map @xmath97 . analogously , @xmath97 is _ reversible _ when there exists a diffeomorphism @xmath184 such that @xmath185 , and then @xmath3 is called a _ reversor _ of the map @xmath97 . in many applications , @xmath3 is an involution , @xmath186 , though this need not be the case @xcite , finally , a function @xmath187 is a first integral of @xmath97 when @xmath188 . [ pro : geometric ] let @xmath0 be a smooth family of diffeomorphisms verifying the assumptions of definition [ def : melnikov ] and let @xmath70 be its melnikov displacement . 1 . [ item : h0 ] given any smooth family of changes of variables @xmath189 , the family @xmath190 also verifies the assumptions of definition [ def : melnikov ] , its melnikov displacement @xmath191 is defined on the saddle connection @xmath192 , and @xmath193 2 . the melnikov displacement is invariant by the pullback and the push - forward of the unperturbed map . that is , @xmath194 3 . if @xmath0 has a smooth family of : 1 . symmetries @xmath195 such that @xmath196 , then @xmath197 2 . reversors @xmath198 such that @xmath199 , then @xmath200 if , in addition , @xmath201 is an involution and its fixed set @xmath202 is a submanifold that intersects @xmath2 transversely at some point @xmath203 , then @xmath204 . first integrals @xmath205 such that @xmath206 , then @xmath207 4 . saddle connections @xmath208 such that @xmath209 , then @xmath210 . if there exists a vector field @xmath211 such that its flow commutes with @xmath0 , then the lie derivative of the melnikov displacement with respect to @xmath212 vanishes , that is , @xmath213 it is very important to stress that the above results make sense as identities on the normal bundle of saddle connections . the relation makes sense because the pullback @xmath214 maps @xmath215 onto @xmath216 , whereas makes sense because @xmath215 is invariant by the pullback @xmath72 and the push - forward @xmath217 . similar arguments apply to and . the identities and make sense because @xmath215 is contained in the kernel of the one - form @xmath218 and the vector field @xmath212 is tangent to @xmath2 , respectively . the hypothesis @xmath206 means that @xmath104 and @xmath219 are contained in the same level set of the first integrals , which can be assumed to be the zero level without loss of generality . this holds , for instance , in the transitive homoclinic case : @xmath220 and @xmath104 is transitive . we can write a geometric proof based on the definition of the melnikov displacement as a canonical section of the normal bundle , or a computational proof using the formulae . we follow the geometric approach . \(i ) the first claims are obvious . for the last one , note that if @xmath221 are any adapted deformations ( for the family @xmath0 ) , then @xmath222 are as well ( for the family @xmath223 ) . differentiating @xmath224 with respect to @xmath21 yields @xmath225 where we used that @xmath226 . hence , @xmath227 , so the difference @xmath228 verifies the relation @xmath229 , which is equivalent to . \(ii ) if we take @xmath230 in , then @xmath231 and @xmath232 . therefore , @xmath233 and follows from . ( iii.a ) if we take @xmath234 in , then @xmath235 and @xmath236 . therefore , @xmath233 and follows from . ( iii.b ) if we take @xmath237 in , then @xmath238 and @xmath239 . moreover , a stable ( respectively , unstable ) manifold of a map becomes an unstable ( respectively , stable ) for the inverse map . therefore , @xmath240 and follows from . next , we assume that @xmath201 is an involution whose fixed set @xmath241 intersects @xmath2 transversely at @xmath203 . that is , @xmath242 , @xmath243 , and @xmath244 . we want to prove that @xmath245 in the normal bundle @xmath11 , or equivalently that @xmath246 . since @xmath243 and @xmath242 , the square of the linear endomorphism @xmath247 is the identity map . this implies that @xmath248 is diagonalizable and its spectrum is contained in the set @xmath249 , so @xmath250 , where @xmath251 . we claim that @xmath252 and @xmath253 . the first claim follows from the fact that involutions are locally conjugate to their linear parts at fixed points . since @xmath254 is invariant under @xmath255 and complementary to @xmath256 in @xmath257 , we get the second claim . finally , the evaluation of at the point @xmath258 , yields @xmath259 , and so @xmath260 . ( iii.c ) let @xmath261 . obviously , @xmath262 . differentiating @xmath263 with respect to @xmath21 yields @xmath264 . thus , @xmath228 verifies . ( iii.d ) in this case , we can take @xmath265 , and so @xmath228 vanishes on @xmath266 . \(iv ) let @xmath267 be the flow of the vector field : @xmath268 and @xmath269 . thus , given any @xmath270 , the ( constant ) family @xmath271 is a smooth family of symmetries of @xmath0 . next , using relation , we get that @xmath272 for any @xmath270 , and follows by definition of the lie derivative . these results have been extensively used in the literature . the invariance of melnikov objects under the unperturbed map gives rise to periodicities when suitable coordinates are used ; examples can be found in @xcite . item ( iii.d ) implies a simple splitting criterion : if the melnikov displacement does not vanish identically , the separatrix splits @xcite . upper bounds on the number of uniform first integrals of the family @xmath0 can be deduced from item ( iii.c ) , see @xcite . this result has also been used in to establish necessary and sufficient conditions for uniform integrability of analytic , exact - symplectic maps @xcite . symmetries have also been extensively used , for example to improve the lower bound obtained by morse theory for the number of critical points of some melnikov potentials @xcite ( we will discuss melnikov potentials in [ sec : exactsymp ] . ) . relation is similar to noether s theorem , since the existence of a continuous symmetry the flow of the vector field @xmath212implies a conservation law for the melnikov displacement . finally , fixed sets of reversors can be used to guarantee the existence of heteroclinic points and zeros of melnikov functions ; this is an old trick , see @xcite . in the classical melnikov method , one uses simple zeros of a melnikov function to predict the transverse intersection of the invariant manifolds . we next show that this result also holds for the melnikov displacement . if @xmath203 is a simple zero of the melnikov displacement , then the perturbed invariant manifolds @xmath173 and @xmath174 intersect transversely at some point @xmath273 for small enough @xmath21 . by definition , the saddle connection @xmath2 is an @xmath1-invariant submanifold of @xmath16 . let @xmath274 be the projection of the normal bundle onto @xmath2 . there is a tubular neighbourhood @xmath34 of @xmath2 that is diffeomorphic to @xmath11 via a diffeomorphism @xmath275 , as illustrated in . since @xmath34 is an open neighbourhood of @xmath2 in @xmath16 , each deformation of @xmath2 occurs inside @xmath34 , for @xmath21 small enough . we note that under the diffeomorphism @xmath276 , and for @xmath21 small enough , deformations of @xmath2 can be thought as deformations of the zero section , @xmath277 , in @xmath11 . indeed , a deformation of @xmath2 can be parametrized by a section @xmath278 . let @xmath279 be an open set that contains @xmath203 . let @xmath280 and @xmath281 be deformations of @xmath282 such that @xmath283 let @xmath284 and @xmath285 be the images in @xmath11 of two deformations of the saddle connection , corresponding to the images of the stable and unstable manifolds . we want to show that @xmath286 and @xmath287 intersect transversely for @xmath21 small enough when @xmath70 has a simple zero . [ ptb ] .,height=192 ] we parametrize each manifold with a section @xmath288 . that is , if @xmath21 is small , then @xmath289 given by @xmath290 are sections of the normal bundle restricted to @xmath282 . notice that the functions @xmath291 given by @xmath292 are adapted deformations of @xmath282 with images @xmath293 . in fact , both are perturbations of the zero section . from the definition of displacements we have that @xmath294 where @xmath295 and @xmath296 , are the displacements of @xmath297 . therefore , each manifold @xmath286 and @xmath298 is the image of @xmath299 and @xmath300 and they intersect transversely at @xmath301 if and only if @xmath203 is a simple zero of the section @xmath302 . now we use the standard `` blow up '' argument so that the implicit function theorem can be applied . let @xmath303 be the section given by @xmath304 notice that @xmath305 is @xmath5 if @xmath306 is @xmath307 , and moreover that , when @xmath308 , @xmath309 is a simple zero of @xmath310 if and only if it is for @xmath311 . finally , since @xmath312 and @xmath286 and @xmath287 are the images of @xmath299 and @xmath300 , the implicit function theorem implies that if @xmath203 is a simple zero of @xmath70 , then @xmath286 and @xmath287 intersect transversely near @xmath203 , for @xmath21 small enough . thus @xmath174 and @xmath173 intersect transversely near @xmath203 , for @xmath21 small enough . as a simple example , we consider the generalized standard map @xmath313 where @xmath314 is a periodic potential . it is easy to see that @xmath97 preserves area and orientation and that its fixed points have the form @xmath315 where @xmath316 . such fixed points are saddles if and only if @xmath317 , because @xmath318 . following mcmillan @xcite , we can find a generalized standard map with a saddle connection between two saddle points if we choose a diffeomorphism @xmath319 such that @xmath320 , and let @xmath321 to see this , first note that , with this choice , the force @xmath322 is periodic with period one . moreover , if @xmath323 is a hyperbolic fixed point of @xmath324 ( that is , @xmath325 and @xmath326 ) , then @xmath327 thus , @xmath315 is a saddle fixed point . moreover , the graphs of the functions @xmath328 are invariant and the dynamics on these sets is very simple : @xmath329 for all @xmath330 . these graphs contain saddle connections if we choose a pair of neighbouring fixed points @xmath331 and @xmath332 of @xmath324 , so that @xmath324 has no fixed points in @xmath333 . suppose further that @xmath331 ( respectively , @xmath332 ) is a stable ( respectively , unstable ) fixed point of @xmath324 , so that @xmath334 and @xmath335 for all @xmath336 . then @xmath337 and @xmath338 are saddle points of the map @xmath97 , and @xmath339 are saddle connections between them : @xmath340 and @xmath341 , see . it is known that a generalized standard map with a potential of the form is typically nonintegrable @xcite . an integrable example , @xmath1 , for each @xmath342 is obtained when the diffeomorphism @xmath324 is given by @xmath343 for @xmath344 . this function , when extended to @xmath345 using @xmath346 , gives the period - one potential @xmath347 a first integral of the map is @xmath348 @xcite . the diffeomorphism is conjugate to a mbius transformation , and it is easy to find an explicit formula for its iterations : @xmath349 the points @xmath350 and @xmath351 are fixed points of @xmath324 . the point @xmath331 is stable and @xmath332 is unstable because @xmath352 and @xmath353 . we now perturb the integrable map by modifying the potential : @xmath354 we compute the melnikov displacement using . for this calculation we must first compute the vector field @xmath355 , which for the generalized standard map with potential is @xmath356 using the relations and , the series for the displacement @xmath70 is easily computed . however , in order to ensure convergence of the series , we must take into account the fact that it is only the normal component of the displacement that is desired ; indeed , the iteration of the tangential component is not bounded . let @xmath357 be the canonical projection onto the normal bundle . due to the fact that @xmath2 is invariant under @xmath1 , we have that @xmath358 . to avoid numerical errors , we project each term in the sum . we display in the projection @xmath359 as a function of @xmath360 along @xmath361 , for two perturbative potentials @xmath362 . the melnikov displacement generalizes the classical methods used to detect the splitting of separatrices due to poincar and melnikov . poincar s method @xcite is based on the existence of first integrals , such as an energy function , for the unperturbed system . assume that @xmath1 has a saddle connection @xmath2 between a pair of hyperbolic fixed points @xmath331 and @xmath332 . for the simplest case , @xmath2 has codimension one , and a single first integral suffices : @xmath363 . saddle connections with a higher codimension can be treated in a similar way if there are sufficiently many integrals . the splitting between the stable and unstable manifolds is measured by the rate of change in the first integral @xmath364 with @xmath21 on the saddle connection . that is , we define the poincar function @xmath365 by @xmath366 where @xmath367 and @xmath368 are deformations adapted to the perturbed invariant manifolds . under the above assumptions , @xmath369 where @xmath70 is the melnikov displacement . the proof is a simple computation . the term @xmath370 makes sense because the first integral is constant on the saddle connection , and so the tangent space @xmath215 in contained in the kernel of @xmath371 . if @xmath372 is nondegenerate , then a point @xmath373 is a simple zero of @xmath374 if and only if it is a simple zero of @xmath70 . in this case , the perturbed invariant manifolds intersect transversely near @xmath203 for @xmath21 small enough . therefore , the melnikov displacement @xmath70 generalizes the estimates of splitting in the poincar style . the classical melnikov method @xcite is based on estimating the movement of a manifold in a direction _ _ n__ormal to the separatrix . to define the normal , the phase space @xmath16 is assumed to have a riemannian inner product @xmath375 . as before , assume for simplicity that @xmath2 is a saddle connection of codimension one . the appropriate normal to @xmath2 is called an adapted normal vector field . [ def : aniv ] let @xmath2 be a @xmath1-invariant submanifold of codimension one . a vector field @xmath376 is adapted normal when 1 . @xmath377 is nondegenerate : @xmath378 , for all @xmath379 , 2 . @xmath377 is normal : @xmath380 , for all @xmath381 and @xmath382 ; and 3 . @xmath377 is invariant : @xmath383 , for all vector field @xmath384 . using this notation , the classical melnikov function @xmath385 is @xmath386 where @xmath367 and @xmath368 are deformations adapted to the perturbed invariant manifolds . consequently , @xmath387 is related to the melnikov displacement by @xmath388 since @xmath377 is nondegenerate , @xmath387 and @xmath70 have the same simple zeros . we also note that makes sense because @xmath377 is normal , and so the component of @xmath70 in the tangent space @xmath215 does not play any role . adapted normal vector fields can be obtained as gradients of nondegenerate first integrals . let @xmath364 be a smooth first integral of @xmath1 such that its gradient @xmath389 does not vanish on @xmath2 . given a riemannian structure , the gradient is the unique vector field such that @xmath390 for any vector field @xmath391 on @xmath16 . therefore , @xmath392 , and @xmath393 is an adapted normal vector field . obviously , the poincar function and the melnikov function coincide when @xmath393 . in this section , we will see how the melnikov displacement generalizes previous theories developed for hyperbolic fixed points of exact symplectic maps . we will also show why general normally hyperbolic invariant manifolds can not be studied in the same way . a @xmath394-dimensional manifold @xmath16 is _ exact symplectic _ when it admits a nondegenerate two - form @xmath14 such that @xmath395 for some liouville one - form @xmath119 . the typical example of an exact symplectic manifold is provided by a cotangent bundle @xmath396 , together with the canonical forms @xmath397 and @xmath398 , in cotangent coordinates @xmath399 . a map @xmath82 is _ exact symplectic _ if @xmath400 for any closed path @xmath401 or , equivalently , if there exists a _ generating _ ( or _ primitive _ ) _ function _ @xmath402 such that @xmath403 . in particular , an exact symplectic map is _ symplectic _ : @xmath404 . a submanifold @xmath34 of @xmath16 is _ exact isotropic _ if @xmath400 for any closed path @xmath405 or , equivalently , if there exists a generating function @xmath406 such that @xmath407 . here , @xmath408 denotes the natural inclusion map . in particular , an exact isotropic submanifold is _ isotropic _ : @xmath409 . the maximal dimension of an isotropic submanifold is @xmath17 , and when the dimension is @xmath17 , the submanifolds is called _ lagrangian_. a vector field @xmath410 is _ globally hamiltonian _ if there exists a _ hamiltonian _ function @xmath411 such that @xmath412 we stress two key properties of the symplectic case . first , the generator of a family of exact symplectic maps is globally hamiltonian and there exists a simple relation between this hamiltonian and the generating function of the maps . second , the stable and unstable invariant manifolds of a connected normally hyperbolic invariant submanifold @xmath83 of an ( exact ) symplectic map are ( exact ) isotropic if and only if @xmath83 is a fixed point , in which case they are lagrangian ( that is , @xmath17-dimensional ) . the second property is an obstruction to develop a symplectic version of our canonical melnikov theory for general normally hyperbolic invariant submanifolds , we shall do it just for fixed points . these properties are well - known , but we prove both for completeness . [ pro : hamiltonian ] let @xmath0 be a family of exact symplectic maps with generating function @xmath413 and generating vector field @xmath8 . then , @xmath8 is globally hamiltonian with hamiltonian @xmath414 by definition , @xmath415 , so the lie derivative with respect to @xmath8 is @xmath416 . using cartan s formula : @xmath417 , and taking the derivative with respect to @xmath21 of the relation @xmath418 , we get @xmath419 rearranging this yields @xmath420 finally , @xmath421 , since @xmath119 is a one - form . [ pro : isotropic ] let @xmath82 be a diffeomorphism with a connected , normally hyperbolic , invariant manifold @xmath83 . if @xmath97 is ( exact ) symplectic , the stable and unstable invariant manifolds @xmath422 are ( exact ) isotropic if and only if @xmath83 is a hyperbolic fixed point . moreover , in this case the submanifolds @xmath423 are ( exact ) lagrangian . since @xmath83 is normally hyperbolic , implies that for each @xmath92 , @xmath424 , @xmath425 and @xmath426 sum to @xmath427 . since @xmath97 is symplectic @xmath428 . therefore @xmath429 . suppose that @xmath423 are isotopic ; then since isotopic manifolds have maximum dimension @xmath17 , we must have @xmath430 . consequently , @xmath431 and since @xmath83 is connected it is a hyperbolic fixed point . conversely , assume that @xmath432 , so that @xmath433 for some hyperbolic fixed point @xmath331 and @xmath434 . to prove that @xmath435 is lagrangian , take any two vectors @xmath436 . we know that @xmath437 and @xmath438 tend to zero as @xmath439 , since the stable directions are uniformly contracted . since @xmath97 preserves @xmath14 , we have @xmath440 as @xmath439 . if , in addition , @xmath97 is exact symplectic , then for every closed loop @xmath441 , @xmath400 . now suppose that @xmath442 , so that @xmath443 as @xmath444 . then @xmath445 finally , the ( exact ) lagrangian character of @xmath116 follows from the fact that it is the stable manifold for @xmath446 . let @xmath15 be a family of exact symplectic maps such that @xmath1 has an exact lagrangian saddle connection @xmath447 between two hyperbolic fixed points @xmath331 and @xmath332 . note that in this case that the stable and unstable manifolds are as smooth as the map @xmath0 . we assume that @xmath448 , where @xmath449 is the hamiltonian . without loss of generality , we can assume that @xmath450 . the natural measure of splitting for this case is a real valued function @xmath451 , the _ melnikov potential _ , whose derivative measures the splitting @xcite . in other words , nondegenerate critical points of the melnikov potential predict transverse splitting . in this subsection , we shall find the relation between the one form @xmath452 introduced by @xcite and the melnikov displacement . for a saddle connection @xmath447 and melnikov displacement @xmath453 , the function @xmath454 implicitly defined by @xmath455 is the melnikov potential .. ] we will show next that @xmath451 is indeed defined by and that its critical points correspond to zeros of the displacement . the pullback of the one - form @xmath456 to the saddle connection @xmath447 is well - defined and exact . in particular , there exists a function @xmath457 , determined uniquely up to additive constants that obeys . moreover , the set of simple zeros of the melnikov displacement @xmath70 coincides with the set of nondegenerate critical points of the function @xmath451 . [ rem : isotropic ] let @xmath34 be a submanifold of @xmath16 and @xmath458 a section of its normal bundle . then the pullback of the one - form @xmath459 to the submanifold @xmath34 is _ well - defined _ if and only if @xmath34 is isotropic . this has to do with the fact that @xmath460 is well - defined if and only if @xmath461 for any vector fields @xmath462 and @xmath463 . the pullback is well - defined because @xmath447 is lagrangian , see remark [ rem : isotropic ] . with regard to the exactness , it suffices to prove that @xmath464 for any closed path @xmath465 . let @xmath441 a closed path contained in the saddle connection , then @xmath466 where , since @xmath1 is symplectic , the melnikov displacement is invariant under the pullback of @xmath1 , and the saddle connection is invariant under @xmath1 . finally , since @xmath467 , we obtain that @xmath468 as @xmath444 . the equivalence between simple zeros of @xmath70 and nondegenerate critical points of @xmath451 follows from the lagrangian character of the saddle connection @xmath447 . an explicit series for @xmath451 can be obtained using the hamiltonian . the melnikov potential is given by the absolutely convergent series @xmath469 since @xmath1 is symplectic , and the generator @xmath355 is globally hamiltonian with hamiltonian @xmath470 : @xmath471 , we deduce that @xmath472 the series converges absolutely because @xmath450 , @xmath473 , and @xmath474 . in fact , it converges at a geometric rate . while the melnikov potential has been used many times for exact symplectic , twist , and hamiltonian maps , the formulation give here , using the power of deformation theory , is more elegant . the melnikov potential introduced here is identical to the one defined in @xcite . this can be checked by direct comparison of the formula with the formula ( 2.7 ) of the cited paper , using to express @xmath470 in terms of the derivative of the generating function @xmath413 at @xmath475 . we also note that it is impossible to define a melnikov potential " on the saddle connection of normally hyperbolic invariant manifolds with nonzero dimension , because then the saddle connection is not isotropic ( proposition [ pro : isotropic ] ) and so the identity makes no sense ( remark [ rem : isotropic ] ) . nevertheless , even in this case the melnikov displacement is defined . in this subsection , we restrict to the two - dimensional case in order to show in a simple way that the melnikov potential and the classical melnikov function are transparently related . moreover , we will see that in some cases there exist geometric obstructions to the non - vanishing of melnikov functions . these points are most easily seen by choosing a special time - like parametrization of the saddle connection . we consider the standard symplectic structure on the plane , @xmath476 , and let @xmath477 be the standard @xmath478 symplectic matrix : @xmath479 . suppose that @xmath480 is a family of diffeomorphisms preserving area and orientation , and @xmath449 is the generating hamiltonian for @xmath8 , the generator for @xmath0 . we assume that the unperturbed map has a saddle connection @xmath481 between two hyperbolic fixed points @xmath331 and @xmath332 such that @xmath482 . note that the unperturbed map not need be integrable . the key point is that , on a one - dimensional saddle connection , there is a parametrization @xmath483 such that @xmath484 in many cases , such parametrizations can be expressed in terms of elementary functions , and the melnikov function can be explicitly computed @xcite . consequently , @xmath485 provides a diffeomorphism between the saddle connection @xmath2 and the real line , so that objects defined over @xmath2 can be considered as depending on the real variable @xmath270 . thus , for example , the melnikov potential can be replaced by @xmath486 to become a function @xmath487 given by @xmath488 here we abuse the notation by not giving the function a new name . our goal is to show that the classical melnikov function , or rather the composition @xmath489 , can be computed by differentiating @xmath451 : @xmath490 in addition we will show that these functions have the properties * periodicity : @xmath491 and @xmath492 . * in each fundamental domain @xmath493 , @xmath16 must vanish . indeed , @xmath494 for any @xmath495 . ( this property will be generalized to volume - preserving maps in the next section . ) * near each simple zero of @xmath496 or nondegenerate critical point of @xmath497 there is transverse intersection of the stable and unstable manifolds . given and , the first two properties are obvious . periodicity is simply a consequence of the invariance of under @xmath498 ; this invariance under the unperturbed map is a property of all of the melnikov objects introduced so far . the second property is a simple consequence of periodicity and integration of . to show that @xmath496 is actually the classical melnikov function we must construct an adapted normal vector field @xmath377 on @xmath447 , recall definition [ def : aniv ] . we claim that , when thought of as a function of @xmath270 , such a vector field @xmath499 is given by @xmath500 this claim is proved in lemma [ lem : alpha ] at the end of this subsection . the relation between @xmath16 and the classical melnikov function follows from a straightforward computation of the derivative using @xmath501 : @xmath502 thus @xmath503 which is the obvious form of under the parametrization @xmath485 . this verifies and the final property . to end this subsection , it remains to prove the claim about the vector field @xmath377 . [ lem : alpha ] the vector field @xmath504 defined by @xmath505 is an adapted normal vector field on the saddle connection @xmath2 . since @xmath1 is symplectic , @xmath506 . consequently , @xmath507 thus , given any vector field @xmath508 , we have @xmath509 moreover , since @xmath477 is antisymmetric , @xmath510 . therefore , according to definition [ def : aniv ] , @xmath504 is an adapted normal vector field . for the case of a volume - preserving mapping with a codimension - one saddle connection , an adapted normal field formulation of the melnikov function also applies @xcite . here we show how to relate this to the melnikov displacement . let @xmath511 be a family of volume - preserving diffeomorphisms on an oriented @xmath17-dimensional manifold @xmath16 with volume form @xmath512 such that @xmath1 has a codimension - one saddle connection @xmath2 between two normally hyperbolic invariant sets @xmath83 and @xmath126 . we start with a simple lemma about the generator for @xmath0 . [ pro : exact ] let @xmath8 be the generator of a volume - preserving smooth family @xmath0 . then 1 . the divergence of @xmath8 with respect to @xmath512 is zero . 2 . the one - form @xmath513 is closed . if @xmath16 is simply connected , then @xmath513 is exact . from ( * ? 2.2.21 ) , we have that @xmath514 . since the divergence is defined by @xmath515 , this implies that the divergence vanishes . moreover , since @xmath516 and @xmath517 , then @xmath518 , implying ( ii ) and ( iii ) . in order to find an invariant nondegenerate @xmath519-form from any adapted normal field , we assume that @xmath16 has a riemannian metric @xmath520 . if @xmath377 is an adapted vector field ( cf . definition [ def : aniv ] ) , then @xmath521 is a nondegenerate @xmath519-form on @xmath2 that is invariant under the restriction @xmath522 . by definition @xmath377 is nonzero , so that @xmath523 is nondegenerate . to prove that @xmath524 on the saddle connection @xmath2 , we introduce the vector field @xmath525 defined by @xmath526 this vector field is tangent to @xmath2 , because @xmath527 . now , we compute the difference @xmath528 hence , it suffices to see that the @xmath519-form @xmath529 vanishes identically on the tangent space @xmath530 . this follows from the fact that @xmath531 is tangent to @xmath2 and @xmath532 . the melnikov function associated with @xmath377 is defined using @xmath523 . [ def : java ] let @xmath70 be the melnikov displacement . given an adapted vector field @xmath377 on @xmath2 , we define the melnikov function @xmath533 as the unique @xmath5 function such that @xmath534 as @xmath519-forms on @xmath2 . the previous definition first appeared in @xcite . note that @xmath535 is an @xmath519-form , but it is possible that it might be degenerate ; if this were the case then zeros of @xmath387 need not correspond to those of @xmath70 . we will show next that this is not the case . [ pro : java ] the melnikov function @xmath387 is invariant under the map @xmath1 and @xmath536 . moreover , a point @xmath203 is a zero of @xmath387 if and only if it is a zero of @xmath70 . using , we obtain @xmath537 where @xmath538 . since @xmath539 , we conclude that @xmath540 , as an @xmath519-form on @xmath2 , and thus @xmath536 . since @xmath377 is an adapted field , and @xmath70 is invariant , @xmath541 . therefore , @xmath542 . finally , @xmath543 only when @xmath204 , since @xmath377 is nonzero and normal to @xmath215 , and @xmath70 is not in the tangent space . we would like to show , as we did for the area - preserving case in , that the volume - preserving melnikov function necessarily has zeros on @xmath2 . to do this , we will show that the integral of @xmath387 with respect to the measure @xmath523 is zero . this can be accomplished by dividing the saddle connection into pieces fundamental domains that are mapped into each other by @xmath1 . [ def : guasa ] let @xmath83 be a compact normally hyperbolic invariant manifold of a diffeomorphism @xmath1 with stable manifold @xmath544 . a proper boundary , @xmath441 , is a submanifold of @xmath544 that bounds an isolating neighbourhood of @xmath83 in @xmath544 . in other words , @xmath441 is proper if there is an closed submanifold @xmath545 such that 1 . @xmath546 , and 2 . @xmath547 . we refer to the closed set @xmath545 as the stable manifold _ starting at _ @xmath441 . similarly , for the unstable manifold , a submanifold @xmath548 is proper it is proper for @xmath549 . however , in this case we define the unstable manifold _ up to _ @xmath550 , denoted by @xmath551 , as the _ interior _ of the local manifold that corresponds to @xmath549 . notice that the definition is not symmetric , because @xmath545 is a closed subset of @xmath544 , while @xmath551 is open in @xmath552 . the asymmetry is just a technicality in order to simplify some proofs . let @xmath83 be a hyperbolic invariant set for @xmath1 . a submanifold with boundary @xmath553 is a fundamental domain of @xmath544 if there exists some proper boundary @xmath554 such that @xmath555 equivalently , a fundamental domain in @xmath552 is a manifold with boundary of the form @xmath556 where @xmath550 is a proper boundary in @xmath552 . in each case , the fundamental domain is a manifold with boundary @xmath557 . an immediate consequence of the definition is that all the forward and backward iterations of a fundamental domain are also fundamental . it is easy to see that proper boundaries always exist , and in fact , the unstable manifold can be decomposed as the disjoint union of fundamental domains : @xmath558 the importance of fundamental domains is that much of the information about the entire manifold can be found by looking only at these submanifolds . for example , as discussed in @xcite , the topology of the intersections of @xmath116 and @xmath435 can be studied by restricting to @xmath553 . [ pro : sony ] let @xmath16 be a simply connected manifold and @xmath15 a family of volume - preserving maps such that @xmath1 has a saddle connection @xmath2 with fundamental domain @xmath553 . if @xmath523 is the @xmath519-form defined in and @xmath387 is the melnikov function , then @xmath559 . the fundamental domain @xmath553 is a submanifold with boundary , such that @xmath560 , where @xmath441 is a closed curve that does not intersect @xmath561 . if we give an orientation @xmath562 $ ] to @xmath553 , the induced orientation on the boundary satisfies @xmath563 = -\left[f_0(\gamma)\right]$ ] . according to proposition [ pro : exact ] , the form @xmath564 is exact . thus there exists an @xmath565-form @xmath566 such that @xmath567 . from we conclude that @xmath568 where @xmath569 . finally , from definition [ def : java ] , the stokes s theorem , and the invariance @xmath570 , we get that the integral @xmath571 vanishes . the previous result implies that the stable and unstable manifolds of a perturbed saddle connection necessarily intersect . examples of such intersections were computed for the case @xmath572where the hypothesis of proposition [ pro : sony ] are satisfied in @xcite . we have studied a general theory of the melnikov method that can be applied to many different settings . the formula for the melnikov displacement @xmath70 generalizes many of the classical methods that use a normal vector field to measure the displacement with respect to a natural direction . for example , when the saddle connection is defined as the level set of a first integral @xmath364 , the classical poincar function @xmath573 measures the splitting as the rate of change of the first integral . if there is a riemannian structure and an associated adapted normal vector field @xmath377 , then the classical melnikov function @xmath574 measures the rate of change of the splitting in this normal direction . for the case of exact symplectic maps with saddle connections between hyperbolic fixed points , the melnikov potential @xmath451 is defined on a lagrangian submanifold and acts as a generator for the displacement : @xmath575 . this melnikov theory can be extended to other situations and can be applied in many problems . for instance , one can study billiard dynamics inside a perturbed ellipsoid , following a program initiated in @xcite . it turns out that the billiard map inside an ellipsoid is an exact symplectic diffeomorphism defined on the cotangent bundle of the ellipsoid , which has a two - dimensional normally hyperbolic invariant manifold with a three - dimensional saddle connection . therefore , ellipsoidal billiards represent a strong motivation for a more detailed study of general normally hyperbolic invariant manifolds in a symplectic framework . this is a work in progress . in discrete volume geometry there are many open melnikov problems , since the application of melnikov methods to volume - preserving maps began just a few years ago @xcite . we plan to continue this program in several ways . as a first step , we plan to obtain bounds on the number of primary heteroclinic orbits in terms of the degree of the polynomial perturbation @xcite . hl was supported in part by asociacin mexicana de cultura . jdm was supported in part by nsf grant dms-0202032 and by the mathematical sciences research institute in berkeley . rr - r was supported in part by mcyt - feder grant mtm2006 - 00478 . useful conversations with guillermo pastor , amadeu delshams , richard montgomery , clark robinson , renato calleja and rafael de la llave are gratefully acknowledged .
we study perturbations of diffeomorphisms that have a saddle connection between a pair of normally hyperbolic invariant manifolds . we develop a first - order deformation calculus for invariant manifolds and show that a generalized melnikov function or _ melnikov displacement _ can be written in a canonical way . this function is defined to be a section of the normal bundle of the saddle connection . we show how our definition reproduces the classical methods of poincar and melnikov and specializes to methods previously used for exact symplectic and volume - preserving maps . we use the method to detect the transverse intersection of stable and unstable manifolds and relate this intersection to the set of zeros of the melnikov displacement .
one of the most enduring questions surrounding active galactic nuclei ( agn ) is the origin of powerful , relativistic , radio - emitting jets in some agn . a closely related question concerns the extent to which we can draw an analogy between jets from stellar - mass black hole systems ( particularly the galactic microquasars ) and agn . in recent years it has become clear that many aspects of black hole accretion and jet formation are directly comparable between agn and the lower - mass ( about 10@xmath0 ) black holes in x - ray binary systems ( xrbs ) @xcite . this is to be expected given the very simple scalings with mass for black holes in general relativity , although there is likely to be a larger diversity of environments and fueling mechanisms in agn as compared with xrbs . xrbs can be observed as they cycle ( secularly ) through different accretion states , and we find that the properties and even the existence of the jets are closely tied to the accretion state ( characterized by the spectral and timing properties of the x - ray emission ) @xcite . importantly , it has been shown that xrb jets manifest themselves in two types depending upon the spectral state of the xrb . the first kind of jet is seen in the hard ( corona dominated ) spectral state as a continuous low - power outflow @xcite . the second type of jet , which is similar to the jets seen in broad line radio galaxies ( blrgs ) , is launched when a low - mass xrb undergoes a transient outburst @xcite . the jet is usually observed close to the time of outburst maximum , as the source moves from the hard state to the soft ( disk dominated ) state @xcite . low - mass black hole xrbs displaying these powerful jets are called microquasars @xcite . assuming that these transient xrb jets and those from blrgs are fundamentally the same phenomenon , the study of jets in agn and xrbs give us different and complementary views of the physics . one major result learned from studying xrb jets is that the jet power , as well as the radiative efficiency of the accretion flow , can change dramatically in the same source at the same overall radiative luminosity on timescales far shorter than those associated with significantly changing black hole mass or angular momentum @xcite . on the other hand , the much longer timescales displayed by agn allow us to follow the complex relationships between individual jet - ejection events and the accretion disk , something which is difficult in xrbs . it is now recognized that both galactic microquasars and luminous radio - loud agn ( rlagn ) display complex cycles . in blrgs , as first found by @xcite in 3c120 , major jet ejection events are preceded by strong dips in the x - ray luminosity . since the x - ray emission in blrgs is thought to be dominated by the corona of the inner accretion disk , as shown for 3c120 by @xcite , this x - ray / radio connection conclusively demonstrates a link between changes in accretion disk structure and powerful ejection events . this behavior , which has parallels in the phenomenology of galactic microquasars is one of the few observational clues that we have to the origin of radio jets . for example , grs 1915 + 105 has been found to launch a relativistic jet knot as the inner - edge of the accretion disk moves all the way to the isco @xcite . this general idea has been confirmed in much more extensive monitoring campaigns of 3c120 and 3c111 by @xcite who also show that an observed correlation between the x - ray and optical flux can be explained in terms of the ( inward or outward ) radial propagation of powerful disturbances within the accretion disk @xcite . we note that @xcite recently reported a possible disk - outflow connection from a _ suzaku _ spectral analysis of 3c 111 and compared this to events in the radio jet . on the basis of these blrgs studies , the current picture is that the jet and disk are linked via a jet cycle " , summarized in cartoon form in fig . the cycle starts with a full accretion disk extending all the way to the isco ( step 1 ) . some instability , possibly associated with a breakdown in the inertial confinement of the magnetic flux bundle threading the black hole @xcite , destroys the innermost regions of the disk and ejects matter out of the disk plane ( step 2 ) , causing an x - ray dip . this triggers a powerful relativistic mass / energy ejection along the jet which , with a time delay corresponding to propagation and optical depth effects , results in a radio flare and the creation of a new vlba knot / shock ( step 3 ) . one possibility is that this ejection event is triggered by the re - accretion of open magnetic field lines onto the rotating event horizon . lastly , the inner accretion disk refills ( step 4 ) and the cycle repeats . relativistic x - ray reflection and , in particular , the broad iron line gives us a powerful tool to study the inner regions of agn accretion disks @xcite . if our hypothesis of the jet - cycle in blrgs is correct we would expect to see relativistic disk reflection during periods when the inner accretion disk is filled . when the inner disk is destroyed , we should see indications for a truncation of the disk . confirming this prediction would provide a major boost to the current jet formation scenario . measuring the inner radius of the accretion disk during several phases of the jet cycle will help to anchor this scenario . in this work we study 3c120 ( @xmath1 ) , a nearby and x - ray bright blrg with a bulge dominated host , showing signs of a recent merger @xcite . it is known to harbor a black hole with a mass of @xmath2 @xcite . 3c120 has been studied in detail at many wavelengths ( e.g. , * ? ? ? * ) and is well known to show a one sided superluminal jet . it also exhibits a prominent jet cycle @xcite , that has been previously studied multiple times in the x - ray range . earlier _ xmm _ observations analyzed by @xcite and @xcite find that the x - ray spectrum above 3kev can be well described by a power law and cold reflection , when accounting for neutral intrinsic absorption in the object . at soft energies they found a soft excess , which can be described by bremsstrahlung , or a power law . @xcite used _ suzaku _ data from 2006 and found indications for relativistic disk emission in the x - ray spectrum . these are confirmed by @xcite , who find evidence for a truncation of the accretion disk in a re - analysis of the same dataset with updated models and calibration . from radio observations of superluminal motion and modeling of the recollimation shock , the jet angle has been estimated to about 16degrees @xcite . with the caveat that the truncation of the thin disk may influence the ability of gr torques to align the inner disk , we assume a bardeen - petterson alignment of the jet and the inner accretion disk @xcite implying that the jet inclination can also be taken as the inner disk inclination . this , together with the fact that the uv / optical flux seems to mostly originate from the accretion disk @xcite , makes 3c120 very suitable for studying potential changes in the accretion disk as expected from a disk - jet connection . it is also important to note that 3c120 was detected with _ fermi _ , allowing a tentative decomposition of the radio - to-@xmath3 spectral energy distribution ( sed ) into jet and disk components @xcite . this very simple decomposition suggests that the jet completely dominates the radio and @xmath3-ray emission as expected , but is only a @xmath4 contributor to the optical , uv and x - ray emission . given that the tentative decomposition suggests that 10% of the x - ray flux comes from the jet , the much more careful treatment we plan to undertake is needed to determine whether the jet s flux can be ignored in fitting the x - ray spectrum or not . in this paper , we test the jet - cycle picture using a detailed spectral analysis in the x - ray band supported by a multi - wavelength analysis in the optical / uv and radio . a multi - epoch analysis of _ xmm - newton _ and _ suzaku _ spectra shows that , judged on purely statistical grounds , there are degenerate phenomenological interpretations of the x - ray spectrum , one of which does not require any relativistic disk reflection component . however , physical considerations lead us to prefer a model in which the soft x - ray excess is a mixture of ( steep ) jet - emission and blurred ionized disk reflection . within the context of this model , we do indeed find evidence for changes in the inner accretion disk structure in the sense expected from the jet - cycle picture . additionally , we investigate the x - ray spectral variability on the timescale of a few hours , enabling us to tap into timescales comparable to the orbital time of the inner accretion disk and unobservable in x - ray binaries . we suggest that the observed rapid x - ray spectral variability corresponds to magnetically mediated disturbances propagating from the disk ( and/or disk corona ) into the jet flow . the paper is organized as follows : first , we describe the datasets used in this work and briefly discuss the data reduction techniques ( [ data ] ) . after placing the observations in the context of the jet cycle ( [ context ] ) and performing a preliminary , basic investigation of spectral shape of the individual data epochs , a multi - epoch analysis is used to investigate the nature of the x - ray spectrum ( [ spectral ] ) . we then investigate the short - term variability and spectral energy distribution to further enhance our understanding of the processes driving the production of x - ray radiation in this agn . we end with a brief discussion of the implications for jet formation that can be drawn from the observations ( [ dis ] ) . throughout this paper , luminosities and distances are calculated using a @xmath5cdm cosmological model with @xmath6 , @xmath7 and @xmath8 @xcite . for a redshift of @xmath1 , this results in a luminosity distance to 3c120 of @xmath9 and an angular size distance of @xmath10 . the goal of this paper is a detailed analysis of the physical processes operating in the central engine of 3c120 . to obtain this complete picture we conduct sensitive spectroscopy with _ xmm - newton _ and _ suzaku _ , which is placed in context by _ rxte _ , _ swift _ and vlba monitoring data . below we describe the datasets that are considered and how they were reduced . c c c c c observatory & instrument & date & obsid & exposure + & & & & [ ksec ] + _ suzaku _ ( b ) & xis & 2012/02/14 & 706042020 & 181.4 + _ suzaku _ ( a ) & xis & 2012/02/09 & 706042010 & 280.2 + _ xmm - newton _ & epic - pn & 2003/08/26 & 0152840101 & 89.2 + the datasets analyzed in this paper are from deep pointings by _ xmm - newton _ and _ suzaku_. the _ xmm _ data have already been analyzed by @xcite and @xcite . a summary of the observations ( re-)analyzed here is presented in table [ tab1 ] . the two _ suzaku _ pointings were taken as part of the _ suzaku agn spin survey _ , a cycle 46 key project ( pi reynolds ) . the suzaku data were reduced with heasoft v6.12 and calibration files dated 2009 september 25 . during our observations , xis was operated in full - window mode . in preparing the xis spectra , we first reprocessed and screened the data using ` aepipeline ` and the standard screening criteria , as listed in the _ suzaku _ abc guide . we then created individual spectra using ` xselect ` and response files using ` xisrmfgen ` and ` xissimarfgen ` tools for each detector and data mode combination . the regions used for source and background are circular with radii of 3.8 and 1.7 respectively for suzaku a and 4.2 and 1.6 for the suzaku b observation . later the spectra of different data modes were combined for each xis . while this is the recommended standard procedure to create xis spectra , one additional complication needs to be accounted for in our case . xis spectra are contaminated by absorption from a hydrocarbon layer residing on the optical blocking filter and we must take measures to correct for this contamination . the contamination is monitored by the _ suzaku _ team , and a correction is part of the construction of the standard effective area file . at the time of our analysis , this contamination model was in the process of being updated due to inconsistencies with recent xis observations . we were able to use a test version of the contamination model in our analysis ; for observations after 2010 , this version is identical to the model incorporated in the caldb files ae_xi?_contami_20120719.fits , released by the xis team as part of the 20120902 caldb release . for the _ suzaku data _ considered in this work , we found that we were unable to utilize data from the pin detector due to the increased thermal noise caused by an increase in the leakage current . while the problem is most apparent for temperatures above -11c , we already notice the effect ( in terms of a distortion of the lower energy portion of the pin spectrum ) at temperatures of about -14c , the pin temperature during our observation . the _ xmm - newton _ dataset was reduced using xmmsas 11.0.0 . the epic data were first reprocessed , using the calibration files as of february 2012 . the spectra were then extracted using the tool ` evselect ` , selecting the default grade pattern . source spectra are taken from a circular region with radius 41.2 `` centered on the source and the background spectra from an also circular region on the same chip with radius 28.9 '' . the effective area files were generated using the xmmsas task ` arfgen ` , and the redistribution matrices were produced using the task ` rmfgen ` . the epic detectors were operated in small window mode during the observation to reduce possible pile - up . during the observation , mos1 was operated in timing mode . mos2 alone has only @xmath111/3 the effective area of epic - pn in the iron k region . assessing possible pile - up effects using the single , double , triple and quadruple event pattern versus the predicted fractions utilizing ` epatplot ` , shows that epic - pn is not affected by pile - up while epic - mos is influenced by pile - up . considering these facts , we only consider epic - pn data for the analysis performed in this paper , as mos2 would not significantly contribute to any constraints . the average observed net pn and mos source count rates are 14.2cts@xmath12 and 5.8cts@xmath12 respectively . for the spectral analysis , all xis / epic - pn spectra were binned to a signal - to - noise ratio of 10 . the considered energy ranges are 0.7 - 1.7kev and 2.3 - 10kev for _ suzaku_-xis , with the energies around 2kev being excluded because of known calibration issues around the mirror ( gold ) and detector ( silicon ) edges . the epic - pn spectrum is used in the 0.5 - 10kev band . we extended all energy grids to energies far beyond the upper energy limit given by the highest data bin considered in fitting , in order to enable a proper model evaluation of the relativistic blurring . in any detailed spectral analysis such as is presented in this paper , we must be cognizant of the possibility of instrumental calibration errors . such errors would directly lead to systematic errors in our spectral fits . the absolute flux calibration of x - ray observatories can be uncertain to @xmath4 , but _ our _ physical conclusions derive purely from the _ shape _ of the spectrum and hence we need only be concerned with the relative calibration . as already mentioned , our _ suzaku _ spectra are clearly inflicted by significant unmodeled calibration features around the silicon k - edge and the gold m - edge ; we remove these energies from consideration by masking data in the 1.72.3kev band . our choice of low - energy cutoffs ( 0.5kev and 0.7kev in _ xmm - newton _ and _ suzaku _ respectively ) is also driven by the need to exclude more poorly calibrated regions of the spectrum . however , due to the lack of a practicable approach for propagating calibration uncertainties into the systematic error budget for our complex ( large parameter space ) spectral models , no further attempt is made in this paper to account for calibration uncertainties . we note that the relevant calibration errors are thought to be small at the time of writing , the _ xmm_/epic has a maximum error in the relative effective area of 5% ( with an rms error that is rather smaller ) and a maximum error in the absolute energy scale of 10ev ; for _ suzaku _ the maximum error on the energy scale is also 10ev and the effective area uncertainty is small outside of the mirror edges . a primary goal of this paper is to investigate changes in accretion disk structure during the jet - cycle in 3c120 . hence , we require monitoring data to diagnose the state of the source at the time of the deep spectroscopic observations by _ xmm - newton _ and _ suzaku_. for the period 20022008 , this monitoring was provided by _ rxte_. the average cadence of the monitoring over this period is 4days , with the exception of a 2-month period each spring when 3c120 lies too close to the sun . we have obtained these data from the heasarc archives and reduced them exactly as described in @xcite . only data from the top layer of proportional counter unit 2 were used . for each pointing , we produce a background - subtracted pca spectrum using the appropriate epoch - dependent background / response files . the fluxes quoted below were obtained from the spectra for each pointing , assuming a simple power - law continuum with a neutral absorption column of @xmath13@xmath14 @xcite . we do not consider any _ rxte _ intra - pointing variability in this work . the resulting _ rxte _ lightcurve is shown in fig . [ rxte_all ] . monitoring from _ rxte _ is not available after 2008 . thus , we established a monitoring campaign with _ swift _ to support the 2012 _ suzaku _ pointings . xrt and uvot data from this campaign are used in this paper . the source can achieve x - ray flux levels that would lead to significant photon piled up in the xrt if operated in photon counting mode . to reduce the impact of pile - up we preferred to use the windowed timing mode data when available and otherwise excluded the central region of the psf to mitigate pile - up . swift_-xrt data were first reprocessed to apply the newest calibration ( xrt calibration files : 20120830 ) . from the resulting event file , a spectrum was extracted using xselect following a standard extraction for the windowed timing mode as described in the xrt user s guide " . to properly account for the rolling of the satellite during the pointings ( especially important in windowed timing mode ) , a spectrum is reduced for each good time interval with a different roll angle . finally , these good time interval spectra are summed to yield the final spectrum for each pointing . as with the _ rxte _ monitoring , x - ray fluxes are derived on the basis of a simple power - law continuum with a neutral absorption column of @xmath13@xmath14 . the uvot analysis is based on level ii products . each individual uvot filter data file contains , in general , a number of exposures which were summed using the tool uvotimsum . the uvotsource tool was then used to extract magnitudes from simple aperture photometry . source and background regions were created for this purpose , with the position of the source region being obtained from the nasa extragalactic database ( ned ) in the first instance . the required source region and background region where both circular with a 4.8 arcsec radius . the resulting uv and x - ray lightcurves from _ swift _ are shown in fig . [ swift_uv ] . we observed 3c120 nine times with the very long baseline array ( vlba ) between 2012 january 27 and 2013 january 15 at a frequency of 43 ghz . the observations , part of the boston university @xmath3-ray blazar monitoring program ( www.bu.edu/blazars ) included ten 3 - 5 minutes scans on 3c120 at each epoch . the raw data were recorded at each antenna and correlated at the national radio astronomy observatory s array operations center in socorro , nm . the resulting data on the _ uv_-plane were edited , calibrated ( including sky opacity corrections and fringe fitting ) , and imaged in a standard manner with routines from the astronomical image processing system and difmap software packages . this involved an iterative procedure alternating imaging the clean algorithm with self - calibration of both phases and intensities , which eventually converged to the final images . the resulting images , with an angular resolution of 0.14 milliarcsec along the direction of the jet , are displayed in figure [ vlba ] . they reveal the appearance of a new superluminal knot , which we designate _ k12 _ , with a proper motion of @xmath15 milliarcsec yr@xmath16 , corresponding to @xmath17 at the distance of 3c120 . the centroid of the knot crossed the centroid of the `` core '' ( the bright , presumed stationary feature at the eastern end of the jet ) on jd @xmath18 ( march , 14th 2012 ) . another development is the enhanced brightness of the jet between 0.2 and 0.5 milliarcsec from the core that is apparent starting with the 2012 august 13 image . this feature is either stationary or moving subluminally . ( reached on 2013 january 15 ) . the easternmost feature , the `` core , '' is presumed to be stationary . k12 _ is marked , with a mean trajectory ( shown by line ) correponding to an apparent speed of and an `` ejection '' date ( when the brightness centroid of _ k12 _ coincided with that of the core ) of 2012 march @xmath19 . the elliptical gaussian restoring beam of fwhm dimensions @xmath20 milliarcsec along position angle @xmath21 is displayed in the bottom left corner . ] as outlined above , we need to estimate the placement of each of our spectroscopic observations in the jet - cycle . the long - term light curve taken by _ rxte _ , previously published by @xcite , clearly shows the pattern of x - ray dips and recovery associated with the jet cycle ( fig . [ rxte_all ] ) . we mark on fig . [ rxte_all ] the time of the _ xmm _ pointing considered in this work . this reveals that the pointing was on the peak of the flux sequence , corresponding , according to the model , to a radiatively - efficient disk extending down close to the black hole ( step 1 of fig . [ jet ] ) . also marked is the position of the 2006-_suzaku _ observation as analyzed by @xcite and @xcite which occurred during a period of rising x - ray flux , as it was already re - analyzed recently we decide to not include it in our spectral analysis . this rising flux may correspond to a refilling accretion disk ( step 4 of fig . [ jet ] ) , in line with the truncated accretion disk ( @xmath22 ) found by @xcite . the _ swift _ monitoring campaign supported the new 2012-_suzaku _ pointings . the monitoring ( fig [ swift_uv ] ) shows that the general trend in the uv band was a rising flux , with some temporary stalling of the rise . the x - ray variability is less coherent , showing indications for short - lived flux dips , but appears to show a rising trend . the x - ray flux is estimated by fitting an absorbed power law with the absorption column fixed to @xmath23@xmath14 ( a value similar as the one found later for the best fit to the _ suzaku _ pointings ) . similar short - timescale x - ray variability appears when zooming into the _ rxte _ x - ray monitoring ; see figure1 of @xcite . the vlba monitoring also helps to place the 2012-_suzaku _ observations into context . the superluminal knot appears approximately one month after our _ suzaku _ pointing . however , the work of @xcite shows that there is approximately a 2 month delay between the probable time of the physical ejection from close to the black hole and the appearance of a new vlba knot , most likely corresponding to the propagation time of the ejector between the black hole and the vlba core . accounting for this time delay , we estimate that the actual ejection occurred approximately one month prior to the _ suzaku _ pointings . hence , the prediction is that _ suzaku _ should find a truncated / refilling accretion disk . in summary , on the basis of the _ rxte _ , _ swift _ and vlba monitoring together with the jet - cycle hypothesis , the 2003-_xmm _ pointing should show a filled accretion disk extending all the way to the isco . during the two _ suzaku _ pointings on the other hand we expect the accretion disk to be still truncated but be in a stage of refilling . we begin by discussing our detailed analysis of the deep spectral _ xmm - newton _ and _ suzaku _ data . our main technique is a multi - epoch analysis " in which all spectral data are fitted together , tying together parameters that must ( on physical grounds ) be common across all epochs . however , we begin by discussing an initial exploration of the spectra and the construction of the spectral models . throughout this work , the spectral analysis was performed with the interactive spectral interpretation system ( isis version 1.6.0 - 7 ; * ? ? ? * ) using the newest _ xspec _ 12.0 models all uncertainties are quoted at the 90% confidence level for one parameter of interest ( @xmath24 ) . systematic errors arising from the physical assumptions made in the models used for describing the spectra are not included in the given errors . investigating the spectral variability by looking at the hardness evolution within each _ xmm _ and _ suzaku _ pointing , we find it is mild , and therefore ignore it in this section . the short - term variability that we do find will be discussed in [ var ] . to get a first look at the spectral shape , we fit a simple absorbed power law to the 24.5kev data and then extend to the full energy range of the observation ( fig . [ basic ] ) . this reveals an iron line and an upturn at higher energies indicative of the presence of reflection in the spectra . below 2kev , additional absorption beyond the galactic column that is already accounted for is apparent . , _ suzaku _ ii [ middle panel ] and _ xmm _ [ bottom panel ] ) fitted from 2 - 4.5kev ; spectra are rebinned for plotting . ] building upon this knowledge and previous works @xcite , we construct three multi - component spectral models to describe our spectra . common to all of these models is a primary powerlaw continuum with photon index @xmath25 together with cold , distant reflection of that powerlaw described by the model ` pexmon ` @xcite . the reflection fraction @xmath26 and the iron abundance characterizing this reflection are left as free parameters . as some parameters for ` pexmon ` can not be determined from our fits we need to fix them to certain values . we assume that the high - energy cutoff of the continuum powerlaw is 300kev , which is well out of our fitting range . since pexmon strictly models reflection from a planar surface , we also need to fix the inclination parameter of the pexmon model . the most likely geometries for the material producing the cold / distant reflection features are either cloud - like or a large - scale torus , and thus a range of inclinations will contribute to the observed spectrum . assuming that this can be approximated as reflection from isotropically oriented planar segments , the most likely inclination angle is 60@xmath27 . thus , we set the inclination of the pexmon component to 60@xmath27 . we have verified that this assumption does not drive any of the conclusions reached in this paper in particular , repeating the analysis presented in [ jetdisk ] using a pexmon inclination of 30@xmath27 yields best fitting parameters that are within the statistical error bars of our canonical analysis , with the exception of the reflection fraction . also common to all of our spectral models is a photoionized emitter producing , most notably , fe xxv / fe xxvi iron - k@xmath28 lines . the photoionized emission is parametrized by ionization parameter and normalization . it is described using a table model calculated using ` xstar2xspec ` from the ` xstar ` model ` photemis ` , assuming irradiation by a power law with @xmath29 and a number density of @xmath30 . the xstar model requires an assumed column density ; we fix this to be @xmath31@xmath14 , noting that there is only a very weak dependence of the spectral shape on this column density . as noted in @xcite there can be a strong degeneracy between fexxv and the blue wing of the broad iron line , therefore the inclusion of this component is crucial . finally , photoelectric galactic absorption ( @xmath32 , the average value for 3c120 from heasarc s nh tool ) is modeled with ` tbnew ` a newer version of ` tbabs ` @xcite with cross sections set to ` vern ` and abundances set to ` wilm ` . we also account for intrinsic cold absorption in the source using ` tbnew ` with column density as a free parameter . our first spectral model is a disk reflection model " . to the above base model , we add a component describing relativistic reflection from the innermost regions of an ionized accretion disk . to describe the ionized reflection associated with reflection of the inner accretion disk , we used a modified version of ` reflionx ` developed by @xcite , see @xcite for details . the iron abundance is fixed to that characterizing the distant reflection ( thus we assume a chemically homogeneous central engine ; see @xcite ) , and the photon index of the continuum irradiating the disk is tied to the one for the primary powerlaw . the ionized reflection component is then relativistically blurred using the model ` relconv ` @xcite ; this naturally gives rise to a relativistic iron line , blurred compton reflection hump and a soft excess . the radial dependence of the emissivity of the reflection component is assumed to have a power - law form with index @xmath33 . the inner edge of the x - ray reflection regime can be at the isco @xcite or further out , and the outer edge was fixed to 400@xmath34 . provided that @xmath35 , the relativistic blurring kernel is only weakly dependent upon this outer radius . the accretion disk inclination @xmath36 , and the black hole spin @xmath37 were left as free parameters . for a radio - loud agn relativistic disk reflection is not the only possible origin for the soft x - ray excess . a contribution from the jet emission could also produce an excess flux at soft energies . thus , our second spectral model is a jet model " . to the base model , we add a second ( steep ) powerlaw component with photon index @xmath38 to characterize a possible contribution from the jet . as compared with the disk reflection model , this decouples the soft excess from the iron - k band structure . this is motivated by similar treatments in @xcite and @xcite . the final spectral model is the disk@xmath39jet model " in which we include both a relativistic disk reflection component and a steep power - law . [ cols="^,^,^,^,^",options="header " , ] driven by the potential deficiencies of the pure - bred " spectral models discussed above , we explore a mixed model in which we include both a jet component ( i.e. an additional steep powerlaw continuum ) and reflection from a relativistic disk . as must be the case ( since this model - space completely encompasses the previous models ) , the goodness of fit parameter shows improvement , @xmath40 ( 0.98 ) ; this improves on the jet - model by @xmath41 ( for 14 additional parameters ) , and the disk - reflection model by @xmath42 ( for 6 additional parameters ) . an example of how the different components of this model are able to describe the data is shown in fig . [ jetdiskflux ] for the first _ suzaku _ pointing as well as the _ xmm _ pointing . we note that the broadening of the disk spectrum in case of _ xmm _ is so severe that the exclusion of the narrow 1.7 - 2.3kev band from our _ suzaku _ spectra does not strongly affect a possible detection of such a disk component . especially when comparing the jet model with the jet@xmath39disk model , an f - test shows that the improvement in fit is not statistically significant . however , this mixed model solves most of the fine - tuning and physical - inconsistencies that have been noted for the pure - bred models . in this spectral solution , the jet component has a more modest photon index ( @xmath43 ) , alleviating the fine tuning problem noted in [ jet_m ] . the disk component fits to an almost face - on inclination ( @xmath44degree ) and the implied ionization state of the disk is approximately constant , alleviating the problems with the pure disk - reflection model . in particular , the large jump in the ionization state of the disk between _ suzaku_-a and _ suzaku_-b that was seen in the disk - reflection model is no longer present . this is straightforward to reconcile . as is evident in fig . [ basic ] , there is a subtle but significant change in the soft excess between these two pointings . in the disk - reflection model , the soft excess is modeled as blurred ionized reflection and so this spectral change drives a strong change in the inferred disk ionization ( even though the rest of the spectrum is very similar ) . in the mixed model , the soft excess change is entirely described by a change in the slope of the soft ( jet ) powerlaw , allowing the accretion disk parameters to remain almost unchanged . more noteworthy is the fact that , in this mixed model fit , the intrinsic neutral absorption approximately halves ( from @xmath45 to @xmath46 ) between the _ xmm - newton _ observation in 2003 to the two _ suzaku _ pointings in 2012 . if we equate this nine year timescale with the dynamical time in the black hole potential ( @xmath47 ) , this implies that the variable absorbing structures are at @xmath48 , suggestive of the optical broad line region ( blr ) . in this scenario , the _ xmm - newton _ spectrum is formally reflection dominated , with the ionized reflection from the inner accretion disk dominating the primary continuum across most of the band . at the same time , the inner edge of the accretion disk is strongly constrained to be at the isco , the black hole is constrained to be rapidly spinning , and the emissivity index @xmath33 becomes extremely steep ( @xmath49 ) . all of this behavior is consistent with the extreme light bending of a primary x - ray source on the spin axis and very close to a rapidly spinning black hole @xcite . after investigating the global long - term variability using long - term light curves ( [ context ] ) and multi - epoch fitting ( [ spectral ] ) , it is important to note that there is shorter timescale ( intra - observation ) variability . the only significant short - timescale variability observed during the pointings considered in this paper was found in the first _ suzaku _ pointing , around the beginning of 12-feb-2012 ( fig [ jump ] ) , in the form of a sudden increase in the count rates below 4kev . to explore the nature of this variability we examine the development of the 410kev/0.72kev hardness ratio ( fig . [ hard_jump ] ) . we see clear indications for spectral changes , starting even before the sudden count rate increase . a significant hardening of the source is observed in the time preceding the count rate jump . an examination of fig [ jump ] shows that the 410kev flux jumps before the 24kev flux which , in turn , peaks before the 0.72kev band . this increase of the 4 - 10kev count rate before the increase of the softer bands explains the observed hardening . two possibilities for this sudden hardening and subsequent softening come to mind . either the absorption column suddenly increases and then progressively decreases ( successively uncovering softer energies ) or there is a real change in the continuum shape ( with the soft energies lagging the hard band ) . to investigate the nature of the variability in more detail , we construct pre - jump and post - jump spectra . the portions of the light curve used for pre - jump and post - jump are indicated in fig . the spectra can be described by an absorbed power law with a gaussian line as neutral iron k. the quality of the data does not allow us to constrain the absorption and the spectral slope to a degree where we could know which of the two is changing . however , varying only one of the two options we find that adjusting only the absorption leads to a worse fit than allowing for a change in photon index . this is also apparent from the contour plot ( fig . [ contr ] ) of @xmath50 versus @xmath25 which suggests that the continuum slope is varying . as discussed in [ me_fitting ] , acceptable spectral models attribute at least some of the soft excess to a steep powerlaw - like component from a jet . thus , while difficult to prove from this one event , we may be witnessing the propagation of a disturbance from the inner accretion disk into the jet . we discuss this possibility further in [ dis ] . and @xmath51 ( in units of @xmath52 ) for the pre - jump ( 1 ) and post - jump ( 2 ) spectra of the _ suzaku _ i pointing . ] our preferred spectral model to describe the multi - epoch _ xmm - newton _ and _ suzaku _ data is the jet@xmath39disk model ( see discussion in [ jetdisk ] ) . the fits of this model to the data directly support the predictions of the jet - cycle model . we find a recessed accretion disk ( @xmath53 ) during the 2012-_suzaku _ observations when the x - ray flux is on a rising trend and the jet - cycle picture would suggest a refilling accretion disk . this scenario is supported by a rising trend in the uv flux . it is likely that most of uv / optical flux in 3c120 is coming from the disk as the measured polarization is low @xcite , so again the uv rise suggests a refilling accretion disk . on the other hand , we find that the disk extends all of the way to the isco and is strongly irradiated by a compact source close to the black hole during the _ xmm - newton _ observation when the x - ray flux is at a peak . to fit this _ xmm - newton _ spectrum , we require a rapidly rotating black hole . the formal limits indicate a truly extreme spin @xmath54 . however , our methodology implicitly assumes a razor - thin accretion disk ( this assumption enters via the termination condition for the ray - tracing underlying the relconv relativistic transfer function , see @xcite ) . for such rapid spins , the isco is very close to the event horizon and finite disk - thickness effects must come into play . while more work is needed on this issue , a first attempt to explore the systematic errors introduced by finite - thickness effects was made by @xcite ( see their fig . 5 ) . this suggests that the formal result of @xmath54 may weaken to @xmath55 once finite disk thickness and emission within the isco is considered . our overall picture is supported by the appearance in the jet of a new superluminal knot in the vlba images about one month after the 2012-_suzaku _ pointings and two months following the middle of the x - ray dip between jd 2455915 ( december , 19th 2011 ) and 2455960 ( february , 2nd 2012 ) . the latter delay is similar to the mean value of @xmath56 yr observed over five years of _ rxte _ and vlba monitoring of 3c120 @xcite . @xcite and @xcite interpret the delay as the travel time of a disturbance in the jet flow from the base of the jet near the accretion disk to the 43 ghz `` core '' about 0.5 pc away . the enhanced emission 0.2 - 0.5 mas from the core and @xmath57 pc from the central engine starting in 2012 august is too bright to be caused by trailing shocks " @xcite behind the superluminal knot . a more likely explanation is either the establishment or strengthening of multiple standing shocks . such shocks are set up by pressure imbalances with the external medium @xcite , which can occur when the energy density injected into the jet varies significantly with time @xcite . @xcite detected such a stationary feature , designated _ , 0.2milliarcsec from the core in 3c120 between 1998 and 2001 . figure [ vlba ] marks the location of _ a1 _ , which is still present in 2012 . the brightening of the core , which could be the first in a series of standing shocks , suggests that the rate of injection of energy did indeed increase starting in july 2011 . another apparently stationary emission feature appeared after this about 0.4milliarcsec from the core . unfortunately , our x - ray , uv , and optical observations ended several months before this occurred , hence we can not determine whether a variation in the disk and corona emission heralded this change in the jet . in our preferred x - ray spectral model , the soft excess is a combination of ionized reflection from the inner disk and a steep ( @xmath58 ) jet continuum . within the context of standard jet emission models , such steep continua suggest that we are seeing the cut - off region of the jet synchrotron spectrum . using the _ xspec _ model srcut , we have verified that our x - ray spectra would not have been sensitive to the curvature of this synchrotron spectrum since the jet dominates a narrow band at the lowest accessible x - ray energies . extrapolating the synchrotron model to lower energies reveals that , in @xmath59 terms , the jet component should be approximately one order of magnitude brighter in the optical - uv band than the soft x - ray . this is inconsistent with the sed of @xcite ; however , this sed was constructed from non - simultaneous data and hence must be viewed with caution . a simultaneous optical / uv / x - ray sed can be constructed from _ swift _ uvot+xrt data . @xcite present exactly such a _ swift _ sed , showing indeed that the optical / uv flux exceeds the x - ray flux by a factor of @xmath60 . the existence of a broad - band spectral component extending from the optical / uv into the soft x - ray band is supported by the detection of a correlation between the uv / optical and soft x - ray flux within the _ xmm _ pointing @xcite . similarly , a weak correlation between the ( 1320-1420 ) band and the soft x - ray band ( 0.3 - 2kev ) has also been found previously by @xcite from simultaneous iue-_exosat _ observations . this uv - soft x - ray correlation makes a comptonization model for the soft excess similar to the one recently discovered in bright seyferts @xcite also a possibility in 3c120 , while at the same time ruling out a solely reflection - based soft excess . in our uv monitoring , we do not find a correlation between the uv flux and the 1 - 2kev/0.3 - 1kev x - ray hardness ratio ( fig . [ hardness ] ) , as it would be expected from a broad - band component . this absence could be explained in two ways : firstly , a correlation on the timescales of the monitoring ( one week ) is not necessarily expected . secondly , the average _ xmm - om _ uvw1 ( 2910 ) flux was @xmath61wm@xmath62 , while our maximum uvot - u flux ( 3465 ) is only @xmath63wm@xmath62 implying that the average flux values must be much lower . if part of this uv / optical flux is in fact a broad - band component its flux contribution could be too low during our monitoring to be detectable . while the spectral analysis enables us to study the spectral components at any given time , variability studies can provide valuable additional information and put constraints on emission regions and the structure of the central accretion flow . previous analysis of the long - term _ rxte _ light curve by @xcite showed that its psd is similar to radio - quiet agn , implying the origin of the x - rays may be similar . in agreement with this hypothesis , the spectrum softens as the source brightens @xcite , as is well established for seyfert galaxies ( e.g. , see * ? ? ? * ) . while this is all well - known , the considered time scales are rather long , making changes in the accretion rate a likely origin . however the _ swift _ and _ rxte _ monitorings also indicate short - time scale variability with a time scale of less than a few days . for example , we find a @xmath64 decrease in the 210kev count - rate between 1-dec-2011 and 4-dec-2011 . however , we find even faster variability in the first of the _ suzaku _ pointings . the sudden increase of flux ( on a time scale of 12hours ) starts in the hard band and propagates into the soft - band . let us examine the classical time scales characterizing the accretion disk . for a black hole mass of @xmath65 the dynamical time scale is @xmath662.6hrs at @xmath67 . changes in the local structure of the accretion disk are governed by the thermal time scale which is @xmath68 at @xmath67 if we assume a disk viscosity parameter of @xmath69 . accretion rate changes occur on the viscous time scale , @xmath70 $ ] where @xmath71 is the disk thickness . if the radiatively - efficient part of the disk really truncates at @xmath72 during this pointing , as suggested by our preferred spectral fit , the flow at @xmath73 would be a hot , advection - dominated , geometrically - thick structure ( @xmath74 ) implying that @xmath75day . in the light of these time scales , we can see that the short time scale _ suzaku _ event must have originated from the advection dominated flow close to the black hole ( within @xmath73 ) . furthermore , given that the onset of the event in a given band is very rapid ( @xmath76hours ) , it seems unlikely that the event is driven by changes in accretion rate ( viscous time scale ) or local disk structure ( thermal time scale ) . as already alluded to in [ var ] , the _ suzaku _ event may be revealing a short - term connection between the accretion disk and the jet . we will discuss this in the context of our preferred jet@xmath39disk spectral model . the _ suzaku _ event starts as a jump in the hard ( 410kev ) band that , in our spectral model , is associated with x - ray emission from the disk corona . over the next 0.5day , it then propagates to the soft ( 0.72kev ) band that , in our model , has a significant jet component . if this is the correct interpretation , it gives us a rare look at the short timescale coupling between the disk and the jet . this coupling is almost certainly magnetic in nature @xcite , and introduces a new timescale into the system the magnetic timescale . the magnetic timescale is the timescale on which the poloidal magnetic fields in the disk can spontaneously align in portions of the disk , potentially changing the dissipation in the disk and its coupling to the jet . this magnetic timescale is hard to estimate ( ultimately relating to the dynamo problem ) but could be very fast @xcite , especially in geometrically - thick flows expected within @xmath77 . in this paper , we present a detailed spectral analysis of deep _ xmm - newton _ and _ suzaku _ pointings of the blrg 3c 120 . these observations are placed into the context of the hypothesized jet - cycle through the use of _ rxte _ , _ swift _ and vlba monitoring campaigns . our main findings are : 1 . using the monitoring data and the jet - cycle picture , we expect that the 2003-_xmm - newton _ spectrum ( taken at a time of peak x - ray flux ) should show a complete accretion disk extending down to the isco , whereas the 2012-_suzaku _ observations should show truncated , refilling accretion disks . 2 . a multi - epoch analysis of the _ xmm - newton _ and _ suzaku _ pointings finds three statistically acceptable spectral models , a disk - reflection model , a jet - model , and a jet@xmath39disk model . while they can not be distinguished on purely statistical grounds , the disk - reflection model strongly violates the radio constraints on the jet / inner - disk inclination and can not explain the observed uv - soft x - ray flux correlation . at the same time the jet - model suffers a severe fine - tuning problem . on the other hand , the jet@xmath39disk model appears physically reasonable in all respects . 3 . adopting this jet@xmath39disk model as our preferred solution , we do indeed find truncated disks during the two _ suzaku _ pointings and a complete accretion disk at the time of the _ xmm - newton _ observation . this is exactly in line with expectations from the jet - cycle picture . the ejection of a new superluminal knot about two months after an x - ray flux dip further supports this scenario . we detect a rapid event in our first _ suzaku _ pointing that starts in the hard band ( 410kev ) and propagates to the soft spectrum ( 0.72kev ) . we interpret this as a disturbance which propagates from the inner ( advective ) disk into the jet on a timescale of @xmath78hours , and suggest that this timescale is set by the dynamics of the magnetic field . our preferred spectral solution has a rapid black hole spin . formally , our spectral fit yields a spin limit of @xmath54 . however , for such rapid spins , finite disk thickness effects must be important and will weaken the limit . employing the toy - model of @xcite to assess the role of these effects suggests a true limit @xmath55 . the authors thank the anonymous referee for comments that improved this manuscript . we thank phil cowperthwaite , javier garca and abdu zoghbi for useful discussions throughout the duration of this work . aml and csr gratefully acknowledge nasa for support under the astrophysical data analysis program ( adap ) grant nnx12ae13 g . the research at boston university was supported by nsf grant ast-907893 and nasa grant nnx11aq03 g . the vlba is operated by the national radio astronomy observatory . the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .. acf thanks the royal society for support . , j. c. a. , sivakoff , g. r. , altamirano , d. , coriat , m. , corbel , s. , dhawan , v. , krimm , h. a. , remillard , r. a. , rupen , m. p. , russell , d. m. , fender , r. p. , heinz , s. , krding , e. g. , maitra , d. , markoff , s. , migliari , s. , sarazin , c. l. , & tudose , v. , 2012 , , 421 , 468
we present a study of the central engine in the broad - line radio galaxy 3c120 using a multi - epoch analysis of a deep _ xmm - newton _ observation and two deep _ suzaku _ pointings ( in 2012 ) . in order to place our spectral data into the context of the disk - disruption / jet - ejection cycles displayed by this object , we monitor the source in the uv / x - ray bands , and in the radio band . we find three statistically acceptable spectral models , a disk - reflection model , a jet - model and a jet+disk model . despite being good descriptions of the data , the disk - reflection model violates the radio constraints on the inclination , and the jet - model has a fine - tuning problem , requiring a jet contribution exceeding that expected . thus , we argue for a composite jet+disk model . within the context of this model , we verify the basic predictions of the jet - cycle paradigm , finding a truncated / refilling disk during the _ suzaku _ observations and a complete disk extending down to the innermost stable circular orbit ( isco ) during the _ xmm - newton _ observation . the idea of a refilling disk is further supported by the detection of the ejection of a new jet knot approximately one month after the _ suzaku _ pointings . we also discover a step - like event in one of the _ suzaku _ pointings in which the soft band lags the hard band . we suggest that we are witnessing the propagation of a disturbance from the disk into the jet on a timescale set by the magnetic field .
the balance property is a notion connected with sturmian words from very first beginning of their investigation . in @xcite , sturmian words were defined as aperiodic words with the smallest possible factor complexity . already in the same article , hedlund and morse observed that sturmian words show also the smallest discrepancy in occurrences of letters . to quote precisely their result , let us denote by @xmath16 the length of the word @xmath5 and by @xmath17 the number of occurrences of letter @xmath7 in @xmath5 . hedlund and morse proved that an infinite aperiodic word @xmath0 over the alphabet @xmath18 is sturmian if and only if for all pairs @xmath5 , @xmath4 of factors of @xmath0 with @xmath19 it holds @xmath20 . let us note that the letter @xmath21 is not preferred as in binary alphabet the relations @xmath19 and @xmath20 imply the inequality @xmath22 as well . during past 70 years many other characterizations of sturmian words have appeared , for their overview see @xcite . each of these characterizations may serve and serves for generalization of sturmian words to multiliteral alphabets , cf . @xcite . nevertheless , the balance property seems to be the most complicated to deal with . only a few results are known about words satisfying the so - called @xmath2-balanced property . let us recall that an infinite word @xmath0 over an alphabet @xmath1 is @xmath2-balanced if for all letters @xmath23 and all pairs @xmath24 of factors of @xmath0 with @xmath25 it holds @xmath26 . note that sturmian words are @xmath27-balanced in this terminology . the set of @xmath2-balanced words differs substantially from all other generalizations of sturmian words . neither generic arnoux - rauzy word nor generic word coding interval exchange transformation are @xmath2-balanced , see @xcite and @xcite . in @xcite , adamczewski studies whether fixed point of a primitive substitution is @xmath2-balanced for some constant @xmath2 . he shows that the existence of such @xmath2 depends only on the spectrum of the incidence matrix of the substitution . however , the minimal value of @xmath2 can not be deduced from the spectrum . in @xcite and @xcite , the minimal value of @xmath2 is determined for binary fixed points of canonical substitutions associated with quadratic pisot numbers . the notion `` canonical substitution associated with a number @xmath28 '' comes from positional numeration systems with the base @xmath29 , see @xcite . generally speaking , it is a very complicated problem to determine minimal value @xmath2 for a ternary balanced word , let alone for words over alphabets of higher cardinalities . despite a common belief that the tribonacci word is @xmath30-balanced , the first proof of this fact has appeared just one year ago in @xcite ( the tribonacci word is the fixed point of the substitution @xmath31 , @xmath32 , @xmath33 ) . in this article we provide minimal value of @xmath2 for a certain class of ternary words , namely for fixed points of substitutions @xmath34 with the parameter @xmath13 . these substitutions are canonical substitutions associated with cubic pisot numbers @xmath35 , roots of polynomials @xmath36 ( cf . let us recall that the tribonacci substitution is associated with a numeration system as well . the definition of a @xmath27-balanced word may be reformulated equivalently using parikh vectors . inspired by this fact , richome , saari and zamboni introduced the abelian complexity @xmath37 of infinite word . in their notation , sturmian words are aperiodic words with @xmath38 for all @xmath39 . the question on existence of words with constant abelian complexity is natural . it was shown in @xcite that for @xmath40 no words with @xmath41 exist . on the other hand , words with @xmath42 can be found in @xcite . the relation between abelian complexity and balance property is not straightforward . it is easy to see that an infinite word @xmath0 is balanced if and only if its abelian complexity is bounded . moreover , if the abelian complexity of @xmath0 is bounded by @xmath43 , then @xmath0 is @xmath44 balanced . the tribonacci case shows that the opposite implication is not valid : according to @xcite , the abelian complexity of the tribonacci word takes all values in the set @xmath45 . the fixed point of the substitutions studied in this article has the same property . let @xmath46 be a finite alphabet . a concatenation of letters in @xmath46 is called a _ word_. the set @xmath47 of all finite words over @xmath46 equipped with the empty word @xmath48 and the operation of concatenation is a free monoid . the _ length _ of the word @xmath49 , denoted by @xmath16 , represents the number of its letters . one may also consider infinite words @xmath50 ; the set of infinite words over the alphabet @xmath46 is denoted by @xmath51 . a word @xmath5 is called a _ factor _ of @xmath52 or @xmath51 if there exist words @xmath53 and @xmath54 or @xmath55 , respectively , such that @xmath56 . the word @xmath5 is called a _ prefix _ of @xmath4 , if @xmath57 . it is a _ suffix _ of @xmath4 , if @xmath58 . let @xmath59 . for @xmath60 , the symbol @xmath61 denotes the concatenation @xmath62 . we set @xmath63 . let a word @xmath52 have the prefix @xmath61 , @xmath60 . then the symbol @xmath64 denotes the word satisfying @xmath65 . similarly , if a word @xmath66 has the suffix @xmath61 for a @xmath60 , then @xmath67 denotes the word with the property @xmath68 . a _ morphism _ on the free monoid @xmath47 is a map @xmath69 satisfying @xmath70 for all @xmath71 . obviously , the morphism @xmath72 is determined if we define @xmath73 for all @xmath74 . a morphism @xmath72 is called a _ substitution _ , if @xmath75 for all @xmath74 and if there is an @xmath76 such that @xmath77 . an infinite word @xmath0 is said to be a _ fixed point _ of the substitution @xmath72 , or invariant under the substitution @xmath72 , if @xmath78 if we naturally extend the action of @xmath72 to infinite words , we may rewrite simply as @xmath79 . an infinite word @xmath0 is @xmath2-balanced , if for every @xmath74 and for every pair of factors @xmath4 , @xmath5 of @xmath0 such that @xmath25 , it holds @xmath80 . this property determines the discrepancy of occurrences of letters in the word @xmath0 . however , it turns out that if the cardinality of @xmath46 is higher than two , it is useful to have more detailed information , namely what is the discrepancy of occurrences of each particular letter . for this purpose we introduce the following notion : let @xmath0 be an infinite word over the alphabet @xmath46 and let @xmath74 . the word @xmath0 is said to be _ @xmath2-balanced with respect to the letter @xmath7 _ , if @xmath81 for all pairs of factors @xmath4 , @xmath5 of @xmath0 of the same length . let us consider an alphabet @xmath46 with @xmath43 elements , i.e. @xmath82 , and an infinite word @xmath0 over @xmath46 . for any factor @xmath5 of @xmath0 , its _ parikh vector _ is the @xmath43-tuple @xmath83 . let the symbol @xmath84 denote the set of all factors of @xmath0 of the length @xmath85 . then the _ abelian complexity _ of the word @xmath0 is a function @xmath86 defined by @xmath87 on the right hand side of there is the cardinality of the set of parikh vectors of all factors of @xmath0 of the length @xmath85 . in the sequel we will denote this set by @xmath88 , i.e. @xmath89 from now on , we will focus on a special class of substitutions on the ternary alphabet @xmath90 . for any integer @xmath13 , we denote by @xmath9 the substitution given by @xmath91 the substitution @xmath9 has a unique fixed point , namely @xmath92 if the results of @xcite are applied on @xmath93 , one finds out that there is a constant @xmath2 such that the word @xmath93 is @xmath2-balanced , but it is not known what the value of @xmath2 is and how it depends on @xmath15 . this is the main aim of this paper to determine @xmath2 . the fact that @xmath93 is balanced immediately implies that the abelian complexity function of @xmath93 is bounded , see introduction . the second aim of this paper is thus to find the optimal bound for @xmath94 . the elements of @xmath46 are usually denoted by numbers : @xmath95 etc . we have considered this notation , but we believe that the paper becomes more transparent if letters are used . the choice of @xmath96 , @xmath97 and @xmath98 has its roots in the fact that the word @xmath93 is a fixed point of a substitution associated with a number @xmath35 , cf . let @xmath99 denote the set of numbers which can be written in the form @xmath100 for non - negative integers @xmath101 . it can be shown ( cf . @xcite ) that when the elements of @xmath99 are drawn on the real line , there are exactly three types of distances between neighbouring points . if we assign the letters @xmath96 , @xmath98 and @xmath97 to the longest , the medium and the shortest distance , respectively , then the order of distances on the real line corresponds exactly to the order of the letters @xmath102 in the infinite word @xmath93 . we begin by the formulation of the main result of the paper . [ balance ] let @xmath93 be the infinite word invariant under the morphism @xmath9 given by . then @xmath93 is * @xmath103-balanced with respect to the letter @xmath96 , * @xmath30-balanced with respect to the letter @xmath97 , * @xmath30-balanced with respect to the letter @xmath98 , and none of these bounds can be improved . the theorem has the following trivial consequence : the infinite word @xmath93 is @xmath103-balanced and this bound is optimal , i.e. it can not be improved . since the proof is long and slightly complicated , we will split it into four sections and proceed in the following way : * we prove that @xmath93 is @xmath30-balanced with respect to the letter @xmath98 . * we prove that @xmath93 is @xmath30-balanced with respect to the letter @xmath97 . * we prove that @xmath93 is @xmath103-balanced with respect to the letter @xmath96 . * we show that none of the bounds can be improved . as we have explained in preliminaries , the word @xmath93 is a fixed point of @xmath9 , i.e. @xmath104 in this sense each letter of @xmath93 can be regarded as the image , or a factor of the image , of another letter of @xmath93 . in view of the definition of @xmath9 , cf . , each segment @xmath105 has the structure @xmath106 for @xmath107 and @xmath108 . the letters @xmath97 and @xmath98 are thus `` terminating symbols '' which cut @xmath93 to images of individual letters . this fact is particularly important when a factor @xmath4 of @xmath93 is given and one needs to find a factor @xmath109 of @xmath93 such that @xmath110 . it holds : [ jednozvzor ] let @xmath111 be a factor of @xmath93 such that @xmath112 and let one of the following conditions be satisfied : * the first letter of @xmath4 is @xmath98 , * @xmath4 has the prefix @xmath113 , * @xmath114 or @xmath115 is a factor of @xmath93 . then there is a unique factor @xmath109 of @xmath93 satisfying @xmath116 . + any of the conditions ( i ) , ( ii ) and ( iii ) together with @xmath112 ensures that @xmath111 is an image of certain factor @xmath109 , and it is obvious from the definition of @xmath9 that @xmath117 . [ pripustnevzor ] _ _ ( i)__let @xmath118 be a letter occuring in @xmath105 for a @xmath119 . then : * if @xmath120 , then @xmath121 and @xmath122 . * if @xmath123 , then either @xmath124 and @xmath125 , or @xmath126 and @xmath127 . _ _ ( ii)__if @xmath128 is a factor of @xmath93 and @xmath129 , @xmath108 , @xmath130 , then @xmath131 and @xmath132 . the following observation describes the possible neighbours of each of the letters @xmath102 in the word @xmath93 . [ pripustne ] the sequence of letters in the word @xmath93 conform to these rules : * each letter @xmath97 in @xmath93 is preceded by @xmath96 and followed either by @xmath96 or by @xmath98 . * each letter @xmath98 in @xmath93 is preceded by @xmath97 and followed by @xmath96 . _ ( i ) _ each @xmath97 is the last letter of @xmath105 for @xmath124 or @xmath126 according to observation [ pripustnevzor ] , i.e. it is the last letter of the block @xmath133 or @xmath134 , thus is preceded by @xmath96 . the letter @xmath97 is followed by the first letter of @xmath135 , which can be either @xmath96 or @xmath98 ( cf . the substitution rule @xmath9 ) . + _ ( ii ) _ each @xmath98 is equal to @xmath105 for @xmath121 . we already know from ( i ) that @xmath136 and @xmath137 , therefore the @xmath98 is preceded by the last letter of @xmath138 ( which is @xmath97 ) and followed by the first letter of @xmath138 or @xmath139 ( which is @xmath96 ) . in order to understand the structure of the word @xmath93 , it is useful to describe possible segments @xmath140 in factors of @xmath93 of the type @xmath141 and @xmath142 . this is done in the next two observations . [ szs ] let @xmath143 be a factor of @xmath93 such that @xmath144 . then one of the following equalities holds : * @xmath145 , * @xmath146 , * @xmath147 . observations [ jednozvzor ] and [ pripustnevzor ] imply that @xmath148 , where @xmath149 is a factor of @xmath93 the last letter of which is either @xmath96 or @xmath98 and which is preceded in @xmath93 by either @xmath96 or @xmath98 . since @xmath150 contains only one @xmath97 , namely its last letter , all letters of @xmath149 except the last one have to be different from @xmath96 and @xmath98 . therefore @xmath151 , where @xmath152 and @xmath153 . taking into account observation [ pripustne ] , we infer that @xmath154 or @xmath155 . therefore only four situations are possible : @xmath156 , @xmath157 , @xmath158 , @xmath159 . moreover , since @xmath149 is preceded by either @xmath96 or @xmath98 , it can not hold @xmath157 , cf . observation [ pripustne ] . therefore @xmath149 is equal to one of the factors @xmath96 , @xmath160 , @xmath161 , which implies that @xmath140 is equal to one of the factors @xmath113 , @xmath162 , @xmath163 . let @xmath142 be a factor of @xmath93 such that @xmath164 . then one of the following equalities holds : * @xmath165 , * @xmath166 , * @xmath167 . consequently , @xmath168 . it follows from observations [ jednozvzor ] and [ pripustnevzor ] that @xmath169 , where @xmath143 is a factor of @xmath93 such that @xmath144 . therefore @xmath170 and , according to observation [ szs ] , @xmath171 . many times we will need to compare the number of letters @xmath102 in a factor of @xmath93 and in its image . the substitution rule leads to the equalities @xmath172 which can be inverted subsequently : [ vzor ] for any factor @xmath4 of @xmath93 it holds @xmath173 balance bound with respect to the letter @xmath98 -------------------------------------------------------- we begin the proof of theorem [ balance ] by its second statement , i.e. we show at first that @xmath93 is 2-balanced with respect to the letter @xmath98 . as we will see , the determination of the balance bound with respect to the letter @xmath98 is by far the most complicated part of the work . [ balancem ] let @xmath4 , @xmath5 be factors of @xmath93 such that @xmath25 . then @xmath174 if we apply observations [ jednozvzor ] and [ pripustnevzor ] , eq . implies that the factor @xmath4 is an image of certain factor of @xmath93 whose first and last letters equal @xmath97 . this allows us to define a factor @xmath179 of @xmath93 in this way : @xmath181 the factor @xmath5 is not ready for a direct application of observation [ jednozvzor ] because of . for that reason we at first extend the factor @xmath5 to both sides up to the closest letter @xmath98 , i.e. we put @xmath182 where @xmath183 is a factor of @xmath93 and @xmath184 . now we can define a factor @xmath180 of @xmath93 by the relation @xmath185 let us show that the factor @xmath180 is shorter than @xmath4 and @xmath5 : [ g kratsi ] it holds @xmath186 . since @xmath187 , the factor @xmath4 contains at least 3 letters @xmath98 , and thus @xmath188 . observation [ mzm ] then implies that @xmath189 . since @xmath19 , it holds @xmath190 , hence necessarily @xmath191 according to observation [ mzm ] . but then @xmath192 and @xmath193 where @xmath194 and @xmath195 for @xmath196 . one more application of observation [ mzm ] gives @xmath197 , hence @xmath198 and necessarily @xmath199 , i.e. @xmath200 where @xmath201 . we deduce from eqs . , , from @xmath195 and from the minimality of @xmath85 that @xmath202 , @xmath196 . the factor @xmath183 ( cf . and ) is given by @xmath203 , hence equality together with observation [ pripustnevzor ] imply @xmath204 where @xmath205 , @xmath206 and @xmath207 . hence @xmath208 , and consequently , using observation [ szs ] , @xmath209 and @xmath210 . the sought inequality @xmath186 follows from these relations : @xmath211 @xmath212 . factors @xmath179 , @xmath180 satisfy latexmath:[\[\begin{gathered } both relations will be proved using the properties of @xmath4 , @xmath183 . let us begin with eq . . considering proposition [ vzor ] , we have @xmath214 . therefore , taking into account eq . , we obtain @xmath215 now we proceed to eq . . it holds @xmath216 for any factor @xmath140 ( cf . the substitution rule ) . the identity @xmath217 allows one to eliminate @xmath218 , hence @xmath219 , we have @xmath220 , equivalently @xmath221 . if we apply to the last inequality , we obtain @xmath222 now we substitute here from which leads to . there is another useful statement , namely proposition [ f - g<=0 ] , but to prove it we need one more observation which gives an estimate of the number of occurences of the letter @xmath98 in a factor of @xmath93 of a given length : [ obsahm ] _ _ ( i)__for every factor @xmath223 of @xmath93 it holds @xmath224 . + _ _ ( ii)__if moreover @xmath223 has a prefix or a suffix of the length greater or equal to @xmath225 that does not contain @xmath98 , it holds @xmath226 . \(i ) observation [ mzm ] says that if @xmath142 is a factor of @xmath93 , then @xmath227 . consequently @xmath228 . the inequality @xmath229 holds trivially . + ( ii ) let @xmath230 and @xmath231 then , employing the result of ( i ) , one has @xmath232 . [ f - g<=0 ] it holds @xmath233 , i.e. the factor @xmath179 is not longer than @xmath180 . let us suppose that the contrary is true , i.e. @xmath234 . let @xmath235 be the suffix of @xmath179 of the length @xmath236 and let @xmath237 . then it holds @xmath238 ( the inequality `` @xmath239 '' is valid due to observation [ g kratsi ] ) and @xmath240 . our goal is to estimate @xmath241 which is equal to @xmath242 . for this purpose an estimate for @xmath243 will be needed ; we obtain it from : @xmath244 . as for @xmath245 , since @xmath179 is followed by @xmath97 , the last letter of @xmath235 is @xmath96 ( cf . observation [ pripustne ] ) . with regard to this fact , observation [ obsahm ] implies @xmath246 . now we distinguish two cases : + @xmath247if @xmath248 , then @xmath249 , hence @xmath250 @xmath247if @xmath251 , then @xmath252 , hence @xmath253 we see that for any value of @xmath254 , the factors @xmath255 and @xmath180 of @xmath93 are of the same length less than @xmath85 and satisfy . this is a contradiction with the minimality of @xmath85 . + at this moment we define another pair of factors . since @xmath257 and @xmath258 are factors of @xmath93 , observation [ jednozvzor ] says that there exist factors @xmath109 and @xmath256 of @xmath93 such that @xmath259 and @xmath260 [ x , y kratsi ] it holds @xmath261 and @xmath262 , i.e. both factors @xmath109 , @xmath256 are shorter that @xmath4 , @xmath5 . we use propositions [ g kratsi ] and [ f - g<=0 ] : @xmath263 it holds latexmath:[\[\begin{gathered } the proof of both statements is straightforward using the definitions of @xmath109 and @xmath256 , proposition [ vzor ] , the identity @xmath265 holding for any factor @xmath4 , and equality . proposition [ vztah xy ] has an immediate corollary : latexmath:[\[\begin{gathered } if we substitute for @xmath243 from to , we obtain . equality and proposition then give . in what follows we split the proof according to the signum of @xmath267 , and we show that whatever the value of @xmath267 is , it always contradicts the minimality of @xmath85 . we set @xmath269 and denote @xmath270 , where @xmath271 is a factor of @xmath93 of the length @xmath236 such that @xmath272 is a factor of @xmath93 . then @xmath273 . moreover @xmath274 due to proposition [ x , y kratsi ] . our goal is to estimate @xmath275 which is equal to @xmath276 . the difference @xmath277 can be estimated using inequality : @xmath278 . let us proceed to @xmath279 . recall at first that @xmath280 ( cf . ) . thus the last letter of the factor @xmath256 is either @xmath96 or @xmath98 , which implies , with respect to observation [ pripustne ] , that the first letter of @xmath271 is either @xmath96 or @xmath97 , i.e. different from @xmath98 . observation [ obsahm ] then gives @xmath281 . now we distinguish two cases : + @xmath247if @xmath248 , then @xmath282 , hence @xmath283 ; + @xmath247if @xmath251 , then @xmath284 , hence @xmath285 the factors @xmath109 and @xmath286 are of the same length less than @xmath85 and for any @xmath254 satisfy @xmath287 . in other words , they contradict the minimality of @xmath85 . equality implies @xmath289 , and we know from proposition [ f - g<=0 ] that @xmath290 . therefore : + @xmath247if @xmath291 or @xmath292 , we have @xmath293 . since moreover @xmath294 , we have arrived at a contradiction with the minimality of @xmath85 . + @xmath247if @xmath295 and @xmath296 , equality gives @xmath297 . this , however , contradicts inequality . the last case to deal with is @xmath298 . since the situation @xmath299 is very complicated to study , we will start with the case @xmath300 and then deal with @xmath299 separately . let us set for simplicity @xmath301 . equality gives @xmath302 . we infer from here and from the minimality of @xmath85 that @xmath303 ( otherwise one could consider the prefix of @xmath179 of the length @xmath304 , denote it by @xmath305 , then @xmath306 , @xmath307 , which is a contradiction with the minimality of @xmath85 ) . similarly , we infer from @xmath308 and from the minimality of @xmath85 that @xmath309 ( otherwise we denote the prefix of @xmath256 of the length @xmath310 by @xmath311 , then @xmath312 , @xmath313 ) . equality then gives an upper bound on @xmath314 , namely @xmath315 . let us denote @xmath316 , where @xmath317 is a factor of @xmath93 and @xmath318 . then @xmath319 and @xmath238 , and it follows from that if @xmath320 , the factor @xmath235 can be chosen such that its first letter is @xmath97 , i.e. different from @xmath98 . now we are going to express @xmath241 which is equal to @xmath321 . let us distinguish the cases @xmath322 and @xmath323 . @xmath247if @xmath322 , we have @xmath324 or @xmath325 , hence @xmath326 . @xmath247let @xmath323 . note that at the same time it holds @xmath327 , thus necessarily @xmath328 . observation [ obsahm ] leads to @xmath329 then @xmath330 we conclude that for any @xmath331 it holds @xmath238 and @xmath332 . this is a contradiction with the minimality of @xmath85 . in this case , equality implies @xmath333 and equality implies @xmath334 . since @xmath335 , necessarily @xmath309 ( the contrary conradicts the minimality of @xmath85 ) , hence gives latexmath:[\[\label{-1fg } @xmath299 , can occur only for @xmath296 , because : * if @xmath15 was greater than 3 , inequality would contradict the inequality @xmath290 , derived in proposition [ f - g<=0 ] . * if @xmath15 was equal to 3 , inequality together with proposition [ f - g<=0 ] would give @xmath295 , which would not conform to . this allows us to restrict our considerations on the case @xmath296 . inequality implies @xmath337 , inequality implies @xmath338 , putting it together we infer @xmath339 . this allows us to use eq . to compute @xmath277 : we obtain @xmath340 . let us sum up the relations between @xmath109 and @xmath256 and between @xmath179 and @xmath180 : @xmath341 it holds : * the word @xmath342 is a factor of @xmath93 and at it holds @xmath343 , * the word @xmath344 is a factor of @xmath93 and at it holds @xmath345 . the proof will be done in ten steps . _ step 1 . _ ( possible prefixes and suffixes of @xmath4 , @xmath183 ) + the following four statements hold : * @xmath346 or @xmath347 , * @xmath348 or @xmath349 , * @xmath350 or @xmath351 , * @xmath352 or @xmath353 . we have found in the proof of observation [ g kratsi ] that @xmath354 where @xmath355 and @xmath202 for @xmath196 . taking into account observation [ mzm ] , we deduce that for @xmath296 , the prefix of @xmath183 and its suffix can be equal to either @xmath356 or @xmath357 , thus ( iii ) and ( iv ) are proved . statements ( i ) and ( ii ) say in fact , with regard to observation [ mzm ] , that @xmath4 can not have the segment @xmath357 as its prefix and suffix , respectively . let us suppose to the contrary that e.g. @xmath358 . we define @xmath359 , then @xmath360 . at the same time we denote @xmath361 , where @xmath362 is the prefix of @xmath5 of the length 9 ( thus @xmath363 ) . since @xmath364 , it follows from observation [ mzm ] that @xmath362 contains exactly @xmath27 letter @xmath98 , hence @xmath365 . therefore @xmath366 , which is a contradiction with the minimality of @xmath85 . _ step 2 . _ ( possible prefixes and suffixes of @xmath179 , @xmath180 ) + it holds * @xmath367 or @xmath368 , * @xmath369 or @xmath370 , * @xmath371 or @xmath372 , * @xmath373 or @xmath374 . it is a trivial consequence of step 1 and of the definitions of @xmath179 and @xmath180 . _ step 3 . _ ( the prefix and suffix of @xmath179 ) + it holds @xmath375 . we show that @xmath368 , the proof of @xmath370 would be similar . step 2 implies that @xmath376 or @xmath377 . let us suppose for a while that @xmath376 . we put @xmath378 ( it is a factor of @xmath93 , because @xmath379 is a factor of @xmath93 , cf . ) . then @xmath380 , @xmath381 . since @xmath382 according to step 2 , we are allowed to set @xmath383 . then it holds @xmath384 , @xmath385 . the factors @xmath255 and @xmath386 satisfy @xmath387 and @xmath388 , see . this is a contradiction with the minimality of @xmath85 . _ step 4 . _ ( the prefixes and suffixes of @xmath4 , @xmath183 and @xmath180 ) + it holds * @xmath389 , * @xmath390 , * @xmath391 . statement ( i ) is a straightforward consequence of step 3 . statement ( ii ) follows from ( i ) , which can be proven by contradiction using similar ideas as in the proof in step 1 . statement ( iii ) is a consequence of ( ii ) . _ ( the prefixes and suffixes of @xmath109 and @xmath256 ) + it holds * @xmath392 or @xmath393 , * @xmath394 . _ ( i ) _ since @xmath395 and @xmath396 according to step 3 , obviously @xmath392 or @xmath393 , cf . observation for @xmath296 . + _ ( ii ) _ it follows from step 4 that @xmath397 . observation [ pripustne ] implies @xmath398 , observation [ szs ] gives @xmath399 . _ step 6 . _ ( two letters after @xmath256 ) + the word @xmath400 is a factor of @xmath93 . it is a consequence of step 5 and observation [ szs ] . _ step 7 . _ ( the suffix of @xmath109 ) + it holds @xmath401 . we consider the result of step 5 and prove that @xmath402 contradicts the minimality of @xmath85 . let @xmath402 . observation [ mzm ] then implies that @xmath109 is followed either by @xmath161 or by @xmath403 . in the first case we define @xmath404 . taking into account , we see that the pair @xmath405 , @xmath256 contradicts the minimality of @xmath85 . in the case when @xmath109 is followed by the group @xmath403 , we define @xmath406 where @xmath407 is the prefix of @xmath109 of the length 2 ( i.e. @xmath408 or @xmath409 ) . also we define @xmath410 ( this is allowed due to step 6 ) . the pair @xmath405 , @xmath286 now contradicts the minimality of @xmath85 with regard to . _ step 8 . _ ( the letter before @xmath109 ) + the word @xmath342 is a factor of @xmath93 . since @xmath257 is a factor of @xmath93 and @xmath411 , it follows from observation [ pripustnevzor ] that @xmath342 or @xmath412 is a factor of @xmath93 . however , if @xmath412 is a factor of @xmath93 , then the factors @xmath412 and @xmath256 contradict the minimality of @xmath85 with regard to . _ step 9 . _ ( the prefix of @xmath109 ) + it holds @xmath413 . since @xmath342 is a factor of @xmath93 according to step 8 and @xmath414 according to step 5 , the word @xmath415 is a factor of @xmath93 . observation [ pripustnevzor ] then gives @xmath416 . _ step 10 . _ ( two letters before @xmath256 ) + the word @xmath344 is a factor of @xmath93 . step 5 and observation [ szs ] imply that @xmath417 or @xmath344 is a factor of @xmath93 . let us suppose for a while that @xmath418 is a factor of @xmath93 . observation [ mzm ] implies immediately that @xmath419 . we introduce the word @xmath420 which is a factor of @xmath93 due to step 6 . then we define @xmath421 , this is a factor of @xmath93 due to step 9 . it follows from that the factors @xmath405 , @xmath286 satisfy @xmath422 , @xmath423 , i.e. they contradict the minimality of @xmath85 . this finishes the proof of proposition [ slozite ] . the statement follows from the results of steps 5 , 7 , 8 , 9 , 10 . proposition [ slozite ] together with observation [ jednozvzor ] allow one to define factors @xmath424 and @xmath425 of @xmath93 such that @xmath426 it is obvious that @xmath427 , @xmath428 , hence @xmath429 , @xmath430 , cf . proposition [ x , y kratsi ] . proposition [ vzor ] and relations , enable us to compute @xmath431 , @xmath432 and also @xmath433 : @xmath434 + 2(|x|_s-|y|_s)=\\ = -(|x|-|y|)+3(|x|_s-|y|_s)+(|x|_m-|y|_m)=\\ = -(|x|-|y|)+3(|fs|_m-|gs|_m)+(|x|_m-|y|_m)=-(-1)+3\cdot(-2)+2=-3\,.\end{gathered}\ ] ] the pair @xmath424 , @xmath425 thus satisfy @xmath435 and @xmath436 . this is a contradiction with the minimality of @xmath85 . the proof of theorem [ balancem ] is completed . once we know that @xmath93 is @xmath30-balanced with respect to the letter @xmath98 , it is easy to prove that it is @xmath30-balanced with respect to the letter @xmath97 as well . [ balances ] let @xmath4 , @xmath5 be factors of @xmath93 such that @xmath25 . then @xmath437 we will proceed by contradiction . let us assume that there exist factors @xmath438 , @xmath16 of @xmath93 such that @xmath25 and @xmath439 . obviously one can suppose that @xmath440 . then , with regard to the substitution rule , one has @xmath441 , @xmath442 , @xmath443 since the factors @xmath4 and @xmath5 are of the same length , they satisfy @xmath444 with regard to theorem [ balancem ] , hence @xmath445 furthermore , since @xmath446 and @xmath447 , one has @xmath448 let @xmath449 be a prefix of @xmath450 of the length @xmath451 . then @xmath452 , hence @xmath453 and since it holds @xmath454 , we have arrived at a contradiction with theorem [ balancem ] . this section is devoted to the proof of the third part of theorem [ balance ] . as we will see , we will use both facts proved in the previous two sections . [ balancel ] let @xmath4 , @xmath5 be factors of @xmath93 such that @xmath25 . then @xmath455 we will again proceed by contradiction . let us assume that there exist factors @xmath438 , @xmath16 of @xmath93 such that @xmath175 and @xmath456 , @xmath177 . the minimality of @xmath85 implies @xmath457 gives @xmath458 and since @xmath459 and @xmath444 by virtue of theorems [ balancem ] and [ balances ] , respectively , we infer @xmath460 note that since @xmath461 , observation [ pripustnevzor ] implies that there are @xmath462 and @xmath463 such that @xmath464 is a factor of @xmath93 . we employ these @xmath43 and @xmath465 and define factors @xmath466 and @xmath183 of @xmath93 in the following way : @xmath467 now we can apply observation [ jednozvzor ] which establishes the existence of factors @xmath109 and @xmath256 of @xmath93 satisfying @xmath468 obviously @xmath469 . we are going to compute @xmath470 and @xmath267 . for that purpose the following relations will be useful : @xmath471 employing these relations and propositon [ vzor ] , one can derive @xmath472 and @xmath473 since @xmath109 is shorter than @xmath256 , we consider the prefix of the factor @xmath256 of the length @xmath310 and denote it by @xmath286 ; it obviously holds @xmath474 . therefore @xmath475 in other words , the factors @xmath109 and @xmath286 contradict the minimality of @xmath85 . to complete the proof of theorem [ balance ] , we have to show that the bounds 3 , 2 and 2 corresponding to @xmath96 , @xmath97 and @xmath98 , respectively , are optimal . to demonstrate this fact it suffices to find three pairs of factors of @xmath93 , let us denote them by @xmath476 , @xmath477 and @xmath478 , such that @xmath479 there are many possibilities , one can take for example : * @xmath480 , @xmath481 + then @xmath482 , @xmath483 . * @xmath484 , @xmath485 + then @xmath486 , @xmath487 . * @xmath488 , @xmath489 + then @xmath490 , @xmath491 . in this section we will determine the optimal bound for the abelian complexity function @xmath94 of the infinite word @xmath93 , taking advantage of the results on optimal balance bounds that have been derived in the previous part of the paper . the set of factors of @xmath93 of the length @xmath85 will be denoted by @xmath84 , and the symbol @xmath88 will stand for the set of corresponding parikh vectors , i.e. @xmath492 . [ mozpar ] for each @xmath493 there are numbers @xmath494 and @xmath495 such that @xmath496 , where @xmath497 consequently , @xmath498 . for each @xmath493 , theorem [ balances ] implies that @xmath499 thus there is an @xmath494 such that @xmath500 similarly , using theorem [ balancem ] , one finds that there is an @xmath495 such that @xmath501 since @xmath502 , we deduce that the set @xmath88 is a subset of @xmath503 , where @xmath504 , @xmath505 , are given by . proposition [ mozpar ] gives an upper bound of @xmath94 , namely @xmath498 for every @xmath493 , but we can even say more . indeed , theorem [ balancel ] implies that @xmath88 can not contain at the same time @xmath506 and @xmath507 , hence @xmath508 . in fact , the optimal bound is even lower , as we will show with the help of the following proposition . [ vylucujici ] there is no pair of factors @xmath4 , @xmath5 of @xmath93 such that their parikh vectors satisfy @xmath509 . we prove the statement by contradiction . let us suppose that there are factors @xmath4 and @xmath5 of @xmath93 satisfying latexmath:[\[\label{aclsm } factors @xmath4 and @xmath5 are obviously of the same length which we denote by @xmath85 . if @xmath511 and @xmath177 , we may assume without loss of generality that @xmath512 ( if e.g. @xmath513 , we can replace the pair @xmath514 by @xmath515 ) . having this assumption and eqs . , we deduce @xmath516 because if e.g. @xmath517 , then the pair @xmath518 and @xmath519 contradicts theorem [ balancel ] . similarly , theorem [ balances ] implies that @xmath520 let @xmath521 and @xmath522 be such numbers and letters that the word @xmath523 is a factor of @xmath93 . it is a trivial fact that such @xmath524 , @xmath525 exist . moreover , observation [ pripustne ] implies @xmath131 , observation [ pripustnevzor ] gives @xmath526 . we take these @xmath524 and @xmath118 and define a factor @xmath466 of @xmath93 , and then define also a factor @xmath183 of @xmath93 : @xmath527 now we apply observation [ jednozvzor ] which says that there are factors @xmath109 and @xmath256 of @xmath93 such that @xmath528 let us write down the following relations between @xmath466 and @xmath4 and between @xmath183 and @xmath5 , @xmath529 and use them to express @xmath277 , @xmath530 ( note that @xmath531 ) , and also to express @xmath267 : @xmath532 _ statement . _ it holds * @xmath533 , * @xmath534 . both ( i ) and ( ii ) can be proved by contradiction : _ ( i ) _ suppose that e.g. the first letter of @xmath256 is @xmath535 . we define @xmath536 , then @xmath273 and @xmath537 , which is a contradiction with theorem [ balancem ] . similarly we prove that the last letter of @xmath256 is @xmath98 . _ ( ii ) _ suppose that @xmath538 . we know from that also @xmath539 , hence @xmath540 . then gives @xmath541 . let @xmath542 be the last letter of @xmath256 and @xmath543 . then it holds @xmath544 and @xmath545 , which is a contradiction with theorem [ balancem ] . statements ( i ) and ( ii ) can be used to determine the first three letters of @xmath5 : @xmath546 however , observation [ szs ] implies that no factor of @xmath93 can contain the segment @xmath547 . this is a contradiction , thus the proposition is proved . _ ( i ) _ let @xmath93 be the fixed point of the substitution @xmath9 defined in . then its abelian complexity satisfies @xmath548 for all @xmath493 and for all @xmath13 . ( ii ) _ the bound @xmath549 is optimal , i.e. it can not be improved . moreover , for any @xmath13 there is infinitely many numbers @xmath493 such that @xmath550 . _ ( i ) _ since @xmath551 and @xmath552 , proposition [ vylucujici ] implies that the set @xmath88 can contain at most one of the vectors @xmath506 , @xmath553 and at most one of the vectors @xmath554 , @xmath507 . therefore @xmath88 has at most 7 elements , hence @xmath548 . _ ( ii ) _ we prove ( ii ) by showing that there are infinitely many values @xmath493 with this property : there are @xmath549 factors of @xmath93 of the length @xmath85 such that their parikh vectors are mutually different . for all @xmath555 , we define two auxiliary words : @xmath556 _ statement . _ for each @xmath555 it holds : _ ( i ) _ @xmath557 and @xmath558 are factors of @xmath93 , _ ( ii ) _ @xmath559 and @xmath560 are factors of @xmath93 , _ ( iii ) _ @xmath557 has the suffix @xmath561 and @xmath559 has the suffix @xmath562 , _ ( iv ) _ @xmath563 . to see ( i ) , we observe that @xmath557 is a prefix of @xmath564 , thus obviously a factor of @xmath93 . moreover , since @xmath565 is a factor of @xmath93 ( cf . observation [ mzm ] ) , the word @xmath566 is a factor of @xmath93 as well . then it suffices to show that @xmath567 has the suffix @xmath562 for each @xmath555 , which can be done easily by induction . the proof of ( ii ) is similar : since @xmath568 is a factor of @xmath569 , @xmath559 is a factor of @xmath93 . to demonstrate that @xmath560 is a factor of @xmath93 , it suffices to realize that @xmath561 is a suffix of @xmath570 for all @xmath555 , which can be shown by induction . let us proceed to ( iii ) . by the definitions above , @xmath557 has the suffix @xmath570 , @xmath559 has the suffix @xmath571 . now we can use the facts known from ( i ) and ( ii ) , namely that @xmath572 and @xmath573 for all @xmath555 . the statement ( iv ) is a consequence of these two equalities : @xmath574 for each @xmath555 , the auxiliary factors @xmath557 and @xmath559 are intrumental to define another set of words that we denote @xmath575 : @xmath576 it follows from statement above that all the words @xmath575 are factors of @xmath93 of the same length @xmath577 . if we compare the numbers of letters @xmath96 , @xmath97 and @xmath98 in these factors , we find that their parikh vectors and mutually different . this means that @xmath550 for every @xmath578 , where @xmath555 , i.e. the function @xmath579 attains the value @xmath549 infinitely many times . we finish this section by two statements describing the lower bound and the range of @xmath94 . it holds @xmath580 for all @xmath493 and for all @xmath13 . at first we show that for any prefix @xmath4 of @xmath93 , both @xmath581 and @xmath582 are factors of @xmath93 . this will be obvious from these three simple facts : + _ ( i ) _ since @xmath583 is a factor of @xmath93 , both @xmath584 and @xmath585 are factors of @xmath93 for any @xmath555 . + _ ( ii ) _ it holds @xmath586 and @xmath587 for any @xmath555 . + _ ( iii ) _ there is an @xmath555 such that @xmath4 is a prefix of @xmath571 , and thus of @xmath570 . statements ( i ) and ( ii ) imply that both @xmath588 and @xmath589 are factors of @xmath93 , and ( iii ) then implies that @xmath582 and @xmath581 are factors of @xmath93 as well . let now @xmath590 be the prefix of @xmath93 of the length @xmath85 and let @xmath591 be its parikh vector . we will show that then there are factors @xmath466 and @xmath592 of @xmath93 such that @xmath593 and @xmath594 . we distinguish three cases according to @xmath595 : + @xmath247 if @xmath596 , then we set @xmath597 , @xmath598 . + @xmath247 if @xmath599 , it holds necessarily @xmath600 ( cf . observation [ pripustne ] ) . in this case we set @xmath601 , @xmath602 . + @xmath247 if @xmath603 , it holds necessarily @xmath604 ( cf . again observation [ pripustne ] ) . we set @xmath597 , @xmath605 . it can be demonstrated that for each value @xmath606 there is an @xmath493 such that @xmath607 , but we omit the proof with regard to the length of the paper . we have studied balance properties and the abelian complexity of a certain class of infinite ternary words . we have found the optimal @xmath2 such that these words are @xmath2-balanced , and also the optimal bound for their abelian complexity functions . we have introduced a new notion , namely the property that a word is `` @xmath2-balanced with respect to the letter @xmath7 '' , which helped us to proceed more effectively from the knowledge of the balance properties to the estimate on @xmath94 . the class of words studied in this paper has one parameter @xmath13 . however , it emerged from our calculations that all the three optimal bounds for balances with respect to particular letters , as well as the optimal bound for the abelian complexity , are independent of the value of @xmath15 . the problem has one more aspect . so far the subword complexity and the abelian complexity are considered as highly independent of each other ( cf . e.g. the work of richome , saari and zamboni ) . however , our result can indicate that there are connections between them , for the present waiting for their discovery . it has been recently shown in @xcite that a fixed point of a canonical substitution associated with a non - simple cubic parry number has affine factor complexity if and only if it belongs just to the class with which we have dealt in this paper . therefore , briefly speaking , `` if the factor complexity is affine , then the abelian complexity has the optimal bound @xmath549 '' holds in the cubic non - simple parry case . we remark that this sort of statement holds as well in the quadratic non - simple parry case , although there it is a trivial fact . the author is grateful to k. binda for performing a numerical experiment and to e. pelantov for helpful discussions on various aspects of the problem . balkov , e. pelantov , o. turek , combinatorial and arithmetical properties of infinite words associated with quadratic non - simple parry numbers . _ rairo _ : theoret . informatics appl . * 41 * 3 ( 2007 ) 307328 .
a word @xmath0 defined over an alphabet @xmath1 is @xmath2-balanced ( @xmath3 ) if for all pairs of factors @xmath4 , @xmath5 of @xmath0 of the same length and for all letters @xmath6 , the difference between the number of letters @xmath7 in @xmath4 and @xmath5 is less or equal to @xmath2 . in this paper we consider a ternary alphabet @xmath8 and a class of substitutions @xmath9 defined by @xmath10 , @xmath11 , @xmath12 where @xmath13 . we prove that the fixed point of @xmath9 , formally written as @xmath14 , is 3-balanced and that its abelian complexity is bounded above by the value 7 , regardless of the value of @xmath15 . we also show that both these bounds are optimal , i.e. they can not be improved . laboratory of physics , kochi university of technology + tosa yamada , kochi 782 - 8502 , japan + email : ` [email protected] `
inspection of bridges , tunnels , wind turbines , and other large civil engineering structures for defects is a time - consuming , costly , and potentially dangerous task . in the future , _ smart coating _ technology , or _ smart paint _ , could do the job more efficiently and without putting people in danger . the idea behind smart coating is to form a thin layer of a specific substance on an object which then makes it possible to measure a condition of the surface ( such as temperature or cracks ) at any location , without direct access to the location . the concept of smart coating already occurs in nature , such as proteins closing wounds , antibodies surrounding bacteria , or ants surrounding food to transport it to their nest . these diverse examples suggest a broad range of applications of smart coating technology in the future , including repairing cracks or monitoring tension on bridges , repairing space craft , fixing leaks in a nuclear reactor , or stopping internal bleeding . we continue the study of coating problems in the context of self - organizing programmable matter consisting of simple computational elements , called particles , that can establish and release bonds and can actively move in a self - organized way using the geometric version of the amoebot model presented in @xcite . in doing so , we proceed to investigate the runtime analysis of our universal coating algorithm , introduced in @xcite . we first show that coating problems do not only have a ( trivial ) linear lower bound on the runtime , but that there is also a linear lower bound on the competitive gap between the runtime of fully local coating algorithms and coating algorithms that rely on global information . we then investigate the worst - case time complexity of our universal coating algorithm and show that it terminates within a linear number of rounds with high probability ( w.h.p . ) , where @xmath0 is the number of particles in the system and @xmath1 is a constant . ] , which implies that our algorithm is optimal in terms of worst - case runtime and also in a competitive sense . moreover , our simulation results show that in practice the competitive ratio of our algorithm is often better than linear . in the _ geometric amoebot model _ , we consider the graph @xmath2 , where @xmath3 is the infinite regular triangular grid graph . each vertex in @xmath4 is a position that can be occupied by at most one particle ( see figure [ fig : graph_handover](a ) ) . each particle occupies either a single node or a pair of adjacent nodes in @xmath2 . any structure a particle system can form can be represented as a subgraph of @xmath2 . two particles occupying adjacent nodes are _ connected _ by a _ bond _ , and we refer to such particles as _ neighbors_. the bonds do not only ensure that the particles form a connected structure but they are also used for exchanging information as explained below . particles move by executing a series of _ expansions _ and _ contractions_. a particle which occupies one node is _ contracted _ and can expand to an unoccupied adjacent node to occupy two nodes . if it occupies two nodes it is _ expanded _ and can contract to occupy a single node . in figure [ fig : graph_handover](b ) , we illustrate a set of expanded and contracted particles on the underlying graph @xmath2 . for an expanded particle , we denote the node the particle last expanded into as the _ head _ of the particle and the other occupied node as its _ tail_. for a contracted particle , the single node occupied by the particle is both its head and its tail . , where nodes of @xmath2 are shown as black circles . ( b ) shows five particles on @xmath2 ; the underlying graph @xmath2 is depicted as a gray mesh ; a contracted particle is depicted as a single black circle and an expanded particle is depicted as two black circles connected by an edge . ( c ) depicts the resulting configuration after a handover was performed by particles @xmath5 and @xmath6 in ( b ) . ] [ fig : graph_handover ] to stay connected as they move , neighboring particles coordinate their motion in a _ handover _ , which can occur in two ways . a contracted particle @xmath5 can initiate a handover by expanding into a node occupied by an expanded neighbor @xmath6 , `` pushing '' @xmath6 and forcing it to contract . alternatively , an expanded particle @xmath6 can initiate a handover by contracting `` pulling '' a contracted neighbor @xmath5 to the node it is vacating , thereby forcing @xmath5 to expand . figures [ fig : graph_handover](b ) and [ fig : graph_handover](c ) illustrate two particles labeled @xmath5 and @xmath6 performing a handover . particles are _ anonymous _ but each has a collection of uniquely labeled _ ports _ corresponding to the edges incident to the nodes the particle occupies . bonds between adjacent particles are formed through ports that face each other . the particles are assumed to have a common _ chirality _ , meaning they share the same notion of _ clockwise ( cw ) direction_. this allows each particle @xmath5 to label its ports counting in the clockwise direction ; without loss of generality , we assume each particle labels its head and tail ports from @xmath7 to @xmath8 . however , particles may have different offsets of the labelings , so they do not share a common sense of orientation . each particle hosts a local memory of constant size for which any neighboring particle has read and write access . with this mechanism particles can communicate by writing into each other s memory . the _ configuration _ @xmath9 of the system at the beginning of time @xmath10 consists of the nodes in @xmath2 occupied by the object and the set of particles , and for each particle @xmath5 , @xmath9 contains the current state of @xmath5 , including whether it is expanded or contracted , its port labeling , and the contents of its local memory . following the standard asynchronous model of computation @xcite , we assume that the system progresses through atomic _ activations _ of individual particles . at each ( atomic ) activation a particle can perform at most one movement and an arbitrary bounded amount of computation involving its local memory and the shared memories of its neighbors . a classical result under this model is that for any asynchronous execution of atomic particle activations , we can organize these activations sequentially still producing the same end configuration @xcite . we count ( asynchronous ) time in terms of the number of activations . a _ round _ is over once each particle has been activated at least once . we assume the activation sequence to be _ fair _ , i.e. , for each particle @xmath5 and any point in time @xmath10 , @xmath5 will eventually be activated at some time @xmath11 . in the _ universal coating problem _ we consider an instance @xmath12 where @xmath13 represents the particle system and @xmath14 represents the fixed object to be coated . let @xmath15 be the number of particles in the system , @xmath16 be the set of nodes occupied by @xmath13 , and @xmath17 be the set of nodes occupied by @xmath14 ( when clear from the context , we may omit the @xmath18 notation ) . for any two nodes @xmath19 , the _ distance _ @xmath20 between @xmath21 and @xmath22 is the length of the shortest path in @xmath2 from @xmath21 to @xmath22 . the distance @xmath23 between a @xmath24 and @xmath25 is defined as @xmath26 . define _ layer @xmath27 _ to be the set of nodes that have a distance @xmath27 to the object , and let @xmath28 be the number of nodes in layer @xmath27 . an instance is _ valid _ if the following properties hold : 1 . the particles are all contracted and are initially in the _ idle _ state . the subgraphs of @xmath2 induced by @xmath17 and @xmath29 , respectively , are connected , i.e. , there is a single object and the particle system is connected to the object . the subgraph of @xmath2 induced by @xmath30 is connected , i.e. , the object @xmath14 has no holes . does contain holes , we consider the subset of particles in each connected region of @xmath30 separately . ] @xmath30 is @xmath31-connected , i.e. , @xmath14 can not form _ tunnels _ of width less than @xmath31 . note that a width of at least @xmath32 is needed to guarantee that the object can be evenly coated . the coating of narrow tunnels requires specific technical mechanisms that complicate the protocol without contributing to the basic idea of coating , so we ignore such cases in favor of simplicity . a configuration @xmath9 is _ legal _ if and only if all particles are contracted and @xmath33 meaning that all particles are as close to the object as possible or _ coat @xmath14 as evenly as possible_. a configuration @xmath9 is said to be _ stable _ if no particle in @xmath9 ever performs a state change or movement . an algorithm _ solves _ the universal coating problem if , starting from any valid instance , it reaches a _ stable legal configuration _ in a finite number of rounds . many approaches have been proposed with potential applications in smart coating ; these can be categorized as active and passive systems . in passive systems , particles move based only on their structural properties and interactions with their environment , or have only limited computational ability and lack control of their motion . examples include dna self - assembly systems ( see , e.g. , the surveys in @xcite ) , population protocols @xcite , and slime molds @xcite . our focus is on active systems , in which computational particles control their actions and motions to complete specific tasks . coating has been extensively studied in the area of _ swarm robotics _ , but not commonly treated as a stand - alone problem ; it is instead examined as part of _ collective transport _ ( e.g. , @xcite ) or _ collective perception _ ( e.g. , see respective section of @xcite ) . some research focuses on coating objects as an independent task under the name of _ target surrounding _ or _ boundary coverage_. the techniques used in this context include stochastic robot behaviors @xcite , rule - based control mechanisms @xcite and potential field - based approaches @xcite . while the analytic techniques developed in swarm robotics are somewhat relevant to this work , those systems have more computational power and movement capabilities than those studied in this work . michail and spirakis recently proposed a model @xcite for network construction inspired by population protocols @xcite . the population protocol model is related to self - organizing particle systems but is different in that agents ( corresponding to our particles ) can move freely in space and establish connections at any time . it would , however , be possible to adapt their approach to study coating problems under the population protocol model . in @xcite we presented our universal coating algorithm and proved its correctness . we also showed it to be worst - case work - optimal , where work is measured in terms of number of particle movements . in this paper we continue the analysis of the _ universal coating algorithm _ introduced in @xcite . as our main contribution in this paper , we investigate the runtime of our algorithm and prove that our algorithm terminates within a _ linear number of rounds _ with high probability . we also present a matching linear lower bound for local - control coating algorithms that holds with high probability . we use this lower bound to show a _ linear lower bound on the competitive gap _ between fully local coating algorithms and coating algorithms that rely on global information , which implies that our algorithm is also optimal in a competitive sense . we then present some simulation results demonstrating that in practice the competitive ratio of our algorithm is often much better than linear . in section [ sec : algo ] , we again present the algorithm introduced in @xcite . we present a comprehensive formal runtime analysis of our algorithm , by first presenting some lower bounds on the competitive ratio of any local - control algorithm in section [ sec : performance ] , and then proving that our algorithm has a runtime of @xmath34 rounds w.h.p . in section [ sec : wcruntime ] , which matches our lower bounds . in this section , we summarize the universal coating algorithm introduced in @xcite ( see @xcite for a detailed description ) . this algorithm is constructed by combining a number of asynchronous primitives , which are integrated seamlessly without any underlying synchronization . the _ spanning forest _ primitive organizes the particles into a spanning forest , which determines the movement of particles while preserving system connectivity ; the _ complaint - based coating _ primitive coats the first layer by bringing any particles not yet touching the object into the first layer while there is still room ; the _ general layering _ primitive allows each layer @xmath27 to form only after layer @xmath35 has been completed , for @xmath36 ; and the _ node - based leader election _ primitive elects a node in layer 1 whose occupant becomes the leader particle , which is used to trigger the general layering process for higher layers . we define the set of _ states _ that a particle can be in as _ idle _ , _ follower _ , _ root _ , and _ retired_. in addition to its state , a particle maintains a constant number of other flags , which in our context are constant size pieces of information visible to neighboring particles . a flag @xmath37 owned by some particle @xmath5 is denoted by @xmath38 . recall that a _ layer _ is the set of nodes @xmath39 that are equidistant to the object @xmath14 . a particle keeps track of its current layer number in @xmath40 . in order to respect the constant - size memory constraint of particles , we take all layer numbers modulo @xmath41 . each root particle @xmath5 has a flag @xmath42 which stores a port label pointing to a node of the object if @xmath43 , and to an occupied node adjacent to its head in layer @xmath44 if @xmath45 . we now describe the coating primitives in more detail . the * _ spanning forest primitive _ * ( algorithm [ alg : spanningforestalgorithm ] ) organizes the particles into a spanning forest @xmath46 , which yields a straightforward mechanism for particles to move while preserving connectivity ( see @xcite for details ) . initially , all particles are _ idle_. a particle @xmath5 touching the object changes its state to _ root_. for any other idle particle @xmath5 , if @xmath5 has a root or a follower in its neighborhood , it stores the direction to one of them in @xmath47 , changes its state to _ follower _ , and generates a complaint flag ; otherwise , it remains idle . a follower particle @xmath5 uses handovers to follow its parent and updates the direction @xmath47 as it moves in order to maintain the same parent in the tree ( note that the particular particle at @xmath47 may change due to @xmath5 s parent performing a handover with another of its children ) . in this way , the trees formed by the parent relations stay connected , occupy only the nodes they covered before , and do not mix with other trees . a root particle @xmath5 uses the flag @xmath48 to determine its movement direction . as @xmath5 moves , it updates @xmath48 so that it always points to the next position of a clockwise movement around the object . for any particle @xmath5 , we call the particle occupying the position that @xmath47 resp . @xmath48 points to the _ predecessor _ of @xmath5 . if a root particle does not have a predecessor , we call it a _ super - root_. a particle @xmath5 acts depending on its state as described below : + lx * idle * : & if @xmath5 is adjacent to the object @xmath14 , it becomes a _ root _ particle , makes the current node it occupies a _ leader candidate position _ , and starts running the leader election algorithm . if @xmath5 is adjacent to a _ retired _ particle , @xmath5 also becomes a _ root _ particle . if a neighbor @xmath49 is a root or a follower , @xmath5 sets the flag @xmath47 to the label of the port to @xmath49 , puts a _ complaint flag _ in its local memory , and becomes a _ follower_. if none of the above applies , @xmath5 remains idle . + * follower * : & if @xmath5 is contracted and adjacent to a retired particle or to @xmath14 , then @xmath5 becomes a _ root _ particle . if @xmath5 is contracted and has an expanded parent , then @xmath5 initiates handover@xmath50 ( algorithm [ alg : handover ] ) ; otherwise , if @xmath5 is expanded , it considers the following two cases : @xmath51 if @xmath5 has a contracted child particle @xmath6 , then @xmath5 initiates handover@xmath50 ; @xmath52 if @xmath5 has no children and no idle neighbor , then @xmath5 contracts . finally , if @xmath5 is contracted , it runs the function forwardcomplaint@xmath53 ( algorithm [ alg : complaint ] ) . + * root * : & if particle @xmath5 is in layer 1 , @xmath5 participates in the leader election process . if @xmath5 is contracted , it first executes markerretiredconditions@xmath50 ( algorithm [ alg : retiredcondition ] ) and becomes _ retired _ , and possibly also a _ marker _ , accordingly . if @xmath5 does not become retired , then if it has an expanded root in @xmath48 , it initiates handover@xmath50 ; otherwise , @xmath5 calls layerextension@xmath50 ( algorithm [ alg : boundarydirectionalgorithm ] ) . if @xmath5 is expanded , it considers the following two cases : @xmath51 if @xmath5 has a contracted child , then @xmath5 initiates handover@xmath50 ; @xmath52 if @xmath5 has no children and no idle neighbor , then @xmath5 contracts . finally , if @xmath5 is contracted , it runs forwardcomplaint@xmath54 . + * retired * : & @xmath5 clears a potential complaint flag from its memory and performs no further action . the * _ complaint - based coating primitive _ * is used for the coating of the first layer . each time a particle @xmath5 holding at least one complaint flag is activated , it forwards one to its predecessor as long as that predecessor holds less than two complaint flags . we allow each particle to hold up to two complaint flags to ensure that a constant size memory is sufficient for storing the complaint flags and so the flags quickly move forward to the super - roots . a contracted super - root @xmath5 only expands to @xmath48 if it holds at least one complaint flag , and when it expands , it consumes one of these complaint flags . all other roots @xmath5 move towards @xmath48 whenever possible ( i.e. , no complaint flags are required ) by performing a handover with their predecessor ( which must be another root ) or a successor ( which is a root or follower of its tree ) , with preference given to a follower so that additional particles enter layer 1 . as we will see , these rules ensure that whenever there are particles in the system that are not yet at layer 1 , eventually one of these particles will move to layer 1 , unless layer 1 is already completely filled with contracted particles . @xmath5 initiates a handover with particle @xmath6 @xmath5 initiates a handover with particle @xmath6 @xmath5 initiates a handover with particle @xmath6 @xmath5 forwards one complaint flag to @xmath6 the * _ leader election primitive _ * runs during the complaint - based coating primitive to elect a node in layer 1 as the leader position . this primitive is similar to the algorithm presented in @xcite with the difference that leader candidates are nodes instead of static particles ( which is important because in our case particles may still move during the leader election primitive ) . the primitive only terminates once all positions in layer 1 are occupied . once the leader position is determined , all positions in layer 1 are filled by contracted particles and whatever particle currently occupies that position becomes the _ leader_. this leader becomes a marker particle , marking a neighboring position in the next layer as a _ marked position _ which determines a starting point for layer 2 , and becomes _ retired_. once a contracted root @xmath5 has a retired particle in the direction @xmath48 , it retires as well , which causes the particles in layer 1 to become retired in counter - clockwise order . at this point , the general layering primitive becomes active , which builds subsequent layers until there are no longer followers in the system . if the leader election primitive does not terminate ( which only happens if @xmath55 and layer 1 is never completely filled ) , then the complaint flags ensure that the super - roots eventually stop , which eventually results in a stable legal coating . the layer number of any node occupied by the object is equal to 0 . let @xmath6 be any neighbor of @xmath5 with smallest layer number ( modulo @xmath41 ) . @xmath56 @xmath57 @xmath58 @xmath59 * extending layer @xmath40 * @xmath5 expands in direction @xmath48 @xmath5 consumes a complaint flag , if it holds one in the * _ general layering primitive _ * , whenever a follower is adjacent to a retired particle , it becomes a root . root particles continue to move along positions of their layer in a clockwise ( if the layer number is odd ) or counter - clockwise ( if the layer number is even ) direction until they reach either the marked position of that layer , a retired particle in that layer , or an empty position of the previous layer ( which causes them to change direction ) . complaint flags are no longer needed to expand into empty positions . followers follow their parents as before . a contracted root particle @xmath5 may retire if : ( i ) it occupies the marked position and the marker particle in the lower layer tells it that all particles in that layer are retired ( which it can determine locally ) , or ( ii ) it has a retired particle in the direction @xmath48 . once a particle at a marked position retires , it becomes a marker particle and marks a neighboring position in the next layer as a marked position . @xmath60 , @xmath61 @xmath62 @xmath63 @xmath64 @xmath65 @xmath5 becomes a _ retired _ particle @xmath5 becomes a _ retired _ particle @xmath5 becomes a _ retired _ particle recall that a _ round _ is over once every particle in @xmath13 has been activated at least once . the _ runtime _ @xmath66 of a coating algorithm @xmath67 is defined as the worst - case number of rounds ( over all sequences of particle activations ) required for @xmath67 to solve the coating problem @xmath12 . certainly , there are instances @xmath12 where every coating algorithm has a runtime of @xmath68 ( see lemma [ lem : lowerbound ] ) , though there are also many other instances where the coating problem can be solved much faster . since a worst - case runtime of @xmath68 is fairly large and therefore not very helpful to distinguish between different coating algorithms , we intend to study the runtime of coating algorithms relative to the best possible runtime . [ lem : lowerbound ] the worst - case runtime required by any local - control algorithm to solve the universal coating problem is @xmath68 . particles ( black dots ) in a line connected to the surface via a single particle @xmath69.,width=226 ] [ fig : worstcaseretiredstructure ] assume the particles @xmath70 form a single line of @xmath0 particles connected to the surface via @xmath69 ( figure [ fig : worstcaseretiredstructure ] ) . suppose @xmath71 . since @xmath72 , it will take @xmath68 rounds in the worst - case ( requiring @xmath73 movements ) until @xmath74 touches the object s surface . this worst - case can happen , for example , if @xmath74 performs no more than one movement ( either an expansion or a contraction ) per round . unfortunately , a large lower bound also holds for the competitiveness of any local - control algorithm . a coating algorithm @xmath67 is called _ @xmath75-competitive _ if for any valid instance @xmath12 , @xmath76 \leq c \cdot { \ensuremath{\text{opt}}}(p , o ) + k\ ] ] where @xmath77 is the minimum runtime needed to solve the coating problem @xmath12 and @xmath78 is a value independent of @xmath12 . [ thm : competitiveness ] any local - control algorithm that solves the universal coating problem has a competitive ratio of @xmath68 . we construct an instance of the coating problem @xmath12 which can be solved by an optimal algorithm in @xmath79 rounds , but requires any local - control algorithm @xmath68 times longer . let @xmath14 be a straight line of arbitrary ( finite ) length , and let @xmath13 be a set of particles which entirely occupy layer 1 , with the exception of one missing particle below @xmath14 equidistant from its sides and one additional particle above @xmath14 in layer 2 equidistant from its sides ( see figure [ fig : opt1borders ] ) . an optimal algorithm could move the particles to solve the coating problem for the given example in @xmath79 rounds , as in figure [ fig : opt2 ] . note that the optimal algorithm always maintains the connectivity of the particle system , so its runtime is valid even under the constraint that any connected component of particles must stay connected . however , for our local - control algorithms we allow particles to disconnect from the rest of the system . . the particles are all contracted and occupy the positions around the object , with the exception that there is one unoccupied node below the object and one extra particle above the object . borders @xmath80 and @xmath81 are shown as red lines . ] [ fig : opt1borders ] now consider an arbitrary local - control algorithm @xmath82 for the coating problem . given a round @xmath83 , we define the _ imbalance _ @xmath84 at border @xmath80 as the net number of particles that have crossed @xmath80 from the top of @xmath14 to the bottom until round @xmath83 ; similarly , the imbalance @xmath85 at border @xmath81 is defined to be the net number of particles that have crossed @xmath81 from the bottom of @xmath14 to the top until round @xmath83 . [ fig : opt2 ] certainly , there is an activation sequence in which information and particles can only travel a distance of up to @xmath86 nodes towards @xmath80 or @xmath81 within the first @xmath86 rounds . hence , for any @xmath87 , the probability distributions of @xmath84 and @xmath85 are independent of each other . additionally , particles up to a distance of @xmath86 from @xmath80 and @xmath81 can not distinguish between which border they are closer to , since the position of the gap is equidistant from the borders . this symmetry also implies that @xmath88=\pr[\phi_r(r)=k]$ ] for any integer @xmath89 . let us focus on round @xmath90 . we distinguish between the following cases . then there are more particles than positions in layer 1 above @xmath14 , so the coating problem can not be solved yet . @xmath92 . from our insights above we know that for any two values @xmath93 and @xmath94 , @xmath95 and @xmath96 = \pr[\phi_l(n/4)=k_2 $ ] and @xmath97 $ ] . hence , the cumulative probability of all outcomes where @xmath98 is equal to the cumulative probability of all outcomes where @xmath99 . if @xmath98 , then there are again more particles than positions in layer 1 above @xmath14 , so the coating problem can not be solved yet . thus , the probability that @xmath67 has not solved the coating problem after @xmath86 rounds is at least @xmath100 , and therefore @xmath101 \geq 1/2 \cdot n/4 = n/8 $ ] . since , on the other hand , @xmath102 , we have established a linear competitive ratio . therefore , even the competitive ratio can be very high in the worst case . we will revisit the notion of competitiveness in section [ sec : experimental ] . in this section , we show that our algorithm solves the coating problem within a linear number of rounds w.h.p .. we start with some basic notation in section [ sec : preliminaryanalysis ] . section [ sec : parallel ] presents a simpler synchronous parallel model for particle activations that we can use to analyze the worst - case number of rounds . section [ sec : firstlayer ] presents the analysis of the number of rounds required to coat the first layer . finally , in section [ sec : higherlayers ] , we analyze the number of rounds required to complete all other coating layers , once layer 1 has been completed . we start with some notation . recall that @xmath28 denotes the number of nodes of @xmath2 at distance @xmath27 from object @xmath14 ( i.e. , the number of nodes in layer @xmath27 ) . let @xmath103 be the the layer number of the final layer for @xmath0 particles ( i.e. , @xmath103 satisfies @xmath104 ) . layer @xmath27 is said to be _ complete _ if every node in layer @xmath27 is occupied by a contracted retired particle ( for @xmath105 ) , or if all particles have reached their final position , are contracted , and never move again ( for @xmath106 ) . given a configuration @xmath9 , we define a directed graph @xmath107 over all nodes in @xmath2 occupied by _ active _ ( follower or root ) particles in @xmath9 . for every expanded active particle @xmath5 in @xmath9 , @xmath107 contains a directed edge from the tail to the head of @xmath5 . for every follower @xmath5 , @xmath107 has a directed edge from the head of @xmath5 to @xmath47 . for the purposes of constructing @xmath107 , we also define parents for root particles : a root particle @xmath5 sets @xmath47 to be the active particle @xmath6 occupying the node in direction @xmath48 once @xmath5 has performed its first handover expansion with @xmath6 . for every root particle @xmath5 , @xmath107 has a directed edge from the head of @xmath5 to @xmath47 , if it exists . certainly , since every node has at most one outgoing edge in @xmath107 , the nodes of @xmath107 form either a collection of disjoint trees or a ring of trees . a ring of trees may occur in any layer , but only temporarily ; the leader election primitive ensures that a leader emerges and retires in layer 1 and marker particles emerge and retire in higher layers , causing the ring in @xmath107 to break . the super - roots defined in section [ subsec : coatingprimitives ] correspond to the roots of the trees in @xmath107 . a _ movement _ executed by a particle @xmath5 can be either a _ sole contraction _ in which @xmath5 contracts and leaves a node unoccupied , a _ sole expansion _ in which @xmath5 expands into an adjacent unoccupied node , a _ handover contraction with @xmath49 _ in which @xmath5 contracts and forces its contracted neighbor @xmath49 to expand into the node it vacates , or a _ handover expansion with @xmath49 _ in which @xmath5 expands into a node currently occupied by its expanded neighbor @xmath49 , forcing @xmath49 to contract . in this section , we show that instead of analyzing our algorithm for asynchronous activations of particles , it suffices to consider a much simpler model of parallel activations of particles . movement schedule _ to be a sequence of particle system configurations @xmath108 . [ defn : parallelschedule ] a movement schedule @xmath108 is called a _ parallel schedule _ if each @xmath109 is a valid configuration of a connected particle system ( i.e. , each particle is either expanded or contracted , and every node of @xmath2 is occupied by at most one particle ) and for every @xmath110 is reached from @xmath109 such that for every particle @xmath5 one of the following properties holds : 1 . @xmath5 occupies the same node(s ) in @xmath109 and @xmath111 , 2 . @xmath5 expands into an adjacent node that was empty in @xmath109 , 3 . @xmath5 contracts , leaving the node occupied by its tail empty in @xmath111 , or 4 . @xmath5 is part of a handover with a neighboring particle @xmath49 . while these properties allow at most one contraction or expansion per particle in moving from @xmath109 to @xmath111 , multiple particles may move in this time . consider an arbitrary fair asynchronous activation sequence @xmath82 for a particle system , and let @xmath112 , for @xmath113 , be the particle system configuration at the end of asynchronous round @xmath27 in @xmath82 if each particle moves according to algorithm [ alg : spanningforestalgorithm ] . a _ forest schedule _ @xmath114 is a parallel schedule @xmath108 with the property that @xmath115 is a forest of one or more trees , and each particle @xmath5 follows the unique path @xmath116 which it would have followed according to @xmath82 , starting from its position in @xmath117 . this implies that @xmath118 remains a forest of trees for every @xmath119 . a forest schedule is said to be _ greedy _ if all particles perform movements according to definition [ defn : parallelschedule ] in the direction of their unique paths whenever possible . we begin our analysis with a result that is critical to both describing configurations of particles in greedy forest schedules and quantifying the amount of progress greedy forest schedules make over time . specifically , we show that if a forest s configuration is `` well - behaved '' at the start , then it remains so throughout its greedy forest schedule , guaranteeing that progress is made once every two configurations . [ lem : expparentchild ] given any fair asynchronous activation sequence @xmath82 , consider any greedy forest schedule @xmath120 . if every expanded parent in @xmath117 has at least one contracted child , then every expanded parent in @xmath109 also has at least one contracted child , for @xmath119 . suppose to the contrary that @xmath109 is the first configuration that contains an expanded parent @xmath5 which has all expanded children . we consider all possible expanded and contracted states of @xmath5 and its children in @xmath121 and show that none of them can result in @xmath5 and its children all being expanded in @xmath109 . first suppose @xmath5 is expanded in @xmath121 ; then by supposition , @xmath5 has a contracted child @xmath6 . by definition [ defn : parallelschedule ] , @xmath6 can not perform any movements with its children ( if they exist ) , so @xmath5 performs a handover contraction with @xmath6 , yielding @xmath5 contracted in @xmath109 , a contradiction . so suppose @xmath5 is contracted in @xmath121 . we know @xmath5 will perform either a handover with its parent or a sole expansion in direction @xmath48 since it is expanded in @xmath109 by supposition . thus , any child of @xmath5 in @xmath121 say @xmath6 does not execute a movement with @xmath5 in moving from @xmath121 to @xmath109 . instead , if @xmath6 is contracted in @xmath121 then it remains contracted in @xmath109 since it is only permitted to perform a handover with its unique parent @xmath5 ; otherwise , if @xmath6 is expanded , it performs either a sole contraction if it has no children or a handover with one of its contracted children , which it must have by supposition . in either case , @xmath5 has a contracted child in @xmath109 , a contradiction . as a final observation , two trees of the forest may `` merge '' when the super - root @xmath122 of one tree performs a sole expansion into an unoccupied node adjacent to a particle @xmath6 of another tree . however , since @xmath122 is a root and thus only defines @xmath6 as its parent after performing a handover expansion with it , the lemma holds in this case as well . for any particle @xmath5 in a configuration @xmath9 of a forest schedule , we define its _ head distance _ @xmath123 ( resp . , _ tail distance _ @xmath124 ) to be the number of edges along @xmath116 from the head ( resp . , tail ) of @xmath5 to the end of @xmath116 . depending on whether @xmath5 is contracted or expanded , we have @xmath125 . for any two configurations @xmath9 and @xmath126 and any particle @xmath5 , we say that @xmath9 _ dominates @xmath126 w.r.t . @xmath5 _ , denoted @xmath127 , if and only if @xmath128 and @xmath129 . we say that @xmath9 _ dominates _ @xmath126 , denoted @xmath130 , if and only if @xmath9 dominates @xmath126 with respect to every particle . then it holds : [ lem : forestdom ] given any fair asynchronous activation sequence @xmath82 which begins at an initial configuration @xmath131 in which every expanded parent has at least one contracted child , there is a greedy forest schedule @xmath114 with @xmath132 such that @xmath133 for all @xmath113 . we first introduce some supporting notation . let @xmath134 be the sequence of movements @xmath5 executes according to @xmath82 . let @xmath135 denote the remaining sequence of movements in @xmath136 after the forest schedule reaches @xmath109 , and let @xmath137 denote the first movement in @xmath135 . a greedy forest schedule @xmath114 can be constructed from configuration @xmath132 such that , for every @xmath113 , configuration @xmath109 is obtained from @xmath121 by executing only the movements of a greedily selected , mutually compatible subset of @xmath138 . argue by induction on @xmath27 , the current configuration number . @xmath117 is trivially obtained , as it is the initial configuration . assume by induction that the claim holds up to @xmath121 . w.l.o.g . let @xmath139 , for @xmath140 , be the greedily selected , mutually compatible subset of movements that @xmath141 performs in moving from @xmath121 to @xmath109 . suppose to the contrary that a movement @xmath142 is executed by a particle @xmath143 . it is easily seen that @xmath144 can not be @xmath145 ; since @xmath145 was excluded when @xmath146 was greedily selected , it must be incompatible with one or more of the selected movements and thus can not also be executed at this time . so @xmath147 , and we consider the following cases : @xmath145 is a sole contraction . then @xmath5 is expanded and has no children in @xmath121 , so we must have @xmath148 , since there are no other movements @xmath5 could execute , a contradiction . @xmath145 is a sole expansion . then @xmath5 is contracted and has no parent in @xmath121 , so we must have @xmath148 , since there are no other movements @xmath5 could execute , a contradiction . @xmath145 is a handover contraction with @xmath6 , one of its children . then at some time in @xmath141 before reaching @xmath121 , @xmath6 became a descendant of @xmath5 ; thus , @xmath6 must also be a descendant of @xmath5 in @xmath121 . if @xmath6 is not a child of @xmath5 in @xmath121 , there exists a particle @xmath149 such that @xmath6 is a descendant of @xmath150 , which is in turn a descendant of @xmath5 . so in order for @xmath145 to be a handover contraction with @xmath6 , @xmath151 must include actions which allow @xmath150 to `` bypass '' its ancestor @xmath5 , which is impossible . so @xmath6 must be a child of @xmath5 in @xmath121 , and must be contracted at the time @xmath145 is performed . if @xmath6 is also contracted in @xmath121 , then once again we must have @xmath148 . otherwise , @xmath6 is expanded in @xmath121 , and must have become so before @xmath121 was reached . but this yields a contradiction : since @xmath141 is greedy , @xmath6 would have contracted prior to this point by executing either a sole contraction if it has no children , or a handover contraction with a contracted child whose existence is guaranteed by lemma [ lem : expparentchild ] , since every expanded parent in @xmath117 has a contracted child . @xmath145 is a handover expansion with @xmath6 , its unique parent . then we must have that @xmath152 is a handover contraction with @xmath5 , and an argument analogous to that of case 3 follows . we conclude by showing that each configuration of the greedy forest schedule @xmath141 constructed according to the claim is dominated by its asynchronous counterpart . argue by induction on @xmath27 , the configuration number . since @xmath132 , we have that @xmath153 . assume by induction that for all rounds @xmath154 , we have @xmath155 . consider any particle @xmath5 . since @xmath141 is constructed using the exact set of movements @xmath5 executes according to @xmath82 and each time @xmath5 moves it decreases either its head distance or tail distance by @xmath156 , it suffices to show that @xmath5 has performed at most as many movements in @xmath141 up to @xmath109 as it has according to @xmath82 up to @xmath112 . if @xmath5 does not perform a movement between @xmath121 and @xmath109 , we trivially have @xmath157 . otherwise , @xmath5 performs movement @xmath145 to obtain @xmath109 from @xmath121 . if @xmath5 has already performed @xmath145 according to @xmath82 before reaching @xmath158 , then clearly @xmath157 . otherwise , @xmath145 must be the next movement @xmath5 is to perform according to @xmath82 , since @xmath5 has performed the same sequence of movements in the asynchronous execution as it has in @xmath141 up to the respective rounds @xmath35 , and thus has equal head and tail distances in @xmath121 and @xmath158 . it remains to show that @xmath5 can indeed perform @xmath145 between @xmath158 and @xmath112 . if @xmath145 is a sole expansion , then @xmath5 is the super - root of its tree ( in both @xmath121 and @xmath158 ) and must also be able to expand in @xmath158 . similarly , if @xmath145 is a sole contraction , then @xmath5 has no children ( in both @xmath121 and @xmath158 ) and must be able to contract in @xmath158 . if @xmath145 is a handover expansion with its parent @xmath6 , then @xmath6 must be expanded in @xmath121 . parent @xmath6 must also be expanded in @xmath158 ; otherwise @xmath159 , contradicting the induction hypothesis . an analogous argument holds if @xmath145 is a handover contraction with one of its contracted children . therefore , in any case we have @xmath157 , and since the choice of @xmath5 was arbitrary , @xmath133 . we can show a similar dominance result when considering complaint flags . [ defn : complaintparallelschedule ] a movement schedule @xmath108 is called a _ complaint - based parallel schedule _ if each @xmath109 is a valid configuration of a particle system in which every particle holds at most _ one _ complaint flag ( rather than two , as described in algorithm [ alg : complaint ] ) and for every @xmath160 , @xmath111 is reached from @xmath109 such that for every particle @xmath5 one of the following properties holds : 1 . @xmath5 does not hold a complaint flag and property 1 , 3 , or 4 of definition [ defn : parallelschedule ] holds , 2 . @xmath5 holds a complaint flag @xmath161 and expands into an adjacent node that was empty in @xmath109 , consuming @xmath161 , 3 . @xmath5 forwards a complaint flag @xmath161 to a neighboring particle @xmath49 which either does not hold a complaint flag in @xmath109 or is also forwarding its complaint flag . a _ complaint - based forest schedule _ @xmath114 has the same properties as a forest schedule , with the exception that @xmath108 is a complaint - based parallel schedule as opposed to a parallel schedule . a complaint - based forest schedule is said to be _ greedy _ if all particles perform movements according to definition [ defn : complaintparallelschedule ] in the direction of their unique paths whenever possible . we can now extend the dominance argument to hold with respect to _ complaint distance _ in addition to head and tail distances . for any particle @xmath5 holding a complaint flag @xmath161 in configuration @xmath9 , we define its complaint distance @xmath162 to be the number of edges along @xmath116 from the node @xmath5 occupies to the end of @xmath116 . for any two configurations @xmath9 and @xmath126 and any complaint flag @xmath161 , we say that _ @xmath9 dominates @xmath126 w.r.t . @xmath161 _ , denoted @xmath163 , if and only if @xmath164 . extending the previous notion of dominance , we say that _ @xmath9 dominates @xmath126 _ , denoted @xmath130 , if and only if @xmath9 dominates @xmath126 with respect to every particle and with respect to every complaint flag . it is also possible to construct a greedy complaint - based forest schedule whose configurations are dominated by their asynchronous counterparts , as we did for greedy forest schedules in lemma [ lem : forestdom ] . many of the details are the same , so as to avoid redundancy we highlight the differences here . the most obvious difference is the inclusion of complaint flags . definition [ defn : complaintparallelschedule ] restricts particles to holding at most one complaint flag at a time , where algorithm [ alg : complaint ] allows a capacity of two . this allows the asynchronous execution to not `` fall behind '' the parallel schedule in terms of forwarding complaint flags . basically , definition [ defn : complaintparallelschedule ] allows a particle @xmath5 holding a complaint flag @xmath161 in the parallel schedule to forward @xmath161 to its parent @xmath6 even if @xmath6 currently holds its own complaint flag , so long as @xmath6 is also forwarding its flag at this time . the asynchronous execution does not have this luxury of synchronized actions , so the mechanism of buffering up to two complaint flags at a time allows it to `` mimic '' the pipelining of forwarding complaint flags that is possible within one round of a complaint - based parallel schedule . another slight difference is that a contracted particle can not expand into an empty adjacent node unless it holds a complaint flag to consume . however , this restriction reflects algorithm [ alg : boundarydirectionalgorithm ] , so once again the greedy complaint - based forest schedule can be constructed directly from the movements taken in the asynchronous execution . moreover , since this restriction can only cause a contracted particle to remain contracted , the conditions of lemma [ lem : expparentchild ] are still upheld . thus , we obtain the following lemma : [ lem : flagforestdom ] given any fair asynchronous activation sequence @xmath82 which begins at an initial configuration @xmath131 in which every expanded parent has at least one contracted child , there is a greedy complaint - based forest schedule @xmath165 with @xmath132 such that @xmath133 for all @xmath113 . by lemmas [ lem : forestdom ] and [ lem : flagforestdom ] , once we have an upper bound for the time it takes a greedy forest schedule to reach a final configuration , we also have an upper bound for the number of rounds required by the asynchronous execution . hence , the remainder of our proofs will serve to upper bound the number of parallel rounds any greedy forest schedule would require to solve the coating problem for a given valid instance @xmath12 , where @xmath166 . let @xmath167 be such a greedy forest schedule , where @xmath117 is the initial configuration of the particle system @xmath13 ( of all contracted particles ) and @xmath168 is the final coating configuration . in sections [ sec : firstlayer ] and [ sec : higherlayers ] , we will upper bound the number of parallel rounds required by @xmath169 in the worst case to coat the first layer and higher layers , respectively . more specifically , we will bound the worst - case time it takes to complete a layer @xmath27 once layers @xmath170 have been completed . for convenience , we will not differentiate between complaint - based and regular forest schedules in the following sections , since the same dominance result holds whether or not complaint flags are considered . to prove these bounds , we need one last definition : a _ forest path schedule _ @xmath171 is a forest schedule @xmath172 with the property that all the trees of @xmath115 are rooted at a path @xmath173 , and each particle @xmath5 must traverse @xmath80 in the same direction . our algorithm must first organize the particles using the spanning forest primitive , whose runtime is easily bounded : following the spanning forest primitive , the particles form a spanning forest within @xmath174 rounds . initially all particles are idle . in each round any idle particle adjacent to the object , an active ( follower or root ) particle , or a retired particle becomes active . it then sets its parent flag if it is a follower , or becomes the root of a tree if it is adjacent to the object or a retired particle . in each round at least one particle becomes active , so given @xmath0 particles in the system it will take @xmath174 rounds in the worst case until all particles join the spanning forest . for ease of presentation , we assume that the particle system is of sufficient size to fill the first layer ( i.e. , @xmath175 ; the proofs can easily be extended to handle the case when @xmath176 ) ; we also assume that the root of a tree also generates a complaint flag upon its activation ( this assumption does not hurt our argument since it only increases the number of the flags generated in the system ) . let @xmath177 be the greedy forest path schedule where @xmath178 is a truncated version of @xmath169 , @xmath179 for @xmath180 is the configuration in @xmath169 in which layer @xmath156 becomes complete , and @xmath181 is the path of nodes in layer @xmath156 . the following lemma shows that the algorithm makes steady progress towards completing layer @xmath156 . [ lem : progresslayer1 ] consider a round @xmath27 of the greedy forest path schedule @xmath182 , where @xmath183 . then within the next two parallel rounds of @xmath182 , @xmath51 at least one complaint flag is consumed , @xmath52 at least one more complaint flag reaches a particle in layer @xmath156 , @xmath184 all remaining complaint flags move one position closer to a super - root along @xmath181 , or @xmath185 layer @xmath156 is completely filled ( possibly with some expanded particles ) . if layer 1 is filled , @xmath185 is satisfied ; otherwise , there exists at least one super - root in @xmath118 . we consider several cases : there exists a super - root @xmath122 in @xmath118 which holds a complaint flag . if @xmath122 is contracted , then it can expand and consume its flag by the next round . otherwise , consider the case when @xmath122 is expanded . if it has no children , then within the next two rounds it can contract and expand again , consuming its complaint flag ; otherwise , by lemma [ lem : expparentchild ] , @xmath122 must have a contracted child with which it can perform a handover to become contracted in @xmath111 and then expand and consume its complaint flag by @xmath186 . in any case , @xmath51 is satisfied . no super - root in @xmath118 holds a complaint flag and not all complaint flags have been moved from follower particles to particles in layer 1 . let @xmath187 be a sequence of particles in layer 1 such that each particle holds a complaint flag , no follower child of any particle except @xmath188 holds a complaint flag , and no particles between the next super - root @xmath122 and @xmath69 hold complaint flags . then , as each @xmath189 forwards its flag to @xmath190 according to definition [ defn : complaintparallelschedule ] , the follower child of @xmath188 holding a flag is able to forward its flag to @xmath188 , satisfying @xmath52 . no super - root in @xmath118 holds a complaint flag and all remaining complaint flags are held by particles in layer 1 . by definition [ defn : complaintparallelschedule ] , since no preference needs to be given to flags entering layer 1 , all remaining flags will move one position closer to a super - root in each round , satisfying @xmath184 . we use lemma [ lem : progresslayer1 ] to show first that layer @xmath156 will be filled with particles ( some possibly still expanded ) in @xmath174 rounds . from that point on , in another @xmath174 rounds , one can guarantee that expanded particles in layer @xmath156 will each contract in a handover with a follower particle , and hence all particles in layer @xmath156 will be contracted , as we see in the following lemma : [ lemma : filled ] after @xmath174 rounds , layer 1 must be filled with contracted particles . we first prove the following claim : after @xmath191 rounds of @xmath141 , layer @xmath156 must be filled with particles . suppose to the contrary that after @xmath192 rounds , layer 1 is not completely filled with particles . then none of these rounds could have satisfied @xmath185 of lemma [ lem : progresslayer1 ] , so one of @xmath193 , or @xmath184 must be satisfied every two rounds . case @xmath51 can be satisfied at most @xmath194 times ( accounting for at most @xmath195 rounds ) , since a super - root expands into an unoccupied position of layer 1 each time a complaint flag is consumed . case @xmath184 can also be satisfied at most @xmath194 times ( accounting for at most @xmath195 rounds ) , since once all remaining complaint flags are in layer 1 , every flag must reach a super - root in @xmath194 moves . thus , the remaining @xmath196 rounds must satisfy @xmath52 @xmath197 times , implying that @xmath197 flags reached particles in layer 1 from follower children . but each particle can hold at most one complaint flag , so at least @xmath198 flags must have been consumed and the super - roots have collectively expanded into at least @xmath198 unoccupied positions , a contradiction . by the claim , it will take at most @xmath191 rounds until layer @xmath156 is completely filled with particles ( some possibly expanded ) . in at most another @xmath194 rounds , every expanded particle in layer @xmath156 will contract in a handover with a follower particle ( since @xmath199 ) , and hence all particles in layer @xmath156 will be contracted after @xmath200 rounds . once layer @xmath156 is filled , the leader election primitive can proceed . the full description of the universal coating algorithm in @xcite uses a node - based version of the leader election algorithm in @xcite for this primitive . for consistency , we kept this description of the primitive in this paper as well . however , in order to formally prove with high probability guarantees on the runtime of our universal coating algorithm , we use a monte carlo variant of the leader election algorithm in @xcite . a description of this variant and its corresponding proofs appear in @xcite . this updated algorithm elects a leader with high probability and gives the following runtime bound . [ lem : leaderelection ] within @xmath174 further rounds , a position of layer 1 has been elected as the leader position , w.h.p . once a leader position has been elected and either no more followers exist ( if @xmath201 ) or all positions are completely filled by contracted particles ( which can be checked in an additional @xmath202 rounds ) , the particle currently occupying the leader position becomes the leader particle . once a leader has emerged , the particles on layer @xmath156 retire , which takes @xmath202 further rounds . together , we get : [ cor : layer1 ] the worst - case number of rounds for @xmath169 to complete layer @xmath156 is @xmath174 , w.h.p . we again use the dominance results we proved in section [ sec : parallel ] to focus on parallel schedules when proving an upper bound on the worst - case number of rounds denoted by @xmath203 for building layer @xmath27 once layer @xmath35 is complete , for @xmath204 . the following lemma provides a more general result which we can use for this purpose . [ lem : tree - time ] consider any greedy forest path schedule @xmath171 with @xmath205 and any @xmath89 such that @xmath206 . if every expanded parent in @xmath117 has at least one contracted child , then in at most @xmath207 configurations , nodes @xmath208 will be occupied by contracted particles . let @xmath122 be the super - root closest to @xmath209 , and suppose @xmath122 initially occupies node @xmath210 in @xmath117 . additionally , suppose there are at least @xmath89 active particles in @xmath117 ( otherwise , we do not have sufficient particles to occupy @xmath89 nodes of @xmath80 ) . argue by induction on @xmath89 , the number of nodes in @xmath80 starting with @xmath209 which must be occupied by contracted particles . first suppose that @xmath211 . by lemma [ lem : expparentchild ] , every expanded parent has at least one contracted child in any configuration @xmath212 , so @xmath122 is always able to either expand forward into an unoccupied node of @xmath80 if it is contracted or contract as part of a handover with one of its children if it is expanded . thus , in at most @xmath213 configurations , @xmath122 has moved forward @xmath214 positions , is contracted , and occupies its final position @xmath215 . now suppose that @xmath216 and that each node @xmath217 , for @xmath218 , becomes occupied by a contracted particle in at most @xmath219 configurations . it suffices to show that @xmath220 also becomes occupied by a contracted particle in at most two additional configurations . let @xmath5 be the particle currently occupying @xmath220 ( such a particle must exist since we supposed we had sufficient particles to occupy @xmath89 nodes and @xmath141 ensures the particles follow this unique path ) . if @xmath5 is contracted in @xmath221 , then it remains contracted and occupying @xmath220 , so we are done . otherwise , if @xmath5 is expanded , it has a contracted child @xmath6 by lemma [ lem : expparentchild ] . particles @xmath5 and @xmath6 thus perform a handover in which @xmath5 contracts to occupy only @xmath220 at @xmath222 , proving the claim . for convenience , we introduce some additional notation . let @xmath223 denote the number of particles of the system that will not belong to layers 1 through @xmath35 , i.e. , @xmath224 , and let @xmath225 ( resp . , @xmath226 ) be the round ( resp . , configuration ) in which layer @xmath27 becomes complete . when coating some layer @xmath27 , each root particle either moves either @xmath227 through the nodes in layer @xmath27 in the set direction @xmath228 ( cw or ccw ) for layer @xmath27 , or @xmath229 through the nodes in layer @xmath230 in the opposite direction over the already retired particles in layer @xmath27 until it finds an empty position in layer @xmath27 . we bound the worst - case scenario for these two movements independently in order to get a an upper bound on @xmath203 . let @xmath231 be the path of nodes in layer @xmath27 listed in the order that they appear from the marker position @xmath232 following direction @xmath228 , and let @xmath233 be the greedy forest path schedule where @xmath234 is a section of @xmath169 . by lemma [ lem : tree - time ] , it would take @xmath235 rounds for all @xmath227 movements to complete ; an analogous argument shows that all @xmath229 movements complete in @xmath236 rounds . this implies the following lemma : [ lem : layeri2 ] starting from configuration @xmath237 , the worst - case additional number of rounds for layer @xmath27 to become complete is @xmath235 . putting it all together , for layers @xmath238 through @xmath103 : [ cor : higherlayers ] the worst - case number of rounds for @xmath169 to coat layers 2 through @xmath103 is @xmath174 . starting from configuration @xmath239 , it follows from lemma [ lem : layeri2 ] that the worst - case number of rounds for @xmath169 to reach a legal coating of the object is upper bounded by @xmath240 where @xmath1 is a constant . combining corollaries [ cor : layer1 ] and [ cor : higherlayers ] , we get that @xmath169 requires @xmath174 rounds w.h.p . to coat any given valid object @xmath14 starting from any valid initial configuration of the set of particles @xmath13 . by lemmas [ lem : forestdom ] and [ lem : flagforestdom ] , the worst - case behavior of @xmath169 is an upper bound for the runtime of our universal coating algorithm , so we conclude : [ thm : chain ] the total number of asynchronous rounds required for the universal coating algorithm to reach a legal coating configuration , starting from an arbitrary valid instance @xmath12 , is @xmath174 w.h.p . , where @xmath0 is the number of particles in the system . in this section we present a brief simulation - based analysis of our algorithm which shows that in practice our algorithm exhibits a better than linear average competitive ratio . since @xmath77 ( as defined in section [ sec : performance ] ) is difficult to compute in general , we investigate the competitiveness with the help of an appropriate lower bound for @xmath77 . recall the definitions of the distances @xmath241 and @xmath242 for @xmath243 and @xmath25 . consider any valid instance @xmath12 . let @xmath244 be the set of all legal particle positions of @xmath12 ; that is , @xmath244 contains all sets @xmath25 such that the positions in @xmath245 constitute a coating of the object @xmath14 by the particles in the system . we compute a lower bound on @xmath77 as follows . consider any @xmath246 , and let @xmath247 denote the complete bipartite graph on partitions @xmath13 and @xmath245 . for each edge @xmath248 , set the cost of the edge to @xmath249 . every perfect matching in @xmath247 corresponds to an assignment of the particles to positions in the coating . the maximum edge weight in a matching corresponds to the maximum distance a particle has to travel in order to take its place in the coating . let @xmath250 be the set of all perfect matchings in @xmath247 . we define the _ matching dilation _ of @xmath12 as @xmath251 since each particle has to move to some position in @xmath245 for some @xmath246 to solve the coating problem , we have @xmath252 . the search for the matching that minimizes the maximum edge cost for a given @xmath246 can be realized efficiently by reducing it to a flow problem using edges up to a maximum cost of @xmath75 and performing binary search on @xmath75 to find the minimal @xmath75 such that a perfect matching exists . we note that our lower bound is not tight . this is due to the fact that it only respects the distances that particles have to move but ignores the congestion that may arise , i.e. , in certain instances the distances to the object might be very small , but all particles may have to traverse one `` chokepoint '' and thus block each other . [ fig : simresults ] we implemented the universal coating algorithm in the amoebot simulator ( see @xcite for videos ) . for simplicity , each simulation is initialized with the object @xmath14 as a regular hexagon of object particles ; this is reasonable since the particles need only know where their immediate neighbors in the object s border are relative to themselves , which can be determined independently of the shape of the border . the particle system @xmath13 is initialized as idle particles attached randomly around the hexagon s perimeter . the parameters that were varied between instances are the radius of the hexagon and the number of ( initially idle ) particles in @xmath13 . each experimental trial randomly generates a new initial configuration of the system . figure [ fig : simresults](a ) shows the number of rounds needed to complete the coating with respect to the hexagon object radius and the number of particles in the system . the number of rounds plotted are averages over 20 instances of a given @xmath253 with 95% confidence intervals . these results show that , in practice , the number of rounds required increases linearly with particle system size . this agrees with our expectations , since leader election depends only on the length of the object s surface while layering depends on the total number of particles . figure [ fig : simresults](b ) shows the ratio of the number of rounds to the matching dilation of the system . these results indicate that , in experiment , the average competitive ratio of our algorithm may exhibit closer to logarithmic behaviors . figure [ fig : simresults](c ) shows the number of rounds needed to complete the coating as the radius of the hexagon object is varied . the runtime of the algorithm appears to increase linearly with both the number of active particles and the size of the object being coated , and there is visibly increased runtime variability for systems with larger radii . this paper continued the study of universal coating in self - organizing particle systems . the runtime analysis shows that our universal coating algorithm , presented in @xcite , terminates in a linear number of rounds , so it is worst - case optimal . this , along with the linear lower bound on the competitive gap between local and global algorithms , further shows our algorithm to be competitively optimal . furthermore , the simulation results show the competitive ratio of our algorithm is better than linear in practice . in the future , we would like to apply the algorithm and analysis to the case of bridging , in which particles create structures across gaps between disconnected objects . we would also like to extend the algorithm to have self - stabilization capabilities , so that it could successfully complete coating without human intervention after occasional particle failure or outside interference .
imagine coating buildings and bridges with smart particles ( also coined smart paint ) that monitor structural integrity and sense and report on traffic and wind loads , leading to technology that could do such inspection jobs faster and cheaper and increase safety at the same time . in this paper , we study the problem of uniformly coating objects of arbitrary shape in the context of _ self - organizing programmable matter _ , i.e. , the programmable matter consists of simple computational elements called particles that can establish and release bonds and can actively move in a self - organized way . particles are anonymous , have constant - size memory and utilize only local interactions in order to coat an object . we continue the study of our universal coating algorithm by focusing on its runtime analysis , showing that our algorithm terminates within a _ linear number of rounds _ with high probability . we also present a matching linear lower bound that holds with high probability . we use this lower bound to show a _ linear lower bound on the competitive gap _ between fully local coating algorithms and coating algorithms that rely on global information , which implies that our algorithm is also optimal in a competitive sense . simulation results show that the competitive ratio of our algorithm may be better than linear in practice .
the snowmass young physicists ( syp ) @xcite was first organized at the community planning meeting held at fermilab in october of 2012 . this group was formed to provide a conduit for young ( nontenured ) particle physicists to participate in the forthcoming community summer study ( known as snowmass on the mississippi ) which took place in the summer of 2013 in minneapolis , mn . the primary charge taken up by the syp was to facilitate and encourage young people to get involved with physics studies and meetings in preparation for snowmass on the mississippi . the syp generated an online platform ( http://snowmassyoung.hep.net ) as well as an offline network for advertising tasks that need to be done and connecting interested syp members with the relevant frontiers . this was accomplished by having our members attend many of the `` pre - snowmass '' meetings across the frontiers . while at these meetings , dedicated parallel sessions were arranged with remote participation to provide a platform for discussion and input . invited speakers from the fermilab directorate , department of energy ( doe ) , national science foundation ( nsf ) , as well as senior scientists and faculty came to speak with a broad section of the young community . in addition to these meetings , over a dozen syp town hall meetings were held prior to snowmass during which plans and concerns were raised . the second charge is related to a `` deliverable '' to the community summer study . the syp gathered information about demographics , career outlook , physics outlook , planned experiments , and general concerns of young physicists in the form of an online survey @xcite . this information is intended to provide a voice to the next generation of leaders in high energy physics ( hep ) and serve as a basis for discussion with senior physicists , politicians , and funding agencies about the current and future state the field . additionally the opinions expressed here represent many people who were unable to attend snowmass on the mississippi personally . the results of this survey consititute the majority of this paper . the next charge of the syp is to become a long term asset to young physicists . this is done through multiple channels such as providing information and resources to people in high energy physics when they are making career decisions . this includes , but is not limited to , information about current and planned scientific experiments and collaborations , having cross - frontier talks and seminars , and providing information and resources for those who decide to take the many skills learned in physics out into the general work force . furthermore , the syp will make efforts to engage members of congress from the many districts in which the syp members live and help lobby for the interests of young people in hep . moreover , the syp aims to provide a chance for young physicists to network and meet each other as well as become known outside of each of our particular subfields . beyond the above listed tasks , syp has made contact with individuals who have recently left hep for non - academic career paths . through gathering information about their experiences we can illustrate the broader positive impact members of hep have on the wider u.s . after leaving the field . moreover this information can serve as an example for career opportunities of current would - be physicists in the future . in addition to the online survey that was taken leading up to snowmass , a follow - up document is availabe for people who know the contact information for those who have left hep ( http://snowmassyoung.hep.net/oa_contacts.pdf ) . the mechanism of young scientists organizing to provide input into the snowmass process is not a new phenomenon , as a group organized during snowmass 2001 and became known as the young particle physicists ( ypp ) @xcite . the ypp also administered a survey in order to collect the opinions and concerns of many in hep @xcite . this work became the basis for the syp in 2013 who were in fact helped a great deal by these original scientists , many of whom have moved into leadership roles across hep . this paper is dedicated to showing a summary of the results of the syp 2013 survey . the data on which these results are based is now publicly available ( http://snowmassyoung.hep.net/syp2013_publicdata.tar.gz ) . section [ sec : method ] will outline the methodology of the survey . section [ sec : demo ] will present the results from the demographic information collected in the survey . in section [ sec : careeroutlook ] we present results pertaining to the opinions of peoples career outlook in hep . more generally , subsection [ sec : spires ] helps frame this discussion in terms of available and filled jobs as reported by spires data @xcite @xcite . section [ sec : physicsoutlook ] presents results pertaining to opinions on the physics prospects in hep including information about which of the planned experiments appear most exciting to the community . in section [ sec : nonacademic ] we present some of the opinions gathered by those who received their training within hep but have chosen to pursue careers outside of academia . finally in section [ sec : surveyquestions ] we present all the survey questions and include some of the direct responses we received from the many participants . the goal of the survey administered by the snowmass young physicists was to collect a broad range of opinions that reflect the physics interests , the career outlook , and the general mood of the field leading up to snowmass 2013 . the survey was administered using google forms @xcite and available from april 1st to july 15th 2013 ( http://tinyurl.com/snowmassyoung ) . this survey contains four distinct sections : 1 . * demographic information : * this includes general information such as the gender , marital status , and citizenship but also information that is specific to the demographics of hep . this information includes current frontier of work , current position within academia ( e.g. graduate student , post - doctoral researcher , tenured faculty , etc ... ) , and number of years in your current position . similar information for those on a non - academic career path was also gathered . * career outlook : * this section asked the respondent questions pertaining to their feelings and outlook on their current career . questions included what type of jobs are most attractive to young scientists , what factors impact their decisions to pursue an academic career , as well as what various external factors ( family , availability of jobs , job location , etc ... ) that impact their career decisions . * physics outlook : * these questions focused on the science that is being planned during snowmass 2013 . for example , respondents were asked : what their outlook for funding of hep in the future is , which frontier will have the greatest impact in the next 10 years on hep , and to indicate which of the planned experiments ( given from a non - exhaustive list ) did the respondents have the highest priority for . 4 . * non - academic career paths : * finally , this section of the survey was dedicated to gathering information from those who had received their training within hep but have since left to pursue a career outside of academia . in this section we asked questions about how their time in hep prepared them for their current career . additionally we gathered information to see how their professional lives compare to those working within hep . the survey was structured such that respondents would only see certain questions based on information about their current employment . figure [ fig : surveyflowchart ] shows a summary of the flow chart for the survey . this allows us to look for similarities across the different career paths without exposing respondents to questions that had little relevance to them . in hindsight , it would have been helpful to do the same for tenured faculty and young people , since many questions asked about preparing for careers outside of hep , and had little relevance for those more senior members of our field . for the survey that is to be administered post - snowmass , http://tinyurl.com/post-snowmass , this issue has been resolved . ] the parsing of responses to the survey allowed us to identify correlations across different frontiers and different career stages . while many of the questions from the career outlook section focused on concerns most relevant to young scientists , most of the questions and answers included options that could be answered by all who were taking the survey . a small amount of data clean up was necessary before the sample could be analyzed . this was mostly to format answers to questions in such a way that they could be analyzed using c++ scripts . an example of one such clean up came from a question asking users to input the average number of hours worked in a week . many users inputted `` @xmath0 45 '' . this tilde was simply removed from the data before being outputted to c++ code . finally results of the survey were outputted to a series of root files which can be read by a simple analysis script . a complete survey response is seen as an `` event '' and this allows the end user to write a simple analysis script to apply selection , make histograms , and output the results . a complete package including the raw data , translated root files , and an example analysis script can be found here : http://snowmassyoung.hep.net/syp2013_publicdata.tar.gz the plots shown in this paper have negligible uncertainties and a full set of plots with uncertainties can be found here : http://snowmassyoung.hep.net/plots.html demographic information composed an important part of the survey and included @xmath0 10 - 12 questions ( depending on if you were on the academic or non - academic path of the survey ) including information about gender , marital status , number of children , and current employment . a total of 1112 responses were collected for the syp 2013 survey , of which 956 fit the `` young '' definition , e.g. non - tenured inside academia , and 74 respondents coming from non - academic career paths . comparing this to the 2001 survey which collected 1508 responses , of which 857 fit the `` young '' definition , we can see that we reached fewer total respondents , but a larger fraction of respondents were young people . below is a summary of some general demographic information collected and , where relevant , comparisons to u.s . 2010 census data @xcite are made . 1 . * what is your gender ? * + male : * 79@xmath1 * ( u.s . 2010 census data : 49@xmath1 ) + female : * 21@xmath1 * ( u.s . 2010 census data : 51@xmath1 ) + 2 . * what is your current marital status ? * + married : * 38@xmath1 * ( u.s . 2010 census data for population @xmath2 18 years old : 51@xmath1 ) + un - married : * 62@xmath1 * ( u.s . 2010 census data for population @xmath2 18 years old : 49@xmath1 ) + * how many children do you have ? * + no children : * 79@xmath1 * ( u.s . 2010 census data for population @xmath2 18 years old : 52@xmath1 ) + @xmath3 1 child * 21@xmath1 * ( u.s . 2010 census data for population @xmath2 18 years old : 48@xmath1 ) + * what is your household salary ( usd / year ) ? * + * making @xmath4 75,000 usd / year : * 69@xmath1 ( u.s . 2010 census data for population @xmath2 18 years old : 80@xmath1 ) + * making @xmath2 75,000 usd / year : * 31@xmath1 ( u.s . 2010 census data for population @xmath2 18 years old : 20@xmath1 ) + exploring the demographic information further , figure [ fig : currentpostion2013 ] shows the breakdown of the respondents whom are currently working in academia by their current position . figure [ fig : currentpostion2001 ] shows this same break down of the hep respondents by their academic positions in the 2001 survey . [ currentpos_2001 ] the snowmass process has identified 7 `` frontiers '' in hep : energy , intensity , theory , cosmic , education / outreach , instrumentation , and computing . figure [ fig : frontier ] shows the number of respondents who primarily work in each of those frontiers . a large number of participants come from the energy frontier , with the intensity , theory , and cosmic frontiers making up the majority of the rest . figure [ fig : citizen ] shows breakdown of survey participants by their current citizenship , with over half coming from the united states . figure [ fig : whereresearch ] shows where participants do the majority of their research . the largest number of respondents are based at universities , followed by cern and fnal . this is a shift from the responses found in the 2001 survey which had the majority of their responses coming from users at fermilab @xcite . [ exitment_2013 ] finally , figure [ fig : attend ] shows the breakdown of the survey participants that plan on attending the snowmass meeting . almost 60@xmath1 of those responding did not plan on attending the snowmass meeting with another 27@xmath1 undecided . figure [ fig : contribute ] shows that fewer than one - fourth of survey respondents were contributing to snowmass prior to taking the survey . the demographic information tells us that the survey reached mostly young scientists in hep , many of whom do their research at universities and who had not been participating in the snowmass process prior to taking this survey . moreover , many of those taking the survey will not be attending the snowmass meeting . this makes the data presented here particularly relevant as it expresses the opinions and concerns of many who otherwise have have not been involved in the snowmass process . career outlook questions were generally focused on the opinions and outlook of those pursuing a career within hep . information about the outlook for future funding and its impact on career decisions as well as questions about the dominant factors that impact decisions to pursue an academic career were covered . some general trends we observed about the career outlook of those in hep responding to our survey : 1 . * on a scale of 1 to 10 ( 1 = funding will stop , 10 = funding will thrive ) how do you feel about the funding within your frontier within the next decade ? * + nearly 60@xmath1 of the respondents believed that funding was more likely to decline ( giving an answer less than or equal to 5 ) in the future . + + figure [ fig : fundingyoungold ] shows the breakdown of this question by the classification of young and senior . this demonstrates that the young respondents are slightly more pessimistic about the future funding than their senior counterparts . + 2 . * do you intend to pursue a permanent career in academia ? * + despite this knowledge about a bleak funding profile being well known amongst the young scientists , @xmath062@xmath1 of the respondents still intend to pursue a career in academia . + 3 . * which of the following career related concerns do you find the most important to you and your future in high energy physics ? * + the two most important career related concerns were the availability of university based jobs followed by the availability of laboratory based jobs . + + + [ fig : contribute ] + + [ fig : contribute ] + + [ fig : contribute ] in addition to the general trends from the questions listed above , figure [ fig : jobsearchall ] shows responses to the question of where young scientists who intend to seek an academic position will search for a permanent position . this shows that the majority intend to search for a permanent position within the united states with the next largest group willing to search wherever they can find at job . figure [ fig : jobsearchusnonus ] shows the breakdown of this question based on whether the person in a u.s . or non - u.s . citizen . clearly , u.s . citizens prefer to search for a job within the u.s . while non - u.s . citizens are more apt to take a job wherever they can find one . one possible interpretation to this breakdown could be that in order to attract the best of the non - u.s . members of the hep community , job availability is already a driving force . figure [ fig : jobsearchgradpostdoc ] suggests that the younger generation of u.s . scientists are more willing to take a job wherever they can find one compared to the post - doctoral respondents . figure [ fig : jobsearchfrontier ] shows the breakdown to this question based on frontier , indicating that people working in the intensity frontier are more apt to search for a job in the u.s . while those from the theory , cosmic , and energy frontiers are more open to accepting a job wherever they can find one . these trends again suggest that a major driving force for where the young and bright physicists will go is where the jobs are available . a further confirmation of this general trend can be seen in figure [ fig : jobsearchacademia ] . this figure shows where our respondents intend to search for a job broken down by whether they intend to pursue an academic job . the results indicate that those who wish to pursue a career in academia are more willing to go wherever the jobs are , with less regard for where those jobs are . to follow - up on this observed trend we show the responses from those who intend to seek an academic career to the question if they will be more inclined to search for a job outside the u.s . if the next major experiment in their frontier is built outside the u.s . to follow - up on this observed trend we ask respondents where they will be more inclined to search for a job if the next major experiement in their frontier is built outside of the u.s . figure [ fig : leavingusall ] indicates that nearly 50@xmath1 would be more likely to search for a job outside the u.s in this if the next major experiment is built outside of the u.s . this trend is greater amongst the non - u.s . members of hep , as shown in figure [ fig : leavinguscitizen ] , and much more heavily present amongst those seeking a job in academia , as shown in figure [ fig : leavingusacademia ] . finally , figure [ fig : leavinggradpostdoc ] indicates that graduate students ( the youngest in our field ) are more inclined than post - docs to search elsewhere if the next major experiment is built outside the u.s . furthering the trends we observe above . figure [ fig : leavingusfrontier ] reinforces this and shows that those in the intensity and energy frontier are more apt to search for a job outside the u.s . if the next major experiment from their frontier is built outside the u.s . taking these two questions together leads to the observation that while most of the respondents to the survey would prefer to seek a job in the u.s . , there could be `` brain drain '' depending on whether or not major experiments take place in the u.s . this effect is particularly apparent among graduate students and the u.s.s ability to attract non - u.s . scientists , but uniformly impacts all of the frontiers . to expand on these job related concerns further , we gathered some data from the inspire high energy physics information system @xcite . this data can be found at http://hoc.web.cern.ch/hoc/jobs_stats2.txt . figure [ fig : spires ] shows the jobs listed on inspire that were filled from 2007 - 2012 . these jobs fall into the category of post - doc , junior , and senior positions . this job data is broken down by location , showing jobs listed in north america and jobs everywhere else ( global ) . a clear downward trend in the last few years can be observed broadly in all markets , with no sustained growth whatsoever since 2007 . a continuation of these trends will almost certainly lead to fewer young scientists remaining in the field of particle physics and exasperate the trends of failing to attract non - u.s . scientists , as indicated in the previous section . questions in this section highlighted the science that is being planned during snowmass 2013 . we asked survey participants such questions as : which frontier will have the greatest impact in the next 10 years on hep , to indicate which of the planned experiments ( given from a non - exhaustive list ) did the they have the highest priority for , and whether or not they would encourage other people to pursue the science within their frontier . some general trends were observed about the physics outlook of those in hep responding to our survey : 1 . * would you encourage other young physicists to pursue a career in your frontier ? * + more than 75@xmath1 of the respondents would recommend other talented young physicists to pursue a career in their frontier . this is a trend that is shown to be true across all the frontiers and for both young and senior members of hep . this particular fact is remarkable given the rather pessimistic outlook for funding and jobs and demonstrates that the science is found to be very compelling . + + 2 . * which of the following frontiers as defined by the snowmass process will have the greatest impact on the landscape of high energy physics in the next 10 years ? * + the results for this question are shown in figures [ fig : impactonhepall ] and [ fig : impactonhepyounold ] . the energy frontier is seen as likely to have the greatest impact in the coming decade . however this fact seems less important than when broken down by frontier of the respondent . our sample contained more energy frontier people , so we received more energy responses . cosmic and intensity frontiers appear in second and third position , respectively . although the trend is the same for both groups , we notice that young scientists mentioned energy frontier and instrumentation less often than seniors , while preferring cosmic , computing and theory slightly more often . + from these responses sub - samples were created based on which frontier people reported to be currently working on . except for theorists , most respondents regarded their own current frontier as the one likely to have the greatest impact in the next 10 years ( 57%,60%,70% for intensity , energy and cosmic frontier , respectively ) . theorists responded with the energy frontier ( 33% ) and cosmic frontier ( 30% ) likely to have the greatest impact . another area we asked respondents was to indicate which of the planned experiments ( given from a non - exhaustive list ) from each of the three major frontiers they gave the highest priority to . section [ sec : intensitysub ] shows the responses for intensity frontier experiments , section [ sec : energysub ] shows responses for the energy frontier , and section [ sec : cosmicsub ] shows the responses for the cosmic frontier . each section is broken down by the current frontier the survey participant lists . figure [ fig : intensityall ] shows a non - exhaustive list of planned experiments from the intensity frontier which the survey respondent was asked to check which of the following experiments they are excited about . the respondent could select more than one experiment . the three intensity frontier experiments receiving the most overall votes are lbne ( 312 votes ) , project - x ( 295 votes ) , and majorana ( 274 votes ) . we looked at which experiments that respondents from different frontiers were the most excited about . figure [ fig : intensityfrontierexp ] shows the excitement for the various intensity frontier experiments broken down by the survey respondents current frontier . the top six intensity frontier experiments respondents were excited about , broken down by their current frontier , is shown in table [ tab : intensityexpfrontier ] . figure [ fig : energyall ] shows a list of planned experiments from the energy frontier which the survey respondent was asked to check which of the following experiments they are excited about . the respondent could select more than one experiment . the three energy frontier experiments receiving the most overall votes are vlhc ( 452 votes ) , muon collider ( 399 votes ) , and linear collider collaboration ( 376 votes ) . figure [ fig : intensityfrontierexp ] shows the excitement for the various energy frontier experiments broken down by the suvery respondents current frontier . the top six energy frontier experiments respondents were excited about , broken down by their current frontier , is shown in table [ tab : energyexpfrontier ] . figure [ fig : cosmicall ] lists the planned experiments from the cosmic frontier which the survey respondent was asked to check which of the following experiments they are excited about . the respondent could select more than one experiment . the three cosmic frontier experiments receiving the most overall votes are icecube ( 454 votes ) , fermi telescope ( 400 votes ) , and dark energy survey ( 381 votes ) . figure [ fig : cosmicfrontierexp ] shows the excitement for the various cosmic frontier experiments broken down by the suvery respondents current frontier . the top six cosmic frontier experiments respondents were excited about , broken down by their current frontier , is shown in table [ tab : cosmicexpfrontier ] . clearly there is excitement about many of the upcoming experiments proposed through snowmass . it is worthwhile remarking that across the frontiers , the experiment seen from within the frontier as having the highest priority never has the highest priority across the other three frontiers . taking the example of the energy frontier , the vlhc is clearly seen as exciting and a high priority within the energy frontier as well as with the cosmic and theory frontier . however , this experiment receives only a third rank within the intensity frontier . similar trends can be said of the intensity frontier and the long baseline experiments as well as the cosmic frontier and the lsst . seeing the priorities vary so much across frontiers indicates that much more work is required to build consensus in the coming months . in addition to our survey reaching those currently working in hep , we endeavoured to reach people who had received their training within hep but have gone on to pursue careers outside of academia . in particular we asked how people on non - academic career paths work lives compare with people pursuing an academic career . in total we received 74 responses from people now working outside of hep . in an effort to expand this sample , a handout was provided at snowmass collecting information from senior scientists about former colleagues who have since left the field . this form can be found at http://snowmassyoung.hep.net/oa_contacts.pdf . we urge any reader who may have contact information to please visit this form and email [email protected] with the information . some general trends we observed from this group responding to our survey : 1 . * before going into your current field of work , did you attempt to find a job in academics ? * + 57@xmath1 of the survey respondents did not attempt to find a job inside academia before pursuing their career . + 2 . * how many hours per week ( weekend ) do you work on average ? * + respondents from non - academic career paths report spending , on average , 50 hours per week at work , of which 7 hours are spent on the weekend working . this trend mirrors the respondents from academic career paths who report , on average , 49 hours per week at work , of which 7 hours are spent on the weekend working as well . this result ran counter to many of our intuitions about who works more both inside and outside of an academic career path . + 3 . * in the last year , how many weeks have you had to travel for work related to your current position ? * the overwhelming majority of respondents from non - academic career paths report typically less than 4 weeks for work related travel in the last year . while the majority of respondents from academic career paths report spending greater than 4 weeks of work related travel . + figure [ fig : oaskills ] shows the main skills that people found most valuable to their current career include programming , data analysis , and statistical analysis . it is also worth noting that skills such as oral communication and technical writing are seen as almost as important in their current job . finally , figure [ fig : oahappy ] shows that the overwhelming majority of the non - academic career path respondents are very happy with their current field of employment . the survey administered by the syp was intended to collect a range of opinions that reflect the physics interests , the career outlook , and the general mood of the field leading up to snowmass 2013 , we collected 1112 total responses , 74 of which came from people who had received their training from hep and have since chosen to pursue a non - academic career path . some broad conclusions we would like to draw from a look at the survey data : 1 . * demographic conclusion : * + the survey contained a large fraction of young people in hep , many of whom had not been participating in snowmass prior to taking this survey and many of which did not plan on attending the snowmass meeting . this makes the viewpoints and opinions expressed even more important as it is likely their voices would go unheard otherwise . * career outlook conclusion : * despite the wide spread perspective that the funding situation in the next ten years will likely be bleak , most young people are excited about the prospect of pursuing a job within hep . moreover , many of the respondents , both u.s . citizen and non - u.s . citizen intend to pursue a job within the u.s . . however , all this can shift if the u.s . misses an opportunity to build the next major physics experiment most relevant to the various frontiers in hep , helping support the idea that the most compelling science will attract the best and brightest minds . 3 . * physics outlook conclusion : * there is a general sense of excitement about the science and the upcoming experiments being proposed at snowmass . this excitement is seen across the frontiers and is shared by both the more senior members as well as the young scientists . however , the breakdown of which experiments that people from other frontiers find exciting shows there is still a lot of work left to do to build consensus . * non - academic career path conclusion : * the self reported work habits seem the mostly the same for both academic and non - academic careers . furthermore , many of the skills learned in hep are seen as valuable skills for those on a non - academic career path . finally , those who have received their training in hep and are now pursuing careers outside of academia are generally very happy with their current careers . 100 * snowmass young physicist movement * http://snowmassyoung.hep.net/ * snowmass young physicist survey * http://tinyurl.com/snowmassyoung * young particle physicists * http://ypp.hep.net/ b. t. fleming , j. krane , g. p. zeller , f. canelli , b. connolly , r. erbacher , g. t. fleming , d. a. harris , e. hawker , m. hildreth , b. kilminster , s. lammers , d. medoff , j. monroe , d. toback , m. sorel , a. turcot , m. velasco , m. o. wascko , g. watts , e. d. zimmerman , * results of the survey on the future of hep * , _ arxiv : hep - ex/0108040 _ august 2001 * inspires * http://inspirehep.net/?ln=en * inspires raw data * http://hoc.web.cern.ch/hoc/jobs_stats2.txt * google forms * http://www.google.com/drive/apps.html * root * http://root.cern.ch/drupal/ * united states census bureau : people and households * http://www.census.gov/people/ * what is your gender ? * what is your current marital status ? * are you a us citizen ? * what is the most advanced degree you currently hold ? * how many children do you have ? * what is your household salary ( usd / year ) ? * in what country did you earn your most recent degree ? * at what type of institution are you currently employed ? * what is your current field of employment ? * how long have you been in your current position ? * in the last year , how many weeks have you had to travel for work related to your current position * before going into your current field of work , did you attempt to find a job in academics ? * what resources did you find useful in finding your first non - academic job ? * have you ever not pursued a job offer because it would require relocating ? * do you maintain a residence in a different location than your spouse / partner because of you work ? * have you found your career options limited due to personal relationships ? * on a scale of 1 to 10 , how well did your hep physics experience prepare you for your current job ? * what skills learned during your hep physics experience were the most valuable to you in your current job ? * would you encourage other young physicists to pursue a career in your field ? * on a scale of 1 to 10 how happy are you with your decision to work in your current field ? * which of the following would you rather be doing now in your career ? * please choose which category best describes your current position * how long have you been in your current category ? * according to the snowmass definitions , which froniter best describes the research you primarily work on ? * according to the snowmass definitions , which frontier do you see yourself working on 5 years from now ? * which of the following institutions do you do the majority of your research at ? * have you been contributing to the pre - snowmass meetings and/or the snowmass process prior to taking this survey ? * do you plan on attending the snowmass on the mississippi in minneapolis , mn in july 2013 ? * do you maintain a residence in a different location than your spouse / partner or children in order to work or study ? * during your current position , which of the following best describes your primary responsibility ? * in the last year , how many weeks have you had to travel for work related to your current position ? * which of the following best describes your primary reason for your work related travel ? * in the last year , how many conference talks did you give ? * select two institutions which you most desire to be employed with after completing your training in high energy physics * do you intend to pursue a permanent career in academia ? * what do you think are the odds that you will obtain a permanent academic position ? * what percentage of postdocs in hep do you think go on to permanent positions in hep ? * how much consideration have you given to pursuing a job outside of academia ? * in the future , will you be searching for a permanent position in the us or abroad ? * if the next major expreiment in your frontier was built outside the us , would you be more inclined to search for a permanent position outside the us ? * would you encourage other young physicist to pursue a career in your frontier ? * have you found your career options limited due to personal relationships ? * * please grade each of the following career related concerns in terms of which you find the most important to you and your future in high energy physics . * * availability of university based jobs * availability of laboratory based jobs * funding for large scale long lead time experiments in the future * funding for small scale short lead time experiments in the future * proximity to your home institution of future planned experiments * bureaucracy and administrative difficulties to conduction research * on a scale of 1 to 10 ( 1 = funding will stop , 10 = funding will thrive ) how do you feel about the funding within your frontier within the next decade * * which of the following statements do you most agree with * * i believe we should invest the majority of our resources in one big project next ( e.g. linear collider , lsst , long baseline neutrino experiment , muon collider , vlhc , etc ... ) * i believe we should invest the majority of our resources in a larger variety of smaller experiments ( e.g. mu2e , project 8 , lar1 , ms - desi etc ... ) * * which of the following frontiers as defined by the snowmass process will have the greatest impact on the landscape of high energy physics in the next 10 years ? * * pick from a list of future planned experiments which you are most excited about : * * intensity frontier experiments * energy frontier experiments * cosmic frontier experiments * how many hours per week do you work ( on average ) ? * how many hours per weekend do you work ( on average ) ? * how many hours per week do you spend in meetings ( on average ) ? * do you think your training in academia is adequate to prepare you for your career path ?
from april to july 2013 the snowmass young physicists ( syp ) administered an online survey collecting the opinions and concerns of the high energy physics ( hep ) community . the aim of this survey is to provide input into the long term planning meeting known as the community summer study ( css ) , or snowmass on the mississippi . in total , 1112 respondents took part in the survey including 74 people who had received their training within hep and have since left for non - academic jobs . this paper presents a summary of the survey results including demographic , career outlook , planned experiments , and non - academic career path information collected .
the distribution of major semiaxes ( separations ) of double and multiple stars is a fossil record of the conditions at star formation , as well as of the processes of dynamical evolution , including the dissociation of wide binaries produced= by encounters with massive objects : molecular clouds , spiral arms , etc . our long - standing interest in these topics has led us to investigate the frequency distribution of major semiaxes ( separations ) of wide binaries as a function of age . in the past , two main distributions of major semiaxes have been proposed : @xmath4(1 ) a power - law frequency distribution _ f(a ) _ @xmath5 _ a_@xmath6 @xmath4(2 ) a gaussian distribution in log _ p _ or log _ when @xmath7 = 1 in the power - law distribution , we have the well known oepik ( 1924 ) distribution ( od ) . on the other hand , kuiper ( 1935 , 1942 ) proposed the gaussian distribution , which was further elaborated by heintz ( 1969 ) . more recently , duquennoy & mayor ( 1991 , dm ) again proposed a gaussian distribution in log _ p _ , valid throughout the interval 1 @xmath2 log _ p _ ( days ) @xmath2 10 . our interest in the subject led us to construct a catalogue of= wide binaries in the solar vicinity , based on the catalogue of nearby stars of gliese and jahreiss ( 1991 , gj ) ; see poveda ( 1994 ) for details . we have also constructed= a list of common proper motion binaries in the orion nebula cluster ( age 10@xmath8 years ) , extracted from the jones & walker ( 1988 ) catalogue of proper motions . ( poveda & hernndez - alcntara 2003 ) . in our search for evolutionary effects in the observed distributions we have also looked at the oldest stars in the galaxy . for this purpose we constructed a catalogue ( allen 2000 ) of common proper motion companions to the lists of high velocity metal - poor stars of schuster ( 1988 ; 1989a ; 1989b ) , with ages of about @xmath9 years . in all our catalogues , as well as in the luyten double star catalogue ( 1940 - 1987 , lds ) and in chanam & gould s catalogue ( 2004 , cg ) , we confirm our previous findings ( poveda 1997 ; poveda & allen 2004 ) , that the separations follow oepik s distribution ( od ) in an interval that is bounded at the lower end ( _ a _ @xmath1 100 au ) by the process of close binary and protoplanetary disk formation , and at large separations ( @xmath10 2500 au ) by the dissociation effects produced by encounters with massive objects . it can be shown that oepik s distribution in the plane ( _ n _ , log _ p _ ) is a horizontal straight line which is quite consistent ( within its error bars ) with dm distribution in the interval 2.44 @xmath2 log _ p _ ( years ) @xmath2 5.44 , which corresponds to 53 @xmath2 _ a _ ( au ) @xmath2 5500 . ( poveda 2004 ) . since a great number of binaries from many different and largely independent sources confirm the validity of oepik s distribution , and since there is no stellar formation or single physical process able to produce a gaussian distribution valid in the interval 1 @xmath2 log _ p _ ( days ) @xmath2 10 , we propose to abandon the gaussian representation for _ a _ @xmath11 100 au . on the contrary , oepik s distribution , which is equivalent to a surface density of secondaries @xmath12(_a _ ) @xmath1 _ a_@xmath13 , has a physical interpretation . in fact , this distribution is similar to the run of surface brightness in globular clusters ( king 1962 ) or to that of elliptical galaxies ( hubble s 1930 , or de vaucouleurs=92 1953 distributions ) . in both cases the physics behind such distributions is well known : it is the result of energy relaxation . the similarity of od to the surface brightness in clusters and elliptical galaxies indicates that binaries are not born alone ; at birth , they must be subject to a process of energy relaxation which can not be produced by two - body encounters , i.e. , stars must be formed in groups of multiplicity @xmath14 . this paper is organized as follows . in section 2 we study the distribution _ f(a ) _ for a volume - complete sample extracted from our catalogue ( poveda 1994 ) . section 3= examines the distribution _ f(a ) _ in our catalogue of very young ( _ t _ @xmath2 10@xmath8 years ) , common proper motion binaries in the orion nebula cluster ( poveda & hernndez - alcntara 2003 ) . in section 4 we examine another sample of wide binaries , namely chanam & gould ( 2004 ) common proper motion binaries from the revised nltt ( salim & gould 2003 ; gould & salim 2003 ) . again , the binaries in this catalogue follow od . section 5 examines the physics behind oepik s distribution , and section 6 presents our conclusions . , i.e. , a volume - complete sample.,width=316,height=211 ] for the 189 binaries of the complete sample in poveda ( 1994 ) , ie , for systems with @xmath15 and luminosity class iv - v . the straight ine is a fit for oepik s relation . the ks test for this fit gives a value of @xmath16 in the interval ( 133au , 2640au ) . ( b)right . comparison between oepik s distribution and the cumulative numbers of log @xmath17 for the binaries in the poveda ( 2004 ) catalogue . the straight line is a fit to oepik s distribution for systems satisfying @xmath18 ( au ) @xmath19 . the ks test gives a value of @xmath20 for the interval @xmath18 ( au)= @xmath21.,width=460,height=211 ] with the purpose of detecting the effects of dissociation of weakly bound binaries with the passage of time , we constructed a catalogue of wide binaries ( 305 double systems , 26 triples and 3 quadruples with @xmath10 25 au ) , ( poveda 1994 ) . a sub - sample of this catalogue i.e. , all those systems with primaries of luminosity class v or iv and brighter than absolute magnitude @xmath22 9 is very important , as it is volume - complete . to show this , in figure 1 we plot _ n_(log _ r _ ) vs. 3 log _ r _ , where _ n_(log _ r _ ) is the number of systems ( @xmath239 ) out to a distance _ r_. as can be seen , with the exception of a few very close systems , the great majority follow the relation _ n _ @xmath1 _ r_@xmath24 right to the limit of the catalogue ( _ r _ = 22.5 pc ) , as expected for a volume - complete sample . having a volume - complete catalogue , we proceed to investigate the frequency distribution of major semiaxes . it can be shown that an equivalent representation of od is the cumulative distribution _ n_(@xmath2 log _ a _ ) @xmath25 log _ a_. in the plane _ n_(@xmath2 log _ a _ ) - log @xmath17 , od is a straight line . in general , we will favor the analysis of the cumulative distribution _ f(a ) _ in order to reduce the noise introduced by small number sampling fluctuations . since most of the wide binaries ( in our catalogues ) do not have reliable orbits because of their long periods , we have to use angular separations _ s _ and distances to give projected separations in astronomical units . however , a statistical relation between the average value of _ a _ for a given projected separation _ s _ , namely @xmath26 = 1.41_s _ ( couteau 1960 ) , can be used to estimate for each binary a value of @xmath26 . figure 2a shows for our volume - complete sample the run of the cumulative distribution @xmath27 . an inspection of figure 2a shows that this sample closely follows oepik s distribution in the interval 2.12 @xmath2 log @xmath26 @xmath2 3.42 , i.e. , 133 @xmath2 @xmath26 ( au ) @xmath2 2,640 . to evaluate quantitatively how reliably oepik = b4s distribution represents the data , we shall use the kolmogorov - smirnov ( ks ) test , which was developed precisely for cumulative distributions and where no arbitrary binning of the data is required , ( as is the case in the popular chi - square test ) . in figure 2a each binary is plotted , thus _ n_(log @xmath26 ) increases one by one . by least squares we fitted to the data points a number of straight lines , each one defined in the interval ( log @xmath28 , log @xmath29 ) , until we found the best straight line that minimized the residuals and maximized the interval ( log @xmath28 , log @xmath29 ) , i.e. , the one giving the largest interval ( log @xmath28 , log @xmath29 ) in which oepik s distribution reliably represents the cumulative distribution of separations . in figure 2a we give the equation of the straight line that best represents the data , in the interval log @xmath28 = 2.1248 , log @xmath29 = 3.4214 ( @xmath28 @xmath30 133 au , @xmath29 @xmath31 2,640 au ) . having found oepik s distribution for the wide binaries in the interval 133 @xmath2 @xmath26(au ) @xmath2 2640 we now proceed to test , via ks , what is the level of significance of the theoretical distribution od . the closer to 1 the estimator @xmath32 is , the better the theoretical representation ( press 1990 ) . for the present case we find _ q _ = 0.99 , i.e. , we can accept oepik s distribution at a very high level of confidence in the interval ( 133 au , 2640 au ) . for the interval ( 133 au , 3100 au ) we find @xmath20 , also representing= a high level of confidence . in figure 2b , we plot in the same plane as in figure 2a , all the binaries ( 305 ) from our 1994 catalogue . even though this sample is not volume - complete , one can argue that this does not seem to introduce an important bias in the distribution of separations . in fact , as can be seen from figure 2b , the cumulative distribution _ n_(@xmath2 log @xmath26 ) again defines a straight line . repeating the statistical analysis for this catalogue of 305 binaries we find that the ks test gives a value _ q _ = 0.96 for @xmath26 in the interval between 60 au and 2965 au . we now seek an explanation for the limits of validity of the od as shown in figures 2a and 2b , particularly for the volume - complete sample . the following hypothesis is proposed : ( 1 ) at the short end of the distribution @xmath26 @xmath33 100 au ) , any primeval od will be quickly modified by the presence of close binaries and protoplanetary disks ; ( 2 ) at the wide end ( @xmath26 @xmath11 3,000 au ) , encounters with massive objects ( molecular clouds , spiral arms , black holes , machos , etc . ) will gradually dissociate the widest binaries . in fact , in a theoretical paper , weinberg ( 1987 ) estimated that a binary with a major semiaxis smaller than 2000 au has a probability of one to survive ( against dissociation by giant molecular clouds ) during 10 billion years , yet a binary with @xmath26 @xmath30 12,600 au after two billion years has a probability of survival of only 0.5 . the consistency between the results of weinberg and the distributions shown in figs . 2a and 2b gives support to the hypothesis that oepik s distribution is primeval , valid up to @xmath26 @xmath30 45,000 au ( see next section ) , but with the passage of time it gets truncated at the large separations . to further examine this hypothesis , we analyze the distribution of major semiaxes of the binaries in a very young group ( @xmath34 years ) . ( au ) @xmath35 ( 60 binaries ) the ks test gives a value of @xmath36.,width=316,height=211 ] taking advantage of the jones & walker ( 1988 ) catalogue of proper motions of the stars in the orion nebula cluster , we have identified 68 candidate common proper motion= binaries ( poveda & hernndez - alcntara , 2003 ) . jones & walker s determinations of proper motions and infrared magnitudes for 1053 stars of the orion nebula cluster are appropriate for our work , because of the proper motions accuracy ( @xmath37 0.1 arcsec / century ) and also because the faint limiting magnitude ( @xmath38 13 ) . moreover , the one- million year age of the cluster allows us to look into the = 93almost=94 primeval distribution of major semiaxes . the cumulative distribution of major semiaxes for this sample of orion binaries follows very neatly oepik s distribution , ( see figure 3 ) . the ks analysis of the data plotted in figure 3 indicates that od fits the data up to major semiaxes of 45,000 au with a @xmath32 = 0.99 . arcsec ( 523 binaries ) . note that in the interval @xmath39 log @xmath40 ( 10 arcsec @xmath41 arcsec ) which includes 285 binaries , this distribution is consistent with oepik s , which in this plane would be a horizontal straight line . for larger separations , the distribution becomes depopulated similarly to what we found to occur for other samples of binaries ( see text ) . we interpret this departure as due to the dissociation of wide binaries by encounters with massive perturbers.,width=316,height=211 ] chanam and gould ( 2004 , cg ) have assembled a catalogue of wide= binaries based on the revised luyten nltt catalogue by gould & salim ( 2003 ) and salim & gould ( 2003 ) . cg identified 999 common proper motion pairs . good photometry allowed them to construct a reduced proper motion diagram for their binaries . the position of the binary components in this diagram helps to separate= disk main - sequence pairs ( 801 ) from halo subdwarfs ( 116 ) . the large number of binaries in the cg catalogue offers an independent sample to test the validity of od for main - sequence and halo binaries , respectively . according to cg , their catalogue is incomplete for separations _ s _ @xmath2 10 arcsec . since we are trying to establish the frequency distribution of separations of binaries in cg we extracted from their list a sample of disk binaries with= separations _ s _ @xmath11 10 arcsec , i.e. , we rejected 276 binaries closer than 10 arcsec out of 800 disk binaries . for the remaining 524 pairs we plot in figure 4 the frequency distribution of separations @xmath42 the large number of binaries allows to display this frequency distribution with small sampling fluctuations . in this figure we see clearly that @xmath42 is essentially constant in the interval 1 @xmath43 log _ s _ @xmath43 1.55 . for larger separations , i.e. , _ s _ @xmath44 35.5 arcsec , @xmath45 is a decreasing function of @xmath46 . cg noticed this behavior , which they found statistically significant . the constancy of @xmath45 in the interval 10 @xmath2 _ s _ ( arcsec ) @xmath2 35.5 is just what we expect from od . since cg estimate the mean distance of their disk binaries to be 60 pc , we can transform into astronomical units the separations listed by cg . the angular interval where od holds transforms into : 600 @xmath47 ( au ) @xmath48 2,129 , which is equivalent to 840 @xmath48 @xmath26 ( au ) @xmath43 2,981 . , @xmath49 , which are moving away from a center with transverse velocities of the order of 26 , 8 and 28 km s@xmath50 , respectively . these velocities are much larger than those expected for secular relaxation . indeed , they correspond to violent relaxation.,width=192,height=211 ] the analysis of the various samples of wide binaries as shown in figs . 2 - 5 , show convincingly that oepik s distribution holds for _ a _ @xmath11 100 au . the distribution _ f(a ) _ @xmath51 is equivalent to a surface density of secondaries @xmath52(_a _ ) @xmath53 . this surface density distribution is reminiscent of the surface density of stars in globular clusters and elliptical= galaxies ( king 1962 ; hubble 1930 ; de vaucouleurs 1953 ) . the physics involved in these distributions is well understood . in the case of globular clusters we have what we may call secular relaxation , i.e. , a process in which two body encounters gradually modify the energy of a given star over a time scale much greater than the crossing time ; in the case of elliptical galaxies , what we have is a case of initially violent= relaxation , in which the energy of the stars changes rapidly because the potential of= the stellar system ( galaxy ) changes rapidly due to an initial stellar collapse ( van albada 1982 ) or because of galaxy mergers . in the first case ( secular relaxation ) there is equipartition of energy and therefore a tendency for the lighter stars to diffuse to the outer parts of the globular cluster ( this has been observationally confirmed ) ; on the contrary , in the case of violent relaxation there is no equipartition of energy and hence no segregation of stellar masses , which indeed is the case in elliptical galaxies ( no color gradients are observed ) . the process of energy relaxation involves the interaction of several close stars ( _ n _ @xmath54 3 ) , very early in the stellar history . in fact , the wide binaries in the orion nebula cluster follow od even though they are still in the pre - main sequence phase ( t @xmath2 10@xmath8 years ) . we conclude that the process of energy relaxation is not related to interactions with stars in the cluster environment , but rather it is the result of very early interactions in multiple systems . this suggests that stars are formed in multiple systems that quickly relax and assume od , with the possible ejection of one or more single stars . the common proper motion binaries in orion show that there is not enough time for the reverse process to take place , i.e. the formation of double and multiple stars by capture from stars in the cluster . in the process of star formation , the transition from gas to stellar dynamics may lead to a virialized multiple system where 2_t _ @xmath55 , or to a non virialized one , depending on the initial conditions prevailing in the transition phase . if initial condensations ( proto - stars ) are not virialized , 2_t _ + @xmath56 , then the proto - stars will collapse towards the center of mass of the multiple systems . here we meet the conditions of violent relaxation , i.e. , the collapse of the proto - stars will produce a rapid change in the gravitational potential experienced by the stars . in this conditions some stars will be accelerated to velocities larger than those associated with a virialized multiple system . an _ n_-body simulation of this scenario was realized by the present authors many years ago ( poveda 1967 ) with the purpose of finding an alternative explanation for the formation of runaway stars . in those simulations we found not only runaway stars but also that the binaries formed followed oepik s distribution ( allen 1968 ; poveda 2004 ) . figure 5 shows a recent and nearby case of violent relaxation ( rodrguez 2005 ; gmez 2006 ) . in the vicinity of the orion trapezium , in the orion molecular cloud , the proper motions of the heavily obscured b star ( becklin - neugebauer object ) , as well as of the infrared objects @xmath57 and _ n _ imply transverse velocities much larger than those expected to be produced in a virialized multiple system . \(1 ) the study of a large number of wide binaries , mostly from independent sources confirms that the frequency distribution of major semiaxes is _ f(a ) _ @xmath58 , i.e. precisely oepik s distribution . this distribution is truncated at the short end ( @xmath59 100 au ) by the presence of close binaries and protoplanetary disks ; at the large separations the distribution is depopulated by the process of dissociation produced by encounters with massive objects . figure 2 clearly exhibits these effects for the volume - complete sample in the solar vicinity . \(2 ) oepik s distribution in the plane _ ( log _ p _ ) is a constant , which is entirely consistent ( within the error bars ) with duquennoy & major s gaussian distribution in the interval 42@xmath2 @xmath26 ( au ) @xmath2 4213 . \(3 ) since there is no single astrophysical process that would generate a gaussian distribution in log _ p _ ( or in log _ a _ ) over such a wide interval ( 1 day @xmath2 _ p _ @xmath60 days ) and since , on the other hand , oepik s distribution can be explained by the process of energy relaxation in few - body interactions , we propose to abandon the gaussian representation for log _ a _ , in favor of oepik s distribution ( for _ a _ @xmath11 100 au . )
the frequency distribution _ f(a ) _ of major semiaxes of double and multiple systems , and their eccentricities and mass ratios , contain valuable fossil information about the process of star formation and the dynamical history of the systems . in order to advance in the understanding of these questions , we made an extensive analysis of the frequency distribution _ f(a ) _ for wide binaries ( @xmath0 au ) in various published catalogues , as well as in our own ( poveda 1994 ; allen 2000 ; poveda & hernndez - alcntara 2003 ) . based upon all these studies we have established that the frequency distribution _ f(a ) _ is function of the age of the system and follows oepik s distribution _ f(a ) _ @xmath1 1/_a _ in the range of 100 au @xmath2 _ a _ @xmath2 _ a_@xmath3(_t _ ) , where _ a_@xmath3(_t _ ) critical major semiaxis beyond which binaries have been dissociated by encounters with massive objects . we argue that the physics behind the distribution _ f(a ) _ @xmath1 1/_a _ is a process of energy relaxation , analogous to those present in stellar clusters ( secular relaxation ) or in the early stages of spherical galaxies ( violent relaxation ) . the existence of runaway stars indicates that both types of relaxation are important in the process of binary and multiple star dynamical evolution .
in recent years , systems of cold atoms have attracted interest of many condensed - matter physicists . in particular , cold atoms put on an optical lattice sometimes regarded as a simulator " to study canonical models of strongly correlated electron systems@xcite . these systems are quite controllable and impurity effects are suppressed . systems of single - species bosonic cold atoms are described by the standard bose - hubbard model@xcite , which explains the mott - superfluid phase transition . for two - component bosonic systems , even the mott phase exhibits rich behavior@xcite . at total filling factor one and with strong inter and intra - species repulsive interactions , the two - component bose - hubbard model reduces to an effective spin model , i.e. heisenberg spin model@xcite , which has a nontrivial phase diagram . the experimental realization of bosonic mixture of @xmath0rb -@xmath1k in a three - dimensional optical lattice is an important development@xcite . the two - component bose - hubbard model at commensurate fillings has been studied in e.g. , refs.@xcite by the mean - field - theory ( mft ) type approximations and the numerical methods , and its one - dimensional counterpart by the tomonaga - luttinger liquid theory in ref.@xcite . phase diagrams of these systems were clarified . from the view point of the strongly - correlated condensed matter physics , it is very interesting to study the two - component bose - hubbard model at _ fractional fillings and with strong repulsions_. then , we have started study of the bosonic t - j model@xcite , which is an effective model of the two - component bose - hubbard model at the above limit and is also a kind of bosonic counterpart of the t - j model for the high-@xmath2 superconducting phenomena@xcite . the bosonic t - j model ( b - t - j model ) describes two - component bosonic cold atoms in an optical lattice with strong inter and intra - species repulsions that can be controlled in the experiments . our previous studies on the b - t - j model mostly employed numerical monte - carlo ( mc ) simulations to clarify its phase diagram at finite temperature ( @xmath3)@xcite . in contrast to the fermionic t - j model , the numerical study can be done without any difficulties for some cases of the b - t - j model . in this paper , we shall derive effective fields theories of the b - t - j model and study them by both analytical and numerical methods . we mostly focus on the b - t - j model on square and triangular lattices at @xmath4 and the phase diagram for quantum phase transitions . the obtained results are compared with the previous findings and numerical study of finite-@xmath3 systems . this paper is organized as follows . in sec.ii , we shall introduce the b - t - j model and derive first - version of an effective field theory by using the path - integral methods . the local constraint is faithfully treated by using the slave - particle representation . by integrating out amplitude modes of the slave particles , we obtained an extended quantum xy model , which describe bose condensation of two atoms and pseudo - spin degrees of freedom . numerical study by the monte - carlo ( mc ) methods can be performed without any difficulties and the phase diagram is obtained for the b - t - j model on both the square and triangular lattices in sec.iii . in sec.iv , we shall derive second - version of the effective field theory from the extended quantum xy model by introducing collective fields for the two bosons and the pseudo - spin degrees of freedom . by using this effective field theory , we obtain phase diagram of the b - t - j model and verify the consistency of the results . furthermore we study low - energy excitations , i.e. , the nambu - goldstone bosons in various phases in the phase diagram . in sec.v , we numerically shall the b - t - j model in the stacked triangular lattice at finite @xmath3 . obtained phase diagram has a similar structure of the model in the triangular lattice at @xmath4 . section vi is devoted for conclusion and discussion . hamiltonian of the b - t - j model , which will be studied in this paper , is given as@xcite , @xmath6 where @xmath7 and @xmath8 are boson creation operators@xcite at site @xmath9 , pseudo - spin operator @xmath10 with @xmath11 , @xmath12 are the pauli spin matrices , and @xmath13 denotes nearest - neighbor ( nn ) sites of the lattice . we shall consider both the square and triangular lattices in the following study . physical hilbert space of the system consists of states with total particle number at each site less than unity ( the local constraint ) . in order to incorporate the local constraint faithfully , we use the following slave - particle representation@xcite , @xmath14 where @xmath15 is a boson operator that _ annihilates hole _ at site @xmath9 , whereas @xmath16 and @xmath17 are bosons that represent the pseudo - spin degrees of freedom . @xmath18 is the physical state of the slave - particle hilbert space . the previous numerical study of the b - t - j model@xcite show that there appear various phases including superfluid with bose condensation , state with the pseudo - spin long - range order , etc . for the most of them , the mc simulations show that density fluctuation at each lattice site is not large even in the spatially inhomogeneous states like a phase - separated state . from this observation , we expect that there appears the following term effectively , @xmath19 where @xmath20 etc are the parameter that controls the densities of @xmath21-atom and @xmath22-atom at site @xmath9 , and @xmath23 controls their fluctuations around the mean values . it should be remarked here that the expectation value of the particle numbers in the physical state @xmath18 are given as @xmath24 and similarly @xmath25 , where @xmath26 is the number of sites of the lattice and @xmath27 denote the trace over the states satisfying the local constraint , i.e. , the physical - state condition ( [ const ] ) . therefore the constraint ( [ const ] ) requires @xmath28 at each site @xmath9 . the values of @xmath29 and @xmath30 are to be determined in principle by @xmath31 and filling factor , but here we add @xmath32 to @xmath33 by hand and regard parameters in @xmath32 as a free parameter . in this sense , we are considering an extended bosonic t - j model . it should be stressed here that the strong on - site repulsions between atoms enhance the effects of spatial lattice and as a result quasi - excitations with lorentz - invariant dispersion can appear in a non - relativistic original model like the bose - hubbard model@xcite . the existence of @xmath32 in eq.([hv ] ) is very useful for study of the quantum many - particle systems . in the path - integral representation of the partition function @xmath34 , the action contains the imaginary terms like @xmath35 , where @xmath36 is for @xmath37 and @xmath38 is the imaginary time , i.e. , @xmath39 \exp\big[-\int d\tau\big(\bar{\varphi}_{1i}(\tau)\partial_\tau \varphi_{1i}(\tau ) \nonumber \\ & & \hspace{1 cm } + \bar{\varphi}_{2i}(\tau)\partial_\tau \varphi_{2i}(\tau ) + \bar{\phi}_i(\tau)\partial_\tau \phi_i(\tau ) \nonumber \\ & & \hspace{1 cm } + h_{\rm tj}+h_{\rm v } \big)\big ] , \label{z}\end{aligned}\ ] ] where @xmath33 is expressed by the slave particles and the above path integral is calculated under the constraint ( [ const ] ) . ( in this paper we set @xmath40 . ) for the existence @xmath32 , we separate the path - integral variables @xmath41 s and @xmath42 as @xmath43 and we integrate out the ( fluctuation of ) radial degrees of freedom . there exists a constraint like @xmath44 on performing the path - integral over the radial degrees of freedom , i.e. , @xmath45 . this constraint can be readily incorporated by using a lagrange multiplier @xmath46 , @xmath47 the variables @xmath45 also appear in @xmath33 , but we ignore them by simply replacing @xmath48 , and then we have @xmath49 where we have ignored the terms like @xmath50 by the periodic boundary condition for the imaginary time . the resultant quantity on the rhs of ( [ integral ] ) is positive definite , and therefore the numerical study by the mc simulation can be done without any difficulty . it should be remarked that the lagrange multiplier @xmath51 in eq.([integral ] ) behaves as a gauge field , i.e. , the rhs of ( [ integral ] ) is invariant under the following gauge transformation " , @xmath52 . in the practical calculation , we shall show that all physical quantities are invariant under the above gauge transformation . various phases can form in the system ( [ htj ] ) at @xmath4 , e.g. , states with long - range spin orders ( fm , af , spiral , etc ) , superfluid with bose condensation of @xmath21 and/or @xmath22 atoms , and superposition of them ( the supersolid ( ss))@xcite . in the following sections , we shall consider several specific cases of the extended b - t - j model @xmath53 and derive effective field - theory model . then we clarify the phase diagram of the b - t - j model by studying the effective model . in this section , we shall study the effective field theory obtained in the previous section on the two - dimensional ( 2d ) square lattice as the first example . one of the simplest case is @xmath54-fm system corresponding to @xmath55 and @xmath56 in eq.([htj ] ) , though an extension to the case @xmath57 is rather straightforward . in this case the pseudo - spin symmetry is @xmath58 , and then we put @xmath59 and @xmath60 . the partition function @xmath61 of the effective theory is obtained as follows from eqs.([z ] ) and ( [ integral ] ) , @xmath62 \exp\big[-a_\tau - a(e^{i\omega_\sigma},e^{-i\omega_\sigma})\big ] , \label{z2}\ ] ] where @xmath63 \label{a1}\end{aligned}\ ] ] where @xmath64 and @xmath65 it is obvious that the integrand in eq.([z2 ] ) is positive definite , and therefore the numerical calculation of @xmath61 by means of the mc simulation is possible without any difficulty . the model ( [ z2 ] ) is a kind of lattice rotor model . for practical calculation , we introduce a lattice for the imaginary - time direction and use the following lattice action @xmath66 corresponding to @xmath67 @xmath68 where @xmath69 denotes site of the space - time cubic lattice , @xmath70 and @xmath71 is the lattice spacing of the imaginary - time direction . we numerically studied the lattice model defined by the lattice action @xmath72 where @xmath73 corresponds to @xmath74 in eq.([a1 ] ) , @xmath75 where @xmath76 and @xmath77 , and @xmath78 denotes the nn sites in the 2d spatial lattice . here it is helpful to notice that the parameters @xmath79 and @xmath80 are dimensionless . we fixed the value of the dimensionless parameter @xmath81 and calculated the partition function @xmath34 , the internal energy " @xmath82 and specific heat " @xmath83 as a function of @xmath84 and @xmath85 , @xmath86e^{-a_{\rm lxy } } , \nonumber \\ e&=&\langle a_{\rm lxy } \rangle / l^3 , \nonumber \\ c&=&\langle ( a_{\rm lxy}-e)^2 \rangle / l^3 , \label{ec}\end{aligned}\ ] ] where @xmath87 is the linear size of the 3d cubic lattice . in order to identify various phases , we also calculated the following pseudo - spin and boson correlation functions , @xmath88 where sites @xmath9 and @xmath89 are located in the same spatial 2d lattice , i.e. , the equal - time correlations . for example , @xmath90 as @xmath91 indicates bose - einstein condensation ( bec ) of the @xmath21-atom . for numerical simulations , we employ the standard monte - carlo metropolis algorithm with local update@xcite . the typical sweeps for measurement is @xmath92 samples ) , and the acceptance ratio is @xmath93 . errors are estimated from 10 samples with the jackknife methods . on 3d cubic lattice for various values of @xmath81 . dots denote the observed phase transition point . thick line denotes first - order phase transition line whereas the others are second - order transition lines . system size @xmath95 . , title="fig:",width=151 ] on 3d cubic lattice for various values of @xmath81 . dots denote the observed phase transition point . thick line denotes first - order phase transition line whereas the others are second - order transition lines . system size @xmath95 . , title="fig:",width=158 ] [ pd1 ] and density of state @xmath96 for the phase transitions shown in fig.[pd1 ] . double - peak shape for @xmath97 and @xmath98 indicates first - oder phase transition . @xmath99 and @xmath95 . , width=264 ] [ fig : ne ] and its finite - size scaling function @xmath100 for the second - order phase transition shown in fig.[pd1 ] . @xmath99 and @xmath101 . , title="fig:",width=188 ] + and its finite - size scaling function @xmath100 for the second - order phase transition shown in fig.[pd1 ] . @xmath99 and @xmath101 . , title="fig:",width=188 ] [ fig : fss ] we numerically studied the model with the system size @xmath102 and show the obtained phase diagram for @xmath94 in fig.[pd1 ] using @xmath95 data . there are three phases and they are separated with each other by first or second - order phase transition lines . to identify the first - order phase transition , we calculated the density of state " @xmath96 defined by @xmath103 step function like behavior of @xmath82 and double - peak shape of @xmath96 at the critical point indicate the existence of a first - order phase transition . we show some of the results in fig.[fig : ne ] . on the other hand to identify second - order phase transition , we used nonsingular behavior of @xmath82 , single - peak shape of @xmath96 and system - size dependence of the specific heat @xmath83 in eq.([ec ] ) . for example , see fig.[fig : fss ] . system - size dependence of the specific heat is parameterized as follows , @xmath104 where @xmath105 and @xmath106 are critical exponents , @xmath107 with @xmath108 the critical coupling of @xmath109 , and @xmath100 is the scaling function . estimated values are @xmath110 and @xmath111 . from fig.[pd1 ] , it is obvious that for small @xmath112 and @xmath113 , there exists a paramagnetic state " ( pm state ) without any long - range orders ( lro s ) and its domain in the @xmath114 plane is decreased for increasing @xmath81 . in this pm state , a superposition of @xmath21-atom and @xmath22-atom is realized at each site , but coherence of the relative phase in the superposition does not exists . it should be notice that the atomic bec always accompanies the pseudo - spin lro . the obtained results are in good agreement with the result of the previous study on some related model on 3d cubic lattice at finite @xmath3@xcite . in this subsection , we shall study the b - t - j model on the 2d triangular lattice . as we also discretize the imaginary time , the model @xmath115 in eq.([al1 ] ) is defined on the 3d stacked triangular lattice . in this section we consider the case of the @xmath54-af case with @xmath58 symmetry . therefore we set @xmath116 and @xmath55 in eq.([htj ] ) . as there exists the frustration , we expect that various phases appear in contrast to the @xmath54-fm case . in later section , we also show the result of the numerical study of the b - t - j model on the stacked triangular lattice _ at finite but low t _ , which is closely related to the 2d model at @xmath4 . this close relation between 3d model at low @xmath3 and 2d model at @xmath4 has been previously observed in various systems@xcite . origin of this similarity is somewhat obvious from @xmath117 in eq.([atau ] ) as it can be regarded as an inter - layer coupling . to perform the numerical calculation , we first assign values of the parameters @xmath118 . here we assume a homogeneous state and put @xmath119 as in the previous case . results of more general cases will be reported in a future publication . we numerically studied the system on the triangular lattice for the system size @xmath120 and @xmath121 . in fig.[pd2 ] , we show the obtained phase diagram for @xmath122 and the hole density@xmath123 . order of the phase transitions were determined as in the case of the fm coupling on the square lattice . there are six phases in the @xmath124 plane , and the order of the phase transitions was determined as in the case of the square lattice . for small @xmath112 ( i.e. , small @xmath85 ) , the pseudo - spin degrees of freedom exhibits the 120@xmath125 long - range order . as we assume a homogeneous hole distribution , a superposed state of atoms and hole is realized at each site , but coherent condensation of atoms does not takes place yet . as @xmath112 is increased , phase transition to the states with the 120@xmath125 long - range order and bose condensation take place . for example in the phase b in fig.[pd2 ] , the both atoms bose condense , as the correlation functions shown in fig.[corf1 ] exhibit . in the mean - field approximation , the wave function of that state is given by @xmath126 \\ \nonumber & & \times\prod_{i\in c}[a^\dagger_i+e^{-i{2\pi \over 3}}b^\dagger_i+c_c]|0\rangle,\end{aligned}\ ] ] where @xmath127 s are some complex number and @xmath128 is the empty state of @xmath21 and @xmath22-atoms . this phase will be discussed by using the effective field theory in sec.iv.c and also by the mc simulations in sec.v . as the parameter @xmath85 is increased further , order of the pseudo - spin is destroyed first ( phase c ) and then changes to the fm one in the @xmath129 plane ( phase d ) . . system size @xmath130 . there are six phases . phase of @xmath131 denotes the state with @xmath131 pseudo - spin long - range order without atomic bec . sf stands for superfluid . , title="fig:",width=207 ] . system size @xmath130 . there are six phases . phase of @xmath131 denotes the state with @xmath131 pseudo - spin long - range order without atomic bec . sf stands for superfluid . , title="fig:",width=207 ] [ pd2 ] and @xmath132 ( left ) , @xmath133 and @xmath132 ( right ) . , title="fig:",width=170 ] and @xmath132 ( left ) , @xmath133 and @xmath132 ( right ) . , title="fig:",width=151 ] [ corf1 ] in the previous sections , we numerically studied lattice quantum xy model that describe the two - component bosons with strong repulsions . the obtained phase diagrams show that there exist bose - condensed phases as well as the state of simple pseudo - spin lro . in this section , we shall derive an effective field theory from the quantum xy model , which describes directly the bose condensation and spin order . this field theory not only explains the numerically obtained phase diagram but also reveals low - energy excitations and interactions between them . this kind of effective field theory has been obtained and discussed for a granular superfluid etc by using hubbard - stratonovichi transformation@xcite . in this section we shall employ a slightly different method@xcite . as a result , we reveal some important point that has been overlooked so far . to illustrate the procedure to derive the effective field theory , we first consider a single rotor model as an example that describes superfluid phase transition . after consideration of the simple model , we shall apply similar methods to the quantum xy model for the b - t - j model . we first rewrite the partition function of the rotor model on the square lattice by introducing source terms , @xmath135e^{-{1 \over v_o } \int d\tau \sum \dot{\omega}^2_{i } -\int d\tau h(e^{i\omega_{i } } , e^{-i\omega_{i } } ) } \nonumber \\ & = & \int [ d\omega]e^{-{1 \over v_o } \int d\tau \sum \dot{\omega}^2_{i } -\int d\tau h({\delta \over \delta\eta_{i } } , { \delta \over \delta\bar{\eta}_{i } } ) } \nonumber \\ & & \times i(\eta_{i } , \omega_{i})|_{\eta=\bar{\eta}=0 } , \label{rotorz } \\ i(\eta_{i } , \omega_{i})&= & e^{\int d\tau\sum ( \eta_{i}e^{i\omega_{i}}+ \bar{\eta}_{i}e^{-i\omega_{i } } ) } , \label{sourcerotor}\end{aligned}\ ] ] where @xmath136 and @xmath137 in eq.([rotorz ] ) , we evaluate the path integral of @xmath138 as @xmath139e^{-{1 \over v_0 } \int d\tau \sum_\sigma \dot{\omega}^2_{i}}i(\eta_{i } , \omega_{i } ) \nonumber \\ & & \hspace{1 cm } = e^{\int d\tau \int d\tau'\bar{\eta}_{i}(\tau ) e^{-v_0|\tau-\tau'|}\eta_{i}(\tau ' ) } , \label{intomega}\end{aligned}\ ] ] where we have used @xmath140 and the fact that other correlators like @xmath141 are vanishing . rhs of eq.([intomega ] ) can be expressed by introducing a complex boson field @xmath142 as @xmath143\exp \big[-{1 \over v_0}\int d\tau \phi^\ast_{i}(-\partial^2_\tau+v^2_0)\phi_{i } \nonumber \\ & & \hspace{0.5 cm } + \int d\tau(\eta_{i}\phi_{i}+\bar{\eta}_{i}\phi^\ast_{i})\big ] . \label{phi}\end{aligned}\ ] ] by inserting eq.([phi ] ) into eq.([rotorz ] ) , we obtain the effective field theory of the rotor model with the following action @xmath144 , @xmath145 , \nonumber \\ z_{\rm rotor}&=&\int [ d\phi]e^{-a_{\rm rotor}}. \label{a_phi}\end{aligned}\ ] ] the above derivation from eq.([rotorz ] ) to eq.([a_phi ] ) seems exact because the integration of @xmath146 is the one - site integral and essentially gaussian integration of the free fields . in fact the above manipulation is exact as long as the boson field @xmath147 does not bose condense . on the other hand for @xmath148 , the bose condensation of @xmath147 takes place , i.e. , @xmath149 . one may wonder if the above manipulation is applicable even for this case because eq.([g ] ) seems to indicate _ nonexistence _ of the long - range oder of @xmath147 . furthermore in this case , the mass term of @xmath147 becomes negative , and the integration of @xmath147 in eq.([a_phi ] ) becomes unstable , e.g. , for the square lattice @xmath150 where @xmath151 is the lattice difference operator and @xmath152 is the number of links emanating from a single site and @xmath153 for the square lattice . this instability comes from the fact that the order of the @xmath154-derivative and the @xmath155-integration is _ not interchangeable _ when the bose condensation , i.e. , a phase transition to an ordered state , takes place . it is obvious @xmath156 in the bose condensed state , and therefore it is plausible to expect term like @xmath157 to appear in the effective field theory to stabilize the integration of @xmath147 for the bose condensed state , though explicit calculation to determine the coefficient is difficult . simple spin - wave like approximation for the rotor model , @xmath158 gives the estimation like @xmath159 , and then the coefficient of the @xmath160-term is estimated as @xmath161 for @xmath162 . hubbard - stratonovich transformation derives a similar effective field theory to the above . but its straightforward application gives a negative coefficient of the @xmath163-term indicating an instability of the system . this means that certain step of the derivation is invalid , e.g. , introduction of the hubbard - stratonovichi field and the @xmath155-integration is not interchangeable . this problem is under study and the result will be reported in a separate publication . similar manipulation to the above can be applied to the present quantum extended xy model . from eqs.([z2 ] ) and ( [ a1 ] ) , @xmath164e^{-{1 \over v_o } \int d\tau \dot{\omega}^2_{\sigma i } -a(e^{i\omega_{\sigma i } } , e^{-i\omega_{\sigma i } } ) } \nonumber \\ & = & \int [ d\omega]e^{-{1 \over v_o } \int d\tau \sum \dot{\omega}^2_{\sigma i } -a({\delta \over \delta\eta_{\sigma i } } , { \delta \over \delta\bar{\eta}_{\sigma i } } ) } \nonumber \\ & & \times i(\eta_{\sigma i } , \omega_{\sigma i})|_{\eta=\bar{\eta}=0 } , \nonumber \\ i(\eta_{\sigma i } , \omega_{\sigma i})&= & e^{\int d\tau\sum ( \eta_{\sigma i}e^{i\omega_{\sigma i}}+ \bar{\eta}_{\sigma i}e^{-i\omega_{\sigma i } } ) } , \label{source}\end{aligned}\ ] ] where @xmath165 and we have omitted the gauge field @xmath51 as we consider the only gauge - invariant objects through @xmath166 . then the integration over @xmath167 can be performed as , @xmath168e^{-{1 \over v_0 } \int d\tau \sum_\sigma \dot{\omega}^2_{\sigma i}}i(\eta_{\sigma i } , \omega_{\sigma i } ) = \tilde{i}(\eta_{\sigma i } , \bar{\eta}_{\sigma i } ) , \label{lsource}\ ] ] and the green function of @xmath169 , @xmath170 , is obtained as @xmath171 typical term of @xmath172 is as follows , @xmath173 and this quantity is expressed by introducing complex scalar fields @xmath174 as @xmath175\exp \big[-{1 \over 4v_0}\int d\tau \phi^\ast_{\sigma i}(-\partial^2_\tau+4v^2_0)\phi_{\sigma i } \nonumber \\ & & \hspace{0.5 cm } + \int d\tau(\eta_{\sigma i}\phi_{\sigma i}+\bar{\eta}_{\sigma i}\phi^\ast_{\sigma i})\big ] . \label{phisigma}\end{aligned}\ ] ] by inserting @xmath172 , which is expressed in terms of the boson fields @xmath174 for @xmath176 , into eq.([source ] ) , the action of the effective field theory is obtained as follows , @xmath177 , \label{a0}\end{aligned}\ ] ] where @xmath178 , @xmath179 and @xmath180 . to describe the bose condensed state , we add the @xmath181-terms in the effective field theory and discuss the phase diagram of the system . the final form of the effective action is therefore given by @xmath182e^{-a_{\rm eff}}. \label{aeff}\end{aligned}\ ] ] in this subsection , we shall apply the field - theoretical approach explained in the previous subsection to the b - t - j model on the square lattice as a simple example . furthermore we focus on the fm parameter region @xmath56 . in this case there exists no frustration and therefore it is rather straightforward to obtain the phase diagram . nevertheless study on this system reveals important aspect of the state with multiple long - range orders and structure of the nambu - goldstone bosons . ( [ v1 ] ) and its minimum . symmetric case @xmath183 . parameters are @xmath184 and @xmath185 . , width=272 ] [ pt1 ] this system was studied in sec.iii.a , and we obtained the phase diagram by mc simulations . the potential @xmath186 of the present system is given as follows from the effective field theory in sec.iv.a , @xmath187 from the first term of eq.([v1 ] ) , it is obvious that spontaneous symmetry breaking occurs for @xmath188 , but the second terms that represent the interplay between the order parameters give nontrivial contribution to the phase diagram . in fig.[pt1 ] , we show typical potential and its minimum obtained from eq.([v1 ] ) , which derives qualitatively the same phase structure with that obtained in sec.iii.a . it is interesting to see how gapless low - energy excitations , i.e. , nambu - goldstone ( ng ) bosons , appear in the present system . in the phase with the fm spin order @xmath189 , we set @xmath190 , and then it is obvious that the field @xmath191 describes a ng boson that appears as a result of the spontaneous symmetry breaking of the u(1 ) pseudo - spin symmetry . interesting point is that how many ng bosons appear in the phase with the fm+2sf . if @xmath192 in eq.([a0 ] ) , it is obvious that there exist three ng bosons . however in the original b - t - j model , the symmetry is u(1)@xmath193u(1 ) for the global phase rotation of @xmath21 and @xmath22-boson operators . in order to study the low - energy excitations in the effective field theory , we shall take the continuum description instead of the lattice one though it is not essential . for the fm b - t - j model on the square lattice , it is straightforward to derive effective lagrangian @xmath194 in the continuum spacetime from the effective field theory on the lattice . for example in eq.([a0 ] ) , we put @xmath195 then @xmath194 in the continuum is derived as follows from eq.([aeff ] ) @xmath196 where the potential @xmath186 is given as @xmath197 we first substitute @xmath198 into @xmath199 , @xmath200 and derive the equations that determine the value of @xmath201 as , @xmath202 where @xmath203 next , we introduce three phase fluctuation fields that represent ng bosons in the fm+2fs phase with @xmath204 , @xmath205 by substituting eq.([ngfield ] ) into @xmath186 in eq.([vphi1 ] ) , the quadratic terms of the fluctuating fields are obtained as @xmath206 by using eq.([n0n1 ] ) , @xmath207 it is straightforward to diagonalize the matrix @xmath208 by using a unitary matrix @xmath209 and obtain the mass gap , @xmath210 therefore there are two gapless modes that correspond to the ng bosons , though one may expect three ng bosons as the u(1 ) pseudo - spin symmetry is spontaneously broken and also both the @xmath21 and @xmath22-atoms bose condense . from the above derivation of the mass gap , it is obvious that the cubic coupling @xmath211 plays an essentially important role . the above result can be understood by returning to the original b - t - j hamiltonian @xmath33 in eq.([htj ] ) , and noticing the following correspondence , @xmath212 furthermore the pseudo - spin @xmath213 is composed of @xmath21 and @xmath22-atoms like @xmath10 with @xmath11 in the original b - t - j model . therefore a global u(1)@xmath193u(1 ) gauge transformation of @xmath21 and @xmath22-atoms @xmath214 naturally induces the u(1 ) transformation of the pseudo - spin , @xmath215 this means that the genuine symmetry of the original b - t - j model is not u(1)@xmath193u(1)@xmath193u(1 ) but u(1)@xmath193u(1 ) . the appearance of the term @xmath211 in the effective field theory eloquently shows this fact . then the maximum number of the ng bosons is not three but two , as we explicitly demonstrated in the above calculation . in this subsection we shall focus on the system on the triangular lattice and study the phase structure and low - energy excitations . the obtained results will be compared with those by the numerical study on the quantum xy model given in the previous section and those of the 3d b - t - j model in the following section . to study the system , we assume a homogeneous state and put @xmath216 . the field @xmath217 represents pseudo - spin degrees of freedom , and it is plausible to assume that its condensation has a uniform amplitude @xmath218constant as holes are distributed homogeneously . on the other hand , its phase degrees of freedom has a nontrivial behavior as a result of the frustration . more general cases will be studied in a future publication . we first study the effective action by a mean - field theory like approximation assuming the @xmath219 symmetry . the triangular lattice is divided into three sublattice labeled @xmath220 and @xmath83 sublattices . each field on the sublattice @xmath221 is denoted as @xmath222 etc . it is not so difficult to search the ground state of the potential . for small @xmath223 ( i.e. , small @xmath112 ) , @xmath224 and @xmath225 do not bose condensate . on the other hand , a nontrivial long - range oder of the pseudo - spin appears for @xmath226 . for constant @xmath227 , one can show that the state with the three sublattice coplanar order like @xmath228 appears as the lowest - energy state . this state obviously corresponds to the phase of the @xmath131 spin order in fig.[pd2 ] . to obtain the expectation value of @xmath227 and to study low - energy excitations , we relabel the lattice site by dividing it into three sublattices as @xmath229 where @xmath230 and @xmath83 . then we parameterize the field @xmath217 as @xmath231 effective hamiltonian of the spin part @xmath232 is readily obtained from @xmath194 in ( [ aeff ] ) ( please notice @xmath233 in the af case ) , @xmath234 and substituting eq.([phiabc ] ) into @xmath232 , @xmath235 where @xmath236 denotes the difference operator on the triangular lattice . from eq.([hspin1 ] ) , bose condensation of @xmath237 takes place for @xmath238 , and the classical expectation value @xmath239 of @xmath237 is easily obtained as @xmath240 to study the low - energy excitations , we introduce a complex scalar field @xmath241 as follows , @xmath242 and then from eq.([hspin0 ] ) , we obtain the quadratic terms of @xmath241 in the hamiltonian as follows @xmath243 \label{hspin2}\end{aligned}\ ] ] where @xmath244 we diagonalize eq.([hspin2 ] ) by the bogoliubov transformation and obtain @xmath245 where @xmath246 and its typical behavior is shown in fig.[dispersion ] . as in the limit @xmath247 , @xmath248 then @xmath249 represents a ng boson corresponding to the spontaneous symmetry breaking of the u(1 ) pseudo - spin symmetry . in eq.([heff2 ] ) . parameters are @xmath250 and @xmath251 , width=188 ] [ dispersion ] finally let us study the low - energy excitations in the phase with the @xmath131 spin order _ plus _ 2sf s that corresponds to the phase b in fig.[pd2 ] . from eq.([aeff ] ) , the effective lagrangian @xmath252 of low - energy excitations in this phase is obtained as @xmath253 @xmath254 , \label{veff2}\end{aligned}\ ] ] where @xmath152 is again the number of links emanating from a single site . to obtain the ground state , we adopt the following ansatz for the various condensations , which respects the three - sublattice symmetry , @xmath255 the values of @xmath256 and @xmath257 are determined by substituting eq.([gscondensation ] ) into ( [ leff ] ) and imposing stationary condition , @xmath258 where @xmath259 . it is verified that nontrivial solutions of @xmath260 and @xmath261 exist for sufficiently large @xmath223 and @xmath262 ( i.e. , negative @xmath263 and @xmath264 ) . correlation functions obtained from a typical solution to eq.([gapeq1 ] ) are show in fig.[corrfs ] , which have similar behavior to those obtained by the previous mc simulations of the quantum xy model . as in the square lattice case , we introduce fields that describe low - energy excitations , @xmath265 @xmath266 @xmath267 substituting the above equations ( [ phi1 ] ) , ( [ phi2 ] ) and ( [ phi3 ] ) into @xmath252 in ( [ leff ] ) , we obtain the mass term of the fluctuating fields as @xmath268 where we have used eq.([gapeq1 ] ) . we can easily diagonalize @xmath269 and get eigenvalues , @xmath270 the above result indicates that there are only two ng bosons , though the pseudo - spin u(1 ) symmetry is spontaneously broken and both the @xmath21 and @xmath22-atoms bose condense . in the effective field theory , and @xmath271 , which is numerically obtained in the quantum xy model . we set @xmath272 and @xmath273 . , width=226 ] [ corrfs ] in sec.iii.b and iv.c , we studied the low - energy effective theories of the b - t - j model on the triangular lattice at @xmath4 and obtained phase diagram and low - energy excitations . as we argued previously , if fluctuations of particle density at each site is not so large , the model at @xmath4 reduces to the system with lorentz invariance " , i.e. , a linear - time derivative term change to a quadratic - time derivative term@xcite . in this case , the imaginary time @xmath38 plays a similar role to the inter - layer dimension of the stacked 3d lattice . , and @xmath274 . system size @xmath130 . , width=226 ] [ pd3 ] -atom and @xmath22-atom ( lower panel ) . @xmath129 is the easy plane of the pseudo - spin ( see fig.[scbc ] ) , and length of arrows indicates magnitude of the pseudo - spin . a stripe oder forms and also there are hole - rich regions . @xmath130 . , title="fig:",width=188 ] + -atom and @xmath22-atom ( lower panel ) . @xmath129 is the easy plane of the pseudo - spin ( see fig.[scbc ] ) , and length of arrows indicates magnitude of the pseudo - spin . a stripe oder forms and also there are hole - rich regions . @xmath130 . , title="fig:",width=226 ] [ solid ] in the present section , we shall study the b - t - j model on the stacked triangular lattice by numerical simulations with a quasi - classical approximation . hamiltonian on the stacked triangular lattice is given as follows , @xmath275 where @xmath78 denotes the nn site of the 2d triangular lattice , and @xmath276 is the unit vector in the inter - layer direction . there are ( at least ) two ways of the mc simulation , i.e. , the grand - canonical and canonical ensemble . as in the previous studies in this paper , we employ the canonical ensemble with the particle number of each atom fixed . to impose the local constraint , we employ the slave - particle representation ( [ slave ] ) . then the partition function @xmath34 is given as @xmath277e^{-\beta h_{\rm 3dtj}}. \label{ztj}\ ] ] the path integral in ( [ ztj ] ) is performed by the mc simulation with local update keeping each particle number fixed . we call the calculation ( [ ztj ] ) the quasi - classical approximation as we ignore the berry phase in the action of the path integral . some detailed discussion on the validity and physical meanings of this approximation was given in the previous papers@xcite . and @xmath278 in fig.[pd3 ] . @xmath279 , @xmath280 , @xmath281 , where sites @xmath9 and @xmath89 are located in the same 2d triangular lattice . the result indicates the phase with 120@xmath125 spin order and 2sf s . , title="fig:",width=188 ] and @xmath278 in fig.[pd3 ] . @xmath279 , @xmath280 , @xmath281 , where sites @xmath9 and @xmath89 are located in the same 2d triangular lattice . the result indicates the phase with 120@xmath125 spin order and 2sf s . , title="fig:",width=188 ] [ scbc ] we first show the phase diagram for @xmath282 and the hole density @xmath123 in fig.[pd3 ] , where @xmath283 and @xmath284 are all dimensionless parameters . numerical study was performed for the system size @xmath134 and @xmath121 . for small hopping amplitude @xmath112 , the system forms a solid with voids whose snapshot is shown in fig.[solid ] . as @xmath285 is increased , phase transition to a state with the 120@xmath125 spin order and 2sf s takes place . in fig.[scbc ] , we show the spin and particle correlation functions that verify this conclusion . the previous numerical study of the quantum xy model and the analytical study by the effective field theory predict the existence of this phase . this result again indicates a strong resemblance of the phase diagram of 2d system at @xmath4 and that of the corresponding model in stacked 3d lattice . as @xmath285 is increased further , phase transition to a phase - separated state takes place . in this phase , the system is divided into @xmath21-atom rich region and @xmath22-atom rich region , and in each region a sf of single atom forms . in this paper we studied the b - t - j model of the two - component bosons with strong on - site repulsions . we used the salve - particle representation to treat the local constraint faithfully and derived the effective field theory by integrating out radial degrees of freedom of the slave bosons . the resultant field theory is a kind of extended quantum xy model , and its phase diagram was investigated by the mc simulations . the results shows that the b - t - j model on the square lattice has a simple phase diagram whereas that with the af pseudo - spin coupling on the triangular lattice has rather complicated structure . then we derived the second version of the effective field theory of the b - t - j model by using a hubbard - stratonovich transformation " with the source terms . the resultant field theory describes the pseudo - spin and bose condensation of the @xmath21 and @xmath22-atoms directly . the effective potential of these order parameters clarifies the phase diagram of the system and verified the existence of the interesting phases like @xmath286(pseudo - spin)+2sf s on the triangular lattice . furthermore , we studied low - energy excitations , in particular the ng bosons , in the effective field theory and found that there exist only two ng bosons even in the phase in which the u(1 ) symmetry of the pseudo - spin is spontaneously broken and the both @xmath21 and @xmath22-atoms bose condense . finally we investigated the b - t - j model on the stacked triangular lattice at finite @xmath3 by the mc simulations and showed that a similar phase diagram to that of 2d system at @xmath4 appears . this result means that the direction perpendicular to the 2d triangular lattice plays a similar role to the imaginary - time direction because of the strong on - site repulsion and the spatial lattice structure . it is rather straightforward to treat @xmath287 and @xmath288 as hard - core bosons as in refs.@xcite . as we impose the local constraint excluding the doubly - occupied state in the following calculation , obtained results are qualitatively the same .
in this paper , we consider the bosonic t - j model , which describes two - component hard - core bosons with a nearest - neighbor ( nn ) pseudo - spin interaction and a nn hopping . to study phase diagram of this model , we derive effective field theories for low - energy excitations . in order to represent the hard - core nature of bosons , we employ a slave - particle representation . in the path - integral quantization , we first integrate our the radial degrees of freedom of each boson field and obtain the low - energy effective field theory of phase degrees of freedom of each boson field and an easy - plane pseudo - spin . coherent condensates of the phases describe , e.g. , a magnetic order " of the pseudo - spin , superfluidity of hard - core bosons , etc . this effective field theory is a kind of extended quantum xy model , and its phase diagram can be investigated precisely by means of the monte - carlo simulations . we then apply a kind of hubbard - stratonovich transformation to the quantum xy model and obtain the second - version of the effective field theory , which is composed of fields describing the pseudo - spin degrees of freedom and boson fields of the original two - component hard - core bosons . as application of the effective - field theory approach , we consider the bosonic t - j model on the square lattice and also on the triangular lattice , and compare the obtained phase diagrams with the results of the numerical studies . we also study low - energy excitations rather in detail in the effective field theory . finally we consider the bosonic t - j model on a stacked triangular lattice and obtain its phase diagram . we compare the obtained phase diagram with that of the effective field theory to find close resemblance . = 10000
there is a sufficient number of numerical and self - consistent methods to compute the one particle spectrum of a solid with good accuracy . however , in addition to this , it is desirable to have analytical solutions of the schrdinger equation . these solutions for multidimensional periodic potentials are useful both for a zero order perturbation theory and for the general theory . this allows us , for example , to gain a deeper understanding of spectral manifolds and corresponding eigenfunctions , as well as their geometrical and analytical properties . among one - dimensional potentials , the finite - gap ones are the most interesting and fruitful in the view of the physical results . the schrdinger equation with the finite - gap potential @xcite has an exact solution , the spectrum and eigenfunctions are defined by analytical functions . the kronig - penny potential @xcite consisting of rectangular wells is usually considered as the most elementary one in the quantum physics of solids . however , in order to determine the band structure that corresponds to this potential , it is necessary to solve transcendental equation and , furthermore , the appropriate eigenfunctions are too cumbersome to use them to calculate the matrix elements of any observables . the finite - gap potentials are free of these limitations . the matrix elements of any observable on the eigenfunctions that correspond to finite - gap potential can be calculated analytically by the residue method . furthermore , one can say that these potentials play the same role in solid - state physics as the kepler problem plays in atomic theory . it is known that generic periodic potential has generally an infinite set of gaps in the spectrum , but their width rapidly decreases as energy increases . if a potential is analytical , then the rate of decrease is exponential . therefore , provided narrow enough gaps in the spectrum are disregarded , any periodic potential can be approximated by a finite - gap potential . @xcite it turns out also that the finite - gap potentials are exact solutions of the peierls - frhlich problem on a self - consistent state of the lattice and conduction electrons . @xcite the simplest finite - gap potentials are the lame ones . @xcite they are expressed through the weierstrass function . the lame potentials are a subset of the darboux potentials , @xcite that are linear superpositions of the weierstrass functions with half - period shifts . whatever the number of gaps , all these potentials are characterized by only two parameters of the elliptic function . treibich and verdier rediscovered the darboux potentials and described them in algebraic - geometric terms of the tangent covers and studied their spectral properties . @xcite krichever suggested considering elliptic finite - gap potentials from the point of view a torus covering . @xcite using this approach , smirnov has presented a large list of these potentials . @xcite its and matveev derived a general formula that expresses an arbitrary finite - gap potential through the multidimensional riemann @xmath2-functions , the parameter number of which coincides with the number of band edges . @xcite there are also multidimensional elliptic calogero - moser potentials , which their integrability was proved by olshanetsky and perelomov . @xcite baryakhtar , belokolos and korostil suggested considering the schrdinger equation with three - dimensional separable potential as a sum of three one - dimensional finite - gap potentials along orthogonal directions . @xcite it is obvious that these potentials correspond to the orthorhombic lattice . the schrdinger equation with such a potential has the exact eigenfunctions and the spectrum is defined by analytical functions . the parameters that this model includes have simple physical sense . these parameters are related to the width of spectral gaps . as an example , let us consider separable elliptic finite - gap potentials . the weierstrass function has two independent parameters : the parameter @xmath3 defines the potential period , the parameter @xmath4 defines the gap width along corresponding direction of dual lattice . it is natural to consider the parameter @xmath5 as a characteristic of the relative width of gaps . as the value of @xmath6 increases , the corresponding gap width increases too . in one of the extreme cases , the width of all the gaps vanishes . it corresponds to the free electron model , while in the other extreme case , the width of all the bands vanishes and this corresponds to the tight binding model . in the model of separable finite - gap potentials , we can also easily study one- and two - dimensional lattices . this can be done by choosing extremely wide gaps along certain directions . finite - gap potentials have successfully been applied to solve different problems in solid - state physics . @xcite in our paper , on one hand , we examine the known one - gap and two - gap separable lame potentials in detail . for the one - dimensional case , we construct the dispersion relation graph @xmath0 and for the three - dimensional case , we construct the fermi surfaces in the first and second bands . the following pictures illustrate a change from one extreme case of free electrons to the other extreme case of tight binding electrons . on the other hand , we use the separable one - gap lame potential to describe the lifshits electron phase transition of @xmath1 kind and derive some exact expressions . at the end , we consider singularities of the second derivative of the magnetic momentum of a normal metal in an external magnetic field . the paper is organized as follows . in section 2 , we present information on the one - dimensional lame potentials using two different approaches and the one - gap and two - gap ones are studied in detail . in section 3 , we consider the separable finite - gap lame potentials and construct the fermi surfaces using the one - gap one . in section 4 , the one - gap separable lame potential is applied to describe the lifshits electron phase transition of @xmath1 kind . at the end of this section we consider singularities of the second derivative of the magnetic momentum of a normal metal in an external magnetic field . the lame potentials can be represented in one of the two forms ( since these potentials are expressed through the elliptic functions which have two independent periods ) @xmath7 @xmath8 were @xmath9 is a number of gaps in the spectrum , and @xmath10 is the elliptic weierstrass function with a real half - period @xmath4 and an imaginary one @xmath3 . we use the standard notation of the elliptic function theory . @xcite if it does nt cause misunderstanding we will write simply @xmath11 . the potential @xmath12 has the period @xmath13 , the period of the potential @xmath14 is @xmath15 . since the weierstrass function is homogeneous and it is nt changed under unimodal transformations , the potentials ( [ 1b ] ) and ( [ 2b ] ) are expressed one through the other : @xmath16 . we put @xmath17 , and @xmath18 ; correspondingly @xmath19 and @xmath20 , also @xmath21 . in order to pass from the spectrum of the schrdinger operator with the potential @xmath14 to the spectrum of the schrdinger operator with the potential @xmath12 , it is necessary to pass from @xmath22 to @xmath23 in all the expressions . this means that we have to write @xmath24 instead of @xmath25 , @xmath26 instead of @xmath27 and @xmath28 instead of @xmath29 in expressions for the spectrum that will be defined later . we will examine the potentials @xmath14 only . the dimensionaless schrdinger equation @xcite @xmath30 has two bloch eigenfunctions which correspond to different signs of the wavevector @xmath31 @xmath32 in order to pass to dimensional values it is necessary to multiply @xmath25 , @xmath14 by @xmath33 and spacial variable @xmath34 by @xmath35 , were @xmath35 is a characteristic length of the crystal potential ; @xmath36 is a particle mass and @xmath37 is the planck constant . w(e ) is the wronskian of the functions @xmath38 , @xmath39 , and @xmath40 is a polynomial with respect to @xmath25 . the roots of this polynomial are spectral edges , @xmath41 the bloch eigenfunction @xmath42 and wavevector @xmath29 are holomorphic functions with respect to complex variable @xmath25 and they are defined on the riemann surface @xmath43 . @xcite this surface will be denoted by @xmath44 . the @xmath45 is a product of the two bloch eigenfunctions of the schrdinger equation ( [ 3b ] ) . since the eigenfunction @xmath46 at @xmath47 , it follows that @xmath45 with respect to the variable @xmath25 is a polynomial of the @xmath9-th order , @xmath48 the roots @xmath49 are disposed inside gaps or on their edges and there is only one root for every gap . from the @xmath45 definition , it follows that functions @xmath50 are periodic with a period @xmath15 . when the variable @xmath34 is changed , the function @xmath49 takes on values in the @xmath51-th gap and extreme values of this function , which are defined by equation @xmath52 coincide with the @xmath51-th gap edges . in every band the @xmath45 maintains the sign . when it passes from one band to another the sign changes . in the first band the @xmath53 it is easy to verify that @xmath45 satisfies the third order equation @xmath54 \partial_{x}\lambda \nonumber \\ & + 2 \ell(\ell+1 ) \wp_{x } \left(i x + \omega \right)\lambda=0 . \label{8b}\end{aligned}\ ] ] functions @xmath55 or coefficients @xmath56 can be found if we insert the @xmath45 into the equation ( [ 8b ] ) . if we know function @xmath45 , we can find the band edges by equation @xcite @xmath57 \lambda^{2}(x , e ) \nonumber \\ & -2\lambda(x , e ) \partial^{2}_{x}\lambda(x , e)+(\partial_{x}\lambda(x , e))^{2}. \label{9b}\end{aligned}\ ] ] the wavevector @xmath58 is defined by expression @xmath59 where @xmath60 is the averaging with respect to the variable @xmath34 . when @xmath25 takes on values in a gap , the function @xmath45 alternates its sign and the integral ( [ 10b ] ) exists only in sense of the principal value . the function @xmath61 is periodic with respect to the variable @xmath34 and has a period @xmath62 . it means that @xmath63+c,\ ] ] where @xmath64 is a certain periodic function with respect to the variable @xmath34 with a period @xmath62 . this function and the function under the integral have poles in the same place . the expression ( [ 10b ] ) can be transformed into a more convenient form . we will use the following designation @xmath65 then , if we pass to the new function @xmath66 in the expression ( [ 9b ] ) and differentiate it with respect to the variable @xmath25 , we obtain @xmath67\frac{\partial \chi}{\partial e } \nonumber \\ & -\frac{\chi ' ' } { \chi } \cdot \frac{\partial \chi}{\partial e}+\frac{\chi ' } { \chi } \cdot \frac{\partial \chi'}{\partial e}. \label{11b}\end{aligned}\ ] ] here and further , we denote a derivative with respect to @xmath34 by prime . the function @xmath68 is periodic with respect to @xmath34 , and @xmath69 using formulas ( [ 9b ] ) and ( [ 11b ] ) we obtain the following equality , @xmath70 using formulas ( [ 10b ] ) , ( [ 12b ] ) and ( [ 13b ] ) we derive the expression for wavevector @xmath71 the expression ( [ 14b ] ) defines a dependence of the electron energy @xmath25 on the wavevector @xmath72 this function is monotonous , but piecewise analytical , it corresponds to the extended band scheme . in the contracted band scheme the energy is periodic analytical function of wavevector in dual lattice with a period of this lattice ( in our case this period is equal to @xmath73 ) . let us introduce @xmath9 parameters @xmath74 by means of the expression @xmath75 then , in terms of these parameters , we can present the functions @xmath76 and @xmath58 in another form , @xcite @xmath77\exp{\left ( i x \sum_{j=1}^{\ell}\frac{\eta'}{\omega'}t_{j } \right ) } \exp(ikx ) , \label{15b}\end{aligned}\ ] ] @xmath78\exp{\left ( -i x \sum_{j=1}^{\ell}\frac{\eta'}{\omega'}t_{j } \right ) } \exp(-ikx ) , \label{16b}\end{aligned}\ ] ] @xmath79 inserting the expression ( [ 17b ] ) into ( [ 8b ] ) , we derive a system of the algebraic equation @xcite to determine the quantities @xmath80 : @xmath81 where @xmath82 and @xmath83 is the symmetric function @xmath84 using the first equation in the system ( [ 19b ] ) , we can express @xmath25 through the quantities @xmath80 and derive the following expression @xmath85 below we describe in detail the spectrum of the schrdinger equation with the one - gap and two - gap lame potentials . the band edges are the following @xmath87 the expression @xmath45 for the one - gap potential is the following @xmath88 the spectrum of the schrdinger equation ( [ 3b ] ) with the one - gap lame potential in parametric form is the following @xmath89{rcl } k&=&\zeta(t)-\frac{\eta'}{\omega'}\:t , \\ e&=&\wp(t ) . \end{array } \right . \label{24b}\ ] ] the expression for dependence @xmath29 on @xmath25 can be written in the form of elliptic integral @xmath90 in fig . [ figure1 ] , the complex plane and the fundamental domain of the weierstrass function @xmath91 are represented . when @xmath25 takes on values in the first band the parameter @xmath92 takes on values in the line @xmath93 . the point b corresponds to the minimum energy value , the points @xmath94 and @xmath95 correspond to the maximum energy value . when @xmath25 takes on values in the second band the parameter @xmath92 takes on values in the line @xmath96 . the points @xmath97 and @xmath25 correspond to the minimum energy value , the point @xmath98 corresponds to the maximum energy value ( @xmath99 ) . the weierstrass function fundamental domain on complex plane . ] let us examine limit cases . in the limit case of free electrons the gap in the spectrum vanishes . let us put @xmath100 then @xmath101 and @xmath102 or @xmath103 . in the limit case of a very wide gap , we pass to another variable @xmath104 in the integral ( [ 25b ] ) . using expression @xmath105 and introducing the designation @xmath106 $ ] we derive the following @xmath107 expanding the expression under the integral to the order of @xmath108 and restricting ourselves to the zero approximation , we obtain the spectrum in the limit case of tight binding electrons . @xmath109 where @xmath110 it is possible to determine effective masses at band edges . according to the expression ( [ 25b ] ) , we have @xmath111 that leads to following expansions , @xmath112 @xmath113 here we denote by @xmath114 and @xmath115 respectively the effective masses at the bottom edge and at the top edge of the first band . the effective mass @xmath114 is positive and @xmath115 is negative . one can see that the both effective masses depend on all the band edges . for the two - gap lame potential the band edges are the following , @xmath117 the expression @xmath45 in this case is presented in such a way , @xmath118 \ : [ \wp(i x + \omega ) -\wp(t_{2 } ) ] \nonumber \\ & = 2\left[\frac{3}{2}\left(\wp(i x + \omega ) - \sqrt{g_{2}-3\wp^{2}(i x + \omega ) } \right ) -e\right ] \nonumber \\ & \times \left[\frac{3}{2}\left(\wp(i x + \omega ) + \sqrt{g_{2}-3\wp^{2}(i x + \omega ) } \right)-e \right ] . \label{29b}\end{aligned}\ ] ] the spectrum of the schrdinger equation ( [ 3b ] ) with the two - gap lame potential in parametric form is the following @xmath89{rcl } k&=&\left ( \zeta(t_{1})-\frac{\eta'}{\omega ' } t_{1}\right ) + \left ( \zeta(t_{2})-\frac{\eta'}{\omega ' } t_{2}\right ) , \\ \wp(t_{1})&=&\frac{e}{6}+\frac{1}{2\sqrt{3}}\sqrt{3g_{2}-e^{2 } } , \\ \wp(t_{2})&=&\frac{e}{6}-\frac{1}{2\sqrt{3}}\sqrt{3g_{2}-e^{2 } } . \end{array } \right . \label{30b}\ ] ] the expression for dependence @xmath29 on @xmath25 can be written in the form of hyperelliptic integral @xmath119 when @xmath25 takes on values in the first band the parameters @xmath120 and @xmath121 take on values in the line @xmath93 ( fig . [ figure1 ] ) . when @xmath25 takes on values in the second band the parameter @xmath120 takes on values in the line @xmath93 and the parameter @xmath121 takes on values in the line @xmath96 . in both cases @xmath122 and @xmath123 are real . the situation changes when @xmath25 takes on values in the third band . in this case @xmath122 and @xmath123 are complex conjugate ( @xmath29 and @xmath25 remain real ) , @xmath120 and @xmath121 take on values in the curves which are depicted inside fundamental domain @xmath91 ( while constructing , we used @xmath124 ) . with the help of the hermite reduction , we can express the integral ( [ 31b ] ) through the standard elliptic integrals of the first and the second kinds @xcite and represent the spectrum in other parametric form . let us put @xmath125 using equalities @xmath126 @xmath127\right|^{e}_{e_{0 } } \\ & = \frac{1}{2\sqrt{3}}\int\limits_{y_{0}}^{y}\frac{y d y}{\sqrt{y^{3}-3ay+b}},\end{aligned}\ ] ] and putting @xmath128 , we can write the integral ( [ 31b ] ) in terms of elliptic functions @xmath89{rcl } \wp ( h)&=&\frac{e^{3}+27g_{3}}{9e^{2}-27g_{2 } } \ ; , \\ k&= & \zeta(h)-\frac{\eta'}{\omega ' } h - \frac{2}{3}\sqrt{\frac{(e+3e_1)(e+3e_2)(e+3e_3)}{e^2 - 3g_{2 } } } \ ; . \end{array } \right . \label{32b}\ ] ] in this form the spectrum of the two - gap potential is provided in the paper . @xcite by the same way as in the case of one - gap potential we derive expressions for effective masses in the first band , @xmath129 where @xmath114 and @xmath115 are effective masses at the bottom and at the top band edges respectively . one can see that the effective masses depend on all the band edges , the same as in an one - gap case . in table [ table1 ] , the effective mass values are provided for different values of the parameter @xmath130 , we put a period of the potential @xmath131 . [ ! h ] .[table1]inverse effective masses for the first band in case of the two - gap lame potential . [ cols="^,^,^",options="header " , ] in table [ table1 ] , one can see that with the increase of the value of @xmath6 , the values of @xmath132 and @xmath133 decrease . when the band degenerates into an energy level that corresponds to the limit case of tight binding electrons , @xmath134 . we obtain so called `` heavy '' electrons . in the other limit case of free electrons , @xmath135 tends to the mass of free electron , i.e. @xmath136 . and since the gap width vanishes , @xmath137 in this case . in fig [ figure2 ] , we provide the graphs @xmath138 and @xmath14 for the two - gap potential corresponding to different values of the parameter @xmath139 . as @xmath6 increases , the width of the gaps increase . there are the limit cases of very narrow gaps and very wide gaps in this figure . with increase of @xmath6 , the second band also degenerates into an energy level . dispersion relation @xmath138 and potential @xmath14 in case of the two - gap lame potential for different values of the parameter @xmath140 . ] the one - electron separable hamiltonian has the form @xmath141 in general case @xmath142 but we consider below only the three dimensional case , @xmath143 we will use the finite - gap lame potentials for the potentials @xmath144 . in general case of the orthorhombic lattice every one - dimensional potential @xmath144 is characterized by its own parameters @xmath22 , @xmath145 in fig . [ figure3 ] , the three dimensional one - gap potential for @xmath146 is represented with the following half - periods : @xmath147 . the three dimensional one - gap lame potential for fixed value of the variable @xmath148 , we put @xmath146 , @xmath147 . ] an eigenfunction of the hamiltonian ( [ 36c ] ) is a product of one - dimensional eigenfunctions of every @xmath149 , @xmath150 where @xmath151 @xmath152 is a band number . the expression for energy consists of three terms , @xmath153 the fermi surface is described by the following equation @xmath154 were @xmath155 is the fermi energy . here the wavevector @xmath156 takes on values in the brillouin zone , every solution of this equation for a prescribed value of dual lattice vector @xmath157 defines one of sheets of the fermi surface . the set of dual lattice vectors corresponds to their set in fourier series expansion of a separable potential . if the real electron potential differs a bit from the separable one , then the corresponding corrections can be taken into account using the perturbation theory . let us examine the limit case of very narrow gaps . since the potential periods @xmath158 are fixed , let @xmath159 . then , potentials @xmath160 tend to constants , the dependence of energy on wavevector becomes quadratic and the fermi surface sheets are similar to pieces of spherical surface . thus , our model is the considerable generalization of the harrison method that was successfully used to construct the fermi surfaces of metals . let us examine the other limit case of very wide gaps . since we fixed the potential periods @xmath161 , let @xmath162 . then , @xmath163 . it is obvious that in this case all the bands degenerate into levels . the finite - gap spectrum of tight binding electrons is similar to the spectrum that is obtained in the well - known lcao method . in this case function @xmath164 corresponds to @xmath165-state . the three functions @xmath166 @xmath167 @xmath168 @xmath169 @xmath170 @xmath171 correspond to @xmath172-state as they are proportional to @xmath34 , @xmath173 , @xmath148 respectively when the space variables tend to zero . the six functions @xmath174 , @xmath175 , @xmath176 , @xmath177 , @xmath178 , @xmath179 ; @xmath180 , @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath185 split into five functions that correspond to @xmath186-state and one function corresponds to the @xmath165-state since these functions are proportional to @xmath187 , @xmath188 respectively when the space variables tend to zero . @xcite the fermi surfaces for cubic lattice are represented in fig . [ figure3]-[figure5 ] . they are constructed using the separable one - gap lame potential for different values of the parameter @xmath6 . in fig . [ figure3 ] , the fermi surfaces in the first band are represented at @xmath189 . the fermi surface sheets do nt appear in the second band at this value of @xmath155 . in fig . [ figure4 ] , the open fermi surfaces are represented at @xmath190 in the first band . in fig . [ figure5 ] , the closed fermi surfaces are represented at the same value of @xmath155 in the second band . the fermi surface in the first band for different values of the parameter @xmath140 , the fermi energy is @xmath189 . ] the fermi surface in the first band for different values of the parameter @xmath140 , the fermi energy is @xmath190 . ] the fermi surface in the second band for different values of the parameter @xmath140 , the fermi energy is @xmath190 . ] it is known that the density of states in a crystal @xmath191 . since @xmath192 is a smooth function of the wavevector @xmath156 , it can be expanded into the taylor series about a critical wavevector @xmath193 restricting ourselves to quadratic terms , we obtain @xmath194 where @xmath195 . depending on signs of the coefficients @xmath196 the four types of critical points are distinguished , they are a minimum value , a maximum value and two types of saddle points . in a three dimensional crystal the function @xmath197 at the critical energy @xmath198 is continuous but its derivative at this point has infinite discontinuity . the function @xmath199 is periodic with respect to every component of the wavevector @xmath200 it means the energy @xmath201 is defined on three dimensional torus . this function has a denumerable set of branches and every one of them corresponds to a certain energy band . the function @xmath201 is differentiable and regular so , according to elementary facts of topology , for each of the branches the number of critical points relates to the betti numbers of the manifold , where the function @xmath25 is being defined . thus , every branch of the function @xmath201 has one maximum value , one minimum value and three saddle points of every type . but in a real crystal the bands may overlap and as a result the singularities of the function @xmath197 , that are related to different branches , may compensate each other . in 1952 van hove examined singularities of the elastic frequency distribution of a crystal ( that are similar to singularities of the electron density of states ) and showed that for a three dimensional crystal the frequency distribution function has at least one saddle point of every types and its derivative takes on the value @xmath202 at the upper end of the spectrum . @xcite in 1960 i.m . lifshits connected singularities of the density of electron states with a variation of the fermi surface topology . @xcite he showed that at zero temperature the electron phase transitions occur at the special energies , they were later named the lifshits electron phase transitions of @xmath1 kind . singularities of the conduction electron spectrum at the point of such a transition lead to anomalies of thermodynamic and kinetic quantities . when temperature increases , this transition smoothes out . at present , the van hove singularities are considered as one of the reasons for the origin of high - temperature superconductivity . @xcite let us consider the schrdinger equation with the one - gap separable lame potential that corresponds to the tetragonal bravais lattice , @xmath203 in order to describe the lifshits electron transformation , it is enough to examine the case when the fermi surface is open only at two opposite sides of the brillouin zone . let us choose half periods and the fermi energy so that the fermi surface has the form of a goffered cylinder which is prolate along the direction @xmath204 , as in fig . [ figure7 ] . in this case we have to put @xmath205 ( while constructing , we used @xmath206 ) . then , in order to simplify computations , we can expand potential with respect to the variables @xmath207 and restrict ourselves to zero approximation . the fermi surface in the first band in the form of a goffered cylinder , we put @xmath208 . ] the spectrum of the schrdinger operator with the potential @xmath209 looks as follows : @xmath89{rcl } e(k_{x},k_{y},h_{z } ) & = & k_{x}^2+k_{y}^2 -e_{3}+\wp\left(h_{z } \right ) , \\ k_{z}(h_{z } ) & = & \zeta(h_{z } ) -\frac{\eta'}{\omega ' } \ ; h_{z } \ ; . \end{array } \right . \label{40d}\ ] ] the function @xmath201 has two critical points in the first band that are a minimum value and a saddle point and one critical point in the second band that is a minimum value . we will consider below only the first band . one can see that @xmath210 . let us put @xmath211 . for the first band @xmath212 $ ] . since , the function @xmath199 is even , it is enough to consider the case when @xmath213 $ ] . the expression for the number of states is the following @xmath214 where @xmath215 is dimensionless volume . it easy to show that at the energy @xmath216 the fermi surface opens on the boundary of the brillouin zone which is orthogonal to the axis @xmath217 in this case we can express the number of states ( [ 41d ] ) in such a form , @xmath218 . \label{e1}\end{aligned}\ ] ] here we have introduced the parameter @xmath219 as follows . * a. * under condition @xmath220 we define the parameter @xmath219 by the equality @xmath221 according to this equality , the function @xmath222 is differentiable and has a derivative @xmath223 * b. * under condition @xmath224 we define the parameter @xmath219 by the equality @xmath225 . differentiating this expression for the number of states in @xmath25 and using the definition of parameter @xmath219 we obtain the electron density of states , @xmath226{rcl } \frac{v}{2\pi^2 } | \mathbf{k}(e ) | , & & e < e_{c } \\ \frac{v}{2\pi } \cdot \frac{1}{2| \omega ' | } , & & e > e_{c } \end{array } \right .. \label{43d}\end{aligned}\ ] ] even though this expression was derived for the one - gap potential , it is correct in a case of any @xmath227gap potential . let us expand the function @xmath197 under the condition @xmath228 in the @xmath229 powers . we derive the following formula , @xmath230 it is obvious that the first term in the expansion ( [ 44d ] ) corresponds to the function @xmath231 . let us calculate the first derivative of the density of states . we should distinguish two cases . * a. * under condition @xmath232 we obtain the expression @xmath233 * b. * under condition @xmath224 we obtain the expression @xmath234 the singularity at the point @xmath235 corresponds to the fermi surface appearance in the second band . the singularity at the point @xmath236 corresponds to the minimum value of @xmath25 in the first band ( the fermi surface appears in the center of the brillouin zone ) . the singularity at the point @xmath237 corresponds to the saddle point ( the fermi surface opens on the boundary of the brillouin zone ) . in all three cases the singularities have positive signs . we will consider further only the case @xmath237 , it is obvious that the other two cases can be considered similarly . let us introduce the following notation : @xmath238.\end{aligned}\ ] ] in order to compute the thermodynamics potential @xmath239 we expand the following expressions @xmath240 @xmath241 @xmath242 in the @xmath243 powers . we derive the following expressions : @xmath244 @xmath245 @xmath246 the coefficient @xmath247 is expressed in terms of the effective mass @xmath115 along the @xmath204 direction at the top edge of the first band , @xmath248 hence the coefficient @xmath247 depends on all the band edges and @xmath249 . when the gap width decreases on boundary of the brillouin zone , which is orthogonal to the @xmath204 axis , the value of the coefficient @xmath247 decreases also . in the limit case of extremely wide gap the parameter @xmath250 . in this case the problem from a three - dimensional one becomes two - dimensional and we should use other speculations . in the other limit case when the gap width vanishes the parameter @xmath251 . it is obvious : when the gap width equals zero the value @xmath198 is not a critical point . thermodynamic potential at a normal metal is defined by the following expression @xmath252 using the expression ( [ e1 ] ) we can calculate the integral ( [ e2 ] ) and find the analytical expression for @xmath253 , but this expression will be quite a complicated one and we do not adduce it . using the formula ( [ 49d ] ) we present the electron thermodynamic potential @xmath254 in a neighborhood of the critical value of energy in such a form , @xmath255{lcl } c+a e_{c}+\frac{1}{2 } b e_{c}^2 -a \mu-\frac{1}{2 } b \mu^2 + \frac{4}{15}\alpha ( e_{c}-\mu)^{5/2}+o \left ( ( e_{c}-\mu)^{7/2 } \right ) , & & \mu < e_{c } , \\ c+a e_{c}+\frac{1}{2 } b e_{c}^2 -a \mu-\frac{1}{2 } b \mu^2 , & & \mu > e_{c}. \end{array } \right . \label{50d}\ ] ] the variation @xmath256 of number of states looks as follows : @xmath257{lcl } \frac{2}{3 } \alpha ( e_{c}-e)^{3/2 } , & & e < e_{c } , \\ 0 , & & e > e_{c}. \end{array } \right.\ ] ] then the variation of the electron thermodynamic potential can be written down in such a way @xmath258 we provide below the asymptotic expansion of the expression @xmath259 under condition @xmath260 @xmath261 a similar expression was presented in the paper . @xcite in fig . [ figure8 ] , the graph @xmath262 on the temperature @xmath263 and the chemical potential @xmath264 is represented ( while constructing , we used @xmath265 , @xmath266 , accordingly @xmath267 ) . one can see that when @xmath268 the third derivative from the thermodynamics potential has an infinite discontinuity and @xmath269 third derivative of thermodynamics potential , we put @xmath265 , @xmath270 and @xmath267 . thermodynamics potential , temperature and chemical potential take on dimensionless values . @xmath271 has an infinite discontinuity when @xmath272 . ] one can see also that the singularity at the point @xmath273 smoothes out with the temperature increase . in order to estimate an appropriate temperature interval , we should pass from dimensionless units to kelvin degrees . in order to do this we have to multiply the dimensionless temperature by the following coefficient , @xmath274 where @xmath275 is the boltzmann constant and @xmath276 is a period of crystal potential . if the potential period equals @xmath277 and effective electron mass is put @xmath278 , where @xmath279 - mass of free electron , we find out that the dimensionless temperature value @xmath280 corresponds to the temperature value in standard units @xmath281 . since the width of the area @xmath282 which contributes to the singularity of @xmath283 is proportional @xmath263 , then in a general case the singularity smoothes out slowly enough with the temperature increase . a magnetization of an electron gas in magnetic field is described by the following expression , @xmath284 where @xmath285 is the bohr magneton , @xmath286 is an external magnetic field , @xmath263 is a temperature . according to this expression the magnetization at @xmath287 can be presented in the following way , @xmath288 let us expand the expression ( [ dd ] ) in the @xmath289 powers . using the formula ( [ 49d ] ) we come to the following result , @xmath290{lcl } \frac{\beta}{2}(2\beta h b + \frac{2}{3}\alpha ( e_{c}-\mu-\beta h)^{3/2 } -\frac{2}{3 } \alpha ( e_{c}-\mu+\beta h)^{3/2 } \\ \hspace{6 cm } + o((e_{c}-\mu\pm\beta h)^{5/2 } ) ) , & & \mu < e_{c } - \beta h , \\ \frac{\beta}{2}(2\beta h b - \frac{2}{3 } \alpha ( e_{c}-\mu+\beta h)^{3/2 } + o((e_{c}-\mu + \beta h)^{5/2 } ) ) , & & e_{c } -\beta h < \mu < e_{c } + \beta h , \\ \beta^{2 } h b , & & \mu > e_{c } + \beta h. \end{array } \right.\ ] ] since @xmath291 has infinite discontinuity when @xmath292 then the second derivative of the appropriate magnetic moment with respect to the chemical potential , @xmath293 has singularity also when @xmath292 . the sign of singularity is defined by the sign ahead of @xmath294 in the formula for magnetization , @xmath295{lcl } \frac{\alpha\beta}{4(e_{c}-\mu-\beta h)^{1/2 } } - \frac{\alpha\beta}{4(e_{c}-\mu+\beta h)^{1/2 } } + o((e_{c}-\mu\pm\beta h)^{1/2 } ) , & & \mu < e_{c } - \beta h , \\ - \frac{\alpha\beta}{4(e_{c}-\mu+\beta h)^{1/2 } } + o((e_{c}-\mu + \beta h)^{1/2 } ) , & & e_{c } - \beta h < \mu < e_{c } + \beta h , \\ 0 , & & \mu > e_{c } + \beta h. \end{array } \right.\ ] ] the expression ( [ d ] ) defines the relation of the coefficient @xmath247 to the effective mass @xmath115 . as mentioned above , with the decrease of gap width on boundary of the brillouin zone a value of the coefficient @xmath247 decreases . in the limit case when the gap width equals zero the coefficient @xmath296 . in this case the value @xmath198 is not critical point and there is no singularities that is related to this point . in fig . [ figure9 ] , the graph @xmath297 on the temperature and the chemical potential is represented ( we put @xmath298 ) . magnetization of an electron gas in magnetic field , we put @xmath299 , @xmath270 and @xmath267 . temperature and chemical potential take on dimensionless values , magnetization is measured in bohr magnetons . ] in fig . [ figure10 ] , the graph @xmath300 on the temperature and the chemical potential is represented ( while constructing , we used @xmath299 , @xmath266 , accordingly @xmath267 ) . one can see that when @xmath301 at @xmath287 the second derivative from the magnetic moment has infinite discontinuities and @xmath302 @xmath303 as in the previous case , the singularities at the points @xmath304 smooth out slowly enough when the temperature increases . second derivative of electron gas magnetization in magnetic field , we put @xmath305 , @xmath306 and @xmath267 . temperature and chemical potential take on dimensionless values , magnetization is measured in bohr magnetons . @xmath307 has infinite discontinuities when @xmath308 . ] in this paper , we have examined the one - gap and two - gap separable lame potentials in detail . we have constructed the dispersion relation @xmath0 for one - dimensional case and the fermi surfaces in the first and second bands for three - dimensional case . the pictures illustrate a passage from the limit case of free electrons to the limit case of tight binding electrons . we have provided also analytical expressions for effective masses of electrons in a metal . these expressions depend on all the band edges . these results have been used to study the singularities of the electron part of the thermodynamic characteristics in metals . we have derived explicit analytical expressions for the density of states in a metal and its derivative . then we have examined the thermodynamic potential and magnetic moment of metal in an external magnetic field and extracted the singular part . we have also obtained the relation between the parameter of singularity and corresponding effective mass . all the expressions we derived contain the parameters of our potential only . therefore , if we fix potential , we can calculate all the coefficients of these expressions . we would like also to say a few words about generalization of the finite - gap potentials to lattices of other spacial symmetry . in the paper @xcite the authors have tried to develop such a generalization for hcp lattice by means of the perturbation theory and calculated the fermi surface of beryllium . the elliptic calogero - moser potentials are also of significant interest concerning their use in solid state physics . these potentials are multidimensional .
one - gap and two - gap separable lame potentials are studied in detail . for the one - dimensional case , we construct the dispersion relation graph @xmath0 and for the three - dimensional case we construct the fermi surfaces in the first and second bands . the pictures illustrate a passage from the limit case of free electrons to the limit case of tight binding electrons . these results are used to describe the lifshits electron phase transition of @xmath1 kind and derive some exact expressions . we also examine the singularities of the second derivative of magnetic momentum in an external magnetic field . the parameter of the singularities depends on corresponding effective mass .
magnetic helicity is an important topological property of solar active regions ( ars ) and is a measure of twist and writhe of the field lines @xcite . it is gauge invariant for a closed volume of space . the sun s outer atmosphere is dominated by magnetic field at all scales . dynamic phenomena , such as , energetic flares and coronal mass ejections ( cmes ) occur due to the loss of equilibrium during the evolution of magnetic fields in solar ars . magnetic helicity has become an important physical parameter in the context of solar transient phenomena . it is one of the few global quantities which is conserved even in resistive magneto - hydrodynamics on a timescale less than that of the global diffusion . there exists no absolute measure of helicity within a sub - volume of space if that sub - volume is not bounded by a magnetic surface . however , a topologically meaningful and gauge invariant relative helicity for such volumes can be defined . there are several methods for estimating helicity in solar ars . by the force - free field assumption of coronal magnetic field , we have : @xmath5 where @xmath6 is the force - free parameter , also known as helicity or twist parameter . assuming @xmath6 to be constant for the whole ar , we can fit observed vector magnetograms to deduce the value of @xmath6 @xcite . latitudinal variation of helicity of photospheric magnetic fields , and statistical significance of the observed temporal variations of the ars hemispheric helicity rule , as measured by the latitudinal gradient of the best - fit linear force - free parameter @xmath6 , etc . , have been discussed by @xcite . the poynting - like theorem for helicity in an open volume as derived by @xcite is given by : @xmath7 where @xmath8 is the vector potential of the potential magnetic field , @xmath9 , which is uniquely specified by the observed flux distribution on the surface(x - y plane ) as @xmath10 where @xmath11 refers to unit vector along vertical direction of cartesian - geometry . equation [ eq_heli_ber ] shows that the helicity of magnetic fields in an open volume may change by the passage of helical field lines through the surface ( first term ) and/or by photospheric footpoint motions of the field lines ( second term ) . the temporal evolution of magnetic helicity flux across the photosphere characterizes the injection of magnetic helicity from the sub - photospheric layers into the solar atmosphere , horizontal flux motions , and the changes in the coronal magnetic field configurations related to eruptive events , such as the cmes , propagating into the interplanetary medium . during the past years , several attempts have been made to estimate magnetic helicity from suitable solar observations . @xcite developed a method for determining the helicity flux ( the second term in equation [ eq_heli_ber ] ) passing through the photosphere . they used a time series of photospheric line - of - sight ( los ) magnetograms to determine horizontal velocities by local correlation tracking ( lct ) technique @xcite . using this method , vector potential @xmath8 was constructed by using photospheric los field ( as an approximation to @xmath12 field ) as boundary conditions with coulomb gauge in terms of fourier transform ( ft ) as : @xmath13 \\ p , y}&=&ft^{-1}\left[\frac{-jk_x}{k_{x}^{2}+k_{y}^{2}}ft\left(b_{\rm z}\right)\right ] \label{eq_fou_sol}\end{aligned}\ ] ] where @xmath14 , @xmath15 are spatial frequencies in the x , y directions , respectively . later , this method was applied to many ars by several authors @xcite . however , @xcite showed that this method of calculation introduced artificial polarities of both signs in the helicity flux density maps with many flow patterns . therefore , they suggested to use relative velocities for calculating the helicity injection rate : @xmath16_n}{|\mathbf{x}-\mathbf{x'}|^2}b'_z(\mathbf{x ' } ) b_z(\mathbf{x } ) ds'ds \label{eq_par}\ ] ] where @xmath17 is the foot - point velocity at the position vector @xmath18 , and @xmath12 is the vertical component of the observed magnetic field . this equation shows that the helicity injection rate can be understood as the summation of relative rotation rates of all the pairs of elementary fluxes weighted with their magnetic flux . furthermore , @xcite has shown that the lct method is inconsistent with the magnetic induction equation , which governs the temporal evolution of the photospheric magnetic fields . tracking methods have serious practical limitations that might result in the failure of detecting significant shear velocity fields and hence in the underestimate of the amount of helicity injected by such velocity fields . @xcite reported that the magnetic energy and helicity fluxes should be computed only from the horizontal motions deduced by tracking the photospheric cross - section of magnetic flux tubes . these authors contend that the apparent horizontal motions include the effect of both the emergence and the shearing motions . they analyzed the observational difficulties involved in deriving such fluxes and in particular , the limitations of the correlation tracking methods . one of the main limitations in the previous studies has been the coarse spatial resolution of the available observations which limits the deduced velocities to the velocity corresponding to the group motion of an unresolved bunch of thin flux tubes covered by a pixel . also , tracking motions faced difficulties in the areas lacking sufficient contrast , such as in the sunspot umbrae . several alternative , improved methods have been developed for inferring plasma velocities consistent with the induction equation at the photospheric level , based on the los , as well as , vector magnetograms . the induction method ( i m ; @xcite ) , induction local correlation tracking(ilct ; @xcite ) , minimum energy fit ( mef ; @xcite ) , differential affine velocity estimator ( dave ; @xcite ) and differential affine velocity estimator for vector magnetograms ( dave4vm ; @xcite ) have been developed for the determination of horizontal component of motion . in contrast , the normal component of velocity can be determined from the doppler observations of ars located near the disk center . dave4vm method requires vector magnetograms . the performance of different techniques has been examined in @xcite which showed that the mef , dave , flct , i m , and ilct algorithms performed comparably . furthermore , they reported that while the dave estimated the magnitude and direction of velocities slightly more accurately than the other methods , mef s estimates of the fluxes of magnetic energy and helicity were more accurate . time series data of photospheric magnetograms have been extensively used to derive magnetic helicity and its evolution in order to examine its role in the level of transient activity of the ars . @xcite reported impulsive variations of helicity during some m and x - class flares . in a survey , @xcite revealed that x - flaring ars have a higher net helicity change with peak helicity rate @xmath19mx@xmath1s@xmath20 with weak hemispheric preference . @xcite have also studied the solar flare productivity in relation to the helicity injection using a large sample of 378 active regions . using soho - mdi magnetograms , they reported variation of helicity injection rates and a significant helicity accumulation of @xmath21 mx@xmath1 over several days around the time of flares above m5.0 . most of the previous studies that used data from michelson doppler imager ( mdi ) onboard soho had the time resolution of 96 minutes . the rather coarse time resolution between two consecutive observations has been a matter of concern in the above calculations because the contribution from the motion of short lived magnetic features in small intervals is difficult to be accounted suitably ( _ e.g. _ , @xcite ) . this underlines the need for observations of magnetic fields with higher temporal resolution . the above mentioned issues can now be addressed with the availability of a better cadence of 12 minutes by the recently launched helioseismic and magnetic imager ( hmi ) onboard _ solar dynamics observatory _ ( sdo ) . our main objective in the present work is to reinvestigate the role of helicity injection in relation with flares and cmes using the high - resolution data obtained from sdo - hmi . we intend to utilize this opportunity to revisit some of the previous studies involving computations of helicity rate for two ars , noaa 11158 and noaa 11166 , that appeared during february and march 2011 , respectively , in the ascending phase of the current solar cycle 24 . using the high quality hmi data obtained for the two ars , we intend to examine whether the variations as reported by @xcite and @xcite for energetic flares occurred also during the flares of lower magnitude . it is of particular interest to investigate if such changes were associated with the cmes as well . for our analysis , we use dave technique for retrieving horizontal foot - point velocity from the los magnetograms . thereafter , using equation [ eq_par ] we determine helicity injection rates and the accumulated helicity in the two ars due to foot - point shearing motions during their disk transit . it has been inferred from the previous studies that most of the helicity injection corresponds to magnetic flux emergence in the ars@xcite . we , therefore , attempt to interpret the variations found in these physical parameters in relation to the occurrence of flares and cmes . in particular , we investigate whether the transport of magnetic helicity plays a role in solar eruptions . we organize this paper as follows . description of the data used in this study and the procedures of the data processing are given in section [ data ] . results obtained for the two selected ars are presented in section [ results ] and the following discussions in section [ disc ] . the summary of the work presented in this article is given in section [ sum ] . for our study , we have used high resolution los magnetograms at a cadence of 12 minutes obtained from the helioseismic magnetic imager ( hmi ; @xcite ) on board _ solar dynamic observatory _ ( sdo ) . hmi observes the full solar disk in the fe i 6173 spectral line with a spatial resolution of 1 arc - second . hmi provides four main types of data : dopplergrams ( maps of solar surface los velocity ) , continuum filtergrams ( broad - wavelength maps of the solar photosphere ) , los and vector magnetograms ( maps of the photospheric magnetic field ) . noaa 11158 ( 19@xmath22s ) and 11166 ( 10@xmath22n ) appeared on the disk during february 11 - 20 , 2011 and march 03 - 16 2011 respectively . these ars were very active , and produced some intense x - class flares associated with cmes in addition to many m- and c- class flares during their disk transits . from the aia observations , intermittent mass expulsions were seen , many of which turned into large , fast moving cmes as further confirmed by stereo . table [ tabflcm ] gives a list of flares ( as recorded by goes ) and cmes . l l l l + ar & date & flares & cmes + ( noaa ) & dd / mm / yyyy & magnitude(time ut ) & ( time ut ) + + 11158 & 11/02/2011 & no flares & no cmes + & 12/02/2011 & no flares & no cmes + & 13/02/2011 & c1.1(12:29),c4.7(13:44),m6.6(17:28 ) & 21:30,23:30 + & 14/02/2011 & c1.6(02:35),c8.3(04:29),c7.1(06:51 ) & 02:40,07:00,12:50,17:30,19:20 + & & c1.8(08:39),c1.7(11:51),c9.4(12:41 ) & + & & c7.0(13:47),m2.2(17:20)*,c6.6(19:23 ) & + & & c1.2(23:14 ) , c2.7(23:40 ) & + & 15/02/2011 & x2.2(01:44),c2.7(00:31 ) & 00:40,02:00*,03:00,04:30,05:00 + & & c4.8(04:27),c4.8(14:32),c1.7(18:07 ) & 09:00 + & & c6.6(19:30),c1.3(22:49 ) & + & 16/02/2011 & c2.0(00:58),c2.2(01:56),c5.9(05:40 ) & 14:35 + & & c2.2(06:18),c9.9(09:02),c3.2(10:25 ) & + & & c1.0(11:58),m1.0(01:32),m1.1(07:35 ) & + & & m1.6(14:19),c7.7(15:27),c1.3(19:29 ) & + & & c1.1(20:11),c4.2(21:06),c2.8(23:02 ) & + + 11166 & 06/03/2011 & c5.1(11:56),c3.9(15:21 ) & 02:00,15:20 + & 07/03/2011 & m1.9(13:45 ) & 14:25,22:10 + & 08/03/2011 & c7.7(23:10 ) & 14:30,19:00 + & 09/03/2011 & c9.4(08:23),m1.7(10:35),m1.7(13:17 ) & 06:40,10:40,21:45 + & & c9.4*(22:03),x1.5(23:13 ) & + & 10/03/2011 & c2.9(03:50),c6.2(07:03),c4.2(13:19 ) & + & & c4.7(13:42),c2.0(14:21 ) & 04:50,07:10 + & & c4.0(19:00),m1.1(22:34 ) & + & 11/03/2011 & c1.4(00:29),c1.1(01:46),c2.8(02:24 ) & 00:50 + & & c5.5(04:15),c4.3(07:22),c1.1(08:13 ) & + & & c2.0(11:10),c3.6(11:43),c1.1(16:04 ) & + & & c1.0(22:20),c1.0(22:50 ) & + + + _ note : _ all flares(cmes ) associated to source region r2(r1 ) of respective ars except those marked by * [ tabflcm ] magnetograms obtained at different times were aligned by the method of @xcite . in this method , an image of the ar taken at the central meridian is considered as the reference image . all other images , in time accounted for differential rotation @xcite along with the latitudinal difference of center of reference image from the ephemeris information , were remapped on to the disk center . this method is intended to reduce errors due to geometrical foreshortening and the ar is transformed to the disk center . since at disk center , normal and vertical components of magnetic fields are same , the difference between the normal and los component was corrected by cosine of the distance of the ar center from the disk center by assuming the horizontal field contribution for the transformation to be negligible @xcite . we followed the transits of the two selected ars from east to west on the solar disk . in order to have negligible errors in geometric correction , we restricted ourselves to a region within @xmath2340@xmath22 longitude from the central meridian . with this constraint , we confined our study of the temporal evolution of the ars to six days s period around their central meridian passage . we derived the horizontal velocities of foot - points on the photosphere by using dave technique @xcite . the dave technique is essentially a local optical flow method that determines the magnetic footpoint velocities within the windowed region . further , it adopts an affine velocity profile specifying velocity field in the windowed region about a point and constrains that profile to satisfy the induction equation . any tracking method depends on two parameters , _ viz . _ , the window size and the time interval . for a given time interval @xmath24 t , the window should be large enough so that tracked features remain confined within the window . also , it should be small enough to be consistent with an affine velocity profile . @xcite presented a way to select an optimal window objectively , using the degree of consistency between change in the observed magnetic field ( @xmath25 ) and the expected magnetic field change based on the flow estimated with several trial windows . they found the best performance of this method at approximately a square window of pixels . since the ars evolved rapidly , we chose a window size of 21@xmath2618 pixel@xmath1 after a careful verification of the physical flux motions and directions of estimated flows . the dependence of helicity injection rate on window size and time difference between the tracked maps using this method were investigated . moreover , as the hmi magnetic field measurement precision is 10 g @xcite , we have set this as the threshold to avoid errors while retrieving velocities . further details of this method are given in a recent work of @xcite . computation of the helicity rate using the method ( direct integration ) proposed by @xcite at each pixel of the ar map ( _ cf . _ , equation [ eq_par ] ) is a tedious , time consuming process . however , we chose to use this method for reducing the effect of fake polarities of helicity flux . restricting the calculations at pixels with magnetic field above the threshold ( @xmath2710 g ) helps to reduce the computation time typically by 15 - 25% . parallelization in integrand computation further reduces the time approximately by a factor of the number of processors used . the same equation as rewritten by @xcite to suit the convolution algorithm by fourier transform is faster than the direct integration method . the intrinsic problem of fourier transform with periodicity could be overcome by padding the array corresponding to the data points with rows and columns of zeros to get results as obtained by direct integration method . in this study , we have implemented the former approach ( direct integration ) to get sufficiently accurate results . the evolution of observed magnetic flux and the computed helicity rates are presented in the following for the two selected ars noaa 11158 and noaa 11166 with the methods and procedures explained in section [ data ] . this ar appeared as small pores at the heliographic location e33s19 on 2011 february 11 as seen in the full disk hmi photoheliograms . thereafter , it grew very rapidly during the next two days as the small pores merged and formed bigger sunspots . it was a newly emerging region which developed to a large ar having @xmath28 magnetic complexity during its rapid evolution . it consisted of four large regions of opposite polarities in quadrupolar configuration . figure [ 58_mos](top row ) shows the evolution of noaa 11158 during 2011 february 13 - 15 in hmi intensity maps . the prominent positive polarity sunspots of the ar are labeled as sp1 , sp2 , sp3 and the negative polarity spots as sn1 , sn2 , sn3 for identification . los contours are overlaid on the intensity image showing the respective polarity distribution . the spatial evolution of the ar shows a large shearing motion of sp2 that rotated around sn2 about its umbral axis during 2011 february 13 - 15 . it then detached and moved towards sp3 along with small patches of both polarities appearing and disappearing over short periods of time . this motion appeared to have created a twist in magnetic fieldlines connecting these spots . a careful examination of the animation made from magnetograms and intensity maps revealed a significant counter clock - wise ( ccw ) rotation of sn1 during the same period , while a small positive polarity region sp1 located to the north of sn1 rotated in the counter clock - wise direction along with a proper motion away from sn1 . the rotations of sn1 and sp1 increased the twist of the field lines , and the magnetic non - potentiality of the sigmoid structure @xcite . several mass expulsions were launched intermittently from this region , as seen from the quick look images in aia . these turned into cmes as confirmed by stereo observations . in order to quantitatively analyze the magnetic complexity or twist contributed by the observed shearing motions of the magnetic foot points , we computed the helicity injection rates using the temporal sequence of magnetograms of the ar . figure [ 58_mos](bottom row ) shows the computed helicity flux density maps corresponding to the hmi continuum intensity images ( top row ) . the dark ( white ) patches in the right panel represent negative ( positive ) helicity flux density according to the usual convention . contours of los magnetic field at [ -150 , 150]g levels are overlaid for a better visualization of helicity flux density with respect to the magnetic polarity . evidently , negative polarity region of sn1 injected negative ( dark ) helicity during 2011 february 14 - 15 which is also consistent with its physical ccw rotation . in contrast , sp2 , sn2 and sn3 injected positive ( white ) helicity along with negative ( dark ) helicity in some small patches . we expect that the nature of motions in these areas could have influenced the helicity pattern there . the photospheric maps of helicity flux ( and its injection rate ) provides spatial information about the basic properties of a link between the activity and its sub - photospheric roots as reflected by the flux emergence process . in a sample of four active regions , @xcite found that helicity was mostly injected while fluxes emerged in the ar , suggesting it to be the major source of helicity injection . the flux cancellation process , on the other hand , resulted in a loss of coronal magnetic helicity , or inverse helicity injection . we thus infer that the ar possessed two main sites , of unstable energy storage systems marked by the rectangular boxes r1 and r2 in figure [ 58_mos ] . these sites had distinctly different injection of helicity flux density corresponding to the flux ( or foot - point ) motions , polarities and activity . in order to show the transient activity of the ar as it evolved , we have plotted the disk integrated goes soft x - ray flux ( 1 - 8 channel ) during february 1117 in figure [ 58_tplot](top ) where the start times of flares of noaa 11158 are marked by arrows . after its birth , the ar gradually evolved during 2011 february 1113 as evident from the monotonic increase of fluxes in both polarities corresponding to @xmath29 mx ( figure [ 58_tplot ] , middle ) . then followed a rapid phase of flux emergence ( of @xmath30 mx ) during february 1314 after which it reached a plateau . also plotted is the flux imbalance , _ i.e _ , the ratio of the net flux and absolute total flux in the ar . the dominance of negative flux during february 1315 , and thereafter of the positive flux , is evident . flux variations occurred in the range of ( 9.512.5)@xmath31 mx with the imbalance within @xmath32 over six days . a significant flux decrease in both polarities by @xmath33mx occurred till the time of the x2.2 flare . we shall discuss more about flux changes during x - flares in section [ disc ] . the unusual rotating sunspots along with the increased fluxes indicated emergence of highly twisted fluxes from the sub - photospheric region @xcite , and not resulting from the surface flows alone . most of the flare and cme activity of this ar occurred only after february 13/12:00ut , indicating that the rapid flux emergence could have played important role in triggering the transients . in figure [ 58_tplot](bottom ) , we have plotted the time profile of helicity injection rate(@xmath34 ) , which is the summation of helicity flux density over the ar . also plotted is the accumulated helicity , _ i.e _ , the integrated helicity change rate over time ( @xmath35 ) . the total accumulated helicity is estimated as 14.16@xmath36mx@xmath1 during the six day period of 2011 february 1116 , with the peak helicity rate of 31.54@xmath37mx@xmath38h@xmath20 . the occurrence times of the cmes associated with the ar are marked by arrows in this panel for reference . an impulsive variation of helicity injection rate due to injection of negative helicity is discernible during the x2.2 flare and the concomitant cme . the helicity injection rate decreased during the period february 14/11:00february 15/13:00 ut , and increased thereafter till february 16 along with fluctuations in the range 24@xmath37mx@xmath38h@xmath20 . we notice a large dip of helicity injection around x2.2 flare with associated cmes . we have smoothed the original time profile at 12 minute interval by a box car window of five data points ( _ i.e. _ , 1 hour ) . similar sudden dips in injection rates during other events can be further analyzed for examining their association . figure [ 58_vel ] shows transverse velocities in the rectangular sub - regions r1 ( top row ) and r2 ( bottom row ) of noaa 11158 overlaid on the corresponding maps of helicity flux density during three flare events . also overlaid are the contours of the los magnetic flux at @xmath39 g levels . maximum rms velocities in the range of 0.60.9 were found over the observed period in the ar . spiral or vortex like velocity patterns are obviously related to the counter rotation of sn1 in figure [ 58_vel](b c ) . a notable observation is that the sub - region r1 possessed negative helicity flux density distribution which is consistent with the chirality associated with the physically observed counter rotation of sn1 whereas r2 possessed mixed helicity flux dominated by positive helicity flux distribution . because of the continued shearing motions at the interface of sp2 and sn2 , the flow field vectors almost aligned with the polarity inversion line ( pil ) as seen in panels ( d f ) . interaction of fluxes with this shear motion can squeeze and converge the flux in both sp2 and sn2 . we hypothesize that the field lines were stressed and twisted by this motion leading to the storage of free energy adequate to account for the release in the energetic x2.2 flare of february 15/01:44ut . as almost all flares ( except m2.2 at 14/17:20 ) occurred in r2 , we examined the spatial distribution of helicity flux before and after the flare events to know whether any sudden changes are found related to the occurrence of flare . during some events , we noticed negative patch of helicity flux in the regions of positive helicity flux . especially , in the panels ( e g ) , a negative helicity flux distribution near the pil during m6.6 , c7.0 , and x2.2 flares can be observed . there may be some concern about these flare - related changes , as it is known that during the impulsive phase of large flares , the spectral line profile itself may undergo some change affecting the magnetic ( and velocity ) field measurements . most of these flares occurred in r2 while the mass expulsions(or cmes ) were associated to r1 . in order to relate helicity rate changes to these events , therefore , we have computed and plotted the total injected quantities for r1 and r2 in figure [ roi_58](a - b ) . injection of helicity in a region of dominant opposite sign can be understood as a sudden dip in the time profile plot . of course , the corresponding spatial information is lost in the averaged quantity . the advantage of using localized analysis of selected sub - areas in the ars is that it reduces complex variations occurring over a much larger area of the entire ar while showing only the variations occurring in the areas - of - interest . it also enhances the dips corresponding to the identified events ( marked by the arrows ) . however , it is important to identify the location of individual event in order to correctly attribute a particular change of helicity rate to it . noaa 11158 was essentially a positive helicity injecting region , while its sub - region r1 had a negative injection rate and accumulated quantity due to the presence of rotational motion . we expect that as the sunspots sn1 and sp1 rotated , the injection rate increased to a maximum of @xmath40mx@xmath1h@xmath20 on february 14/18:00ut . a total helicity accumulation of @xmath41mx@xmath38 occurred during the six day period in this region . noticeably , a steep accumulation occurred during feb 1415 along with many observed mass expulsions shown by arrows . this could be interpreted as shedding of excess helicity from the corona in the form of eruptive events . the steep accumulation of helicity by monotonic injection rate , therefore , is suggested to be a cause of expulsions . accumulated helicity amounting to @xmath42mx@xmath38 in sub - region r2 with steep accumulation observed from february 13 onwards , could be mostly associated with the observed large shear motion of sp2 . for a quantitative study of the association of short term variations in helicity rate to the flaring or cme , the following analysis is carried out . the absolute time difference of the helicity flux ( @xmath43 , having units same as dh / dt ) averaged over start and stop times of goes flares above c2.0 is computed . this is compared to that of randomly selected but equal length time intervals containing no flares . a significantly higher mean of @xmath43 during flares compared to quiet times would indicate a robust association between flaring and helicity fluxes . a similar analysis is undertaken for time windows around cmes to look for a cme - helicity flux association . we assume that there is no time lag between flaring and helicity flux signal while carrying out this analysis . we first interpolated the signal at 1 min interval from 12 min interval to get values as required by the goes flare times , then it was smoothed to a boxcar width of 30 minutes . within start and stop times of flares , the averaged value of absolute variation was computed to compare with that calculated during randomly selected , constant interval(30 min ) quiet times . the time difference of helicity rate in r1-r2 is shown in figure [ roi_58](c - d ) with cmes and flares marked by arrows . large amplitude variations are discernible during m6.6 , x2.2 and the cme at 12:30ut indicating some association , but similar variations are present around the mean position even in quiet times . from the above described analysis , we found a significantly higher mean during cme s ( @xmath44 ) compared to quiet times(@xmath45 ) . the difference in cme versus quiet time helicity fluctuations are marginally statistically significant , at better than one - sigma . similarly , a mean of @xmath46(@xmath47 ) during flare ( quiet ) times indicate poor or no association of flaring to helicity flux variations . the same analysis for the helicity flux over the entire ar improved the association ( in terms of mean absolute helicity variation ) slightly for cmes but worsened it for flaring . we shall further discuss these helicity variations during flare / cmes in view of the involved flare - related effects in section [ flreff ] . ar noaa 11166 appeared on the east limb of solar disk on 2011 march 03 at the location n10e64 . we monitored its activity during the period of 2011 march 611 in which it produced a large x1.5 flare , two m - class flares and several c - class flares , some of which were also associated with plasmoid ejections or cmes . table [ tabflcm ] lists the flares and cmes of this ar . daily evolution of the ar in the period of march 8 - 11 , 2011 is shown in figure [ 66_mos](top row ) . the major sunspots of the ar are labeled as sp1 , sp2 , sn1 and sn2 . the identification of sp2 was somewhat unclear before march 10 as several small umbrae were spread over its location . they moved and coalesced to form sp2 . polarities of the respective sunspots are identified by the overlaid los magnetic field contours . this ar also consisted of a complex magnetic configuration with two positive ( sp1 , sp2 ) and two negative ( sn1 , sn2 ) polarity sunspots located within the surrounding diffused fluxes . emerging and moving flux regions , fp3 and fn2 , were identified in the course of the evolution in the sunspot periphery ( march 11/22:00ut panel ) , having opposite sign to that of their native sunspots . however , there were no intrinsic rotating sunspots or flux patches as observed in the case of ar noaa 11158 . we computed the helicity flux density for ar noaa 11166 during its evolution in the period 2011 march 611 . the corresponding maps for three successive days are plotted in figure [ 66_mos](bottom row ) . locations of helicity flux density of mixed sign were distributed all over the ar through out the evolution period . the peripheral sites of the sunspots exhibited helicity flux density predominantly of negative sign . however , patches of negative helicity flux were also observed embedded in the positive helicity flux site of the flare ( march 09/23:00ut panel ) . for further close examination , we consider two sub - areas r1 and r2 , as marked by the boxes in this panel . the disk integrated goes soft x - ray flux ( 1 - 8 channel ) during 2011 march 6 - 11 is plotted in figure [ 66_tplot](top ) . the arrows in this panel indicate the start time of flares in noaa 11166 . during the disk transit of the ar , fluxes of both polarities increased corresponding to @xmath48mx , with the imbalance varying below 6% ( figure [ 66_tplot ] , middle ) . as observed for noaa 11158 , a rapid flux emergence occurred in this ar too during march 79 . thereafter , only small variations associated with local cancellations / emergence of about @xmath33mx took place pertaining to the gradual evolution of the ar . positive flux dominated in the ar during march 7 - 11 , and then a near balance was established . it is worth noticing that magnetic fluxes in both polarities decreased by @xmath49mx while evolution of fluxes leading to the occurrence of a cme following the x1.5 flare . however , it is not clear whether this decrease in flux six hours before the flare / cme has some role in these events . but , the flux imbalance , increasing prior to the flare , reduced significantly after the flare consistent with observations reported by @xcite . most of the flares and cme activity of this ar occurred only after march 8 , suggesting that the rapid emergence of fluxes could be an important factor for triggering of these transients . temporal evolution of helicity injection rate and the accumulated helicity for noaa 11166 are shown in figure [ 66_tplot](bottom ) with arrows marking the times of the cmes . a five magnetogram boxcar was used to smooth the profile to reduce fluctuations in the profile . as expected , these parameters increased in the first phase corresponding to the flux emergence , in agreement with @xcite that helicity is mostly injected while the fluxes emerged . total helicity accumulated during the six days period of the ar s evolution was estimated to @xmath50mx@xmath38 . the maximum helicity injection occurred during 2011 march 8 at the rate of @xmath51mx@xmath1h@xmath20 . thereafter , it reduced gradually to the minimum rate at @xmath52mx@xmath53 on 2011 march 10 . the coronal helicity of the ar is likely to be positive as a result of this positive helicity injection . horizontal , or transverse , velocity vectors corresponding to the tracked flux motions are plotted in figure [ 66_vel ] separately for r1 ( top row ) and r2(bottom row ) . the rms velocities of flux motions are found to have the maximum values in the range 0.50.9 . strong moat flows were systematically dominant in both regions from the peripheral regions of sunspots in addition to the shear flows . persistent strong shear motions due to the merging sp2 group were identified in r2 . these flows appear to collide head on with those from sp1 resulting in the flux submergence / cancellation . flux emergence was also identified from the diverging flow field observed in animated flows from r1 . from this region , flux moved towards r2 as the ar evolved . such motions appear to be associated with injection of negative helicity into a region with predominantly positive flux , increasing the complexity of the magnetic flux system as shown in panels ( d)(f ) of r2 . further , these negative helicity injections often coincided with some observed events , such as the three of them shown in this plot . for the x1.5 flare the distribution of helicity flux is shown in panel ( e ) on march 09/23:36ut . the injection rates and accumulated helicities deduced from sub - regions r1 and r2 are plotted in figure [ roi_66](a b ) . also the contribution of each signed helicity flux in the net helicity flux is plotted separately . the time profile of r1 shows it to have positive helicity injection with a steep increasing phase during march 79 at a peak rate of @xmath54mx@xmath1h@xmath20 . thereafter , gradual decrease in the rate of injection is evident from the plot . as mentioned earlier , r1 was a site of emerging flux that resulted in contributing to accumulation of helicity amounting to @xmath55mx@xmath1 . while r2 exhibited mixed sign injection rates during its evolution . as in the previous ar , continuous injection of dominant positive helicity from r1 is suggested to be the cause of observed mass expulsions , whereas the injection from r2 is of mixed signs suggested to result in flares . an enhanced peak of helicity rate was seen around the time of the x1.5 flare in r2 of ar 11166 that was not obvious in figure [ 66_tplot](bottom panel ) since we reduced fluctuations occurring over entire ar by selecting small area . after this event , the negative injection rate increasingly dominated on march 10 , turning the net injection of the entire ar negative . the implication of this transition of injection rate from positive to negative sign over a day is not clear in the observed events shown by the arrows . the time variation of helicity flux in both r1 and r2 are plotted in figure [ roi_66](c d ) along with the arrows pointing start times of cmes and flares in the ar . some of the large amplitude variations of helicity flux about the mean position appear to be related to these events . as in the previous ar , we have analyzed the association of flare / cmes that originated from the sub - regions r1 and r2 of this ar with the respective helicity flux . the calculated mean of variation in helicity flux ( @xmath56 ) during flaring ( @xmath57 ) is marginally statistically different at about two - sigma level over that during quiet times ( @xmath58 ) , reflecting a robust association of flaring and helicity fluxes . the mean of @xmath43 obtained in quiet times do not have any information or bias of flaring or cme , therefore higher mean during the flare / cmes implies some impact of helicity flux variations in them . a similar analysis undertaken for cmes also showed the similar association ( during cmes of @xmath59 dominated over quiet times of @xmath60 , but not statistically significant difference ) . however , the association strengthened for flaring and weakened for cmes when the helicity flux over the entire ar was considered in the analysis . it is well known that the photospheric magnetic ( and doppler ) field measurements are affected by flares . during an energetic flare , the profile of spectral line used for the measurement was reported to change from absorption to emission , resulting in a change of sign in the deduced magnetic polarity ( * ? ? ? * and references therein ) . this abnormal polarity reversal was observed to last for about a few minutes during the impulsive phase of the flare ( typically 3 - 4 minutes ) . similar abnormal , transient changes have also been reported for some other large , white light flares @xcite . the change in the line profile may arise due to both thermal effects and non - thermal excitation and ionization by the penetrating electron jets produced during the large flares . we term these as flare - related transient changes , considered to be artifacts as they do not correspond to real magnetic field changes . there is increasing evidence that flares may change the magnetic field more significantly on a persistent and permanent manner @xcite . the persistence of the observed field changes implies that they are not artifacts of changes in the photospheric plasma parameters during the flare , and the temporal and spatial coincidences between flare emission and the field changes suggest the link of the field changes to the flare . we term these as permanent flare - related changes . with these known transient and permanent flare - related effects on magnetic fields , it would not be clear , particularly during the impulsive phase of the flare , if the change in helicity flux can be interpreted as genuine transport of helicity across the photosphere . in addition , an implicit assumption made in our approach of calculating helicity injection is the ideal evolution of photospheric magnetic fields in the induction equation used to derive velocities of flux motions . moreover , the same assumption is involved in the derivation of helicity injection from the relative helicity formula @xcite . this assumption is valid and reasonable outside the flaring time intervals ( at least during permanent changes of fields ) as the typical observed photospheric velocities are far less than the alfven velocities . in the real conditions of rapid , transient changes in photospheric magnetic fields spanning impulsive period of the flare , the assumption of ideal magnetic evolution may not be applicable . therefore , there is theoretical uncertainty regarding the interpretation of helicity fluxes during flares . in order to inspect these aspects in the signal of the helicity change rate , we procured 45s cadence magnetograms for some selected flare events and averaged them to 3min cadence after processing as the previous data set . a mosaic of distribution of helicity flux around the x2.2 flare is shown in figure [ mosx22 ] . during the impulsive period ( 01:48 - 02:02ut ) of this flare , negative helicity flux is distributed about the pil which we believe to be due to the transient flare - related effect . the magnetic ( and doppler ) transients and locations of spectral line reversal associated with this flare are already reported by @xcite , which are spatially and temporally consistent with this negative helicity flux distribution . therefore , the observed negative helicity flux distribution in the dominant positive site can be attributed to the transient flare - effect , and is likely to be artifact , i.e. , not a true transfer of helicity . similar mosaics of helicity flux distribution maps were made and examined for other events . the computed magnetic and helicity fluxes are plotted with time in figure [ ploteve ] . the flare start time is shown in vertical dotted line labeled with magnitude of the flare . it should be noted that we have not applied any smoothing to the computed helicity rate signal in these panels . magnetic fluxes of both signs decreased abruptly with a dip during the impulsive period following with injection of negative helicity flux in the dominant positive helicity flux , during the m6.6 , x2.2 flare events . magnetic field measurements could also be underestimated by 18 - 25% due to enhanced core emission of spectral line by the heating of the impulsive flare @xcite as a result of which the integrated flux profile could show such a dip during peak phase of the flare . interpretation of flux annihilation through reconnection during this peak phase might be ambiguous due to this fact , although it could be a possible consideration . in the post - flare phase , fluxes increased in both polarities as field lines reorganized as a post - reconnection process . this falls under the `` permanent '' real change related to the flare . for smaller magnitude flares , transient effects may be absent or not be prominent in the impulsive phase . therefore the measurements of magnetic fields and the computed helicity rate signal are not expected to be affected during the flare . hence , they may indicate true transfer of helicity flux , except for the theoretical uncertainties as mentioned above . in the case of the 14 february/13:47ut ( c7.0 ) flare , shown in panels ( b1)-(b2 ) , indeed the variation of helicity signal occurs without the variations in magnetic fluxes associated to the flare - related effects . this may be an example of true transfer of helicity of the flux system , but with the theoretical uncertainty in our approach . there are no significant variations in magnetic and helicity fluxes corresponding to the 09 march/09:23ut ( c9.4 ) , and 10:35ut ( m1.7 ) flares . large amplitude fluctuations in both sign of helicity signals during the cme just before the 09 march/22:03ut ( c9.4 ) flare are apparent in panels ( e1)-(e2 ) . we speculate that these fluctuations subsequently led to the initiation of the prominent cme that followed the 09 march/23:13ut ( x1.5 ) flare an hour later . similarly , the transient flare effects might be responsible for the abrupt changes in magnetic fluxes resulting in variations of helicity injection signal during the x1.5 flare ( panels ( e1)-(e2 ) ) . during the 10 march/13:19ut ( c4.2 ) , 13:42ut ( c4.7 ) flares , the transfer of helicity flux from positive to negative , negative to positive sign is clear from the panels ( f1)-(f2 ) , respectively . these flares are of small magnitude , with no obvious flare - related artifacts . therefore , the observed helicity flux changes are expected to be true ( with the implicit theoretical uncertainty in the approach ) . a point to be noted is that all large flares ( m and x - class ) may be involved with transient flare effects . therefore , it is better to look for helicity variations in small flares where magnetic fields are expected to be less affected , making it easier to examine the possible role of transfer of helicity flux . thus , we consider the 14 february/13:47ut ( c7.0 ) , 10 march/13:19ut ( c4.2 ) and 13:42ut ( c4.7 ) flares to be the best examples here , supporting the true transfer of helicity . it is not clear that whether the helicity transfer in these cases is related to permanent flare - effects . at present , it is difficult to say much about the physical significance of these variations over the ar in the corona , i.e. , at the primary sites of the flares . it would be particularly interesting to study the physical significance of such injection along with the information of coronal connectivities ( _ e.g. _ , @xcite ) as suggested by @xcite for understanding the possible role of transfer of helicity flux during the flares / cmes . computation of helicity injection rate involves the measurement of magnetic field and the inferred horizontal velocities . apart from the errors in the measurements , the computations involving the dave method for deriving velocities depend on two main parameters _ viz . _ , the time interval between two successive magnetic maps , @xmath24 t , and the dave window size . for obtaining optimized results , horizontal displacements of features during the time interval @xmath24 t should be large enough to be well determined by dave . also , these displacements should be smaller than the selected window size . to check our results for consistency , we carried out the dave calculations using the time intervals @xmath24 t = 12 , 24 and 36 minutes , while keeping the window size fixed at @xmath61 pixels . then , calculations were carried out for different window sizes , _ viz . _ , 21@xmath2618 , 15@xmath2612 , 9@xmath266 while keeping @xmath24 t fixed at 36 minutes . furthermore , to avoid the effect arising from noise , we used a threshold of magnetic field at 10 g , which is the hmi precision . as the hmi provides 12 minute averaged data products , we averaged them corresponding to our calculations at 24 ( 2 maps ) and 36 ( 3 maps ) minutes . the dependence of helicity injection rates on time interval @xmath24 t is shown in figure [ depen_158](top row ) for noaa 11158 . the scattered data are fitted by straight line in the least square sense . due to the large volume of data , this computation is tedious and time consuming . therefore , results are shown here only for noaa 11158 , but , we expect they are also valid for other ars observed by the hmi . there is an additional issue of unequally spaced data points to be addressed in case , for example , we intend to plot the results for @xmath24t=36 with @xmath24 t = 24 minutes . for such cases , we used a cubic spline interpolation ( _ cf . _ , @xcite ) , to get corresponding abscissa values for the ordinate points or vice - versa . essentially , this algorithm employs cubic polynomial between each pair of data points with the constraint that the second and first derivatives of that polynomial are same at the end points so that the resulting values are smooth . table [ tabheldep ] lists the minimum and maximum values of helicity injection rates ( dh / dt , in units of @xmath62mx@xmath1h@xmath20 ) and the accumulated helicity ( @xmath63 , in units of @xmath64mx@xmath1 ) for the computational runs carried out with various dave parameters as mentioned above . .helicity injection rates and accumulated helicities at different dave parameters [ cols="^,^,^,^,^ " , ] + units of dh / dt are @xmath62mx@xmath1h@xmath20 and @xmath63 are @xmath64mx@xmath1 [ tabheldep ] it can be observed from the scatter plots that the helicity rates decreased slightly as the time interval @xmath24 t is increased from 12 min to 36 min . the fitted straight line deviates at a slope of 0.87 and 0.91 corresponding to @xmath65 versus 24 and @xmath66 versus 36 min indicating that helicity injection decreases by 13% and 9% respectively . this implies that short - lived features and their dynamics have considerable contribution to helicity rates . the helicity rates at intervals of 36min are lower by a factor of 21% than that at 12 min with worst correlation coefficient of 0.79 . these effects in turn reflected in the variation of accumulated helicity by 9% . this implies that averaging in time between 12 - 36 min has significant effect on injected helicity rates up to 13% corresponding to 9% of variation in accumulated helicity . the dependence of helicity injection rate on window size by keeping the time interval @xmath24 t fixed at 36 minutes is shown in figure [ depen_158](bottom row ) . the slopes of 1.09 and 1.05 for the dave windows @xmath3 versus @xmath67 and @xmath67 versus @xmath4 respectively , show increasing trend of helicity rates with decreasing window size . indeed , a scalable factor of 14% reduction of helicity rate is evident for windows @xmath3 versus @xmath4 . accumulated helicity also showed this increased trend with decreased window size . a total variation of @xmath32 is found , however , with the same trend of helicity injection rate profiles which is discernible in correlation coefficient with the plots . a maximum velocity of 1 km - s@xmath20 during the time interval of 12 min corresponds to a plasma displacement of an arc - sec . hence , for the window size of 4.5@xmath263(9@xmath266 pixel@xmath1 ) , the issue of features overflowing out of the window should not pose problem . these results are consistent with those reported by ( * ? ? ? * their figure 7 ) . they deduced and compared velocity and helicity rates by combinations of time difference between magnetograms and lct window size . their rms velocity values varied up to 0.6km / s at time interval of 5min . they found that smaller values of lct parameters result in larger amplitude fluctuations of the rate of helicity , with variation within 10% . we , in our computations , found maximum rms velocities for 12min , 24min and 36min in the ar as 0.95 , 0.85 and 0.8km / s respectively . however for the window sizes @xmath3 , @xmath67 and @xmath4 , we obtained the rms velocities as 0.8 , 0.9 and 1.5km / s respectively . these are higher by a factor of 2 compared to their values probably due to the higher resolution and sensitivity of hmi as against the coarser spatial resolution of mdi of 1.98/pixel . nevertheless , the variation in accumulated helicity found in our analysis is within @xmath32 ; consistent with their result . we thus , found the measured helicity injection rate to depend on the time interval between the two successive magnetograms , i.e. , the observational cadence . the selected window size also influenced the measured quantities . our analysis suggests that it is better to use images averaged over up to 24 minutes with relatively small dave window size subjected to the overflow condition as mentioned above . these are important considerations to derive reasonable and meaningful results in addition to optimizing the computations involving large data - sets . free energy storage and release are some of the most important problems in the eruption physics of the sun . there are essentially two effects that can supply magnetic free energy and helicity from below the solar surface to the corona . flux emergence is the process in which vertical motions carry magnetic fluxes through the photosphere . if the sub - surface fluxes emerging through the photosphere are already twisted , then it will contribute to the injection of helicity ( _ cf . _ , the 1st term in equation [ eq_heli_ber ] ) . computation of this term requires the knowledge of the vertical component of velocity and the horizontal or transverse component of magnetic field . flux motions in the form of rotation or proper motions are another process that may efficiently supply helicity injection ( _ cf . _ , the 2nd term in equation [ eq_heli_ber ] ) . the helicity injected by solar differential rotation is rather small , less than 10% of that contributed by the flux motions @xcite , and has only a much longer term effect on helicity accumulation @xcite . magnetic helicity is a physical quantity having a positive or negative sign , representing a right - handed or left - handed linkage of magnetic fluxes , respectively . this means that if positive and negative helicities co - exist in a single domain , magnetic reconnection can cancel magnetic helicity by merging magnetic flux systems of opposite helicities . helicity densities are not gauge - invariant . it is only area - integrated relative helicity flux that is gauge - invariant . in order to define true helicity flux density , the coronal linkage needs to be provided @xcite , so the helicity flux density inferred from tracking will not be precisely accurate . our computations of magnetic helicity injection in both ars revealed that the distribution of helicity flux is highly complicated in time and space . even the sign of helicity flux often changed within the ar . it has been suggested earlier by several workers that magnetic helicity must play an important role in flares as a substantial amount of helicity accumulation is found before many events ( @xcite ) . however , the correlation between various magnetic field parameters and the flare index of an ar is not high irrespective of the method used . this is an intrinsic problem for flare forecasting as the occurrence of a flare depends not only on the amount of magnetic energy stored in an ar , but also on how it is triggered . thus , it appears that helicity accumulation might be a necessary , but insufficient condition for the flares requiring a trigger even if a magnetic system has enough non - potentiality . for instance , @xcite suggested that coexistence of positive and negative helicities may be important for the onset of flares . careful three - dimensional simulations have been carried out by @xcite to explore the physics of flux tube interaction for the co - helicity ( same sign ) or counter - helicity ( opposite sign ) . according to them , counter - helicity presented the most energetic type of slingshot interaction in which flux is annihilated and twist is canceled . in contrast , co - helicity exhibited very little interaction , and the flux tubes bounced off resulting in negligible magnetic energy release . magnetic helicity in the solar corona is closely related to the photospheric magnetic shear , which is usually defined as the extent of alignment of the transverse component of magnetic field along the neutral or polarity inversion line ( pil)@xcite . based on this idea , @xcite performed a numerical simulation by applying a slow footpoint motion . this motion can reverse the preloaded magnetic shear at the pil resulting in a large scale eruption of the magnetic arcade through a series of two different kinds of magnetic reconnections . they proposed a model for solar flares in which magnetic reconnection converts oppositely sheared field into shear - free fields . we interpret our observations according to the above observational and simulation aspects as follows . we have found flux interactions during the x - class flares and associated cmes as seen in figure [ 58_vel ] in the form of continued shearing motion of sp2 around sn2 in ar 11158 . similar motions are also associated with sp2 in ar 11166 . in both ars cases , the flare prone regions ( r2 ) had inhomogeneous the helicity flux distribution with mixed helicities of both signs . correspondingly , sudden impulsive peaks appeared in the profiles of helicity injection due to the injection of negative signed helicity during some flare events . these were also spatially correlated with the observed flares . opposite helicity flux tubes can interact easily leading to reconnection , thereby unleashing explosive release of magnetic energy . impulsive variations of the magnetic helicity injection rate associated with eruptive x- and m- class flares accompanied with cmes were reported also by @xcite . recently , @xcite conjectured that the occurrence of the x3.4 flare on 2006 december 13 was involved with the positive helicity injection into an existing system of negative helicity . further , a solar eruption triggered by the interaction of two opposite - helicity flux systems @xcite , and occurrence of flares in relation to spatial distribution of helicity flux density @xcite were reported . the main drawback of these findings is that the time span between two magnetograms is more than the duration of the flare(@xmath68 m ) , so the time rate of helicity could not be easily resolved at the onset time of the flare . therefore , our results appear to be consistent with the reports of opposite helicity flux tubes reconnecting to trigger transient events . however , it should be cautioned that we have not found such variations of helicity flux clearly in all flare / cme events . from a quantitative analysis , we found poor association of difference in helicity rate during flares to that of quiet times in ar noaa 11158 . this indicates such variations are not prominent or present during all flares . moreover , statistically significant association of such impulsive variations was found during cmes compared to quiet times . there are many possible reasons for this poor association ; one of them is time duration of helicity flux change . we first interpolated the signal at 1 min interval from 12 min interval to get values as required by the goes flare times . then , it was smoothed to a boxcar - averaging window of 30 minutes to reduce fluctuations arising due to interpolation . within start and stop times of flares , the averaged values of absolute variation were computed . here , averaging might have diluted the original helicity variation , so comparison with the helicity variation during quiet times might not be valid . in any case , there is no better way to find appreciable variation in the helicity flux over background fluctuations to incorporate into the correlation analysis , unless individual events are monitored manually to get variation timings . despite these difficulties , statistically significant association of helicity flux is found during flares , but dominant association that is not statistically significant during cmes in the ar 11166 by following the same approach . further , there are concerns about the flare - related effects on magnetic field measurements resulting in misleading interpretation of helicity flux transfer , in addition to the theoretical uncertainty with the assumption of ideal magnetic field evolution in the approach . we therefore investigated this issue using 3 min interval time sequence magnetograms . we found transient flare effects resulting in spurious negative helicity flux distribution during the x2.2 , m6.6 , and x1.5 flare events . also , we indeed observed the true transfer of helicity flux with variations of opposite sign helicity without such flare - related effects in small flares such as the c7.0 on 14 february , c4.2 at 13:19ut , c4.7 at 13:42ut on 10 march . the important point to note is that we found statistically significant association of helicity flux variations with flares / cmes in above cases of ars at zero time lags . also these variations are clear during the flare events ( see figure [ ploteve ] ) and not before their commencement . therefore , it is difficult to suggest that these variations triggered the flares . a study with the information of fieldline connectivity from coronal observations may be expected to reveal the physical significance of the role of helicity transfer during these events . our computed helicity rates involving photospheric flux motions include the flux emergence term as explained by @xcite . by a simple geometrical argument , horizontal foot - point velocity ( @xmath17 , here the dave velocity ) can be written in terms of horizontal and vertical plasma velocities , @xmath69 , @xmath70 , respectively : @xmath71 from this relation , it is not possible to infer as to which term , _ viz . , _ the flux emergence or flux motions , governs the level of activity of the ars . to resolve this difficulty , we have plotted the integrated absolute flux and accumulated helicity computed over the ars , as shown in figure [ flx_vs_hel ] . evidently , the accumulated helicity increased monotonically with the emergence of magnetic flux in the ar in its first phase ( marked by the vertical dashed line for noaa 11158 ) . after this phase followed the next , the active phase , where an appreciable increase of helicity occurred with only small variation in the flux , i.e. , where little emergence of fluxes occurred . this rapid increase in helicity in the second phase could be interpreted as the dominant contribution of the flux motions . intermittent mass expulsions in the form of cmes transferred away the excess helicity . the extent of this transfer , however , is not clear from this plot , although one can make plausible conclusions from the timings of the flares and cmes . the x - class flares with associated cmes in both ars occurred at a slowing phase of helicity accumulation by negative helicity injection . these facts add to the cases as reported by @xcite . moreover , it can be inferred for ar 11158 , that less than 25% of the total helicity flux accumulated with the emergence of the first 75% of the magnetic flux . most of the helicity flux ( from about @xmath72mx@xmath1 ) was accompanied by very little flux emergence ( about @xmath29mx out of the @xmath73mx ) . therefore , more than 75% of the helicity flux came with only 10% of the total magnetic flux . similarly , the first 60% ( @xmath74mx ) of total magnetic flux was associated to less than 30% ( @xmath75mx@xmath1 of @xmath2mx@xmath1 ) of the total helicity flux in ar 11166 . this implies that more than 70% of total helicity flux was accompanied with less than 40% of total magnetic flux . these two cases are thus contrary to the findings of @xcite stating that most of the helicity flux occurs during flux emergence . our study suggests that flux emergence may not always play a major role in accumulating helicity flux . it is also evident that although flux emergence is necessary but horizontal motions also played crucial and dominant role over emergence term in increasing the complexity of magnetic structures contributing to the helicity flux . therefore , we suggest that the horizontal flux motions contributed further , in addition to the emergence term , in creating more complex magnetic structures that caused the observed eruptive phenomena . we have studied the evolution of magnetic fluxes , horizontal flux motions , helicity injection and their relationship with the eruptive transient events in two recent flare ( cme ) productive ars , noaa 11158 and noaa 11166 of 2011 february and march , respectively . we have used high resolution , high cadence data provided by sdo - hmi for these ars which were in their emerging and active phases . the emerging ar consisted of rotating sunspots with increasing flux indicating emergence of twisted flux from the sub - photospheric layers . this indicated the transfer of twist or helicity injection through the photosphere to the outer atmosphere . we suggest that strong shear motions that include rotational and proper motions played significant role in most of the events in addition to the flux emergence . such motions are crucial in twisting or shearing the magnetic field lines and for further flux interactions . ar noaa 11158 consisted of a cme - prone site of rotating main sunspot along with emerging flux of opposite sign and moving magnetic feature . it also had a flare - prone site consisting of self - rotating sunspot(sp2 ) moving about a sunspot of opposite sign(sn2 ) , leading to flux interaction . these motions are likely to form the sigmoidal structures , which are unstable , and more likely to produce eruptive events . a huge expulsion as cme on 2011 february 14/17:30ut occurred in the former site and a white light , energetic x2.2 flare on 2011 february 15/01:44ut occurred in the later site . the other case , ar noaa 11166 was already in its active phase with further increasing content of flux as it evolved . group motions of diffused fluxes merging to form a bigger sunspot manifested major shear motions in addition to outward flows from sunspot . a large cme on 2011 march 09/21:45ut , followed by an x1.5 flare , was one of the major events in this ar . ar noaa 11158 injected @xmath0mx@xmath1 while ar noaa 11166 injected @xmath2mx@xmath1 helicity during the six days period of their evolution . these are consistent with the previously reported order of helicity accumulation ( _ e.g. _ , * ? ? ? it appears that due to the presence of rotational motions , the former ar accumulated larger amount of helicity accounting for its greater activity in the form of flares and cmes . it is also evident that flux emergence is necessary and their motions are crucial in additionally accounting for the accumulated amount of helicity to the emergence term . in both ars , x - class flares with associated cmes were observed in the decreasing phase of helicity accumulation by the injection of opposite helicity . apart from the instrumental and computational errors , the estimation of helicity injection rates are also affected by the choice of dave parameters used to track the motion of the fluxes . helicity injection rates are found to decrease up to 13% by increasing the time interval between magnetograms from 12 to 36 min whereas an increasing trend upto 9% resulted by decreasing the window size from @xmath3 to @xmath4 pixel@xmath1 , with a total variation of 10% in the deduced value of accumulated helicity . the time profile of helicity rate exhibited sudden sharp variations during some flare events due to injection of opposite helicity flux into the existing system of helicity flux . in both ars , the flare prone regions ( r2 ) had inhomogeneous helicity flux distribution with mixed helicities of both signs and that of cme prone regions had almost homogeneous distribution of helicity flux dominated by single sign . a quantitative analysis was carried out to show the association of these variations to the timings of flares / cmes . for the ar 11158 , we find a marginally significant association of helicity flux with cmes but not flares , while for the ar 11166 , we find marginally significant association of helicity flux with flares but not cmes . moreover , these variations of helicity flux may not reflect true transfer ; there exists flare - related transient effects and theoretical uncertainties resulting to these variations . we believe the helicity transfer in the cases of c7.0 on 14 february , c4.2 at 13:19ut , c4.7 at 13:42ut on 10 march to be true , without flare - related transient effect but with theoretical uncertainty in the approach . therefore , to further strengthen the above evidences of true helicity transfer , it would be worthwhile to scrutinize more flare / cmes cases using 3 min cadence magnetic observations , over a period of a day or so . this will enable one to find detectable changes in helicity flux signal during smaller magnitude flares with less transient - flare effects . interpreting the physical significance of such variations using the information of coronal connectivities will be another important aspect to add further to the present knowledge of helicity physics . our study reveals that the spatial information of helicity injection is a key factor to understand its role in the flares / cmes . the data have been used here courtesy of nasa / sdo and hmi science team . we thank dr . etienne pariat for checking our helicity program with comments and suggestions . the author expresses his gratitude to prof . p. venkatakrishnan for some useful discussions on the concept of helicity . we thank an anonymous referee for carefully reading the manuscript and making valuable comments which led to improved clarity and readability of the manuscript .
an investigation of helicity injection by photospheric shear motions is carried out for two active regions(ars ) , noaa 11158 and 11166 , using line - of - sight magnetic field observations obtained from the helioseismic and magnetic imager on - board solar dynamics observatory . we derived the horizontal velocities in the active regions from the differential affine velocity estimator(dave ) technique . persistent strong shear motions at the maximum velocities in the range of 0.60.9km / s along the magnetic polarity inversion line and outward flows from the peripheral regions of the sunspots were observed in the two active regions . the helicities injected in noaa 11158 and 11166 during their six days evolution period were estimated as @xmath0mx@xmath1 and @xmath2mx@xmath1 , respectively . the estimated injection rates decreased up to 13% by increasing the time interval between the magnetograms from 12 min to 36 min , and increased up to 9% by decreasing the dave window size from @xmath3 to @xmath4 pixel@xmath1 , resulting in 10% variation in the accumulated helicity . in both ars , the flare prone regions ( r2 ) had inhomogeneous helicity flux distribution with mixed helicities of both signs and that of cme prone regions had almost homogeneous distribution of helicity flux dominated by single sign . the temporal profiles of helicity injection showed impulsive variations during some flares / cmes due to negative helicity injection into the dominant region of positive helicity flux . a quantitative analysis reveals a marginally significant association of helicity flux with cmes but not flares in ar 11158 , while for the ar 11166 , we found marginally significant association of helicity flux with flares but not cmes , providing evidences of the role of helicity injection at localized sites of the events . these short - term variations of helicity flux are further discussed in view of possible flare - related effects . this study suggests that flux motions and spatial distribution of helicity injection are important to understand the complex nature of magnetic flux system of the active region leading to conditions favorable for eruptive events .
let @xmath24 be some fixed finite connected graph with @xmath25 , and let @xmath26 be the eigenvalues of its adjacency matrix . we think of the edges as being oriented , though the results do not depend on the orientation chosen . we recall that @xmath27 denotes the probability space of @xmath4-lifts of @xmath0 ( i.e. , graphs that have an @xmath4-fold cover map onto @xmath0 ) . a graph @xmath28 , has vertex set @xmath29 . for every ( oriented ) edge @xmath30 , we choose independently and uniformly a random permutation , @xmath31 , and introduce an edge between @xmath32 to @xmath33 for all @xmath34 . for background on lifts and random lifts , see @xcite . in particular @xcite shows how to construct regular graph lifts with a nearly optimal spectral gap . the projection @xmath35 given by @xmath36 is the cover map associated with @xmath28 . every eigenfunction @xmath37 of ( the adjacency matrix of ) @xmath0 is pulled back by @xmath38 to an eigenfunction @xmath39 of @xmath5 . therefore , every eigenvalue of @xmath0 is also an eigenvalue of @xmath5 . such an eigenvalue of @xmath5 is considered `` old '' while all other ones are `` new '' ( including , possibly , duplicates of the old eigenvalues corresponding to `` new '' eigenfunctions ) . let @xmath1 be the ( infinite ) universal cover of @xmath0 ( and @xmath5 ) . we consider @xmath40 , the hilbert space of square - summable real functions on @xmath1 s vertices , i.e. the functions @xmath41 with @xmath42 . if @xmath43 is the ( infinite ) adjacency matrix of @xmath1 , we consider the linear operator corresponding to @xmath43 . namely , @xmath44 where @xmath45 . this is a bounded linear operator on @xmath40 , and we denote its spectral radius by @xmath46 . ( recall that , by definition , @xmath47 ) . our main result is a new bound on @xmath5 s _ new _ eigenvalues in terms of @xmath2 ( the largest eigenvalue of @xmath0 ) , and @xmath3 . let @xmath48 . in @xcite , friedman showed that @xmath49 for almost every @xmath5 . we improve this bound as follows : [ ther:2/3 ] almost every random @xmath4-lift @xmath5 of @xmath0 satisfies : @xmath50 more specifically , @xmath51 almost surely for every @xmath52 . + [ rem : rho<=lambda1 ] let @xmath53 be a ( not necessarily finite ) connected graph , and let @xmath54 be a vertex in @xmath53 . we denote by @xmath55 the number of closed paths of length @xmath56 that start and end at @xmath54 . it is well known that the spectral radius of @xmath53 equals @xmath57 . in particular , this value is independent of the choice of @xmath54 . ( these facts may be proven by an easy variation on the proof of proposition 3.1 in @xcite . ) returning to our notation , observe that a path that starts and ends at a vertex @xmath58 is projected to a path of the same length that starts and ends at the corresponding vertex of @xmath0 . consequently , @xmath59 always holds . our proof of theorem [ ther:2/3 ] suggests an approach that may lead to an even better ( nearly optimal ) bound @xmath60 . this plan depends on an unresolved conjecture that we present shortly . it follows from lubotzky and greenberg @xcite that this statement can not hold with any bound smaller than @xmath61 . it is shown in @xcite that for every infinite tree @xmath1 and for every @xmath52 , there exists a constant @xmath62 , such that if @xmath1 is the universal covering space of a finite graph @xmath53 , then at least @xmath63 . thus for @xmath0 fixed , and for @xmath52 there exists an @xmath64 such that @xmath65 for every @xmath66 and every @xmath28 . ( since the infinite @xmath10-regular tree @xmath67 has spectral radius @xmath68 ( @xcite ) , this extends the alon - boppana bound @xcite that @xmath69 for every @xmath4-vertex @xmath10-regular graph ) . the `` permutation model '' of random @xmath10-regular graphs ( for @xmath10 even ) is a special case of random lifts of graphs . in the permutation model , @xmath4-vertex @xmath10-regular graphs are generated through a random @xmath4-lift of a bouquet of @xmath9 loops . thus , our result , as well as friedman s , extend earlier work on random @xmath10-regular graphs . namely , friedman s result states that @xmath70 for almost every @xmath10-regular graph , which is a slight improvement of an old result of broder and shamir @xcite . in this special case , theorem [ ther:2/3 ] states that @xmath71 holds almost surely , and the tentative proof strategy mentioned above would yield @xmath72 almost surely . in particular , we obtain the following corollary ( which is , of course , substantially weaker than the one proven in @xcite ) : [ cor : d>=107 ] if @xmath0 is @xmath10-regular and @xmath73 , then @xmath74^{1/3 } + \epsilon.\ ] ] almost surely for every @xmath52 . a major tool in this area is the _ trace method _ which goes back to wigner @xcite . it is based on a natural connection between graph spectra and word - maps . this approach underlies the work of broder - shamir @xcite and of friedman @xcite . let @xmath13 be a ( not necessarily reduced ) formal word in the letters @xmath75 . for every @xmath76-tuple @xmath77 of permutations in @xmath78 , we form the permutation @xmath79 , by replacing @xmath80 with @xmath81 in the expression of @xmath13 . for instance , if @xmath82 , then @xmath83 . the correspondence between @xmath13 and the permutation @xmath84 is called a _ word map_. such maps can be evaluated in groups other than @xmath85 as well ( we refer to this briefly in section [ sec : word - maps ] ) . the study of word maps has a long history in group theory ( see @xcite and the references therein ) . our perspective is mostly combinatorial and probabilistic . + for fixed formal word @xmath13 we denote by @xmath19 a random variable on @xmath86 which is defined by : @xmath87 + now let @xmath5 be an @xmath4-lift of @xmath0 and let @xmath88 be the adjacency matrices of @xmath89 resp . we denote by @xmath90 a running index for the `` new '' eigenvalues of @xmath5 . for every @xmath91 , the trace of @xmath92 equals the number of closed paths of length @xmath93 in @xmath5 . this number can also be expressed as @xmath94 . therefore , for @xmath93 even we obtain : @xmath95 every closed path in @xmath5 is a lift of a closed path in @xmath0 . since the edges of @xmath0 are labeled @xmath96 , every ( closed ) path in @xmath0 corresponds to some formal word @xmath13 in @xmath97 . the closed lifts of this path are in @xmath98 correspondence with the fixed points of @xmath99 , so that their number is @xmath100 . let @xmath101 denote the set of all closed paths of length @xmath93 in @xmath0 ( i.e. , @xmath102 ) . the above inequality now becomes : @xmath103\ ] ] taking expectations , we obtain : @xmath104\ ] ] equation shows the significance of @xmath105 in the study of spectra in random lifts . if we let @xmath106 equal @xmath107 , it turns out that for every @xmath13 , @xmath108 can be expressed as a power series in @xmath109 . namely , @xmath110 where the @xmath111 are integers . ( this fact appears in @xcite , but we present ( lemma [ lem : phi ] ) a new and simpler proof ) . this induces a categorization of words in @xmath112 , the free group with generators @xmath96 . namely , @xmath113 is the smallest index @xmath16 for which @xmath114 , or @xmath115 if @xmath116 . + we consider next ( section [ sbs : beta ] ) another categorization of the words in @xmath112 , which does not depend on a word - map to specific groups such as @xmath85 . to every @xmath117 we associate @xmath118 which is a non - negative integer or @xmath115 . the categorizations induced by both @xmath113 and @xmath118 extend the dichotomy between primitive and imprimitive words ( recall that @xmath13 is called _ imprimitive _ if @xmath119 for some @xmath120 and @xmath121 ) . without going into the ( somewhat lengthy ) definition , let us say that the main step in both @xcite and @xcite can be viewed as the observation that for @xmath122 , @xmath123 iff @xmath124 . our aforementioned conjecture states in this language that @xmath125 for every word @xmath13 ( conjecture [ conj : beta = phi ] ) . these relations between @xmath126 and @xmath127 allow us to bound the sum in the r.h.s of : we can bound the contribution of @xmath13 to this sum in terms of @xmath113 . this is complemented by bounding the number of words @xmath128 with a given value of @xmath118 which bound is stated in terms of @xmath3 . indeed , a key step in the present paper ( lemma [ lem : beta>=3 ] ) can be interpreted as a partial proof of the claim that @xmath129 iff @xmath130 . as an aside to our work we obtain a new conceptual and relatively simple proof of a theorem of a. nica @xcite , which determines for every fixed @xmath13 the limit distribution of @xmath19 as @xmath131 ( see theorem [ ther : limit - xwln ] ) . we carry out a similar analysis for all higher moments of @xmath19 , and use the method of moments to derive nica s result . a surprising aspect of this theorem is that the limit distribution depends _ only _ on the largest integer @xmath10 such that @xmath22 for some @xmath132 . nica s full result ( which we derive by the same argument ) concerns not only fixed points but applies just as well to the number of @xmath133-cycles for any fixed @xmath134 . the paper is arranged as follows . we begin ( section [ sec : word - maps ] ) with our analysis of word maps and introduce the two new categorizations of formal words . based on this analysis , we prove theorem [ ther:2/3 ] in section [ sec : lifts ] . in section [ sec : nica ] we deal with the distribution of the number of @xmath133-cycles in @xmath135 and present our new proof for nica s theorem . for the reader interested only in this new proof , this section is mostly self - contained with only occasional references to earlier parts of the paper . there are numerous open problems and conjectures raised in this paper , some of which we collect in section [ sec : open_prob ] . we begin with some notation . we denote by @xmath136 the set of all finite words in letters @xmath14 ( though we occasionally use the letters @xmath137 instead ) . the quotient of @xmath136 modulo reduction of words is @xmath112 , the set of elements of the free group on @xmath76 generators . for instance , the set @xmath101 introduced before equation is a subset of @xmath136 , so it may contain different words which are equivalent as members of @xmath112 . for every group @xmath138 and every word @xmath139 , the _ word map _ @xmath140 is defined by substitutions and composition . for @xmath141 , the element @xmath142 is obtained by substituting @xmath143 for each occurrence of @xmath144 in @xmath13 and this for every @xmath145 . clearly , the word map of @xmath13 is invariant under reductions , so we can regard @xmath13 as an element in @xmath112 . most research on word maps concerns the range of certain fixed words @xmath13 in a group @xmath138 . more specifically , for @xmath138 finite , it is of interest to understand the distribution induced on @xmath138 by the word map @xmath140 and the uniform distribution on @xmath146 . this perspective makes it natural to consider an equivalence relation on words ( beyond that of reduction ) . in order to introduce this equivalence relation , we now recall some simple terminology from combinatorial group theory . there are three _ elementary nielsen transformations _ defined on the free group @xmath112 : ( i ) exchanging any two generators @xmath144 and @xmath147 for some @xmath148 , ( ii ) replacing some @xmath144 with @xmath149 , ( iii ) replacing any @xmath144 by @xmath150 , for some @xmath151 . we recall ( e.g. @xcite , theorem 3.2 ) that these transformations generate the automorphism group @xmath152 of @xmath112 . we say that two words @xmath153 are equivalent , and denote @xmath154 , if they belong to the same orbit of @xmath152 . obviously , @xmath155 is an equivalence relation . it is quite clear that for every finite group @xmath138 , every two equivalent words @xmath153 induce the same distribution on @xmath138 . we do not know whether the converse is true as well , and we state a specific problem ( conjecture [ nielsen_conj ] ) in this vein . given a word @xmath13 and the distribution it induces on a group @xmath138 , it is of interest to consider how far this distribution is from the uniform distribution . the two gradings of words @xmath156 and @xmath157 can be viewed as our attempts to capture this intuition . both parameters associate a non - negative integer or @xmath115 with every @xmath158 , and they tend to grow as the above - mentioned distance decreases . of course , the distribution furthest away from the uniform distribution corresponds to the word @xmath159 . indeed , @xmath160 iff @xmath161 iff @xmath159 ( lemma [ lem : phi = beta=0 ] ) . also , @xmath162 iff @xmath163 iff @xmath13 is imprimitive ( lemma [ lem : phi = beta=1 ] ) . in this case the range of @xmath13 contains only powers of certain exponent . ( recall that @xmath158 is called imprimitive if @xmath22 for some word @xmath23 and @xmath164 . ) at the other end of the scale , both @xmath118 and @xmath113 equal @xmath115 for words that are @xmath165-equivalent to a single - letter word . clearly , such words always induce the uniform distribution on @xmath138 . another important property is that both @xmath127 and @xmath126 are invariant under @xmath155 . the definition of @xmath113 depends on the word map for the symmetric group @xmath85 ( see section [ sbs : exp - fix - points ] ) . the definition of @xmath118 is more involved and is based on a certain analysis of @xmath13 as a formal word without reference to groups ( see section [ sbs : beta ] ) . in fact , we have arrived at our definition of @xmath118 through our study of @xmath113 . some proven results and extensive numerical simulations suggest that @xmath125 for every @xmath13 ( section [ sbs : phi = beta ] ) . one advantage of the parameter @xmath127 over @xmath126 is that we can bound the number of words with fixed value of @xmath127 - see section [ sbs : n_words_with_beta= ] . since both @xmath113 and @xmath118 offer an extension of the primitive - imprimitive dichotomy for words , we tend to think of them as quantifying _ `` the level of primitivity '' _ of a word ( this level is 0 if @xmath159 , it is 1 if @xmath13 is imprimitive , and @xmath166 for primitive words - see section [ sbs : phi = beta ] ) . in this section we present a method to calculate the expectation of @xmath19 , the number of fixed points in @xmath168 ( defined in ) . we count the fixed points in @xmath169 for _ all _ @xmath76-tuples @xmath170 and divide by @xmath171 . this calculation is carried out through a certain categorization of all fixed points . we note that similar considerations appear in @xcite and in @xcite . we begin with some technicalities . let @xmath172 , where @xmath173 and @xmath174 , and let @xmath175 . assume that @xmath176 is a fixed point of @xmath169 . associated with @xmath177 is the following closed trail : @xmath178 with @xmath179 , and @xmath180 ( @xmath181 ) . note that for the sake of convenience , we compose permutations from left to right . this is inconsequential for the analysis of the variables @xmath19 since @xmath169 with left - to - right composition is the inverse of @xmath182 with right - to - left composition , and thus both have the same cycle structure . we categorize fixed points according to their associated trails . let @xmath183 and @xmath184 be the trails of the fixed points @xmath177 and @xmath185 in @xmath186 and @xmath187 respectively . these two trails are placed in _ the same category _ , if they have the same coincidence pattern , that is , if for every @xmath188 , @xmath189 . each closed trail consists of @xmath190 integers , or points , possibly with repetitions , and each category of trails uniquely corresponds to some _ partition _ of these @xmath190 points . consequently , there are at most @xmath191 categories , where the @xmath190-th bell number , @xmath191 , is the number of partitions of an @xmath190 element set . this bound is , however , not tight . for instance , if @xmath192 and @xmath193 , then @xmath194 . therefore , not every partition corresponds to a realizable category of trails . it is convenient to associate a directed edge - colored graph @xmath53 with each category . vertices in @xmath53 correspond to blocks in the partition that defines @xmath53 s category . in other words , @xmath53 has as many vertices as the number of distinct integers among the @xmath195 s . there is a directed edge labeled @xmath34 from one vertex ( = block ) to another , whenever the trails include an arrow labeled @xmath196 ( resp . @xmath197 ) from a point in the first ( second ) block to a point in the second ( first ) one . of special importance is the graph associated with the finest possible partition which we call _ the universal graph_. two points in the trail are merged in this partition if and only if they are merged in every realizable partition . if @xmath13 is cyclically reduced ( i.e. , no two consecutive letters are inverses , nor are the first and last letter ) , this is the partition where all @xmath190 points in the trail are distinct . to illustrate , we draw in figure [ fig : univ - graphs ] the universal graph of three different words . ( the commutator word ) , of @xmath198 , and of @xmath199.,scaledwidth=90.0% ] all other graphs are now easily derived as quotients of the universal graph , or partitions of its vertices . a quotient graph has one vertex per each block in the partition . it has a @xmath34-labeled directed edge ( @xmath34-edge for short ) from block @xmath200 to block @xmath201 , if the universal graph contains a @xmath34-edge from a vertex in @xmath200 to a vertex in @xmath201 . a quotient is not _ realizable _ if it contains two distinct @xmath34-edges with common head and different tails or vice - versa . we denote by @xmath202 the set of all realizable quotients . to illustrate , we draw ( figure [ fig:7graphs ] ) all the realizable quotient graphs of the universal graph of the commutator word ( one of the graphs in figure [ fig : univ - graphs ] ) . note that a four element set has 15 partitions ( the fourth bell number , @xmath203 ) , of which only 7 are realizable in this case . , the set of graphs of categories of fixed points when @xmath204 . each graph is drawn together with the partition of the universal graph which yields it . ( we do not specify the block corresponding to each vertex).,scaledwidth=80.0% ] these graphs suggest a simple formula for the number of fixed points in each category . let @xmath205 , ( @xmath206 ) be the number of vertices ( edges ) in the graph @xmath53 , and @xmath207 be the number of @xmath34-edges ( and so @xmath208 ) . to count the number of fixed points in @xmath53 s category , or the number of _ realizations _ of @xmath53 , we first label @xmath53 s vertices by distinct numbers from @xmath209 ( i.e. , specify the values of @xmath210 ) . this can be done in @xmath211 ways . for each @xmath212 there are @xmath213 permutations that are consistent with the @xmath207 values in the permutation @xmath196 that are already determined . thus , the number of realizations of @xmath53 is : @xmath214!\ ] ] a formula for the expectation of @xmath19 is now at hand : @xmath215 ( the third equality holds only for @xmath4 that is @xmath216 for all @xmath34 and @xmath53 . ) we illustrate these calculations for @xmath13 the commutator word @xmath217 . if we go over the graphs in figure [ fig:7graphs ] in clockwise order starting at the upper - left graph , becomes : @xmath218[n(n-1 ) ] } + \frac{n(n-1)(n-2)}{[n(n-1)][n(n-1 ) ] } + \\ & & + \frac{n(n-1)}{[n(n-1)][n ] } + \frac{n(n-1)}{[n][n(n-1 ) ] } + \frac{n}{[n][n ] } + \\ & & + \frac{n(n-1)}{[n(n-1)][n(n-1 ) ] } + \frac{n(n-1)(n-2)}{[n(n-1)][n(n-1 ) ] } = \frac{n}{n-1}\end{aligned}\ ] ] in we defined @xmath106 to be @xmath107 . the following lemma states an important property of @xmath108 : [ lem : phi ] associated with every @xmath158 is a power series @xmath219 that has a positive radius of convergence where the coefficients @xmath111 are integers . in particular , for every @xmath13 and for every sufficiently large @xmath4 , there holds @xmath220 . as we saw in , for large enough @xmath4 ( e.g. , @xmath221 suffices ) , @xmath222 since every @xmath53 is a connected graph , we have @xmath223 . the lemma follows when we individually consider the expression corresponding to each @xmath53 . ( see the proof of lemma [ lem : analyze - o(1 ) ] for a thorough analysis of these expressions ) . by lemma [ lem : phi ] , the contribution of every @xmath128 in , is @xmath224 for some nonnegative integer @xmath16 . this induces the following useful grading of words : @xmath225 recall that although the construction of @xmath108 depends on the actual representation of @xmath13 ( a reduction of @xmath13 usually changes @xmath202 ) , this function captures some features of the distribution of the image of @xmath13 on @xmath85 . thus @xmath108 , as well as @xmath113 , are invariant not only under reduction , but also under @xmath155 . with this new terminology we can reinterpret both @xcite and @xcite as follows : both papers rely on the fact that @xmath226 iff @xmath13 reduces to the empty word , and that @xmath227 iff @xmath13 is imprimitive ( see lemmas [ lem : phi = beta=0 ] and [ lem : phi = beta=1 ] ) . to study the new spectrum of random lifts , one proceeds as follows : the number of words of these two kinds can be bounded in terms of @xmath3 ( the spectral radius of the universal cover ) alone ( and does not depend on @xmath2 , the spectral radius of the base graph ) . finally , the rest of the words ( which are , in fact , the vast majority ) contribute to the summation in only @xmath228 each . here we extend these ideas and seek ( with partial success ) a similar characterization for all words with @xmath123 for fixed @xmath16 . our analysis of the new spectrum extends these arguments and refines them . we further split the above - mentioned third set and attain an improved bound on the contributions of these subsets to the sum in equation . we now start our second attempt at capturing the level of primitivity " of @xmath13 by means of the parameter @xmath118 . we begin with some definitions . recall the notion of a trail from section [ sbs : exp - fix - points ] . consider then the following trail through @xmath229 ( this time , the @xmath230 s should not be thought of as numbers but rather as abstract symbols , and the trail deliberately ends with @xmath231 and not with @xmath177 ) : @xmath232 as in our former definition of a realizable category of trails , we say that a partition of @xmath233 is realizable if the following conditions hold : whenever @xmath234 and @xmath235 , @xmath236 is in the same block with @xmath237 ( we denote @xmath238 ) iff @xmath239 . likewise , whenever @xmath234 and @xmath240 , @xmath241 . ( in other words , a partition is realizable whenever it traces a trail of some point , fixed or not , through @xmath169 for some @xmath242 and some @xmath4 . ) as before , to each realizable partition of @xmath233 corresponds a directed edge - colored graph @xmath53 , which is a quotient of the graph of the trail . according to our former notation , @xmath243 whenever @xmath244 . we now concentrate on the number of pairs of @xmath230 s that should be merged in order to yield a specific @xmath53 . [ df : generating - set ] let @xmath53 be the quotient graph corresponding to some realizable partition of @xmath233 . we say that set of pairs @xmath245 * generates * @xmath53 , if @xmath53 corresponds to the finest realizable partition in which @xmath246 . + let us return to the commutator word @xmath247 . the trail here is @xmath248 in figure [ fig : minimal - gps ] we revisit the seven quotient graphs from figure [ fig:7graphs ] ( the seven graphs in @xmath202 ) , and specify a smallest generating set for each of them . , the set of graphs of categories of fixed points when @xmath204.,scaledwidth=80.0% ] we denote by @xmath249 the euler characteristic of @xmath53 . it turns out that there is a tight connection between @xmath250 and the smallest size of a generating set of @xmath53 : [ lem : char = mgps ] let @xmath53 be the quotient graph corresponding to some realizable partition of @xmath233 . the smallest cardinality of a generating set for @xmath53 is @xmath250 . it is quite easy to construct a set @xmath251 of @xmath250 pairs that generates @xmath53 . to this end , we adopt the original terminology of @xcite . as we follow the path of @xmath13 through @xmath53 , each move has one of three types . in a * free * step we traverse a new edge and reach a new vertex , and so one vertex and one edge are added to the partial graph . in a * coincidence * a new edge leads us to an `` old '' vertex , so we gain one new edge and no new vertices . in a * forced * step we traverse an old edge ( necessarily to an old vertex ) , so the numbers of vertices and edges remain unchanged . consequently , @xmath250 equals the number of coincidences in this walk . we introduce into @xmath251 one pair for each coincidence . if the @xmath34-th step is a coincidence in which we reach a vertex in @xmath53 representing the block @xmath252 , we add @xmath253 to @xmath251 . clearly , the cardinality of @xmath251 is @xmath250 and it generates @xmath53 . to see that @xmath53 has no generating set smaller than @xmath251 , consider @xmath254 , the collection of all generating sets of @xmath53 of the smallest possible cardinality . we claim that @xmath251 is the lexicographically first member of @xmath254 . concretely , write each pair under consideration as @xmath255 with @xmath256 . now sort the pairs in each @xmath257 in increasing lexicographic order and let @xmath258 be the lexicographically first member of @xmath254 . our claim is that @xmath259 . observe that if @xmath260 then there is no index @xmath261 with @xmath262 with @xmath263 . otherwise we could replace the pair @xmath255 with the pair @xmath264 and generate the same quotient as does @xmath1 with a lexicographically smaller set of pairs . it is helpful to consider for each @xmath265 the graph @xmath266 that is the quotient of @xmath267 generated by the initial segments of @xmath1 that includes only those pairs in @xmath1 where both indices are @xmath268 . we claim that @xmath269 is a subgraph of @xmath266 for all @xmath16 . if there is no pair in @xmath1 where @xmath230 is the larger member , this is clear . if @xmath255 is in @xmath1 then the index @xmath34 is uniquely defined by the above remark . in this case we need to show that the identification of @xmath230 with @xmath270 does not entail any additional identification ( which would violate the inclusion @xmath271 ) . how can such an identification occur ? only if the label of the edge @xmath272 agrees with that of an edge @xmath273 incident with the block that contains @xmath270 ( and has the correct orientation ) . let @xmath274 be a member of the block at the other end of @xmath273 . clearly @xmath275 . but now again we can generate the quotient generated by @xmath1 by a lexicographically smaller set , i.e. , replace the pair @xmath255 by @xmath276 . we can again recognize the three types of steps by observing how the graphs @xmath266 grow at each step . if @xmath266 stays unchanged , this is a forced move . if only an edge is added , this is a coincidence and in a free step one vertex and one edge are added . it is exactly at each coincidence step that vertices at @xmath1 get merged . but this is precisely what we did in constructing @xmath251 , so that @xmath277 , as claimed . the following categorization of the quotient graphs in @xmath278 turns out to be very useful : [ df : ab ] let @xmath13 be a word in @xmath136 . we say that a quotient graph @xmath243 * has type a * , if one of the smallest generating sets for @xmath53 contains the pair @xmath279 . otherwise , we say @xmath53 * has type b*. given a word @xmath13 , we classify the graphs in @xmath202 according to their characteristics and type . note that @xmath280 purposely include @xmath281 ) . table [ tab : chi - type ] shows the whole census . .the number of graphs in @xmath202 in any specified character and type , when @xmath204 . [ cols="^,^,^,^,^,^",options="header " , ] we now select @xmath93 to minimize the resulting upper bound on @xmath282 . it is not hard to see that the optimal @xmath93 is @xmath283 . consequently , all terms @xmath284 are @xmath285 and can be ignored . the information in table [ tab : split - by - beta ] is now combined with to yield [ for @xmath286 $ ] : @xmath287^t\end{aligned}\ ] ] in order to balance the first and the last terms , we set @xmath288 ( recall that @xmath93 is an even integer , so we can not guarantee exact equality here ) . since @xmath289 with @xmath4 , the constant and polynomial factors can be replaced by @xmath290 . we obtain @xmath291^t \\ & = & \left[\lambda_1^{\;1/3 } \rho^{2/3 } \cdot \left(1+o_n(1)\right ) \cdot max\left(1 , 3\left(\frac{\rho}{\lambda_1}\right)^{2/3}\right)\right]^t\end{aligned}\ ] ] ( the equality holds because @xmath59 is always true , see remark [ rem : rho<=lambda1 ] ) . the statement of theorem [ ther:2/3 ] now follows from a standard application of markov s inequality . obviously , for every @xmath52 , @xmath292 may serve as an absolute upper bound . here is a sketch of our proposed approach to friedman s conjecture . say we seek to prove that all new eigenvalues are , almost surely @xmath293 for every @xmath52 . to prove a bound of @xmath294 one would have to show that @xmath295 for every @xmath296 . we believe that this can be shown in a way similar to our proof of theorem [ ther:2/3 ] . in addition , it would be necessary to follow up on remark [ rem : bound_for_beta = r ] , and establish a bound on the number of words @xmath297 with given @xmath118 . in this section we introduce a new conceptual and relatively simple proof of a theorem of a. nica @xcite . in we defined the random variable @xmath19 which counts the number of fixed points in @xmath169 for fixed @xmath13 . we extend this concept and for every integer @xmath299 denote by @xmath300 a random variable on @xmath301 which is defined by : @xmath302 ( @xmath303 is a new notation for @xmath19 ) . + nica s theorem says that the variables @xmath300 have , for fixed @xmath13 and @xmath133 and for @xmath304 , a limit distribution which can be computed explicitly . ( unless otherwise stated , the distribution on @xmath86 is always uniform . ) [ ther : limit - xwln ] let @xmath305 and suppose that @xmath119 , with @xmath23 primitive . then for every integer @xmath134 , the random variable @xmath300 defined in ( [ eq : xwln ] ) has , for @xmath304 , a limit distribution , which is given by : @xmath306 where @xmath307 is the set @xmath308 @xmath309 ( a variable with poisson distribution with parameter @xmath190 ) , and `` @xmath310 '' denotes convergence in distribution . in particular , this limit distribution depends only on @xmath10 and @xmath133 ( and not on @xmath23 ) . note that in the case that @xmath13 is primitive ( i.e. @xmath311 ) , the limit distribution is simply poisson with parameter @xmath312 . our proof of theorem [ ther : limit - xwln ] is based on the method of moments and provides , in particular , explicit expressions for the moments of @xmath300 : [ cor : moments ] for every @xmath313 , @xmath134 and @xmath314 there exists a rational function @xmath315 that is defined on a neighborhood of @xmath316 such that @xmath317^r)=\psi_{w , l , r}(1/n)$ ] for sufficiently large @xmath4 . consequently , the limit @xmath318^r)$ ] exists and equals @xmath319 . in addition , if @xmath119 , with @xmath23 primitive , then this limit equals the corresponding moment of the sum in . in particular , @xmath320 . ( note that this is a generalization of the function @xmath106 defined in : @xmath321 . ) the contents of nica s work is theorem [ ther : limit - xwln ] and corollary [ cor : moments ] for the first moment . + the main idea of the proof was already illustrated in section [ sec : word - maps ] , where we analyzed the expectation of @xmath303 . recall equation where @xmath106 is expressed as a power series @xmath322 . we already know ( lemma [ lem : phi = beta=0 ] ) that when @xmath323 , @xmath324 , and so @xmath325 . but this is exactly the number of graphs @xmath243 with @xmath326 ( see , for instance , lemma [ lem : phi = beta=1 ] ) . if @xmath13 is cyclically reduced , then the only graphs in @xmath202 with @xmath327 are cycles ( again , see the proof of lemma [ lem : phi = beta=1 ] ) . such a cycle consists of a cyclic concatenation of several copies of @xmath23 ( the primitive word such that @xmath22 ) . the number of copies of @xmath23 in the cycle has to divide @xmath10 , hence @xmath328 but we can indeed restrict our discussion to cyclically reduced words . the justification for this is the following . let @xmath329 , and let @xmath330 be its cyclic reduction . we have already mentioned that @xmath331 and so they induce the same distribution on @xmath85 . in particular , @xmath300 and @xmath332 are equally distributed . it is also quite evident that @xmath13 and @xmath330 share an identical exponent of their primitive root ( i.e. , if @xmath333 for some @xmath334 and @xmath335 with @xmath23 primitive , then @xmath336 is primitive too , and @xmath337 ) . thus the validity of theorem [ ther : limit - xwln ] and corollary [ cor : moments ] for @xmath330 , yields their validity for @xmath13 as well . below , we extend this argument to obtain the limit of the expectation of @xmath338^r$ ] for every @xmath133 and @xmath339 . we essentially use the same way we counted fixed points in @xmath169 for all @xmath76-tuples @xmath170 , to count @xmath133-cycles and _ sequences _ of @xmath133-cycles , which we call _ lists _ of cycles . the point is that the total number of lists of length @xmath339 of @xmath133-cycles , divided by @xmath171 , equals the expectation of @xmath338_r$ ] , the @xmath339-th _ factorial _ moment of @xmath300 . once we know how to calculate the limits of the factorial moments , the limits of the regular moments are at easy reach . to finish , we show that these limits equal the corresponding moments in the r.h.s of , and use the method of moments to conclude the proof . we begin by generalizing some of the notions from section [ sbs : exp - fix - points ] . let @xmath340 be cyclically reduced , @xmath341 an integer , @xmath342 , and @xmath242 . the _ trail of @xmath177 through @xmath169 _ is the sequence of images of @xmath177 under @xmath169 . namely , if @xmath343 ( with @xmath344 ) in reduced form , then associated with @xmath177 is the following path : @xmath345 with @xmath346 , and @xmath347 ( @xmath181 ) . ( recall that we compose permutations from left to right . ) likewise , we can speak of the trail through some power of @xmath13 . for example , the trail of @xmath177 through @xmath348 is @xmath349 with @xmath350 as before , and @xmath351 satisfying the obvious constraints . we recall that two trails were placed in the same category if they have the same coincidence pattern . this notion can be extended to our present , more general context , in an obvious way . namely , every category is associated with some directed edge - colored graph . moreover , we can define categories not only of single trails , but also of _ lists _ of trails , and again , associate a graph to each category . the nature of this graph is exactly as described in section [ sbs : exp - fix - points ] . to illustrate , let @xmath204 be the commutator word , @xmath352 an integer and @xmath353 such that the following trails are realized by @xmath354 and @xmath355 , respectively : @xmath356 we denote the nodes of the trail through @xmath13 by @xmath357 and the nodes through @xmath358 by @xmath359 . then the graph associated with the category of this list of trails through @xmath360 is shown in figure [ fig : first - graph ] . + , where @xmath204 is the commutator word . , scaledwidth=80.0% ] although the notions here have a wider scope , we limit our discussion to categories of trails which represent _ cycles _ in @xmath169 : an @xmath133-cycle is represented by a closed trail through @xmath361 . in fact , we are interested in counting cycles of a given length , so for our purposes we can confine ourselves to categories of a list of @xmath339 trails through @xmath361 , for some @xmath362 . first , let us analyze the categories that represent a single @xmath133-cycle . a closed trail through @xmath361 represents an @xmath133-cycle only if it does not represent any smaller cycle . that is , in the closed trail @xmath363 the @xmath133 labels @xmath364 must all be distinct . once again , the graph of each category is a quotient of the _ universal graph_. in this case , the universal graph is @xmath365 , the cycle of length @xmath366 . a partition of its vertices corresponds to some category of an @xmath133-cycle if and only if it does not create `` collisions '' of same - color edges , and keeps apart all @xmath133 vertices corresponding to @xmath367 . we demonstrate this in figure [ fig : l - univ - graph ] . + . in a realizable partition of the vertices that represents a 3-cycle , the 3 circled vertices ( @xmath368 ) must belong to distinct blocks.,scaledwidth=60.0% ] the most general case in our discussion comes up when we turn to calculate the @xmath339-th moment of @xmath300 . to this end we consider categories of lists of @xmath339 @xmath133-cycles through @xmath369 . the universal graph @xmath370 which represents @xmath339 ordered cycles of length @xmath133 each , consists of a disjoint union of @xmath339 copies of @xmath365 . we name the vertices of the first cycle in @xmath370 by @xmath371 , the vertices of the second cycle by @xmath372 and so on until @xmath373 for the @xmath339-th cycle . we are interested in quotients ( or partitions of the vertices ) of @xmath370 that represent @xmath339 distinct @xmath133-cycles . this means that the vertices @xmath374 ( a total of @xmath375 vertices ) should be kept apart in every partition , as illustrated in figure [ fig : lr - univ - graph ] . of four 3-cycles where @xmath204 . in a realizable partition of the vertices that represents four 3-cycles , the 12 circled vertices must belong to distinct blocks . ( we omit the blocks corresponding to the rest of the vertices.),scaledwidth=80.0% ] the method we use to calculate the expectation of @xmath376^r$ ] is quite similar to the one we used for @xmath377 . we first calculate the `` factorial moments '' of @xmath300 and derive the regular moments from them ( the @xmath339-th factorial moment of a random variable @xmath378 is defined as @xmath379_r \big)$ ] , where @xmath380_r$ ] is the `` falling factorial '' , namely @xmath380_r = x(x-1 ) \ldots ( x - r+1)$ ] ) . @xmath381_r$ ] counts lists of @xmath339 @xmath133-cycles in @xmath369 . as in the case of the first moment , we calculate its expectation by counting the total number of lists of @xmath339 @xmath133-cycles in @xmath369 for _ all _ @xmath76-tuples @xmath382 and dividing by @xmath171 . the counting is carried out by classifying these lists into categories . each list of @xmath339 @xmath133-cycles _ with specified starting point for each cycle _ , belongs to some category . these categories are the quotients of the universal graph @xmath370 , which we denote by @xmath383 ( e.g. , @xmath384 is the same set as @xmath278 ) . to recap , the set @xmath385 can be generated as follows : we first draw @xmath370 , the universal graph of @xmath339 ordered @xmath133-cycles of @xmath13 , which consists of @xmath339 disjoint cycles each of which has @xmath386 vertices . @xmath385 consists of quotient graphs that are generated by partitions of the vertices of @xmath370 . a quotient graph is included in @xmath385 if it is realizable , and if in the corresponding partition each of the @xmath387 vertices that represent the @xmath339 @xmath133-cycles is in a different block . + let @xmath53 be some graph in @xmath385 . realization _ of @xmath53 is a @xmath76-tuple of permutations @xmath242 , a list of @xmath339 @xmath133-cycles of @xmath169 and a specified starting point for each cycle , such that they belong to @xmath53 s category . the number of realizations of @xmath53 is the same as in , namely : @xmath388 since every list of @xmath339 @xmath133-cycles is counted @xmath389 times ( there are @xmath389 ways to choose the starting points ) , we have : @xmath390_r\big ) & = & \frac{1}{(n!)^k } \sum_{(\sigma_1,\ldots,\sigma_k)\in s_n^{\;k } } [ x_{w , l}^{(n)}(\sigma_1,\ldots,\sigma_k)\big]_r = \nonumber \\ & = & \frac{1}{(n!)^k } \cdot \frac{1}{l^r } \sum_{\gamma \in { \cal q}_{w , l , r}}n_\gamma(n ) = \\ \label{eq : moment - formula2 } & = & \frac{1}{l^r } \sum_{\gamma \in { \cal q}_{w , l , r } } \frac{n(n-1)\ldots(n - v_\gamma+1)}{\prod_{j=1}^k n(n-1)\ldots ( n - e^j_\gamma+1 ) } = \\ & = & \frac{1}{l^r } \sum_{\gamma \in { \cal q}_{w , l , r } } [ n]_{v_\gamma } \prod_{j=1}^k \frac{1}{[n]_{e^j_\gamma } } \nonumber\end{aligned}\ ] ] note that the equality between and holds only for @xmath4 large enough . indeed @xmath391 if @xmath392 for some @xmath393 and @xmath394 . this holds for @xmath395 . for every @xmath134 and @xmath314 , thus yields a rational function in @xmath4 , which , for sufficiently large @xmath4 , is the @xmath339-th factorial moment of @xmath300 . it is convenient to rewrite as a function of @xmath109 : @xmath396_r\big ) = \frac{1}{l^r}\sum_{\gamma \in { \cal q}_{w , l , r } } \left(\frac{1}{n}\right)^{e_\gamma - v_\gamma } \frac{\prod_{t=1}^{v_\gamma-1}(1-\frac{t}{n } ) } { \prod_{j=1}^k\prod_{t=1}^{e^j_\gamma-1}(1-\frac{t}{n})}\ ] ] we can now define a rational function @xmath397 by : @xmath398 and so @xmath399_r\big ) = \psi_{w , l , r}(\frac{1}{n})$ ] for @xmath400 . the following lemma shows that @xmath397 is well defined on a neighborhood of @xmath316 . [ lem : edges ] for each @xmath393 , @xmath401 . note that in @xmath370 every vertex has degree 2 , since @xmath13 is cyclically reduced . this degree can not decrease in a quotient . therefore , every vertex in every @xmath393 has degree at least 2 , and the lemma follows . it is well - known how to express the regular moments as linear combinations of the factorial moments . thus we can derive ( an efficiently computable ) rational function @xmath315 ( which is a linear combination of @xmath402 ) , such that @xmath317^r)=\psi_{w , l , r}(1/n)$ ] for sufficiently large @xmath4 . this function @xmath315 is obviously defined on a neighborhood of @xmath316 , which proves the first part of corollary [ cor : moments ] . the proof of theorem [ ther : limit - xwln ] is based on the method of moments . under certain mild conditions , a probability distribution is determined by its moments , or as here , a limit distribution is determined by the limits of the moments . a probability measure @xmath90 on @xmath403 is said to be _ determined by its moments _ if it has finite moments @xmath404 of all orders , and @xmath90 is the only probability measure with these moments . we quote theorem 30.2 from @xcite : [ metod - of - moments ] let @xmath378 and @xmath405 be random variables , and suppose that the distribution of @xmath378 is determined by its moments , that the @xmath406 have moments of all order , and that @xmath407 for every @xmath408 . + then @xmath409 note that if @xmath378 and the @xmath406 are integer - valued then @xmath410 is equivalent to @xmath411 for every integer @xmath76 . the relation between regular moments and factorial moments implies : [ cor : method - of - factorial - moments ] the statement of theorem [ metod - of - moments ] holds where moments are replaced by factorial moments . in this section we use corollary [ cor : method - of - factorial - moments ] to prove theorem [ ther : limit - xwln ] . that @xmath300 has moments of all order is evident . we still need to show that the r.h.s of is determined by its moments , and that the @xmath339-th factorial moment of @xmath300 indeed converges to the @xmath339-th factorial moment of this random variable . + theorem 30.1 from @xcite provides a sufficient condition for a probability measure @xmath90 to be determined by its moments , namely , that the power series @xmath412 where @xmath413 is the @xmath339-th moment of @xmath90 has a positive radius of convergence . ( this series is the moment generating function of @xmath90 , when the latter exists . ) for @xmath90 a poisson distribution ( with any parameter ) , this power series converges for all real @xmath93 ( e.g. , @xcite , section 4 ) , so @xmath90 is determined by its moments . a convolution ( a summation ) of several poisson distributions is itself poisson ( whose parameter is the sum of parameters ) , and thus satisfies the condition as well . in particular , recall that the r.h.s . of is @xmath414 , where @xmath309 . if we omit the constant @xmath415 of every term in this sum , we obtain @xmath416 , whose distribution is simply @xmath417 , which is determined by its moments . according to the definition of @xmath307 in , each @xmath418 satisfies @xmath419 . thus , if we denote by @xmath413 the @xmath339-th moment of @xmath420 and by @xmath421 the @xmath339-th moment of @xmath414 , then @xmath422 . consequently , the series @xmath423 has radius of convergence that is @xmath424 that of the series @xmath412 . but the latter converges for all real @xmath93 , hence so does @xmath425 , and the distribution of @xmath426 is determined by its moments . + + to conclude the proof of theorem [ ther : limit - xwln ] , we need to show that the ( factorial ) moments of @xmath300 indeed converge to the respective moments of the r.h.s . of . but what is the limit of @xmath399_r\big)$ ] ? by lemma [ lem : edges ] , the limit of each term in the r.h.s . of is either @xmath316 ( if @xmath427 , or @xmath428 ( if @xmath326 ) . therefore , @xmath429_r\big ) = \frac{1}{l^r } \big|\{\gamma \in { \cal q}_{w , l , r}~:~ \chi(\gamma)=1 \ } \big|\ ] ] as explained in the proof of lemma [ lem : edges ] , the equality @xmath326 holds for some @xmath393 , iff every vertex in @xmath53 has degree 2 , i.e. , iff @xmath53 is a disjoint union of cycles . we denote by @xmath430 the subset of @xmath385 consisting of all graphs which are a disjoint union of cycles . now becomes : @xmath431_r\big ) = \frac{1}{l^r}\big| { \cal c}_{w , l , r } \big|\ ] ] let @xmath158 be cyclically reduced and equal @xmath432 with @xmath23 primitive and @xmath433 . a graph @xmath434 has a very specific structure : each cycle @xmath435 in @xmath53 must be a cyclic concatenation of several copies of @xmath23 ( every cycle in @xmath53 looks like @xmath436 for some positive integer @xmath437 ) . to see this , recall that each cycle @xmath435 in @xmath53 represents a closed trail through @xmath438 ( at least one closed trail ) . hence there is some vertex @xmath439 in @xmath435 and some orientation on @xmath435 , such that if we leave @xmath439 in this orientation and go exactly @xmath440 times through @xmath23 , we get back to @xmath439 ( possibly after tracing @xmath435 several times ) . since @xmath23 is primitive , it can not be invariant under cyclic shift of any length @xmath441 , and the size of @xmath435 must divide @xmath442 . ( in fact , the integer @xmath443 divides @xmath444 ) . moreover , all closed trails that are represented in @xmath435 , go in the same direction ( and thus also start in one of the @xmath443 head vertices of @xmath23 ) . we already saw in the proof of lemma [ lem : phi approx beta=2 ] that for @xmath23 primitive , @xmath445 is not a subword of @xmath446 . this rules out the possibility of `` finding @xmath23 in the opposite direction '' . this analysis of the structure of the graphs in @xmath430 yields an important corollary , which ultimately explains why the limit distribution of @xmath447 depends solely on @xmath10 and @xmath133 , and not on @xmath23 : [ cor : u - not - matter ] let @xmath448 be cyclically reduced and equal @xmath449 and @xmath450 , respectively , with @xmath451 and @xmath452 primitive and @xmath433 . then @xmath453 the analysis above shows that the inner structure of @xmath23 is completely irrelevant to the graphs in @xmath430 , and there is a natural bijection between @xmath454 and @xmath455 : simply replace each copy of @xmath451 by a copy of @xmath452 . the material in this section is not necessary for the proof and deals with the special case of primitive words . our hope is that this spacial case makes it easier to follow the general proof . so let us assume that @xmath456 and @xmath311 . the universal graph @xmath370 is a disjoint union of @xmath339 cycles , each of which consists of @xmath133 copies of @xmath23 . we have a total of @xmath457 copies of @xmath23 , and their @xmath387 initial vertices are kept separated in each quotient . it is easily verified that the only quotient of @xmath370 which is a disjoint union of cycles is @xmath370 itself , and thus @xmath458 . by we have @xmath459_r\big ) = \frac{1}{l^r}$ ] . on the other hand , when @xmath13 is primitive , the r.h.s . of is simply @xmath460 . let @xmath378 be an integer - valued non - negative random variable and let @xmath461 be its generating function @xmath462 under certain mild conditions and in particular if @xmath461 is analytic on @xmath403 , @xmath463_r ) = f_x^{\;(r)}(1 ) \qquad \forall r\in\mathbb{n}.\ ] ] the generating function of @xmath460 is @xmath464 . thus , @xmath465_r ) = f^{(r)}(1 ) = \frac{1}{l^r}\ ] ] which completes the proof of theorem [ ther : limit - xwln ] for @xmath13 primitive . to complete the proof of theorem [ ther : limit - xwln ] , we show that for every @xmath466 , @xmath134 and @xmath467 , the series @xmath399_r\big)$ ] converges , for @xmath21 , to @xmath468_r\big)$ ] , where @xmath469 is the r.h.s . of , i.e. @xmath470 . by , the series @xmath399_r\big)$ ] converges to @xmath483 , so we proceed to analyze the set @xmath430 . let @xmath53 be a graph in @xmath430 . as explained in section [ sbsbs : cwlr ] , @xmath53 is a disjoint union of cycles , each of which consists of several copies of @xmath23 . if we let @xmath437 denote the number of copies of @xmath23 in some cycle , what are the possible values of @xmath437 ? to begin with , @xmath484 , as this cycle represents a closed path that consists of @xmath485 copies of @xmath23 . secondly , as this cycle represents an @xmath133-cycle , @xmath486 and @xmath487 for any @xmath488 . if we let @xmath489 , then the constraints on @xmath437 translate into the following conditions : @xmath490 and @xmath491 , precisely as in the definition of @xmath307 note that a cycle in @xmath53 with @xmath492 copies of @xmath23 can represent up to @xmath415 distinct @xmath133-cycles in @xmath169 . as before , let @xmath493 . for @xmath494 consider those @xmath133-cycles which are associated in the quotient @xmath53 to a cycle of length @xmath495 . ( the @xmath16-th @xmath133-cycle belongs to the cycle @xmath435 in @xmath53 if the blocks containing @xmath496 correspond to vertices in @xmath435 . ) let @xmath497 be the number of such @xmath133-cycles whence @xmath498 , and there are @xmath499 ways to choose which @xmath133-cycles go where ( recall that @xmath370 consists of an _ ordered _ list of @xmath339 @xmath133-cycles ) . now let @xmath500 denote the subset of @xmath430 consisting of all quotient graphs where all disjoint cycles are of equal length of @xmath501 each . then we have : @xmath502_r\big ) & \stackrel{\eqref{eq : lim = c}}{= } & \frac{1}{l^r}\big| { \cal c}_{w , l , r } \big| = \frac{1}{l^r } \sum_{\substack{r_1 + \ldots + r_p = r \\ r_j \geq 0 } } \binom{r}{r_1 \ldots r_p } \prod_{j=1}^p \big| { \cal c}_{w , l , r_j}^{h_j } \big| \nonumber \\ & = & \sum_{\substack{r_1 + \ldots + r_p = r \\ r_j \geq 0 } } \binom{r}{r_1 \ldots r_p } \prod_{j=1}^p \frac{1}{l^{r_j}}\big| { \cal c}_{w , l , r_j}^{h_j } \big|\end{aligned}\ ] ] ( @xmath503 denotes the singleton of the empty graph , and therefore @xmath504 ) . by combining and , we conclude that theorem [ ther : limit - xwln ] will follow if we show @xmath505 for every @xmath506 and @xmath418 . + + we begin with the l.h.s . of . recall that by definition , each graph @xmath507 consists of a disjoint union of cycles of length @xmath501 each . thus , each cycle represents up to @xmath415 distinct @xmath133-cycles of @xmath169 , and @xmath53 can represent up to @xmath508 distinct @xmath133-cycles . but @xmath53 represents only @xmath339 distinct @xmath133-cycles , whence there are @xmath509 `` free spots '' in @xmath53 that can contain new @xmath133-cycles . we illustrate these notions in figure [ fig : c_hwlr_example ] . , where @xmath510 for some primitive @xmath511 , @xmath512 , @xmath513 and @xmath514 . ( we use the labels @xmath515 introduced in section [ sbs : categories ] . ) this @xmath53 contains two cycles , each of which consists of six copies of @xmath23 . each of the cycles can correspond to up to three distinct @xmath133-cycles ( indeed , @xmath512 ) , so @xmath53 can contain up to six @xmath133-cycles . as it already contains three , it has @xmath516 free spots.,scaledwidth=100.0% ] now let @xmath517 $ ] denote the number of graphs in @xmath518 with @xmath34 free spots . we can define the generating function of the @xmath517 $ ] : @xmath519 + \alpha_{w , l , r}^{h}[1]t + \alpha_{w , l , r}^{h}[2]t^2 + \ldots\ ] ] and obviously @xmath520 . before we derive a recursion formula for this function , we want to illustrate by writing explicit expressions for @xmath521 . every connected component ( = cycle ) in every @xmath507 , realizes at least one @xmath133-cycle . thus , when @xmath522 and there are no @xmath133-cycles at all , we have only the empty graph which has no free spots , and so @xmath523 . when @xmath524 , we have a single graph in @xmath525 , with @xmath526 free spots ( a single cycle of @xmath527 copies of @xmath23 ) , and therefore @xmath528 . for @xmath529 there is always a two - cycle graph with one @xmath133-cycle in each cycle . this graph has @xmath530 free spots . in addition , if @xmath531 , there are also graphs consisting of a single cycles that corresponds to two @xmath133-cycles . there are @xmath532 ways to place the two @xmath133-cycles and such graphs have @xmath533 free spots . thus , for @xmath534 , @xmath535 and for @xmath531 @xmath536 . we now want to derive the functions @xmath537 by recursing on @xmath339 . let @xmath53 be a graph in @xmath538 with @xmath34 free spots . in what manners can we add another @xmath133-cycle and make it a graph in @xmath539 ? we have two options : we can put the new @xmath133-cycle in one of the @xmath34 free spots , in @xmath133 possible ways ( @xmath133 possible cyclic shifts ) , resulting in @xmath540 different graphs in @xmath539 , each of which has @xmath541 free spots . alternatively , we can add one new cycle to @xmath53 and put there our new @xmath133-cycle , which yields a single graph in @xmath542 with @xmath543 free spots . thus , we have : @xmath544 we now go back to the right side of . recall that @xmath545 . if we write @xmath546 where @xmath547 is the appropriate polynomial , then @xmath548 and @xmath549 thus @xmath550 , and we can conclude : @xmath551 when the last equality comes from the fact that @xmath552 . + + the proof of theorem [ ther : limit - xwln ] is now complete . corollary [ cor : moments ] is also proved . for completeness sake , here is a short proof of the last sentence in the corollary , namely , that @xmath553 . @xmath554 in fact , the technique presented here are likely to yield further results . the method of moment applies as well to random vectors ( for distributions that are determined by their moments , see , e.g. , @xcite , theorem 6.2 ) . the joint moments of @xmath555 for some @xmath76 and positive integers @xmath556 can be analyzed similarly to the way we analyzed the moments of @xmath300 for some @xmath133 , and the limit joint distribution of these variables is probably determined by its moments . it is of great interest to study word maps for other groups or for non - uniform distributions on @xmath85 . for results of this nature see @xcite . many interesting questions and conjectures were raised in this paper . we collect them here . * let @xmath23 and @xmath13 be two words such that for any finite group @xmath0 , the distribution of the two word maps on @xmath0 are identical . is it true that @xmath557 ? * ( conjecture [ conj : beta = phi ] ) @xmath558 for every word @xmath13 . * ( a consequence of conjecture [ conj : beta = phi ] : ) for every word @xmath13 , and sufficiently large @xmath4 , a random permutation in the image of @xmath13 in @xmath85 has , on average , at least one fixed point . * friedman s conjecture : for every base graph @xmath0 , almost surely all new eigenvalues in lifts of @xmath0 are @xmath559 . * nica s theorem determines the behavior of the number of @xmath133-cycles in the @xmath85-image of any formal word @xmath13 . there are numerous other parameters of such permutations ( e.g. the number of cycles ) whose typical behavior is still not understood . p. cartier . fonctions harmonique sur un arbre . in _ symposia mathematica _ , vol . ix ( _ convegno di calcolo delle probabilita , indam , rome , 1971 _ ) , pages 203 - 270 . academic press , london , 1972 . mr0353467 ( 50:5950 )
we study here the spectra of random lifts of graphs . let @xmath0 be a finite connected graph , and let the infinite tree @xmath1 be its universal cover space . if @xmath2 and @xmath3 are the spectral radii of @xmath0 and @xmath1 respectively , then , as shown by friedman @xcite , in almost every @xmath4-lift @xmath5 of @xmath0 , all `` new '' eigenvalues of @xmath5 are @xmath6 . here we improve this bound to @xmath7 . it is conjectured in @xcite that the statement holds with the bound @xmath8 which , if true , is tight by @xcite . for @xmath0 a bouquet with @xmath9 loops , our arguments yield a simple proof that almost every @xmath10-regular graph has second eigenvalue @xmath11 . for the bouquet , friedman @xcite has famously proved the ( nearly ? ) optimal bound of @xmath12 . central to our work is a new analysis of formal words . let @xmath13 be a formal word in letters @xmath14 . the _ word map _ associated with @xmath13 maps the permutations @xmath15 to the permutation obtained by replacing for each @xmath16 , every occurrence of @xmath17 in @xmath13 by @xmath18 . we investigate the random variable @xmath19 that counts the fixed points in this permutation when the @xmath18 are selected uniformly at random . the analysis of the expectation @xmath20 suggests a categorization of formal words which considerably extends the dichotomy of primitive vs. imprimitive words . a major ingredient of a our work is a second categorization of formal words with the same property . we establish some results and make a few conjectures about the relation between the two categorizations . these conjectures suggest a possible approach to ( a slightly weaker version of ) friedman s conjecture . as an aside , we obtain a new conceptual and relatively simple proof of a theorem of a. nica @xcite , which determines , for every fixed @xmath13 , the limit distribution ( as @xmath21 ) of @xmath19 . a surprising aspect of this theorem is that the answer depends _ only _ on the largest integer @xmath10 so that @xmath22 for some word @xmath23 .
in recent years the investigation of two - dimensional frustrated heisenberg model is of great interest . this is mainly caused by studies of magnetic properties of superconducting cuprates . the so called @xmath1 model was studied by different methods in the case of completely antiferromagnetic interactions @xmath2 [ 1 - 8 ] . an existence of disordered phases at the @xmath3 and nontrivial ground states in this papers is generally assumed . the model with ferromagnetic interactions of nearest neighbors and antiferromagnetic interactions of next nearest neighbor spins has been investigated much less . the hamiltonian of this model has a form : @xmath4 where vectors @xmath5 and @xmath6 connects nearest sites and nearest sites along the diagonal line respectively , @xmath7 - a site number and @xmath8 . the model ( 1 ) is a special case of the @xmath1 model with @xmath9 and @xmath10 ( @xmath11 - is the exchange integral of the next nearest neighbors along x and y axes ) . this model is the simplest example of the 2d frustrated system . the ground state of the hamiltonian ( 1 ) is ferromagnetic at small @xmath12 and the energy @xmath13 . but this state becomes unstable at some critical point @xmath14 . in the classical approximation @xmath15 is equal to @xmath16 and the ground state at @xmath17 corresponds to two independent sublattices with neel order . however , quantum fluctuations can change this situation . this question will be discussed in this paper . to clarify the situation let us compare the 2d model ( 1 ) with its 1d version which has been studied in detail previously [ 9 - 11 ] . in the 1d model the transition from the ferromagnetic ground state to the spiral one takes place at @xmath18 . in a recent work by two of present authors [ 11 ] the character of this transition has been investigated by the perturbation theory in the small parameter @xmath19 and the classical state has been used as zeroth - order approximation . the ground state in all cases turned out to be either with @xmath20 or with @xmath21 , and the transition from the ferromagnetic state to the state with @xmath20 occurs by passing the states with intermediate spins . quantum fluctuations do not change the critical point @xmath15 , which coincides with its classical value . ( 11,7 ) this method has been applied also to the study of the 2d model with @xmath22 and @xmath23 in the vicinity of the phase boundary @xmath24 ( fig.1 ) , which determines the region of stability of the ferromagnetic state in the classical approximation . at @xmath25 the situation is similar to the 1d case [ 12 ] . however , at @xmath26 the second order of the perturbation theory in the small parameter describing a deviation from the phase boundary diverges . it proves that quantum fluctuations change the classical phase diagram itself . the energies of one- and two - magnon states are shown in fig.2 as functions of the parameter @xmath12 for model ( 1 ) ( @xmath27 and @xmath10 ) . the critical values @xmath28 and @xmath29 of the instability of the ferromagnetic state with respect to a creation of one and two magnons are equal to @xmath30 and @xmath31 respectively ( two - magnon state energy was found by the numerical solution of the corresponding schrdinger equation ) . the difference between @xmath28 and @xmath29 shows that the classical approach is inapplicable to the study of the stability of the ferromagnetic state . in this respect the situation is essentially different from those in the 1d or in the 2d cases at @xmath25 , when the critical point @xmath15 is independent on the number of magnons and coincides with its classical value . it is natural to assume , that the critical value @xmath32 for model ( 1 ) is less than @xmath29 , and the critical value @xmath33 corresponding to the state with @xmath20 is a true point of the phase transition . so , the classical approximation has proved to be unsuitable for the study of the phase transition in the model ( 1 ) and for the determination of the critical value @xmath15 and , therefore , another approach is needed . ( 11,7 ) one can obtain a crude estimate of the energy and the critical point @xmath33 by using a product of the ground state wave functions of two independent antiferromagnetic sublattices as a variational wave function ( vwf ) . this vwf gives for the critical point @xmath33 a value : @xmath34 where @xmath35 is the ground state energy per site of the 2d antiferromagnetic heisenberg model , and we use the most accurate numerical estimations of @xmath36 [ 13,14 ] . of course , this approach is too poor , and the obtained value @xmath33 is greater than @xmath29 . in the present work we propose a new type of vwf , which allows us to get more accurate estimations of @xmath15 and to study the character of the phase transition . this approach is based on a boson representation of the hamiltonian ( 1 ) which is different from the holstein - primakoff one . in contrast with the spin wave theory ( swt ) the proposed method is variational . this article is organized as follows . in the next section we demonstrate the method by the application it to the 2d heisenberg antiferromagnetic model . in section 3 the stability of the ferromagnetic phase of the frustrated model is studied and the data are discussed . to illustrate the main features of our approach , let us consider the 2d haf model : @xmath37 where @xmath7 is a site number of the 2d lattice and a vector @xmath5 connects nearest sites . it is convenient to rotate the local coordinate system of one of sublattices by the angle @xmath38 in xz plane : @xmath39 the transformation from spin - operators to bose - ones is defined by : @xmath40 where @xmath41 are bose - operators , @xmath42 , and the operator function @xmath43 is : @xmath44 it is evident that this transformation preserves all commutation relations for the spin operators . the states with different numbers of bosons on each site are effectively separated into equivalent unconnected pairs : @xmath45 as a result of the transformation ( 5 ) the hamiltonian ( 4 ) takes the form : @xmath46 this hamiltonian , as well as the original one , can not be solved exactly . as a trial wave function for hamiltonian ( 8) we choose : @xmath47 where the function @xmath48 will be found by the minimization of the total energy . the vacuum state in eq.(9 ) corresponds to the state of the hamiltonian ( 4 ) with all spins pointing down ( or corresponds to a chess arrangement of spins for the hamiltonian ( 3 ) ) . to calculate the ground state energy we need to calculate expectation values of all terms in the hamiltonian ( 8) with respect to vwf ( 9 ) . at first we calculate the terms corresponding to the spin interactions along the horizontal line ( x axe ) . owing to the translational symmetry of vwf ( 9 ) , all of these terms give equal contributions to the energy , and , therefore , it is sufficient to calculate terms corresponding to the interaction @xmath49 in the original hamiltonian ( 3 ) , where 1 and 2 are nearest neighbors along the horizontal line . at first we represent all factors @xmath50 in the hamiltonian ( 8) in the form : @xmath51 then the expectation value of the second term of the hamiltonian ( 8) , corresponding to the term @xmath52 in the hamiltonian ( 4 ) , takes the form : @xmath53 \label{11}\ ] ] thus , we need to calculate expectation values of the type : @xmath54 and @xmath55 where @xmath56 and @xmath57 can be : @xmath58 in order to decompose the boson quadratic form in eq.(9 ) we use the hubbard - stratanovich transformation . now , the vwf ( 9 ) takes the form : @xmath59 where @xmath60 is a number of lattice sites . now we can calculate the expectation values ( 12,13 ) : @xmath61 @xmath62 where we use the notations : @xmath63 @xmath64 using the identity : @xmath65 _ { p_3=0}\ ] ] eqs.(16,17 ) can be written as : @xmath66 @xmath67 _ { p_3=0 } , \label{21}\ ] ] where @xmath68 and @xmath69 it is convenient to use hubbard - stratanovich transformation for ( 23 ) : @xmath70 where @xmath71 @xmath72 and @xmath73 so , there are only linear terms on @xmath74 and @xmath75 in exponent of eq.(25 ) , and , therefore , we can compute the expectation value @xmath76 by diagonalizing @xmath77 . using fourier transformation for @xmath78 , we obtain : @xmath79 \right ) , \label{27}\ ] ] where @xmath80 are eigenvalues of the matrix @xmath81 , and @xmath82 . diagonalizing the quadratic form in eq.(27 ) , we find : @xmath83 where we use the notation : @xmath84 substituting ( 28 ) into eq.(24 ) , we arrive to : @xmath85 where matrix @xmath86 has the form : @xmath87 substituting eqs.(30,31 ) into ( 21 ) , we find : @xmath88 where @xmath89 ^{-\frac 12 } \label{33}\ ] ] there are four terms in ( 11 ) , corresponding to the different values @xmath56 , @xmath90 and @xmath91 , @xmath92 in ( 14 ) and ( 19 ) . @xmath93 using the notation ( 34 ) , eq.(32 ) and the definition of functions @xmath94 and @xmath95 in ( 33 ) , eq.(11 ) can be written as : @xmath96 @xmath97 @xmath98 we note , that the expectation value @xmath99 can be obtained from eq.(30 ) by setting @xmath100 and @xmath101 . hence , we obtain : @xmath102 the contribution to the energy of the terms corresponding to nearest neighbors along the horizontal line in the hamiltonian ( 8) is : @xmath103 one can verify by rotating the local coordinate systems and repeating the calculations ( 5 - 30 ) for the transformed hamiltonian , that the values @xmath104 and @xmath105 correspond to expectation values : @xmath106 the calculation of the expectation values of the terms of the hamiltonian ( 8) corresponding to the nearest neighbor interactions along the vertical line ( y axe ) @xmath107 can be carried out in the analogy with ( 9 - 30 ) . in this case , we must only substitute the functions @xmath108 for @xmath109 into eqs.(36 - 38 ) . then we obtain the contribution to energy of the vertical neighbor terms : @xmath110 the total energy of hamiltonian ( 8) can be written as : @xmath111 thus , we find energy as a function of parameters @xmath112 . these parameters are functions of @xmath113 [ or of @xmath114 in vwf ( 9 ) ] . now we need to minimize the energy with respect to @xmath115 : @xmath116 where @xmath117 we can define the functional form of @xmath115 from eq.(43 ) : @xmath118 where @xmath119 and @xmath120 @xmath121 are variational parameters . thus , we reduce the problem of the energy minimization over the variational function @xmath115 to the energy minimization with respect to five variational parameters @xmath120 . this procedure has been performed numerically . it gives the final result for 2d haf : @xmath122 besides , both terms in eq.(42 ) give equal contributions to the energy : @xmath123 obtained ground state energy for the 2d haf ( 3 ) is 4 - 4,5% higher than the most accurate results obtained by various numerical techniques [ 13,14 ] and is in a good agreement with energies obtained by different vwf methods [ 15 - 17 ] . now we apply the proposed approach to the frustrated model ( 1 ) . the ground state of the hamiltonian ( 1 ) is ferromagnetic at small @xmath12 . in the classical approximation two - sublattice neel state is realized at @xmath124 . the energy of this state is : @xmath125 a spiral state ( the incommensurate phase with the momentum @xmath126 , @xmath127 ) has higher energy at @xmath124 : @xmath128 although it tends to zero at @xmath129 as well . energies of other phases are considerably higher . the energy of the dimer phase , for example , equals to @xmath130 at @xmath131 . hence , it is natural to consider neel two - sublattice and the spiral states as main candidates for the ground state at @xmath17 . we will calculate quantum corrections to the classical energy for these states using the vwf ( 9 ) . before that , we make the following remark . the vwf ( 9 ) has no rotational symmetry ( it is not the eigenfunction of @xmath132 ) , and , therefore , the ground state energy calculated in this approximation depends on the choice of the local coordinate system for the hamiltonian ( 1 ) . in other words , if we rotate spin operators @xmath133 and perform a transformation to bose - operators according to eq.(5 ) , then , generally speaking , obtained energies will be different in spite of the equivalence of the hamiltonians @xmath134 and @xmath135 . hence , parameters of the rotation can be considered as variational ones . for the consideration of the two - sublattice neel phase let us rotate the coordinate system so that there are the angle @xmath38 in @xmath136 plane between two nearest neighbors on each sublattice ( as in the 2d haf case ) and the angle @xmath137 between sublattices in @xmath136 plane . we will keep the angle @xmath138 as a variational parameter ( we note , that the energy is infinitely degenerated with respect to @xmath139 in the classical approximation ) . in this case , the hamiltonian ( 1 ) takes the form : @xmath140 \nonumber \\ & + & j \sum_{\bf n , d}\left ( -\varsigma _ { { \bf n}}^x\varsigma _ { { \bf % n}+{\bf d}}^x+\varsigma _ { { \bf n}}^y\varsigma _ { { \bf n}+{\bf d}}^y-\varsigma _ { { \bf n}}^z\varsigma _ { { \bf n}+{\bf d}}^z-\frac 14\right ) , \label{49}\end{aligned}\ ] ] where vector @xmath6 connects nearest sites on diagonal line . expectation values of the terms in the hamiltonian ( 49 ) , such as @xmath141 , @xmath142 and @xmath143 are determined by eqs.(36,37,38 ) respectively , where @xmath144 and @xmath145 are defined by eq.(29 ) . the vwf ( 9 ) contains only terms that involve even numbers of boson operators and , therefore , all expectation values such as @xmath146 in ( 49 ) are equal to zero : @xmath147 in this case there are nine variational parameters in eq.(45 ) : @xmath148 ( 11,8 ) the critical value @xmath15 is defined by the condition that the ground state energy is negative for @xmath17 . to find @xmath15 we minimize the following ratio : @xmath149 where @xmath150 and @xmath151 are the contributions to the energy from the first and the second terms in the hamiltonian ( 49 ) : @xmath152 the dependence of @xmath15 on @xmath137 obtained by a minimization with respect to parameters @xmath120 is shown on fig.3 . as it can be seen from fig.3 , the minima of @xmath15 are reached at @xmath153 or @xmath154 . the corresponding states belong to the so called collinear phase . at @xmath155 we have : @xmath156 the calculation of the energy with use of the vwf ( 9 ) for the spiral phase can be carried out in a similar way . it turns out that quantum fluctuations do not shift the transition point and change the coefficient in the quadratic dependence in ( 48 ) only . the dependences of the energies of the collinear and spiral states are shown on fig.4 . for comparison , we show the spin wave theory results as well . the swt energy is not a variational one and is defined up to @xmath157 , where the sublattice magnetization vanishes in the swt approximation . ( 11,8 ) thus , the singlet collinear phase is the ground state of the model ( 1 ) for @xmath158 . a characteristic feature of the considered model ( 1 ) is the fact , that the critical value @xmath15 , ( the point of the instability of the ferromagnetic state ) , depends on @xmath159 and @xmath160 . in this respect this model differs from the 1d case where @xmath15 does not depend on @xmath159 . such 1d behaviour takes place also in a more general 2d @xmath161 model in the vicinity of the classical phase boundary and at @xmath25 . in this case , quantum effects do not change the classical boundary of the instability of the ferromagnetic state . the region near the boundary can be considered in the framework of the perturbation theory , and the transition from the ferromagnetic state to the spiral one occurs in this case . the character of the transition changes at @xmath162 , and the critical parameter @xmath15 , corresponding to @xmath163 , depends on @xmath159 . in this case the transition from the ferromagnetic to the collinear state is realized . the boundary of the stability of the ferromagnetic phase has a form shown on fig.1 . in conclusion , we have studied the transition from the ferromagnetic to the singlet state in the 2d frustrated spin model . the transition region is characterized by strong quantum fluctuations and can not be described by the classical approximation . to study the behaviour of the system close to the ferromagnetic boundary we have proposed new approach based on the bozonization of the spin operators . this approach is different from the holstein - primakoff method and is variational . we believe that the proposed method can be used to the study of other heisenberg models with the frustration . the authors are grateful to prof . m.ya.ovchinnikova for stimulating discussions . this work is supported by the istc under grant no.015 and in part by rffr under grant no.96 - 03 - 32186 .
the stability of the ferromagnetic phase of the 2d quantum spin-@xmath0 model with nearest - neighbor ferro- and next - nearest neighbor antiferromagnetic interactions is studied . it turns out that values of exchange integrals at which the ferromagnetic state becomes unstable with respect to a creation of one and two magnon are different . this difference shows that the classical approximation is inapplicable to the study of the transition from the ferromagnetic to the singlet state in contrast with 1d case . this problem is investigated using a variational function of new type . it is based on the boson representation of spin operators which is different from the holstein - primakoff approximation . this allows us to obtain the accurate estimate of the transition point and to study the character of the phase transition .
since the famous work of berry @xcite , the geometric phase has been widely investigated and its generalizations have been made in many ways @xcite@xcite@xcite@xcite . recently much concern has been concentrated on the geometric phase of entangled states as well as mixed states @xcite@xcite . coherent state is an important physical concept both theoretically and experimentally @xcite@xcite . it can be generated from an arbitrary reference state , and in this brief report landau levels are chosen to be such reference states . though coherent states of landau level have been studied in refs . @xcite@xcite and berry s phase for coherent states as well as squeezed coherent states of a one - dimensional harmonic oscillator has been illustrated in ref . @xcite , berry s phase for coherent states of landau levels which is highly degenerate and with an additional parameter , i.e. , the magnetic field @xmath0 , is still worthy of further investigation . this paper is organized as follows . in sec . ii , we show how to get the coherent states of landau levels , and these states can be regarded as a result of a magnetic flux moved adiabaticly from infinity to a finite place on the plane . in sec . iii , we calculate the abelian and non - abelian berry s phase for coherent states of landau levels . the abelian berry s phase is just like an alternative version of the aharonov - bohm ( ab ) effect ; the difference between them is that in our case the cyclic motion of the magnetic flux results in the phase shift . in sec . iv , we provide the explicit form of the hamiltonian and berry s connections for squeezed coherent states of landau levels . conclusion and discussion are made in the last section . the motion of a free electron in a two - dimensional @xmath1-plane in a static magnetic field along the @xmath2-direction is described by the following hamiltonian @xmath3,\ ] ] where @xmath4 is the mass of the electron , @xmath5 is the planck constant , @xmath6 is the electron charge , @xmath7 is the speed of light in the vacuum , @xmath8 and @xmath9 are the linear momentums , @xmath10 and @xmath11 are vector potentials of the magnetic field satisfying @xmath12 . for simplicity the zeeman s term is not included . we introduce the following operators @xmath13 from which one can form a pair of operators @xmath14 and @xmath15 which satisfy the commutation relation @xmath16=1 $ ] : @xmath17 in this case we can rewrite the hamiltonian in a more simpler form @xmath18 where @xmath19 , and @xmath14 , @xmath15 are raising and lowering operators between landau levels with @xmath20 , @xmath21 . the energy of landau levels is @xmath22 . one knows that landau levels are highly degenerate . in the degenerate space of landau levels we can introduce another pair of raising and lowering operators @xmath23 we choose the symmetric gauge @xmath24 and @xmath25 , and then @xmath26 , @xmath27 are commutative with @xmath15 , @xmath14 and @xmath28=1 $ ] . therefore the hamiltonian commutes with @xmath26 and @xmath27 . the ground state is defined as @xmath29 $ ] , and all other eigenstates of this system can be generated from @xmath30 state with raising operators @xmath31 . the states @xmath32 are also orthogonal and normalized bases for this system . states with the same @xmath33 are in the same energy level , and states with the same @xmath33 but different @xmath34 stand for the different degenerate states on the same energy level . the coherent states of landau levels are generated in the following way as in @xcite : @xmath35 where @xmath36 . the hamiltonian for the coherent states is : @xmath37,\ ] ] where @xmath38 , and the eigenstates of this hamiltonian are always combinations of the degenerate states with the same energy : @xmath39 where @xmath40 are arbitrary complex numbers which make @xmath41 normalized . we put eq . ( [ b ] ) back to eq . ( [ csh ] ) , this would make the hamiltonian easier to understand , we get @xmath42.\ ] ] we found that the magnetic vector potential is added by a constant vector potential . this can be regarded as a result of a magnetic flux perpendicular to the plane moving adiabaticly from infinitely far to a finitely far position on the plane . in the following we would like to show how we get the result . we can assume that the added magnetic flux is a gaussian form magnetic field centered @xmath43 , @xmath44 $ ] , where @xmath45 is refered to as the spread or standard deviation for the gaussian function . and we may choose the symmetric gauge with respect to @xmath46 , i.e. , @xmath47 , the nonsingular vector potential for this added magnetic field is @xmath48 - 1\}(y - y_0)}{2\pi[(x - x_0)^2+(y - y_0)^2]},\nonumber \\ a'_y&=&-\frac{\phi_0 \{\exp[-\frac{(x - x_0)^2+(y - y_0)^2}{\delta^2}]-1\}(x - x_0)}{2\pi[(x - x_0)^2+(y - y_0)^2]}.\end{aligned}\ ] ] one may observe that @xmath49 has nothing to do with the hamiltonian ( [ h ] ) when @xmath50 . now we assume that the electron in the plane is in a certain eigenstate , for example , @xmath51 , in this case the electron is localized near the origin because of @xmath52 , @xmath53 . let @xmath54 , so when the flux moves adiabaticly to a place @xmath46 which is finitely far from the electron ( i.e. @xmath55 ) , we assume that the electron is still distributed around the origin and @xmath56 near the origin of the plane can be regarded as constants . then the hamiltonian for the electron will be of the form ( [ cshb ] ) with @xmath57 } { 2\pi(x_0 ^ 2+y_0 ^ 2)},\nonumber \\ x_2=\sqrt{\frac{e}{2\hbar b c}}\frac{\phi_0 x_0 [ \exp\left(-\frac{x_0 ^ 2+y_0 ^ 2}{\delta^2}\right)-1]}{2\pi(x_0 ^ 2+y_0 ^ 2)}.\end{aligned}\ ] ] this modification of the hamiltonian also corresponds to the following transformation @xmath58 , where @xmath59}{\pi b(x_0 ^ 2+y_0 ^ 2 ) } , \nonumber \\ \delta y=\frac{\phi_0 y_0 [ \exp(-\frac{x_0 ^ 2+y_0 ^ 2}{\delta^2 } ) -1]}{\pi b(x_0 ^ 2+y_0 ^ 2)}.\end{aligned}\ ] ] so the state @xmath60 will become @xmath61 . since the distribution of the electron is near the origin , one also makes sure that @xmath62 . the coherent states of landau levels is nothing but the shifted eigenstates of landau levels in the phase space , here we assume such a shift happens in real space @xmath63 . when the conditions above are satisfied , this assumption is reasonable . it is @xmath49 who causes this small shift . we may see from eq . ( [ dx ] ) that the direction of @xmath49 , i.e. , the sign of @xmath64 is related to the direction of the shift . when @xmath65 the shift is parallel to the direction of the flux , and vice versa . interestingly , if the flux circles the electron once , the electron will also circle the origin in a much smaller loop once . this is the reason why the berry s phase emerged as we will show in the next section . we know that the landau levels are highly degenerate , and so is the coherent states of landau levels . berry s phase for degenerate states was presented in @xcite and may have a non - abelian nature . we calculated the berry s connections as follows : @xmath66 in the degenerate space , i.e. , between states with the same @xmath33 , the berry s connections become @xmath67 we found that @xmath68 and @xmath69 are abelian . with the adiabatic theorem for degenerate states proved in @xcite and to a higher order in @xcite , we know that the @xmath70 in eq . ( [ dcs ] ) will not change during the arbitrary slow evolution of @xmath71 and @xmath72 . also because the berry s connections of @xmath71 and @xmath72 are abelian , @xmath40 s will gain a berry s phase factor , which is the same of all @xmath34 , after an adiabatic evolution in the @xmath71-@xmath72 plane . the berry s phase is @xmath73 where @xmath74 is the path of the adiabatic evolution of @xmath75 in @xmath71-@xmath72 plane and @xmath76 is the area of @xmath74 . this result appeared in @xcite for non - degenerate coherent states . the result is also the same as the phase in the paper @xcite . however , in our case it is the moving magnetic flux that moves the electron instead of a moving potential well . with the interpretation in the above section , we can see from eq . ( [ x1 ] ) that when the magnetic flux circles the electron for one loop , the @xmath75 will also enclose an area , and this gives the berry s phase . for example , we let @xmath46 moves around the origin in a circle with the radius @xmath77 for one loop . the berry s phase will be @xmath78 this can be viewed as an alternative version of the ab effect . the different between them is that we move the magnetic flux instead of the electron . one may see from eq . ( [ me3 ] ) that @xmath79 is non - abelian , so the change of @xmath0 will give non - abelian berry s phase . as the non - abelian berry s phase in such a system has not been shown in the literature before , in the following we would like to give a simple examples to illustrate it . we assume @xmath80 during the evolution and the other two parameters undergo the loop in fig . [ loopfig ] . we can get the eigenvalues of matrix @xmath81 as @xmath82 and the corresponding eigenstates @xmath83 . the states before evolution eq . ( [ dcs ] ) can be rewritten in the new base as @xmath84 after the system undergoes an evolution as shown in fig . [ loopfig ] , the state may become @xmath85 where @xmath86 is the area enclosed by @xmath72 and @xmath87 as in fig . [ loopfig ] . however , if @xmath88 , the calculation will involve path ordered integral and beome very complicated . the hamiltonian for squeezed coherent state is @xmath89 where @xmath90 is defined in eq . ( [ h0 ] ) , @xmath91 and @xmath92 the eigenstates for this hamiltonian , i.e. , the squeezed coherent states are @xmath93 in the same way , we put eq . ( [ b ] ) to eq . ( [ scsh ] ) , and we can get @xmath94 ^ 2 \nonumber \right . \\ & & \left . + e^{2r } \left [ -\sin(\theta/2)\pi'_x + \cos(\theta/2 ) \pi'_y \right]^2 \right\},\end{aligned}\ ] ] where @xmath95 . for @xmath96 , ( [ scshb ] ) reduces to eq . ( [ cshb ] ) . for @xmath97 , one can see from this hamiltonian that the squeezing operation @xmath98 caused an anisotropy in the plane . more clearly if we set @xmath99 eq . ( [ scshb ] ) will become @xmath100 , in other words , the kinetic energies @xmath101 , @xmath102 are squeezed by the factors @xmath103 and @xmath104 respectively . now we consider the berry s connections of squeezed coherent states . @xmath105\delta_{m , m ' } , \nonumber\\ \left < n(\alpha,\beta),m \right| \frac { \partial}{\partial \theta } \left| n'(\alpha,\beta),m ' \right > & = & \frac{i \sinh(2r)}{4 } \left [ e^{-i \theta } \sqrt{n'(n'-1)}\delta_{n , n'-2 } + e^{i \theta } \sqrt{(n'+1)(n'+2 ) } \delta_{n , n'+2}\right]\delta_{m , m'}\nonumber \\&+ & \frac{i \sinh^2 r}{2 } ( 2n'+1)\delta_{n , n'}\delta_{m , m'},\nonumber\\ \label{pb}\left < n(\alpha,\beta),m \right| \frac{\partial } { \partial b } \left| n'(\alpha,\beta),m ' \right > & = & \frac{1}{2b } \left [ 2 \sinh^2 \frac{r}{2 } \sqrt{n'm ' } \delta_{n , n'-1}\delta_{m , m'-1 } -2\sinh^2\frac{r}{2 } \sqrt{(n'+1)(m'+1 ) } \delta_{m , m'+1}\delta_{n , n'+1 } \right.\nonumber \\ & + & \left.e^{i \theta}\sinh r \sqrt{m'(n'+1 ) } \delta_{n , n'+1 } \delta_{m , m'-1 } - e^{-i \theta } \sinh r \sqrt{(m'+1 ) n ' } \delta_{m , m'+1 } \delta_{n , n'-1 } \right]\nonumber \\ & + & \frac{1}{2b } ( \alpha \sqrt{m ' } \delta_{m , m'-1}- \alpha^ * \sqrt{m'+1}\delta_{m , m'+1})\delta_{n , n'}.\end{aligned}\ ] ] to our knowledge , the berry s connection with respect to @xmath0 has not been appeared in the literature before . with these , the berry s phase is not hard to obtain . if anisotropy exist in 2-dimensional electron gas systems , its hamiltonian would be of the form of eq . ( [ scshb ] ) . in this brief report , we have calculated the berry s phase for coherent states as well as squeezed coherent states of landau levels . the hamiltonian of the coherent states of landau levels is interpreted as a result of a magnetic flux perpendicular to the plane , and it is moved adiabaticly from infinity to a distance away from the electron so that some approximations are satisfied . the cyclic adiabtic motion of this magnetic flux caused the berry s phase of coherent states of landau levels , this is an analogue of the ab effect . and the non - abelian phase is also of interest , the magnetic field @xmath0 appears in the berry s phase in the form of @xmath87 . so when @xmath106 the berry s phase will be very sensitive to @xmath0 and becomes indefinite , and the reversion of the magnet field is prohibited if we want to get this phase . @xcite also states this phenomenon that near the level crossing point the berry s phase sometimes vanishes .
the berry s phase for coherent states and squeezed coherent states of landau levels are calculated . coherent states of landau levels are interpreted as a result of a magnetic flux moved adiabaticly from infinity to a finite place on the plane . the abelian berry s phase for coherent states of landau levels is an analogue of the aharonov - bohm effect . moreover , the non - abelian berry s phase is calculated for the adiabatic evolution of the magnetic field @xmath0 .
a nuclear recoil moving at very low energies can be used to detect rare events such as neutrino coherent interactions or non - baryonic dark matter signatures . the direct detection of these non - baryonic particules is based on the detection of nuclear recoils coming from elastic colisions on different targets . ionization is one of the most important channel to detect such nuclear recoils . the ionization quenching factor is defined as the fraction of the kinetic energy released through ionization by a recoil in a medium . in the last decades , an important effort has been made to measure the iqf in different materials : gases @xcite , solids @xcite and liquids @xcite , using different techniques . however , in the low energy range , the measurements are rare or absent for many targets due to ionization threshold of detectors and experiment contraints . + we have developped an experimental setup devoted to the measurement of low energy ( a few kev ) ionization quenching factor for the mimac project @xcite . the measurements reported in this work have been done with a micromegas detector filled with a gas mixture of 95% of @xmath0he and 5% of isobutane ( c@xmath1h@xmath2 ) at different pressures . the micromegas detector was calibrated using x - rays produced by fluorescence on different targets . to produce nucleus moving with a controlled energy in the detection volume , we have developped calibrated electron cyclotron resonance ion source ( ecris ) @xcite which produced ions with energy from a fraction of one kev up to 50 kev . the ecris source was coupled to a micromegas detector @xcite . the micromegas used in this work was a standard bulk micromegas @xcite , in which the mesh and the anode are built and integrated with a fixed 128 @xmath3 gap . the gap between the copper anode and stainless steel mesh is maintained by cylindrical pilars every 3 mm . this gap is well adapted for working gas pressure between 350 and 1300 mbar . the drift distance between the cathode and the mesh was 3 cm , large enough to include the tracks of nuclei of energies up to 50 kev at 350 mbar . the active area was @xmath4 . we used an individual power supply for each electrode ( mesh and anode ) with cathode drift to the ground , allowing to independently vary the different fields . the nim high voltage module ( iseg 223 m ) presents a high stability and high precision of the output voltage setting and measurement ( about 10 mv ) . the setting parameters , the voltage and the current controle are performed by remote control via a labview interface . typical applied field were @xmath5 for the drift and a voltage of 450 v for the avalanche . with these values , an optimum collection efficiency is achieved and the electron drift velocity is comparatively low . the electronic system is composed of a dedicated charge pre - amplifier and of a commercial spectrocopy amplifier . due to the low drift velocity of electron in such a low drift field ( @xmath5 ) , the integration time constant of this pre - amplifier should be large enough to assure a total colection charge and an output proportionnal to the total energy of the event . the calibration constant for the preamplifier is measured by applying test pulse on a test capacity . a pci - bus multichannel analyzers with internal 8k adc is used to convert incoming signals . for each measurement , the gas vessel is previously pumped using a turbomolecular pump and a dry primary vacuum pump . then the detector is filled with gas mixture at the working pressure and operates with a circulating and controlled flux . to correct this low gain variation , a calibration is performed just before each energy measurement with the ion source . he + 5% c@xmath1h@xmath2 micromegas to aluminium ( 1.486 kev ) and iron ( 6.4 kev ) fluorescence x - rays at 1000 mbar , with a resolution ( fwhm ) of 29,4 % and 15,3 % respectively . ] due to the high transparency of @xmath0he gas for x - rays energy above 2 kev , the detector shows a very low efficiency to x - ray from a @xmath6fe source used for calibration purpose , especially at low pressure . so , to avoid using high intensity radioactive source and to calibrate in a realistic time , we have developped a high intensity multi energies x - ray system based on the production of fluoresence photons on different high purity targets . the system used a pulsed miniature x - rays generator inside the gas vessel with maximum flux equivalent to a 2 mci radioactive source and the following targets with the corresponding x - ray energies : aluminium ( 1.486 kev ) , titanium ( 4.504 kev ) , iron ( 6.4 kev ) , copper ( 8.1 kev ) . figure [ fg.calibration ] shows a typical spectrum for aluminium ( 1.486 kev ) and iron ( 6.4 kev ) targets with a resolution ( fwhm ) of 29,4 % and 15,3 % respectively . the energy threshold is close to 300 ev . the ecri ion source is a plasma device designed and built by the lpsc ssi team to provide charged ions at low velocities from a fraction of a kev to 50 kev . one of the important features of this facility is the availability of very low and stable currents of a few picoamp . in the present configuration , the beam is adapted to the 45 degree spectrometer by an einzel lens which has been tuned with the faraday cup . a micrometer aperture ( @xmath7 in diameter ) selects a part of this beam which is injected in the micromegas detector and produced about 25 ions per second . according to the gas injected is the plasma chamber , the source is able to produce many type of beams : proton , @xmath8he , @xmath0he , @xmath9f . + to calibrate the ion source , a 50 nm thick silicon nitride membrane ( si@xmath10n@xmath1 ) was used as interface between the ion source and the gas chamber . a time of flight measurement has been performed under vacuum , using two channeltrons . one of them detecting the low energy electrons extracted from the si@xmath10n@xmath1 menbrane by the ions and the other one detecting the ions for 6 different known positions . this setup allows to measure the energies of the ions just after the foil . with this configuration a first energy measurement with micromegas was performed . in such a way , we could verify that the energy measured by tof were the same as those indicated by the extraction potentiel values in kv for the 1 + charge state ions with the @xmath7 diameter aperture interface . he ions in a @xmath0he + 5% c@xmath1h@xmath2 micromegas detector at 700 mbar . ] using the photoelectron produced by fluorescence x - rays as calibration , we can measure ionization energies produced by the recoils . the ionization quenching factor of a recoil will be the ratio between this energy and its kinetic energy . the measurement reported @xcite have been focused on the low energy @xmath0he iqf @xcite . + at the same time , it was possible to extract the micromegas energy resolution for a wide range of incoming ion energies . ionization energies have measured for ions with kinetic energy down to 1 kev . figure [ fg.resol ] presents the evolution of the resolution ( fwhm ) at 700 mbar for measured ionization energies of @xmath0he ions from 300 ev to 35 kev . we can note that the resolution is slightly better for x - rays . at low energy , down to 1 kev , the energy resolution is about 50% , but this does not affect the number of expected events for dark matter search @xcite . + using the possibility to produce recoils of different states of charge ( 0,+1,+2 for @xmath0he ) with the same kinetic energy , it is also possible to confirm that the ionization energy does not depend of the state of charge of the recoil . this was clearly demonstrated with @xmath0he recoil with 50 kev of kinetic energy . measurement with neutral recoils have been obtained with the si@xmath10n@xmath1 window because ions are neutralized when passing through the membrane . moreover , this experimental setup allow to measure the quenching factor for other different recoils wich could be used for dark matter search such as : @xmath8he in @xmath8he or @xmath9f in cf@xmath1 . + concerning @xmath8he , which is a rather expensive gas , a clean close loop gas system is under construction to allow gas recovery and storage after the measurements . in summary , we have demonstrated the possibility to measure ionization energy for @xmath0he recoils down to 1 kev with a compatible energy resolution for dark matter search . for the next step , this measurement will be extended to low gas pressure and to different recoil such as @xmath8he of @xmath9f . this measurement is particularly important to better understand the ionization response of gas detectors .
the measurement of the ionization produced by particles in a medium presents a great interest in several fields from metrology to particule physics and cosmology . the ionization quenching factor is defined as the fraction of energy released by ionisation by a recoil in a medium compared with its kinetic energy . at low energy , in the range of a few kev , the ionization falls rapidly and systematic measurement are needed . we have developped an experimental setup devoted to the measurement of low energy ( kev ) ionization quenching factor for the mimac project . the ionization produced in the gas has been measured with a micromegas detector filled with helium gas mixture . # 1#2#3#4#1 * # 2 * , # 3 ( # 4 ) # 1#2#3nucl . instr . and meth . * a#1 * ( # 2 ) # 3 # 1#2#3nucl . instr . and meth . * # 1 * ( # 2 ) # 3 # 1#2#3phys . rev . * a#1 * ( # 2 ) # 3 # 1#2#3phys . rev . * b#1 * ( # 2 ) # 3 # 1#2#3phys . rev . * c#1 * ( # 2 ) # 3 # 1#2#3phys . rev . * d#1 * ( # 2 ) # 3 # 1#2#3jhep * # 1 * ( # 2 ) # 3 # 1#2#3phys . rev . lett . * # 1 * ( # 2 ) # 3 # 1#2#3phys . lett . * b#1 * ( # 2 ) # 3 # 1#2#3astropart . phys . * b#1 * ( # 2 ) # 3 # 1#2#3astrophys . j. * # 1 * ( # 2 ) # 3 # 1#2#3astrophys . j. suppl . * # 1 * ( # 2 ) # 3 #1#2#3astron . & astrophys . * # 1 * ( # 2 ) # 3 # 1#2#3jcap * # 1 * ( # 2 ) # 3
the present calculation with @xmath3 mev and @xmath26 and @xmath27 volumes extends earlier domain wall fermion results with @xmath28 mev and @xmath29 , @xmath30 and @xmath26 volumes @xcite . we use the same combination of the iwasaki gauge action and dislocation suppressing determinant ratio ( dsdr ) exploited to reduce residual chiral symmetry breaking in these earlier studies . however , to enable calculations at @xmath31 mev with available computing resources we have changed the shamir domain wall formulation to mbius @xcite . by choosing the mbius parameters @xmath32 and @xmath33 of ref . @xcite so that @xmath34 , we insure that our mbius green s functions will agree at the 0.1% level with those of shamir evaluated at a much larger @xmath35 . thus , our @xmath28 and 135 mev calculations are equivalent , including all lattice artifacts , except for the intended reduction in quark mass . .a summary of the @xmath3 mev ensembles . the units are mev for the temperature @xmath8 and @xmath36 for the masses @xmath37 , @xmath38 and @xmath15 . @xmath39 , @xmath40 and @xmath41 label the number of independent streams , the total equilibrated time units and the number of sites in each spatial direction , respectively . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] our most dramatic result is the temperature - dependent , disconnected chiral susceptibility @xmath2 , plotted in fig . [ fig : chi_disc ] . three of the four lower curves show earlier results with @xmath28 mev on @xmath42 , @xmath43 and @xmath44 volumes . a significant decrease in @xmath2 is seen for temperatures below 165 mev as the volume is increased above @xmath42 , a volume dependence anticipated in earlier scaling @xcite and model @xcite studies . the two higher curves show a large increase in @xmath2 in the entire transition region for @xmath3 mev and both @xmath44 and @xmath45 volumes . the ratio of peak heights for the @xmath3 and 200 mev , @xmath44 data is 2.1(0.2 ) , which is consistent with the ratio 1.86 predicted by universal @xmath10 scaling @xmath46 , only if the regular , mass - independent part of @xmath2 is small . this comparison of @xmath2 with universal @xmath10 scaling neglects the connected part of the chiral susceptibility . we find that the connected chiral susceptibility has a mild dependence on both the temperature and quark mass ( as is expected if the @xmath47 screening mass remains non - zero at @xmath7 ) and so does not contribute to the singular part of the chiral susceptibility . also shown in this figure are hisq results for @xmath48 and a goldstone pion mass of 161 mev @xcite . if scaled to @xmath3 mev assuming this same @xmath49 behavior , the hisq value for @xmath2 is 50% smaller than that seen here . this discrepancy reaffirms the importance of an independent study of the order of the transition and calculation of @xmath7 using chiral quarks . and 200 mev . the @xmath3 mev data shows a near @xmath50 increase over that for @xmath28 mev . hisq results for @xmath51 mev @xcite are also plotted . ] the peak shown in fig . [ fig : chi_disc ] implies a pseudo - critical temperature of 155(1)(8 ) mev . the central value and statistical error are obtained by fitting the @xmath52 , 154 and 159 mev values of @xmath53 to a parabola . the second , systematic error reflects the expected 5% discretization error . we do not include a systematic error caused by our finite volume . while typically neglected when @xmath54 , we lack the data needed for an empirical estimate . this result for @xmath7 is consistent with the continuum limit for this quantity obtained using staggered fermions @xcite . for four streams with @xmath55 mev . streams beginning with an ordered or disordered configuration are labeled `` ord '' and `` dis '' . each point is the average of measurements made with 10 random sources on each of 20 configurations , separated by one time unit . ] the order of the qcd phase transition can now be studied using the time - history of the chiral condensate for @xmath56 . figure [ fig : pbp_evol ] shows four time histories of @xmath57 at @xmath55 mev . all four streams fluctuate over the same range of values , showing no metastable behavior and no difference between those streams starting from ordered versus disorder configurations . this and the failure of @xmath58 to grow as @xmath59 when the volume is increased from @xmath44 to @xmath45 provide strong evidence that for @xmath31 mev , the qcd phase transition is not first - order but a cross - over , a conclusion consistent with previous staggered work @xcite . symmetry of qcd and our chiral fermion formulation . below @xmath7 this symmetry is spontaneously broken . for @xmath60 mev we see accurate chiral symmetry . ] -dependence of the anomalous @xmath6-breaking difference @xmath61 , which remains non - zero and becomes mass independent for @xmath62 mev . ] in fig . [ fig : sua ] we show the @xmath5-breaking differences between the susceptibilities @xmath63 and @xmath64 and between @xmath65 and @xmath66 . each pair of fields , @xmath67 and @xmath68 forms a 4-dimensional representation of @xmath5 symmetry . these @xmath5-breaking differences are large below @xmath7 but have become zero for @xmath60 mev . in fig . [ fig : ua ] we show the difference @xmath61 . this pair of quantities is related by the anomalous @xmath6 transformation , a symmetry of the classical field theory that is broken by the axial anomaly . figure [ fig : ua ] shows that this symmetry is not restored until at least @xmath69 mev . also shown in this figure is the result from our earlier @xmath28 mev calculation @xcite . the expected increase in @xmath61 with decreasing pion mass is seen for @xmath70 . however , above @xmath71 mev this difference is still non - zero and has become mass independent , confirming our previous conclusion that this non - zero value is a result of the axial anomaly , not the small quark mass . we have presented results from the first study of the qcd phase transition using chirally symmetric lattice fermions , physical quark masses and therefore three degenerate pions with @xmath72 mev . we find @xmath73 mev , similar to previous staggered fermion results , and see cross - over behavior , consistent with a second order critical point at zero quark mass . we show that anomalous symmetry breaking extends to temperatures approximately 30 mev above @xmath7 . finally , we see a factor of two increase in the disconnected chiral susceptibility , @xmath2 near @xmath7 as @xmath11 decreases from 200 to 135 mev , similar to the expectations for critical @xmath10 scaling , provided the regular part of @xmath2 is small . however , in this region we find @xmath2 50% larger than that suggested by staggered fermion results . these results may close a chapter in the study of the qcd phase transition . the cross - over character and pseudo - critical temperature of the transition have now been obtained using a formulation which respects the symmetries of qcd , uses physical values for both the strange and light quark masses and is performed for values of the inverse lattice spacing @xmath74 gev where 5% discretization errors are to be expected . this is a challenging calculation with 5-dimensional lattice volumes as large as @xmath75 and a physically light quark mass . this study was made possible by the use of the dsdr action @xcite , mbius fermions @xcite , highly efficient code @xcite and substantial resources provided by the lawrence livermore national laboratory . of course , it remains important to continue to explore these questions at larger spatial volume and smaller lattice spacing when adequate resources become available . computing support for this work on the llnl vulcan bg / q supercomputer came from the lawrence livermore national laboratory ( llnl ) institutional computing grand challenge program . we thank llnl for funding from ldrd13-erd-023 . this work has been supported by the u. s. department of energy under contract de - ac52 - 07na27344 ( llnl ) . nhc , zl , rdm , gm , dm and hy were supported in part by doe grant de - fg02 - 92er40699 . tb and rg were supported in part by doe grant de - ka-1401020 . h - td , fk , sm and pp were supported in part by doe grant de - ac02 - 98ch10886 . mib was supported in part by doe grant de - fg02 - 00er41132 this work and the development of software used in this work has been supported in part through the scientific discovery through advanced computing ( scidac ) program funded by the u.s . department of energy , office of science , advanced scientific computing research and nuclear physics . 17ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevd.29.338 [ * * , ( ) ] link:\doibase 10.1103/physrevd.88.105018 [ * * , ( ) ] , @noop ( ) , @noop * * , ( ) , link:\doibase 10.1103/physrevd.87.094514 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.86.094503 [ * * , ( ) ] , @noop ( ) , link:\doibase 10.1016/s0370 - 2693(01)00798 - 5 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.85.094506 [ * * , ( ) ] , @noop * * , ( ) link:\doibase 10.1103/physrevd.85.054503 [ * * , ( ) ] , link:\doibase 10.1007/jhep09(2010)073 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.71.034504 [ * * , ( ) ] , link:\doibase 10.1038/nature05120 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.74.054507 [ * * , ( ) ] , @noop * * , ( ) , @noop * * , ( )
we report on the first lattice calculation of the qcd phase transition using chiral fermions at physical values of the quark masses . this calculation uses 2 + 1 quark flavors , spatial volumes between ( 4 @xmath0 and ( 11 @xmath0 and temperatures between 139 and 196 mev . each temperature was calculated using a single lattice spacing corresponding to a temporal euclidean extent of @xmath1 . the disconnected chiral susceptibility , @xmath2 shows a pronounced peak whose position and height depend sensitively on the quark mass . we find no metastability in the region of the peak and a peak height which does not change when a 5 fm spatial extent is increased to 10 fm . each result is strong evidence that the qcd `` phase transition '' is not first order but a continuous cross - over for @xmath3 mev . the peak location determines a pseudo - critical temperature @xmath4 mev . chiral @xmath5 symmetry is fully restored above 164 mev , but anomalous @xmath6 symmetry breaking is non - zero above @xmath7 and vanishes as @xmath8 is increased to 196 mev . as the temperature of the qcd vacuum is increased above the qcd energy scale @xmath9 mev , asymptotic freedom implies that the vacuum breaking of chiral symmetry must disappear and the familiar chirally - asymmetric world of massive nucleons and light pseudo - goldstone bosons must be replaced by an @xmath5 symmetric plasma of nearly massless up and down quarks and gluons . predicting , observing and characterizing this transition has been an experimental and theoretical goal since the 1980 s . general principles are consistent with this being either a first - order transition for sufficiently light pion mass or a second - order transition in the @xmath10 universality class at zero pion mass with cross - over behavior for non - zero @xmath11 . while second order behavior is commonly expected , first - order behavior may be more likely if anomalous @xmath6 symmetry is partially restored at @xmath7 resulting in an effective @xmath12 symmetry @xcite . the importance of the @xmath5 chiral symmetry of qcd for the phase transition has motivated the widespread use of staggered fermions in lattice studies of qcd thermodynamics because this formulation possesses one exact chiral symmetry at finite lattice spacing , broken only by the quark mass . however , the flavor symmetry of the staggered fermion formulation is complicated showing an @xmath13 `` taste '' symmetry that is broken by lattice artifacts and made to resemble the physical @xmath5 symmetry by taking the square root of the dirac determinant , a procedure believed to have a correct but subtle continuum limit for non - zero quark masses . because of these limitations , it is important to study these phenomena using a different fermion formulation , ideally one which supports the full @xmath5 chiral symmetry of qcd at finite lattice spacing . it is such a study which we report here . we use mbius domain wall fermions @xcite , a formulation in which the fermions are defined on a five - dimensional lattice . the extent in the fifth dimension , @xmath14 or 24 , making the calculation at least 16 to 24 times more costly . however , the resulting theory possesses an accurate @xmath5 symmetry , broken only by the input quark mass and the highly suppressed mixing between the left and right four - dimensional boundaries , where the low - energy fermions propagate . this residual chiral asymmetry is a short - distance phenomenon whose leading long - distance effect is to add a constant @xmath15 to each input quark mass , @xmath16 , giving a total mass @xmath17 . here the residual mass @xmath18 mev . additional residual chiral symmetry breaking is @xmath19 smaller @xcite . because of the computational cost of this formulation , the calculation reported here uses only one lattice spacing , @xmath20 , at each temperature , corresponding to a single temporal extent of @xmath1 . the good agreement with experiment for @xmath21 and @xmath22 computed at our largest lattice spacing and a comparison of zero temperature results at our @xmath23 lattice spacing with results from two smaller lattice spacings @xcite , suggest discretization errors of @xmath24 in our results . in contrast , the less costly staggered fermion calculations are performed using @xmath1 , 10 , 12 and 16 . however , to make a controlled continuum extrapolation , the staggered fermion discretization errors are assumed to behave as @xmath25 . potential non - linearities in the taste - breaking effects , which in zero - temperature staggered fermion calculations are handled using staggered chiral perturbation theory , are ignored because of the absence of a corresponding theory of finite - temperature taste breaking .
las estrellas de carbono ( css ) son estrellas gigantes fras evolucionadas que presentan material circunestelar en forma de @xmath1 , granularidad amorfa , discos o nubes . uno de sus fenmenos caractersticos es la variabilidad y el anlisis de la misma permite explicar las propiedades fsicas y los procesos que tienen lugar en sus atmsferas , como as tambin determinar su estado evolutivo ( alksne et al . 1991 ) . el relevamiento vvv provee una excelente oportunidad para llevar adelante estudios precisos de variabilidad , ya que nos brinda fotometra infrarroja profunda y multi - pocas que permiten construir curvas de luz de alta calidad ( merlo 2015 , 2016a , b ) . debido a la gran cantidad de datos que el mismo brinda , elaboramos un cdigo que permite obtener para tal fin las magnitudes yzjhks de cada objeto de inters . en la figura [ fig : ab1 ] se muestra el procedimiento general del cdigo . el mismo consta de tres mdulos : _ descarga @xmath2 identificacin @xmath2 tabulacin _ ( merlo 2016a ) . el procedimiento se inicia introduciendo en un @xmath3 genrico el cdigo numrico que permite la descarga de los catlogos , previamente gestionado en la plataforma casu , el @xmath4 de trabajo y las coordenadas de catlogo de la fuente de inters . posteriormente se lo ejecuta en consola y automticamente lleva a cabo cada uno de los mdulos para lo cual requiere conectividad internet permanente . las corridas del cdigo fueron realizadas en grupos de 8 a 10 fuentes , debido a imprevisibles cortes de la conexin a internet . y cdigos @xmath5 ) . ver detalles en merlo ( 2016a ) . ] los catlogos fotomtricos utilizados fueron obtenidos a partir de las imgenes vvv ya procesadas puestas a disponibilidad del grupo colaborador por casu ( _ cambridge astronomical survey unit _ ) . las mismas contienen las posiciones , los flujos y algunas mediciones de forma obtenidas con diferentes aperturas , que incluye la clasificacin morfolgica ms probable . cada unidad de observacin que brinda el telescopio vista , de cuatro metros de dimetro , se denomina `` @xmath4 '' , y consiste de 6 campos individuales de calado denominados `` @xmath6 '' ( ligeramente desplazados ) , cada uno de stos a su vez conformados por las 16 imgenes obtenidas en cada uno de los ccds del instrumento , convenientemente superpuestos para producir la imagen final . toda la fotometra para este trabajo est basada en los catlogos @xmath7 @xmath8 casu 1.3 . estos fueron elegidos en vez de los catlogos @xmath9 ms profundos ya que , en estos ltimos , se han realizado correcciones fotomtricas a travs del _ casu vista data flow system _ con el objetivo de aplanar diferencias espaciales que pudieran aparecer en cada uno de los detectores . estas correcciones se han determinado para degradar aleatoriamente la precisin fotomtrica de algunas fuentes puntuales ( irwin 2011 ) . la ventaja del uso de los catlogos @xmath6 radica en el hecho de disponer de mayor cantidad ( 6x ) de datos fotomtricos ( pocas ) para la fuente de inters , lo cual redunda en una mejor calidad de las curvas de luz construidas a partir de ellas . se han desarrollado una familia de versiones , las cuales se diferencian principalmente en el algoritmo de bsqueda de fuentes como as tambin en el tipo de archivos de catlogo utilizado . las versiones 1.x utilizan el procedimiento de bsqueda por aproximaciones sucesivas , en el cual se va restringiendo el campo alrededor de la coordenada de catlogo y contando las fuentes incluidas , detenindose el proceso cuando no se encuentra ninguna fuente ( ver fig . [ fig : ab2]a ) . en cambio las versiones 2.x hacen uso de un procedimiento de minimizacin de distancia a las coordenadas de catlogo . en el mismo se identifican las fuentes estelares halladas en un campo centrado en dichas coordenadas y se calculan las distancias angulares respectivas a ellas , seleccionndose aquella fuente que disponga del menor valor ( ver fig . [ fig : ab2]b ) . a su vez , las versiones x.0 utilizan los catlogos provenientes de las baldosas o @xmath9 ( merlo 2016a ) , mientras que las versiones x.5 los catlogos @xmath6 ( merlo 2016b ) en relacin a los tiempos de procesamiento , y a modo de ejemplo , en un campo con 396 catlogos @xmath6 , el nuevo proceso de identificacin ( versin 2.5 ) insumi 7@xmath10 ( @xmath01 seg / cat ) , mientras que en otro campo , con 77 catlogos @xmath9 , el anterior proceso de identificacin ( versin 1.0 ) necesit 1,5@xmath11 ( @xmath070 seg / cat ) . los grficos de la fig . [ fig : ab3 ] muestran , en trazo continuo , los nuevos valores ( * s2 * ) de las magnitudes ks de las cs halladas , comparados con las versiones anteriores 1.0 y 1.5 ( * s * ) . las estrellas pertenecen a los @xmath9 del bulbo galctico @xmath12 y @xmath13 . los tiempos de procesamiento obtenidos disminuyeron significativamente , logrndose adems optimizar el proceso de deteccin e identificacin eliminando muchos falsos positivos . este cdigo se encuentra en etapa de depuracin y ampliacin , las cuales una vez implementadas el mismo ser puesto a disposicin de toda la comunidad interesada en utilizarlo . recientemente se ha finalizado el procedimiento de descarga y obtencin de las magnitudes yzjhks - vvv de todas las estrellas de carbono pertenecientes al bulbo de la va lctea , por lo que se ha iniciado el proceso de anlisis de variabilidad . una vez finalizado , se llevar adelante un procedimiento similar con las css pertenecientes a la zona del disco galctico cubierto por el relevamiento vvv . cabe destacar , finalmente , que este relevamiento tendr prximamente una extensin denominada vvv - x , el cual ampliar un poco ms la zona cubierta por el vvv . por ello tenemos previsto continuar nuestro estudio utilizando los nuevos y valiosos ctalogos que surgirn del mismo . alksne , z. et al . , 1991 , `` properties of galactic carbon stars '' , orbit book co. alksnis , a. et al . , 2001 , balt.a . , 10 , 1 . irwin , m.j . , 2011 , comunicacin privada . merlo , d. , 2015 , baaa , 57 , 111 . merlo , d. , 2016a , 7vvv workshop , 29 - 1/03 , antofagasta , chile . [ link ] merlo , d. , 2016b , fof2016 , 29 - 1/04 , crdoba , argentina . [ link ] saito , r. et al . , 2012 , a&a , 544 , a147 .
an improved version of the identification stellar sources module in the quasi - automatic sacaman code is presented , which allows to obtain yzjhks - vvv magnitudes for a set of predefined objects . the procedure uses a proximity algorithm @xmath0 70 less time - consuming than the previous method of successive and counting approximation . this code is being applied to the study of variability of carbon stars belonging to the galactic bulge . 1.0 cm nuevo algoritmo de identificacio en sacaman se presenta una versin mejorada del mdulo de identificacin de fuentes estelares en el cdigo cuasi - automtico sacaman , el cual permite obtener magnitudes yzjhks - vvv para un conjunto de objetos de inters . el procedimiento utiliza un algoritmo de proximidad , insumiendo un tiempo de procesamiento @xmath0 70 veces menor que el mtodo de aproximaciones y conteos sucesivos de la versin anterior . este cdigo est siendo utilizado en el estudio de variabilidad de estrellas de carbono pertenecientes al bulbo galctico .
the most recent and relatively accurate test of the general relativistic gravitomagnetic lense - thirring effect on the orbit of a test particle ( lense and thirring 1918 ; barker and oconnell 1974 ; cugusi and proverbio 1978 ; soffel 1989 ; ashby and allison 1993 ; iorio 2001 ) in the gravitational field of the earth on average ) test of the lense - thirring effect was recently reported by iorio ( 2006a ) in the gravitational field of mars . ] was performed by ciufolini and pavlis ( 2004 ) , who analyzed the laser data of the lageos and lageos ii satellites according to a suitable combination of the residuals of their nodes proposed in ( ries et al . 2003a ; 2003b ; iorio and morea 2004 ) ^lageos+ c_1^lageos ii ~48.1 . let us briefly recall the linear combination approach from which originates . the combinations are obtained by explicitly writing down the expressions of the residuals of @xmath5 orbital elements ( the nodes of different satellites in our case ) in terms of the classical secular precessions induced by the mismodelled part of @xmath6 even zonal harmonic coefficient @xmath7 of the multipolar expansion of the terrestrial gravitational potential ( see also section [ gravzon ] and for the meaning of the coefficients @xmath8 ) and the lense - thirring effect @xmath9 considered as an entirely unmodelled feature of motion ^(i)=_=2 ^ 2(n-1)^(i)_.j_+^(i)_lt_lt , i=1,2 ... n , and solving the resulting algebraic non - homogeneous linear system of @xmath5 equations in @xmath5 unknowns of with respect to the scaling parameter @xmath10 which is 1 in the einsteinian theory and 0 in newtonian mechanics . the obtained coefficients weighing the satellites orbital elements depend on their semimajor axes @xmath11 , eccentricities @xmath12 and inclinations @xmath13 : they allow to cancel out the impact of the @xmath6 even zonal harmonics considered . in the value of the secular trend predicted by the general relativity theory is 48.1 milliarcseconds per year ( mas yr@xmath14 ) , and @xmath15 . the coefficient @xmath16 makes the combination of insensitive to the biasing action of only the first even zonal @xmath17 and its temporal variations . the other even zonal harmonics @xmath18 , along with their secular variations @xmath19 , do affect inducing a systematic error in the measurement of the lense - thirring effect , whose correct and reliable evaluation is of crucial importance for the reliability of such an important test of fundamental physics . ciufolini and pavlis ( 2004 ) , who used the grace - only earth gravity model eigen - grace02s ( reigber et al 2005a ) , claimed a total error of @xmath20 at 1-sigma and @xmath21 at 3-sigma . such estimates were criticized by iorio ( 2005 ; 2006b ) for various reasons . his evaluations , based on the analysis of different gravity model solutions and on the impact of the secular variations of the uncancelled even zonals , point toward a more conservative @xmath22 total error at 1-sigma . the major drawbacks of the combination of are as follows * it is mainly affected by the low - degree even zonal harmonics @xmath23 . the combination of is practically insensitive to the even zonal harmonics of degree higher than @xmath24 in the sense that the error induced by the uncancelled zonals does not change if the terms of degree higher than @xmath24 are neglected in the calculation , as fully explained in section [ gravzon ] . unfortunately , the major improvements from the present - day and forthcoming grace models are mainly expected just for the medium - high degree even zonal harmonics which do not affect . instead , the low - degree even zonals should not experience notable improvements , as showed by the most recent long - term models like eigen - cg01c ( reigber et al . 2006 ) , eigen - cg03c ( frste et al . 2005 ) , eigen - grace02s , ggm02s ( tapley et al . 2005 ) . moreover , the part of the systematic error due to them is still rather model - dependent ranging from @xmath1 to @xmath25 . * another source of aliasing for the combination of is represented by the secular variations @xmath26 and @xmath27 whose signal grows quadratically in time . their bias on the measurement of the lense - thirring effect with the combination of was evaluated to be of the order of @xmath28 ( iorio 2005 ) . they are , at present , known with modest accuracy and there are few hopes that the situation could become more favorable in the near future . moreover , also interannual variations of @xmath29 and @xmath30 may turn out to occur thus , it seems unlikely that relevant improvements in the reliability and accuracy of the tests conducted with the adopted node - node combination of the lageos satellites will occur in the foreseeable future . in iorio and doornbos ( 2005 ) the following combination ^lageos+k_1^lageosii+k_2^ajisai+k_3^jason-1=_lt49.5,with k_1=0.347,k_2=-0.005 , k_3=0.068,was designed : it comes from applied to the nodes of lageos , lageos ii , jason-1 and ajisai . a similar proposal was put forth by vespe and rutigliano ( 2005 ) : however , the less accurate champ - only earth gravity model eigen3p ( reigber et al . 2005b ) was used in that exhaustive analysis . such a combination involves the nodes of the geodetic ajisai satellite and of the radar altimeter jason-1 satellite . their orbital parameters , together with those of the lageos satellites , are listed in table [ sat_par ] . the combination of allows cancellation of the first three even zonal harmonics @xmath31 along with their temporal variations . the resulting systematic error of gravitational origin is of the order of @xmath32 . the practical implementation of the proposed test would consist in the following three stages * the best possible nodes from independent arcs of data ( for example , weekly ) will be assembled as a time - series for the four satellites * correspondingly , an integrated long - term node time - series will be constructed for each satellite with the best available dynamical models not using such force models derived empirically from the same tracking data determining the observed nodes ( otherwise , also the lense - thirring effect would be removed . see section [ cazzo ] ) * from such two time - series a residual time - series will be built up for each satellite , combined according to and analyzed for both secular and periodic terms ; the secular component will be used to extract the lense - thirring effect the goal of the present paper is to analyze in detail some important critical aspects of the use of such a combination . they are * the impact of the higher degree even zonal harmonics introduced by the lower orbiting satellites ajisai and jason-1 * the impact of the realistically obtainable accuracy of a truly dynamical orbital reconstruction for ajisai and jason-1 * the impact of the atmospheric drag and of the other non - gravitational perturbations on ajisai and , especially , jason-1 the even zonal harmonics @xmath33 of the newtonian multipolar expansion of the earth s gravitational potential induce on the node of an artificial satellite a classical secular precession which can be cast in the form ^class = _ 2_.j_.the coefficients @xmath8 depend on the earth s @xmath34 and mean equatorial radius @xmath35 , and on the semimajor axis , the eccentricity and the inclination of the satellite . they were analytically calculated up to degree @xmath36 in iorio ( 2003 ) and their numerical values , in mas yr@xmath14 , for lageos , lageos ii , ajisai and jason-1 can be found in table [ prec_coef ] . the coefficients @xmath37 and @xmath38 of the combinations of and are built up with @xmath8 . the precessions of are much larger than the lense - thirring rates . this is the reason why the combinations of and are , by construction , designed in order to cancel out the precessions induced by the first low - degree even zonals . this approach was proposed for the first time by ciufolini ( 1996 ) with a combination involving the nodes of lageos and lageos ii and the perigee of lageos ii . one of the major objections about the combination also involving ajisai and jason-1 is that such satellites , which orbit at much lower altitudes with respect to lageos and lageos ii , ( table [ sat_par ] ) , would introduce much more even zonals in the systematic error of gravitational origin than the node - node lageos - lageos ii combination . indeed , the classical precessions depend on the satellite s semimajor axis as _ .a^-(+).in fact , this criticism would better fit the case of the other existing low - orbit geodetic satellites like , e.g. , starlette and stella . indeed , iorio ( 2006c ) showed that , according to eigen - cg03c , a combination involving such spacecraft is not yet competitive with other combinations just because of the systematic bias due to the even zonals . in the case of the combination of , it turns out that only about the first ten even zonals are relevant for a satisfactorily estimate of the systematic error of gravitational origin . indeed , it can be evaluated as _ lt_8|(_.^lageos+k_1_.^lageosii+k_2_.^ajisai+k_3_.^jason-1)|j_,where @xmath39 are the errors in the even zonal harmonics according to a given earth gravity models . they can be found in table [ geopot_models ] for eigen - cg03c , eigen - cg01c , eigen - grace02s and ggm02s . note that yields a conservative upper bound of the bias induced by the mismodelling in the even zonal harmonics . the individual terms of , calculated for the various gravity solutions considered here , are listed in table [ comb_err ] . it can be noted that the resulting total error @xmath40 , which is of the order of @xmath0 of the lense - thirring effect for the latest models combining data from champ , grace and ground - based measurements , does not significantly change if the terms of degree higher than @xmath36 are not included in the calculation . another important feature is that such an error is much less model - dependent than that of the combination of ; moreover , it is likely that the forthcoming gravity models based on champ and grace will further ameliorate the situation because they should especially improve the medium - high degree even zonal harmonics to which the combination of is mainly sensitive . other distinctive features of are that the secular variations @xmath41 do not affect it by construction and no long - period harmonic perturbations of tidal origin would corrupt the measurement of the lense - thirring effect . indeed , the most powerful uncancelled tidal perturbation is due to the @xmath42 component of the solar @xmath43 tide whose period is equal to the satellite s node period : the periods of the nodes of lageos , lageos ii , ajisai and jason amount to 2.84 , -1.55 , -0.32 and -0.47 years , respectively . in conclusion , the systematic error of gravitational origin of the combination of can safely be evaluated as @xmath0 : the forthcoming earth gravity models from champ and , especially , grace should be able to reduce such error below the @xmath44 level . another possible criticism about the use of ajisai and jason-1 for measuring the lense - thirring effect is the following . while the cm - level accuracy in reconstructing the orbits of the lageos satellites is based on a truly dynamical , very extensive and accurate modelling of the various accelerations of non - gravitational origin affecting them , it would not be so for ajisai and , especially , jason-1 . indeed , too many empirical accelerations which could sweep out also the lense - thirring effect of interest are used to reach the cm level for ajisai and jason-1 ( lutchke et al . 2003 ) . without resorting to such a reduced - dynamic approach , the genuine obtainable orbit accuracy would be worst . let us assume , very conservatively , that the root mean square ( rms ) of the recovered orbits of ajisai and jason-1 , amount to 1 m over , say , 1 year . then , the error in the nodal rates can be quantified as 26.2 mas and 26.7 mas for ajisai and jason-1 , respectively . thus , their impact on the combination would amount to 1.6 mas , i.e. about @xmath45 of the lense - thirring effect over 1 year . in view of the fact that the temporal interval of the analysis should cover some years and that a more realistic estimate of the orbital accuracy amounts to some tens of cm , it can be concluded that the impact of the orbital reconstruction errors on the combination of is at the few percent level . however , it must be stressed that this is a very pessimistic evaluation because , even at this level for jason-1 or ajisai such independent errors for weekly orbits would result in totally negligible secular rate errors on fitting the weekly time - series over a number of years . the impact of the non - gravitational perturbations on the lageos - type satellites has been the subject of numerous recent papers ( ries et al . 1989 ; lucchesi 2001 ; 2002 ; 2003 ; 2004 ; lucchesi et al 2004 ) ; according to the most recent works , it turns out that the systematic bias induced by them on the combination of through lageos and lageos ii would be of the order of @xmath44 . undoubtedly the major concern about the use of the combination of , the non gravitational perturbations affect ajisai and , especially , jason-1 much more severely than lageos and lageos ii . indeed , the area - to - mass ratios , to which such kind of perturbing effects are proportional , of ajisai and jason-1 are larger than those of lageos and lageos ii by one or two orders of magnitude . they are listed in table [ sat_par ] . moreover , jason-1 is not spherical in shape , is endowed with steering solar panels and is regularly affected by orbital maneuvers due to its primary altimetric and oceanographic tasks . for example , the pointing of the solar panels to the sun is not perfectly normal to the solar direction , so that there may be small systematic cross - track forces whose impact is difficult to be reliably assessed . there are also many complicated reflective and emissive surfaces on the spacecraft bus . as the solar panels produce power for the onboard instruments and heaters , some part of it is routinely dumped into space by heat radiators on the side of the spacecraft , in particular when the batteries are fully re - charged . since the heat can not be dumped to the side of the spacecraft exposed to the sun , part of the yaw - steering algorithm is intended to keep this side away from solar exposure . such heat re - radiation acceleration tends to have a component in the cross - track direction which probably has some orbital period dependence . another point to be considered is that the radiators are typically only on one side of jason-1 , so that the satellite performs a ` yaw - flip ' each time the orbital plane passes through the solar direction to keep that side away from the sun . this fact might yield to a non - symmetric pattern of the resulting accelerations . as a consequence , an entirely reliable and accurate modelling of the perturbations of non gravitational origin acting on it is not an easy task . nonetheless , in the next sections we will show that the situation for the combination of is less unfavorable than it could seem at a first sight , provided that some simplifying assumptions are made . ajisai is a spherical geodetic satellite launched in 1986 . it is a hollow sphere covered with 1436 corner cube reflectors ( ccrs ) for slr and 318 mirrors to reflect sunlight . its diameter is 2.15 m , contrary to lageos which has a diameter of 60 cm . its mass is 685 kg , while lageos mass is 406 kg . then , the ajisai s area - to - mass ratio @xmath46 , which the non conservative accelerations are proportional to , is larger than that of the lageos satellites by almost one order of magnitude resulting in a higher sensitivity to surface forces . however , we will show that their impact on the proposed combination should be less than 1@xmath2 . an important non conservative force affecting the orbits of low - earth satellites is the atmospheric drag . its acceleration can be written as * _ a_*_d =- c_d()v * _ v_*,where @xmath47 is a dimensionless drag coefficient close to 2 , @xmath48 is the atmospheric density and * _ v _ * is the velocity of the satellite relative to the atmosphere ( called ambient velocity ) . let us write * * _ v_**=**_v_**@xmath49 * _ r _ * where * _ v _ * is the satellite s velocity in an inertial frame . if the atmosphere corotates with the earth @xmath50 is the earth s angular velocity vector @xmath51**_k _ * * , where * _ k _ * is a unit vector . however , it must be considered that there is a 20@xmath2 uncertainty in the corotation of the earth s atmosphere at the ajisai s altitude . indeed , it is believed that the atmosphere rotates slightly faster than the earth at some altitudes with a 10 - 20@xmath2 uncertainty . we will then assume @xmath52**_k _ * * , with @xmath53 in order to account for this effect . regarding the impact of a perturbing acceleration on the orbital motion , the gaussian perturbative equation for the nodal rate is = a_n()u , where @xmath54 is the keplerian mean motion , @xmath55 is the out of plane component of the perturbing acceleration and @xmath56 is the satellite s argument of latitude . the out of plane acceleration induced by the atmospheric drag can be written as ( abd el - salam and sehnal 2004 ) a_n^(atm ) = -k_dvriu , with k_d = c_d()and k_r=1-+()^2 . the quantities @xmath57 and @xmath58 are the orbital angular momentum per unit mass and the satellite s declination , respectively . by inserting in and evaluating it on an unperturbed keplerian ellipse it can be obtained ^(atm ) -k_d ( 1-e^2)a.it must be pointed out that the density of the atmosphere @xmath48 has many irregular and complex variations both in position and time . it is largely affected by solar activity and by the heating or cooling of the atmosphere . moreover , it is not actually spherically symmetric but tends to be oblate . a very cumbersome analytic expansion of @xmath48 based on the td88 model can be found in ( abd el - salam and sehnal 2004 ) . in order to get an order of magnitude estimate we will consider a typical value @xmath59 g @xmath60 at ajisai altitude ( sengoku et al . 1996 ) . by assuming @xmath61 yields a nominal amplitude of 25 mas yr@xmath14 for the atmosphere corotation case and 5 mas yr@xmath14 for the 20@xmath2 departure from exact corotation . the impact of such an effect on would be @xmath62 . the action of the thermal forces due to the interaction of solar and terrestrial electromagnetic radiation with the complex physical structure of ajisai has been investigated in sengoku et al . the temperature asymmetry on ajisai caused by the infrared radiation of the earth produces a force along the satellite spin axis direction called the yarkovsky - rubincam effect . this thermal thrust produces secular perturbations in the orbital elements , but no long - periodic perturbations exist if the spin axis of ajisai is aligned with the earth s rotation axis . in fact , the spin axis was set parallel to the earth rotation axis at orbit insertion . the analogous solar heating ( yarkovsky - schach effect ) is smaller than the terrestrial heating . a nominal secular nodal rate of 15 mas yr@xmath14 due to the earth heating has been found . it would affect at a @xmath63 level . the effect of the direct solar radiation pressure on ajisai has been studied in sengoku et al . for an axially symmetric , but not spherically symmetric , satellite like ajisai , there is a component of the radiation pressure acceleration directed along the sun - satellite direction @xmath64 and another smaller component perpendicular to the sun - satellite direction . by assuming for the isotropic reflectivity coefficient its maximum value @xmath65 , the radiation pressure acceleration @xmath66 experienced by ajisai amounts to @xmath67 m s@xmath68 . its nominal impact on the node , proportional to @xmath69 , can be quantified as 5.7 mas yr@xmath14 ; it yields a @xmath70 relative error on . the anisotropic component of the acceleration would amount , at most , to 2@xmath2 of the isotropic one , so that its impact on would be totally negligible . as already previously noted , the complex shape , varying attitude modes and the relatively high area to mass ratio of jason-1 , suggest a more complex modelling and higher sensitivity to the non gravitational accelerations than in the case of the spherical geodetic satellites . on the other hand , important limiting factors in the non - gravitational force modelling of the spherical satellites are attitude and temperature knowledge . these parameters are actually very well defined and accurately measured ( marshall et al . 1995 ) on satellites such as jason1 . a lot of effort has already been put into the modelling of non gravitational accelerations for topex / poseidon ( antreasian and rosborough 1992 , marshall et al . 1994 ; kubitschek and born 2001 ) , so that similar models ( berthias et al . 2002 ) have been routinely implemented for jason1 . these so called box wing models , in which the satellite is represented by eight flat panels , were developed for adequate accuracy while requiring minimal computational resources . a recent development is the work on much more detailed models of satellite geometry , surface properties , eclipse conditions and the earth s radiation pressure environment for use in orbit processing software ( doornbos et al . 2002 ; ziebart et al . 2003 ) . it should be noted that such detailed models were not yet adopted in the orbit analyses by lutchke et al . in fact , their results were based on the estimation of many empirical 1-cycle - per - revolution ( cpr ) along - track and cross - track acceleration parameters , which absorb all the mismodelled / unmodelled physical effects , of gravitational and non - gravitational origin , which induce secular and long - period changes in the orbital elements . due to the power of this reduced dynamic technique , based on the dense tracking data , further improvements in the force models become largely irrelevant for the accuracy of the final orbit . such improved models remain , however , of the highest importance for the determination of @xmath71 . from it can be noted that , since we are interested in the effects averaged over one orbital revolution , the impact of every acceleration constant over such a timescale would be averaged out . as previously noted , the major problems come from 1cpr out of plane accelerations of the form @xmath72 , with @xmath73 and @xmath74 constant over one orbital revolution . we have analyzed both the output of the non - gravitational force models described in berthias et al . ( 2002 ) and the resulting residual empirical 1cpr accelerations estimated from doris and slr tracking over 24hour intervals . solar radiation pressure , plotted in figure [ fig : solrad ] , is by far the largest out of plane non gravitational acceleration , with a maximum amplitude of 147 nm s@xmath68 . it is followed by earth radiation pressure at approximately 7 nm s@xmath68 . the contributions of aerodynamic drag and the thermal imbalance force on the cross track component are both estimated to have a maximum of approximately 0.5 nm s@xmath68 . as can be seen in figure [ fig : solrad ] , the cross track solar radiation pressure acceleration shows a sinusoidal long - term behavior , crossing zero when the sun satellite vector is in the orbital plane , roughly every 60 days . it is modulated by the long term seasonal variations in sun earth geometry , as well as by eclipses and the changing satellite frontal area , both of which contribute 1cpr variations . in fact , the shading of certain areas in figure [ fig : solrad ] is due to the effect of the eclipses , which , at once per orbit , occur much more frequently than can be resolved in figure [ fig : solrad ] . as mentioned before , the empirical 1cpr accelerations absorb the errors of almost all the unmodelled or mismodelled forces . now note the systematic way in figure [ fig : solrad ] , in which the empirical 1cpr cross - track acceleration drops to values of below 1 nm s@xmath68 near the end of each eclipse free period , and has its maximum level of 56 nm s@xmath68 only during periods containing eclipses . the fact that the amplitude , but also the phase ( not shown in figure [ fig : solrad ] ) of the 1cpr accelerations show a correlation with the orientation of the orbital plane with respect to the sun , indicates that it is for a large part absorbing mismodelled radiation pressure accelerations . by averaging over one orbital revolution and from the orbital parameters of table [ sat_par ] it turns out that a 1cpr cross track acceleration would induce a secular rate on the node of jason proportional to @xmath75 s m@xmath76 m s@xmath68 . this figure must be multiplied by the combination coefficient @xmath77 . by using the average value of the empirical 1cpr acceleration from the above analysis @xmath78 2.3 nm s@xmath68 as an estimate for the mismodelled non gravitational forces , it can be argued that the impact on @xmath79 would amount to 77.4 mas yr@xmath14 . however , it must be pointed out that our assumed value of @xmath73 can be improved by adopting the aforementioned more detailed force models or by tuning the radiation pressure models using tracking data . in addition , it must be pointed out that @xmath73 experiences long term variations mainly induced by the orientation of the orbital plane with respect to the sun , and the related variations in satellite attitude . for jason1 such a periodicity amounts to approximately 120 days ( the @xmath80 cycle ) . let us , now , evaluate what would be the impact of such a long - periodic perturbation on our proposed measurement of the lense thirring effect . let us write , e.g. , a sinusoidal law for the long periodic component of the weighted nodal rate of jason1 k_3=(77.4 mas yr^-1);then , if we integrate over a certain time span @xmath81 we get k_3=()(77.4 mas yr^-1 ) .then , the amplitude of the shift due to the weighted node of jason1 , by assuming @xmath82 days , would amount to k_3 4 mas.the maximum value would be obtained for = , j=1,3,5, ... 30 , 90 , 150 , ... days . so , the impact on the proposed measurement of the lense thirring effect would amount to .|_srp;for , say , @xmath83 years yields an upper bound of @xmath84 . moreover , it must also be noted that it would be possible to fit and remove such long periodic signals from the time - series provided that an observational time span longer than the period of the perturbation is adopted . in this paper the use of a suitable linear combination of the nodes of lageos , lageos ii , ajisai and jason-1 to measure the lense - thirring effect in the gravitational field of the earth is examined . below we list the major sources of errors along with our evaluations of their impact on the proposed measurement . they are also summarized in table [ tabwag ] . it turns out that the systematic error of gravitational origin due to the even zonal harmonics can be presently evaluated to be @xmath32 , according to the latest earth gravity models based on the combined data of champ , grace and ground - based measurements . such an estimate is rather model - independent and will be likely further improved when the new , forthcoming solutions for the terrestrial gravitational potential will be available . the temporal variations of the even zonal harmonics do not represent a major concern because the secular and possible interannual variations of the first three even zonal harmonics are cancelled out , by construction , along with their static components . moreover , the uncancelled tidal perturbations , like the solar @xmath43 tide , vary with relatively high frequencies , so that they could be fitted and removed from the time - series or averaged out over an observational time span of at least 3 years ( the longest period is that of the lageos node amounting to 2.84 years ) . our largely conservative evaluation for the measurement errors amounts to @xmath85 , where @xmath5 is the number of years of the experiment duration , by assuming a really pessimistic 1 m error in a truly dynamical orbit reconstruction for ajisai and jason-1 over the adopted time span . in regard to the non - gravitational perturbations , which especially affect jason-1 , it is worthwhile noting that no secular aliasing trends should occur , but only high - frequency harmonic perturbations however , particular attention should be paid to an as accurate as possible truly dynamical modelling of the non - gravitational accelerations acting on the node of jason-1 . also a careful choice of the observational time span of the analysis would be required in order to reduce the uncertainties related to the orbital maneuvers which are mainly in plane , although a small , unknown , part of them affects also the out - of - plane part of the orbit . we evaluate the error due to the non - gravitational accelerations as large as @xmath1 over 2 years . in conclusion , the use of the proposed combination , although undoubtedly difficult and demanding , seems to be reasonable and feasible ; we give a total root - sum - square uncertainty of @xmath86 4 - 5@xmath2 over at least 3 years required to average out the uncancelled tidal perturbations . moreover , the efforts required to perform the outlined analysis should be rewarding not only for the relativists community but also for people involved in space geodesy , altimetry and oceanography . i thank e doornbos for figure 1 , many useful references and important discussions and clarifications . i am also grateful to c wagner , e grafarend and j ries for their useful and critical remarks and observations . i am also grateful to the anonymous referee whose comments and observations greatly improved the manuscript . frste c , flechtner f , schmidt r , meyer u , stubenvoll r , barthelmes f , knig r , neumayer kh , rothacher m , reigber ch , biancale r , bruinsma s , lemoine j - m , raimondo jc ( 2005 ) a new high resolution global gravity field model derived from combination of grace and champ mission and altimetry / gravimetry surface gravity data . egu general assembly 2005 , vienna , austria , 24 - 29 , april 2005 ( http://icgem.gfz-potsdam.de/icgem/icgem.html ) lense j , thirring h ( 1918 ) ber den einfluss der eigenrotation der zentralkrper auf die bewegung der planeten und monde nach der einsteinschen gravitationstheorie . phys z 19 156163 . english translation and discussion by mashhoon b , hehl fw , theiss ds ( 1984 ) on the gravitational effects of rotating masses : the thirring - lense papers . gen rel gravit 16 : 711750 lucchesi dm , ciufolini i , andrs ji , pavlis ec , peron r , noomen r , currie dg ( 2004 ) lageos ii perigee rate and eccentricity vector excitations residuals and the yarkovsky schach effect . plan space sci 52 : 699 - 710 marshall ja , zelensky np , klosko sm , chinn ds , luthcke sb , rachlin ke , williamson rg ( 1995 ) the temporal and spatial characteristics of topex / poseidon radial orbit error . j geophys res 100 : 2533125352 reigber ch , schmidt r , flechtner f , knig r , meyer u , neumayer k - h , schwintzer p , zhu sy ( 2005a ) an earth gravity field model complete to degree and order 150 from grace : eigen - grace02s . j geodyn 39 : 110 ( http://icgem.gfz-potsdam.de/icgem/icgem.html ) reigber ch , jochmann h , wnsch j , petrovic s , schwintzer p , barthelmes f , neumayer k - h , knig r , frste ch , balmino g , biancale r , lemoine j - m , loyer s , perosanz f ( 2005b ) earth gravity field and seasonal variability from champ in : reigber ch , lr h , schwintzer p , wickert , j. ( ed ) earth observation with champ - results from three years in orbit . springer , berlin , pp 25 - 30 reigber ch , schwintzer p , stubenvoll r , schmidt r , flechtner f , meyer u , knig r , neumayer h , frste ch , barthelmes f , zhu sy , balmino g , biancale r , lemoine j - m , meixner h , raimondo jc ( 2006 ) a high resolution global gravity field model combining champ and grace satellite mission and surface gravity data : eigen - cg01c . paper presented at the joint champ / grace science meeting , gfz , july 5 - 7 , 2004 ( page 16 , no . 24 in solid earth abstracts ( pdf file ) ) . j geod in press ( http://icgem.gfz-potsdam.de/icgem/icgem.html ) . ries jc , eanes rj , watkins mm , tapley bd ( 1989 ) joint nasa / asi study on measuring the lense - thirring precession using a second lageos satellite . csr-89 - 3 , center for space research , the university of texas at austin . ries jc , eanes rj , tapley bd ( 2003a ) lense - thirring precession determination from laser ranging to artificial satellites . in : ruffini r , sigismondi c ( eds . ) , nonlinear gravitodynamics . the lense thirring effect , world scientific , singapore , pp . 201 - 211 . ries jc , eanes rj , tapley bd , peterson ge ( 2003b ) prospects for an improved lense - thirring test with slr and the grace gravity mission . in : noomen r , klosko s , noll c , pearlman m ( ed ) proc . laser ranging workshop , nasa cp 2003 - 212248 , nasa goddard , greenbelt . preprint http://cddisa.gsfc.nasa.gov/lw13/[email protected]@xmath88science tapley b , ries j , bettadpur s , chambers d , cheng mk , condi f , gunter b , kang z , nagel p , pastor r , pekker t , poole s , wang f ( 2005 ) ggm02 - an improved earth gravity field model from grace . j geod 79 : 467 - 478 ( http://icgem.gfz-potsdam.de/icgem/icgem.html ) ziebart m , adhya s , cross p , bar sever y , desai s ( 2003 ) pixel array solar and thermal force modelling for jason : on orbit test results . paper presented at the from topex - poseidon to jason science working team meeting , arles , france , november 2003 . @xmath11 ( km ) & 12270 & 12163 & 7870 & 7713 + @xmath12 & 0.0045 & 0.014 & 0.001 & 0.0001 + @xmath13 ( deg ) & 110 & 52.65 & 50 & 66.04 + @xmath46 ( m@xmath89 kg@xmath14 ) & @xmath90 & @xmath91 & @xmath92 & @xmath93 + @xmath9 ( mas yr@xmath14 ) & 30.7 & 31.4 & 116.2 & 123.4 + lllll @xmath94 & eigen - cg03c & eigen - grace02s & eigen - cg01c & ggm02s + 2 & 2.341@xmath95 & @xmath96 & @xmath97 & @xmath98 + 4 & @xmath99 & @xmath100 & @xmath101 & @xmath102 + 6 & @xmath103 & @xmath104 & @xmath105 & @xmath106 + 8 & @xmath107 & @xmath108 & @xmath109 & @xmath110 + 10 & @xmath111 & @xmath112 & @xmath113 & @xmath114 + 12 & @xmath115 & @xmath116 & @xmath117 & @xmath118 + 14 & @xmath119 & @xmath120 & @xmath121 & @xmath122 + 16 & @xmath123 & @xmath124 & @xmath125 & @xmath122 + 18 & @xmath126 & @xmath127 & @xmath128 & @xmath122 + 20 & @xmath129 & @xmath130 & @xmath131 & @xmath122 + 22 & @xmath132 & @xmath133 & @xmath134 & @xmath122 + 24 & @xmath135 & @xmath136 & @xmath137 & @xmath138 + 26 & @xmath139 & @xmath140 & @xmath141 & @xmath138 + 28 & @xmath142 & @xmath143 & @xmath144 & @xmath118 + 30 & @xmath145 & @xmath146 & @xmath147 & @xmath148 + 32 & @xmath149 & @xmath150 & @xmath151 & @xmath148 + 34 & @xmath152 & @xmath153 & @xmath154 & @xmath155 + 36 & @xmath156 & @xmath157 & @xmath158 & @xmath159 + 38 & @xmath160 & @xmath161 & @xmath162 & @xmath163 + 40 & @xmath164 & @xmath165 & @xmath166 & @xmath163 + lllll @xmath94 & eigen - cg03c & eigen - grace02s & eigen - cg01c & ggm02s + 2 & - & - & - & - + 4 & - & - & - & - + 6 & -&- & -&- + 8 & 0.141 & 0.178 & 0.216 & 0.338 + 10 & 0.170 & 0.417 & 0.261 & 0.397 + 12 & 0.093 & 0.168 & 0.014 & 0.246 + 14 & 0.011 & 0.023 & 0.017 & 0.031 + 16 & 0.027 & 0.052 & 0.041 & 0.084 + 18 & 0.027 & 0.058 & 0.042 & 0.092 + 20 & 0.013 & 0.032 & 0.020 & 0.047 + 22 & @xmath167 & @xmath167 & @xmath167 & @xmath167 + 24 & 0.007 & 0.014 & 0.011 & 0.025 + 26 & 0.008 & 0.016 & 0.012 & 0.026 + 28 & 0.004 & 0.008 & 0.006 & 0.013 + 30 & @xmath167 & 0.001 & 0.001 & 0.003 + 32 & 0.003 & 0.006 & 0.005 & 0.009 + 34 & 0.002 & 0.005 & 0.004 & 0.008 + 36 & @xmath167 & 0.002 & 0.001 & 0.003 + 38 & @xmath167 & 0.001 & 0.001 & 0.002 + 40 & 0.001 & 0.002 & 0.002 & 0.003 + @xmath168 & 1.0 & 1.9 & 1.3 & 2.6 +
in this paper we analyze in detail some aspects of the proposed use of ajisai and jason-1 , together with the lageos satellites , to measure the general relativistic lense - thirring effect in the gravitational field of the earth . a linear combination of the nodes of such satellites is the proposed observable . the systematic error due to the mismodelling in the uncancelled even zonal harmonics would be @xmath0 according to the latest present - day champ / grace - based earth gravity models . in regard to the non - gravitational perturbations especially affecting jason-1 , only relatively high - frequency harmonic perturbations should occur : neither semisecular nor secular bias of non - gravitational origin should affect the proposed combination : their maximum impact is evaluated to @xmath1 over 2 years . our estimation of the root - sum - square total error is about 4 - 5@xmath2 over at least 3 years of data analysis required to average out the uncancelled tidal perturbations . # 1 # 1 # 1#2_#1^#2 # 1eq . ( [ # 1 ] ) # 1#2([#1])-([#2 ] ) # 1eq . ( [ # 1 ] ) # 1#2([#1])-([#2 ] ) # 1 # 1#2 # 1#2 # 1#2 } \def\ear{\end{aligned}\ ] ] # 1#2#1#2 # 1 # 1[#1 ] 2@xmath3 # 1 * * + + + l. iorio , + _ viale unit@xmath4 di italia 68 , 70125 + bari , italy + tel./fax 0039 080 5443144 + e - mail : [email protected]_ keywords : gravitation ; relativity ; pacs : 04.80.cc
strongly focused laser beams , produced by passage through a high numerical aperture ( na ) microscope objective , have been employed increasingly often in recent years . a widespread application is their use for trapping micro - particles @xcite or atoms @xcite in optical tweezers , employed in a variety of fields @xcite , ranging from cell biology @xcite to quantum information processing @xcite . in the standard optical tweezers setup , the objective entrance port is usually illuminated by a gaussian linearly polarized beam , overfilling the entrance aperture in order to fully utilize the high na of the objective . a proper description of such a tightly focused laser beam must include diffraction at the aperture edge as well as nonparaxial effects . the vectorial debye - type integral representation developed by richards and wolf ( rw ) @xcite takes both effects into account . this model has been thoroughly verified directly , by measuring the electric energy density distribution in the focal region @xcite , and also indirectly , by testing the mie - debye theory for the trapping force in optical tweezers ( derived from the rw model @xcite ) against experimental results @xcite . novel features , with potential applications in micromechanics , microfluidics and biotechnology , have been developed by employing laguerre - gaussian ( lg ) beams @xcite . these beams have helical wave fronts and a phase singularity along the axis ( optical vortex ) @xcite . since paraxial lg beams have enhanced gradient forces and reduced radiation pressure , they produce more efficient trapping @xcite . more importantly , since they carry orbital angular momentum @xcite ( see also @xcite for reviews ) , they allow rotational control of the trapped particles . by transferring orbital angular momentum to absorptive particles , such particles were set to rotate @xcite . the photon spin , which is associated with the beam circular polarization , can be employed to add or subtract from the effect of the orbital angular momentum @xcite . in some cases , the particle can be trapped off - axis , near the ring of maximum intensity , and set to rotate about the beam axis @xcite . by using a spatial light modulator , it is possible to scan the orbital index @xmath3 up to values of order @xmath5 and analyze the resulting rotation about the beam symmetry axis as a function of @xmath3 @xcite . we consider a circularly polarized ( paraxial ) laguerre - gaussian @xmath6 ( radial index @xmath7 ) model for the beam incident along the positive @xmath8-axis _ before _ the objective , and then apply the rw approach to calculate the resulting strongly focused beam beyond the objective . ganic , gan and gu calculated the electric energy density on the focal plane for a linearly - polarized lg beam @xcite and showed that the intensity at the focus does not vanish when @xmath9 or @xmath10 . this effect becomes stronger for a circularly polarized incident beam @xcite , with the photon spin anti - parallel to the orbital angular momentum ( the intensity at the focus vanishes when they are parallel ) . the experimental verification of this remarkable effect provided an additional test of the rw approach @xcite . here we present a detailed theoretical analysis of the focused beam , elucidating the spatial variations of field polarization , energy density and poynting vector . we find some unexpected results when the spin and orbital angular momenta of the incident beam are antiparallel . in this situation , the interplay between polarization and angular dependence is particularly interesting , resulting in a strong modification of the focused beam polarization near the beam axis . the discussion of angular momentum is of particular theoretical relevance . the identification of separate spin and orbital contributions to the angular momentum of electromagnetic fields is somewhat controversial @xcite , at least in the sense that the corresponding operators do not satisfy the commutation relations of ` true ' angular momenta in the full quantum theory @xcite . nevertheless , the separation presents itself in a very compelling way in the paraxial approximation , with spin and orbital contributions corresponding to the field polarization and the angular spatial dependence , respectively @xcite . with @xmath11 corresponding to left(right)-handed circular polarization , the spin and orbital angular momenta along the propagation direction per photon are @xmath12 and @xmath13 , respectively . in the framework of the classical ( i.e. non - quantal ) paraxial theory , these important results are inferred from the values of @xmath14 and @xmath15 for the ratios between the linear densities of angular momenta and energy . in the nonparaxial regime the separation is not so straightforward , as far as local quantities are concerned @xcite . however , a natural separation was proposed by barnett @xcite in terms of the overall angular momentum flux along the propagation direction . we calculate the angular momentum flux for the focused nonparaxial beam , and compare the separate spin and orbital contributions , as defined by barnett , with the corresponding values for the paraxial beam before the objective . recently , zhao and co - workers have argued that the focusing effect leads to an interconversion between spin and orbital angular momenta @xcite . we derive quantitative results for the spin and orbital flux modifications , which turn out , however , to have signs in disagreement with zhao _ et al _ s qualitative discussion since their arguments , as well as their experimental demonstration , involve local quantities , we attribute the disagreement to the fact that we calculate the _ global _ overall flux across a plane perpendicular to the propagation direction , rather than local results . the paper is organized as follows . we analyze the local polarization on the focal plane in sec . ii , and discuss the energy and energy flux densities in sec . the optical angular momentum is analyzed in sec . iv , and concluding remarks are presented in sec . v. we assume that the beam waist ( radius @xmath16 ) is precisely at the position of the objective entrance port . the incident circularly polarized electric field is then given by ( the factor @xmath17 is omitted ) @xmath18 the minimum spot size @xmath19 is typically of the order of or larger than the entrance aperture radius ( overfilling ) , so as to take full advantage of the high na of the objective . the focused beam ( focal length @xmath20 ) is written as a superposition of plane waves . according to kirchhoff s approximation of classical diffraction theory @xcite , the amplitude and phase of each plane wave component @xmath21 ( @xmath22 wavenumber in the isotropic medium of refractive index @xmath23 ) are determined by the field at the entrance port @xmath24 @xcite , with : @xmath25 where @xmath26 ( abbe sine condition ) and @xmath27 . we have assumed that the objective transmission amplitude @xmath28 is uniform . radial dependence of objective transmittance , which may be present , can be taken into account by introducing an effective beam waist size @xcite . if the beam is gaussian at the entrance port ( @xmath29 ) , all plane wave components in ( [ focusedfield ] ) have the same phase at the focal position @xmath30 and hence interfere constructively to produce a maximum intensity at this point . on the other hand , for a laguerre - gaussian beam , each component contains the additional phase factor @xmath31 , and then the intensity at the focus vanishes except for some special cases discussed below ( see also refs . @xcite ) . in order to compute the field from ( [ focusedfield ] ) , we need the unit vectors @xmath32 and @xmath33 . they are defined in the plane perpendicular to @xmath34 by the condition that the angles between @xmath32 and @xmath33 and the meridional plane at the angular position @xmath35 are @xmath35 and @xmath36 respectively @xcite . after integration over @xmath35 , we find @xmath37 the dependence on @xmath38 and @xmath8 is contained in the coefficients @xmath39 @xmath40 which can be expressed in terms of the cylindrical bessel functions @xmath41 @xcite : @xmath42 the angular functions @xmath43 are given by @xmath44 the magnetic field is given by ( @xmath45 vacuum magnetic permeability , @xmath46 electric permittivity of dielectric medium ) @xmath47 the expressions above for @xmath48 and @xmath49 are _ exact _ solutions of the maxwell equations and approximate solutions of the boundary conditions corresponding to kirchhoff s classical diffraction theory @xcite . this is in sharp contrast with the usual paraxial models , where approximate solutions of the maxwell equations are employed and diffraction is completely neglected ( the beam waist size is usually assumed to be much smaller than the transverse sizes of the optical elements ) . the electric and magnetic fields are linear combinations of terms of the form @xmath50 . since the phase factor is ill - defined on the beam axis ( phase singularity ) , consistency requires that the functions @xmath51 vanish at @xmath52 , _ except _ when @xmath53 . this important property is verified by replacing the result @xmath54 into ( [ i ] ) . the electric and magnetic field components @xmath55 and @xmath56 have no phase singularity when @xmath57 and @xmath58 ( anti - parallel incident beam orbital and spin angular momenta ) . in this situation , the vectors @xmath48 and @xmath59 oscillate along the beam axis in phase quadrature . there is a second situation with non - vanishing axial intensity : @xmath60 @xmath61 . from ( [ ex ] ) and ( [ ey ] ) , we find , on the axis , @xmath62 and @xmath63 , so that the field on the axis is circularly polarized as in the entrance port . however , the sense of rotation is opposite to that at the entrance port . to provide geometrical insight into the origin of the reversed rotation , we analyze the effects of strong focusing ( @xmath64 ) on vector interference at the focal point . [ pattern ] shows the electric field vectors for the incident paraxial @xmath65 beam on a plane parallel to the @xmath66 plane . the objective maps each of these vectors to the focal point by three - dimensional transport , constrained by the condition that its angle with respect to the plane of incidence ( containing @xmath34 and the axis ) is conserved . thus , the cylindrical azimuthal components ( which are perpendicular to the incidence plane ) are transported parallel to themselves , whereas the cylindrical radial components yield a @xmath8 component proportional to @xmath67 and a component on the @xmath66 plane proportional to @xmath68 . in the following qualitative discussion for @xmath69 we consider a large value of @xmath70 , so that the latter contribution is small . at @xmath71 vectors 3 and 7 in fig . [ pattern ] contribute @xmath72 along the negative @xmath73 direction , whereas 1 and 5 contribute @xmath74 along the positive @xmath73-direction . the contributions to the @xmath8-component from opposite points ( 1 and 5 , 2 and 6 , etc ) are exactly cancelled . the diference between @xmath75 corresponding to figs . 1(a ) and 1(b ) , lies in the contributions of vectors 2 , 4 , 6 and 8 . when adding their azimuthal components , we find a vector pointing along the positive @xmath73-direction in case ( a ) , and then the field resulting from interference with 3 and 7 vanishes . on the other hand , they add to the contributions of 3 and 7 in case ( b ) , and the resulting electric field , containing a factor @xmath76 in agreement with ( [ ex ] ) , ( [ ey ] ) and ( [ f2 ] ) , points along the negative @xmath73-direction at @xmath77 moreover , while the electric field at each spatial position on the incident beam rotates clockwise in fig . [ pattern]b , the resulting overall pattern rotates counter - clockwise , with the same angular frequency . thus , the resulting vector sum at the focal point also rotates counter - clockwise . beam ( before the objective ) , with ( a ) left - handed ( @xmath78 ) and ( b ) right - handed ( @xmath79 ) circular polarizations , at times @xmath80 @xmath81 and @xmath82 the resulting spatial patterns counter - rotate with respect to the circular polarization of the incident beam.,height=245 ] these remarkable effects disappear in the paraxial focusing limit , which may be obtained from our more general results by assuming that @xmath83 ( small na ) @xcite . in this limit , the angular functions given by ( [ f0])-([f2 ] ) satisfy @xmath84 . then , we may neglect @xmath85 and @xmath86 in ( [ ex])-([ez ] ) , yielding @xmath87 : the polarization of the incident beam is preserved in this case . moreover , the intensity along the beam axis is very small even if @xmath88 or if @xmath89 ( specially in the second case ) . polarization thus plays a minor role in the paraxial limit . on the other hand , there is a nontrivial interplay between polarization and the field spatial dependence in the nonparaxial regime . to fully picture the polarization of the focused beam , it is convenient to derive the cylindrical field components @xcite from ( [ ex ] ) and ( [ ey ] ) : @xmath90 thus , all cylindrical components , including @xmath55 given by ( [ ez ] ) , have the same dependence on @xmath91 , determined by the total axial incident beam angular momentum @xmath92 . note , however , that the electric and magnetic fields depend separately on the spin and orbital indexes @xmath4 and @xmath3 , since their dependence on @xmath38 and @xmath8 is determined by the functions @xmath93 [ eq . ( [ i ] ) ] . these functions obey the important symmetry relation @xmath94 it follows that , when we reverse _ both _ incident spin and orbital indexes ( @xmath95 ) the electric field changes as @xmath96 whereas corresponding results for the magnetic field have the opposite sign due to multiplication by @xmath4 in eq . ( [ h ] ) . these transformation rules allow us to extend to negative values of @xmath3 results from the discussion for @xmath97 , presented below . for @xmath98 @xmath79 . along the circle @xmath99 , the polarization ellipse is perpendicular to the focal plane ( as illustrated at point @xmath100 ) . inside the circle , @xmath101 the projection of the electric field onto the @xmath66 plane rotates counterclockwise , so that the rotation is reversed with respect to the polarization of the paraxial incident beam . at the focal point @xmath102 itself , the polarization is circular . , width=340 ] on the focal plane @xmath103 all functions @xmath104 are real according to ( [ i ] ) . then , from ( [ ez ] ) and ( [ erho])-([ephi ] ) we conclude that @xmath105 and @xmath55 are in phase or in phase opposition depending on the sign of @xmath106 , whereas @xmath107 is in phase quadrature . the polarization ellipse is at an angle @xmath108 $ ] with respect to the focal plane , with its principal axis along the @xmath109 direction . this is illustrated at point @xmath110 in fig . [ fig - pola ] , for @xmath98 @xmath111 for a detailed picture of the polarization in this interesting case , we plot in fig . [ amplitudes ] the field amplitude cylindrical components as functions of @xmath112 the amplitudes are divided by the maximum incident electric field amplitude at the entrance port , obtained from eq . ( [ port ] ) on the circle of maximum intensity @xmath113 @xcite . these dimensionless amplitudes are given by @xmath114 the pre - factor @xmath115 in ( [ etilderho])-([etildephi ] ) arises from the focused beam intensity enhancement by the geometric factor @xmath116 which is of order of the ratio between the beam transverse areas at the entrance port and at the focal plane @xcite , as expected by energy conservation . for the numerical example shown in fig . [ amplitudes ] , we take na@xmath117 , @xmath118 , @xmath119 , @xmath120 , @xmath121 and @xmath122 . as expected , the field is localized in a region around the focus , with dimensions of the order of the wavelength @xmath123 . when the amplitudes @xmath124 and @xmath125 have the same sign , the projection of the electric field on the @xmath66 plane rotates clockwise , like the incident beam . this happens in most spatial regions , in particular around the ring of maximum intensity ( to be discussed in the next section ) . on the circle @xmath126 ( for the above numerical values ) , we have @xmath127 , so that the polarization ellipse is perpendicular to the @xmath66 plane ( @xmath128 ) . this is represented at point @xmath100 in figure [ fig - pola ] . inside the disk ( @xmath129 ) , @xmath124 and @xmath125 have opposite signs ; thus , the projection of @xmath48 on the @xmath66 plane now rotates counter - clockwise . at the focal point ( @xmath52 ) , the polarization is circular with reversed rotation as already explained . these properties are illustrated by fig . [ cpoeletrico ] , showing the projection of the electric field vectors onto the focal plane at different times . , with @xmath98 @xmath130 na@xmath117 , @xmath118 , @xmath120 , @xmath119 , @xmath121 and @xmath122 . the vertical dotted line indicates the circle @xmath99.,width=359 ] ( b ) @xmath131 , ( c ) @xmath132 and ( d ) @xmath133 we take @xmath98 @xmath111 the red ( gray ) line indicates the circle @xmath99 . the electric field at a given spatial position outside this circle rotates clockwise like the incident beam . inside the circle , however , the rotation is counterclockwise , as illustrated by the vector at the focal point . , width=302 ] the same reversion takes place for @xmath134 and the radius @xmath135 increases linearly with @xmath3 . however , there is a phase singularity on the beam axis in these cases , so that the field vanishes at @xmath52 instead of being circularly polarized . on the other hand , no reversion occurs when @xmath3 and @xmath4 have the same sign , nor with @xmath57 . thus , we may interpret the reversion as a nontrivial interplay between polarization and spatial structure , as already discussed in connection with fig . [ pattern ] . near the beam axis ( @xmath136 ) , for @xmath2 , the spatial angular dependence dominates over the polarization inherited from the incident beam , leading to a reversed rotation when @xmath3 and @xmath4 have opposite signs . in the next section , we show that the direction of energy flow near the axis also gets _ inverted _ with respect to the overall propagation direction ( defined by the paraxial incident beam ) when @xmath3 and @xmath4 have opposite signs and @xmath137 the ( time - averaged ) electric energy density @xmath138 is symmetrical with respect to the focal plane @xmath139 @xmath140 independent of @xmath91 ( rotational symmetry ) and identical to the magnetic energy density at each point : @xmath141 ( these last two properties would not hold for linear polarization of the incident beam ) . from ( [ ez ] ) , ( [ erho ] ) and ( [ ephi ] ) , we find for the total energy density @xmath142 @xmath143 from ( [ transf1 ] ) or ( [ transf2])-([transf4 ] ) , it is clear that the energy density is invariant under change of signs of both @xmath3 and @xmath144 thus , results for negative values of @xmath3 are easily obtained from those for @xmath145 ( discussed below ) by simply inverting @xmath144 in fig . [ energy ] , we plot the ratio @xmath146 as a function of @xmath147 , for the same parameters as in fig . [ amplitudes ] ( @xmath148 represents the maximum density of the incident paraxial beam ) . we also show , for comparison , corresponding results for @xmath10 , @xmath149 , with vanishing density at the axis . although they have the same spatial profile at the objective entrance port , the strong interplay with polarization results in different distributions on the focal plane . for @xmath10 , @xmath79 , the bright ring is thicker and slightly displaced inwards ( see also the insets in fig . [ energy ] ) . the density value at the focus for @xmath79 is only @xmath150 of its maximum on the circle @xmath151 , but it is very large in comparison with the incident beam energy density : @xmath152 this arises partly from focusing the available energy to a much tighter region , with an area of the order of @xmath153 for the large na objective considered in this example . but increasing na has a stronger effect on the focal energy density for this particular case ( and also for @xmath88 ) . this is shown in fig . [ na ] , where we plot the ratio @xmath154 as a function of na for @xmath89 . we also show the results for @xmath29 : in this case the effect of increasing na is less dramatic because the axial density does not vanish in the paraxial limit . versus @xmath147 at the focal plane ( same parameters as in fig . [ amplitudes ] ) . , width=359 ] at the focal point versus na . same parameters as in fig . [ amplitudes].,width=359 ] for all @xmath155 the peak value of the energy density on the focal plane is reached at a radius that increases with @xmath156 for a paraxial laguerre - gaussian beam , this radius grows as @xmath157 ( for given values of @xmath19 and @xmath123 ) . [ energy ] shows that the radius @xmath158 of a tightly focused beam also depends on polarization . remarkably , for a given polarization it grows linearly with @xmath3 for large @xmath3 , as shown experimentally by curtis and grier from the rotational dynamics of optically trapped particles @xcite . in fig . [ rmax ] , we show that the beam radius is in fact well fitted by a linear function for @xmath159 , with @xmath149 . the result of a linear fit for @xmath160 is also shown ( @xmath161 ) . , width=302 ] the poynting vector @xmath162 yields the energy flux density at position @xmath163 . from ( [ ez ] ) and ( [ erho])-([ephi ] ) we find @xmath164 \\ \label{sphi } s_{\phi } & = & 2\sqrt{\frac{\epsilon}{\mu_0}}\,e_0 ^ 2\;{\rm re}\left [ \left ( i^{(\ell)}_{0 } + i^{(\ell)}_{2\sigma } \right ) i^{(\ell)}_{\sigma}{}^ * \right ] \\ \label{sz } s_{z } & = & \sqrt{\frac{\epsilon}{\mu_0}}\,e_0 ^ 2\,\left ( as expected by symmetry , the cylindrical components do not depend on @xmath165 since the functions @xmath85 are real at the focal plane @xmath103 we get @xmath166 on this plane . this is also expected , since @xmath167 must change sign as the beam converges and then diverges from the focal plane . the polarization ellipses represented in fig . [ fig - pola ] already provide the directions of the poynting vector field at the focal plane : on each point , @xmath168 is perpendicular to the corresponding ellipse . when @xmath3 and @xmath4 have opposite signs and @xmath2 , eq . ( [ sz ] ) yields @xmath169 at @xmath99 ( see fig . [ fig - pola ] ) , so that the poynting vector is parallel to @xmath170 on this circle ( or anti - parallel if @xmath171 ) . moreover , @xmath172 is _ negative _ inside the disk @xmath101 so that the local energy current density flux near the axis is antiparallel to the incident beam direction . note that the sense of the electric field rotation with respect to the local energy flux direction is everywhere the same , because within the disk @xmath129 both are reversed . three - dimensional insight into the beam energy flow is provided by fig . the projection of the poynting vector on the focal plane is represented by arrows , while its axial component is represented in magnitude and sign by the false color map , for @xmath173 ( a ) and @xmath174 ( b ) . the axial sign reversal within the circle @xmath175 is apparent . and ( a ) @xmath79 , ( b ) @xmath149 ( same parameters as in fig . [ amplitudes ] ) . the values of @xmath176 are represented by the false color map ( @xmath177 maximum poynting vector of the incident beam ) . in case ( a ) , @xmath172 is negative within the circle @xmath178 . , width=283 ] the global energy flux is of course directed along the positive @xmath8 axis . using ( [ i ] ) , ( [ sz ] ) and the result @xmath179 we find that the global energy flux across a plane corresponding to some fixed value of @xmath8 is independent of @xmath8 ( as expected by energy conservation ) and given by @xmath180 we may check this result against the incident energy flux at the objective entrance port ( radius @xmath181 ) : @xmath182 from ( [ port ] ) , ( [ e0 ] ) and ( [ energy_flux_focal ] ) , we find @xmath183 as expected . as a final remark about the poynting vector , we note the transformation rules under inversion of both spin and orbital indices ( @xmath95 ) obtained from ( [ transf1 ] ) and ( [ srho])-([sz ] ) . we find that @xmath184 and @xmath185 are invariant whereas @xmath186 this sign change is expected since @xmath187 is directly related to the optical angular momentum as discussed in the next section . from the minkowski linear pseudo - momentum density @xcite @xmath188 we get the pseudo - angular momentum density @xcite @xmath189 which may be numerically calculated from ( [ i ] ) and ( [ sphi ] ) . in fig . [ angularmomentum ] , we plot the dimensionless density ( @xmath190 energy density at the focus for @xmath29 ) @xmath191 as a function of @xmath147 for several values of @xmath3 and @xmath144 according to eq . ( [ tsphi ] ) , negative values of @xmath3 for a given polarization @xmath4 are be obtained simply by changing the sign of @xmath192 calculated for @xmath193 on the focal plane versus @xmath147 ( same parameters as in fig . [ amplitudes]).,width=264 ] the results shown in fig . [ angularmomentum ] are qualitatively similar to those for a paraxial beam , with @xmath123 playing the role of the paraxial beam waist @xmath19 as a transverse length scale for our diffraction - limited beam . for a paraxial beam , the orbital contribution is proportional to @xmath194 , whereas the spin is proportional to @xmath195 @xcite . for @xmath130 fig . [ angularmomentum ] shows that the pseudo - angular momentum density is in fact negative near the outer edge of the ring of maximum intensity ( @xmath196 ) as in the paraxial case . closer to the beam axis , on the other hand , the density is always positive for @xmath145 , so that one might be tempted to conclude that the orbital contribution dominates near the beam axis . however , the identification of spin and orbital contributions is not straightforward ( see refs . @xcite and @xcite for a general discussion ) , particularly in the nonparaxial regime , as shown by barnett and allen @xcite . they took a general nonparaxial cylindrically symmetric beam containing a phase factor @xmath197 . besides the ambiguity in the identification of spin and orbital contributions , the ratio between the total angular momentum per unit length and the total energy per unit length was found not to be @xmath198 as in the case of a paraxial beam . the nonparaxial focused beam model , as derived here , does not coincide with the barnett - allen model . the reason is that the angular position phase factor is not simply @xmath199 the cartesian electric field components contain instead all three phase factors @xmath197 , @xmath200 and @xmath201 according to ( [ ex])-([ez ] ) . nevertheless , we find the same ambiguity concerning the separation between spin and orbital contributions to the pseudo - angular momentum density in the dielectric medium , and no simple relation between pseudo - angular momentum and energy per unit length . a more convenient description of optical angular momentum flow outside the paraxial regime is provided by the angular momentum current flux density , represented by a second order tensor @xmath202 @xcite . with the pseudo - angular momentum density given by ( [ jz ] ) in terms of the minkowski linear pseudo - momentum @xcite , the flux density of the axial pseudo - angular momentum component along the @xmath8 direction is @xmath203.\ ] ] it is useful to write @xmath55 and @xmath56 in ( [ mzz ] ) in terms of the curl of @xmath59 and @xmath204 by using the maxwell equations . the factor @xmath205 appearing in ( [ mzz ] ) is canceled , and the resulting expression is further simplified by using eq . ( [ h ] ) to write the magnetic field in terms of the electric field . barnett has shown that the total angular momentum flux @xmath206 through a plane parallel to the @xmath66 plane can be separated into two gauge - independent contributions respectively associated to the polarization and to the spatial field structure @xcite . these two contributions provide natural definitions for spin and orbital angular momentum fluxes , respectively . after integration by parts , one finds : @xmath207 @xmath208 @xmath209 since the component @xmath210 already contains a factor @xmath4 according to ( [ port ] ) and ( [ ey ] ) , the pre - factor @xmath4 in the expression on the r .- h .- s . of ( [ morbpre ] ) is canceled as expected . if one applies ( [ mspinpre ] ) and ( [ morbpre ] ) to a paraxial lg beam , one finds , after division by the energy flux @xmath211 , @xmath212 and @xmath213 thus , the pseudo - angular momentum per photon in the dielectric medium does not depend on @xmath205 and is exactly the same as in vacuum , as already verified experimentally @xcite . we now apply ( [ mspinpre ] ) and ( [ morbpre ] ) to the nonparaxial focused beam . by writing the electric field components in terms of @xmath214 as in ( [ ex ] ) and ( [ ey ] ) , we find @xmath215\end{aligned}\ ] ] using the definition ( [ i ] ) and the result ( [ jj ] ) , it is straightforward to show that both @xmath216 and @xmath217 are independent of @xmath218 so that the spin and orbital fluxes are independently conserved as the focused beam propagates in the region beyond the objective . however , their values do not separately coincide with the spin and orbital fluxes per unit power for the incident paraxial beam before the objective . because of the angular dependence @xmath201 of both @xmath219 and @xmath210 [ cf . ( [ ex ] ) and ( [ ey ] ) ] , the angular derivative contained in the orbital flux in ( [ morbpre ] ) leads to the term proportional to @xmath220 in ( [ morb ] ) . only the total pseudo - angular momentum flux per unit power is conserved by focusing , as expected given the cylindrical symmetry of the objective . in fact , when adding the spin and orbital fluxes given by ( [ mspin ] ) and ( [ morb ] ) , we find , by comparison with eq . ( [ sz ] ) , @xmath221 the relative change of the spin flux per unit power with respect to the incident beam is @xmath222 whereas the relative change of orbital flux per unit power is simply @xmath223 from ( [ jj ] ) , ( [ mspin ] ) and ( [ morb ] ) , we find @xmath224 where @xmath225 by inspection of ( [ r ] ) and ( [ g ] ) , we conclude that @xmath226 is positive and does not depend on the sign of @xmath3 nor on the value of @xmath144 hence the variation of @xmath227 has the sign of @xmath4 and is such that its absolute value always increases . at first sight , values for @xmath227 outside the range from @xmath228 to @xmath229 might seem to contradict the fact that the photon has spin one . note , however , that the quantities @xmath216 and @xmath230 represent overall net flux balances of the local spin angular momentum and energy across a plane of constant @xmath218 in a situation where the polarization and the poynting vector vary strongly from point to point . as an illustration , consider the case @xmath98 @xmath79 discussed at length in sec . ii . as shown by fig . 1 , the electric field rotates counterclockwise ( as seen from @xmath231 ) near the axis , and , as discussed in sec . iii , the energy locally flows along the negative @xmath8-direction . a picture in terms of ray contributions then goes as follows : the flux of rays @xmath232 crossing the plane far from the axis contributes @xmath233 to @xmath230 and @xmath234 to @xmath235 on the other hand , rays crossing near the axis ( flux @xmath236 ) provide a negative contribution @xmath237 to @xmath238 as well as a _ negative _ contribution @xmath239 to @xmath216 because they propagate along the negative @xmath8 direction with a positive @xmath8 component of spin angular momentum ( the crucial point here is that the angular momentum is a vector quantity whereas the energy is a scalar ) . the net overall ratio is then @xmath240 the interplay between spin and orbital fluxes is clearly a nonparaxial effect which becomes enhanced as @xmath241 increases , because larger values of @xmath241 reduce the contribution of paraxial angles @xmath242 in ( [ r ] ) . the integrands in ( [ r ] ) are equal at @xmath243 , which corresponds to a numerical aperture na@xmath244 for @xmath245 typical high - na values @xmath246 and @xmath247 are such that for all values of @xmath70 in ( [ r ] ) the integrand in the numerator is considerably smaller than that in the denominator , leading to small values for the relative change @xmath226 . in table i , we show the values of @xmath226 for na@xmath248 and @xmath247 , for different values of @xmath156 the orbital flux per unit power @xmath249 increases ( decreases ) when @xmath250 this seems to be in contradiction with the concept of ` conversion ' between spin and orbital angular momenta @xcite . note , however , that the connection between the global angular momentum flux calculated here and the mechanical effects on local probe particles is not straightforward , particularly when local quantities vary strongly , as discussed in the previous paragraph . to our knowledge , a consistent identification of spin and orbital contributions to the _ local _ angular momentum or flux densities of nonparaxial beams is not available @xcite @xcite ( this is the reason why we employ the term ` interplay ' rather than ` exchange ' or ` interconversion ' ) . .[table]relative spin flux change @xmath226 for different values of @xmath156 additional parameters as in fig . 3 . [ cols="^,^,^",options="header " , ] the diffraction - limited optical beam that results from focusing a circularly - polarized lg beam by a high - na objective has some remarkable properties , particularly when @xmath3 and @xmath4 have opposite signs . in this situation , the interplay between polarization and angular spatial dependence leads to a strong modification of the local polarization and of the energy propagation direction in the region near the beam axis : the energy locally propagates along the negative @xmath8-direction near the focal point for any @xmath137 also , the rotation of the electric field projection onto the focal plane is reversed ; e.g. , when @xmath251 the polarization at the focal point is circular with the reversed rotation . all these effects disappear in the paraxial limit , which can be obtained from our results by taking low values of na or a small waist size at the entrance port , thereby underfilling the objective aperture ( @xmath252 ) . the total angular momentum flux per unit power across a plane perpendicular to the @xmath8-axis is conserved by the focusing effect , but not the separate spin and orbital contributions . on the other hand , they are separately conserved as the beam freely propagates beyond the objective . we have presented quantitative results for the modification of the spin and orbital fluxes . even for the highest possible values of na , the relative spin flux modification is typically small , but it grows with @xmath156 interplay between light orbital and spin angular momentum is an unfamiliar concept . geometrical insight concerning this effect was provided in connection with fig . [ pattern ] ( cf . @xcite ) : in the paraxial case , spin has to do with the change in rotational orientation of the electric field vector at a given point , whereas orbital angular momentum is associated to its change around a circle at a given time . when focusing a circularly - polarized lg beam , the interesting patterns emerging from the combination of these two concepts give rise to surprising results near the beam axis of the nonparaxial beam . from a physical point of view , this may be regarded as a vectorial interference effect among plane waves in the angular spectrum produced by diffraction at the objective . an analogy is provided by a ( te ) or ( tm ) mode in a parallel - plate waveguide : it may be described in terms of multiple reflections of ordinary ( tem ) plane waves against the walls ( boundary effects ) , but vectorial interference leads to the appearance of longitudinal field components . the results presented in this paper have possible applications to optical trapping . a calculation of the optical force and torque on dielectric spheres along the lines presented here is currently under way . we thank nathan viana for useful discussions . this work was supported by the brazilian agencies cnpq , faperj , fujb and institutos do milnio de nanocincias e de informao quntica . p. a. maia neto and h. m. nussenzveig , europhys . lett . * 50 * , 702 ( 2000 ) ; a. mazolli , p. a. maia neto , and h. m. nussenzveig , proc . london , ser . a * 459 * , 3021 ( 2003 ) ; r. s. dutra , n. b. viana , p. a. maia neto and h. m. nussenzveig , j. opt . a : pure appl * 9 * , s221 ( 2007 ) . for a linearly polarized beam at the objective entrance , the polarization of each plane wave component follows from the condition that the angle between the polarization direction and the incidence plane at each interface in the objective region is conserved @xcite . the result for circular polarization follows by linearity .
we investigate in detail the focusing of a circularly polarized laguerre - gaussian laser beam ( @xmath0 orbital angular momentum per photon ; @xmath1 for left / right - handed polarization ) by a high numerical aperture objective . the diffraction - limited focused beam has unexpected properties , resulting from a strong interplay between the angular spatial structure and the local polarization in the non - paraxial regime . in the region near the beam axis , and provided that @xmath2 and @xmath3 and @xmath4 have opposite signs , the energy locally counter - propagates and the projection of the electric field onto the focal plane counter - rotates with respect to the circular polarization of the incident beam . we explicitly show that the total angular momentum flux per unit power is conserved after focusing , as expected by rotational symmetry , but the spin and orbital separate contributions change .
supermassive black hole growth and galaxy growth have long been thought to be correlated , as supported by the magorrian and m@xmath3 relations @xcite . both active galactic nuclei ( agn ) activity and star formation peak at @xmath4 ( e.g. , @xcite ) , and the processes appear to be correlated in both agn host and normal galaxy samples ( e.g. , @xcite ) . however , the connection between star formation activity and black hole mass accretion in galaxies is not well - understood . several studies suggest that agn activity is needed to explain the cutoff at the bright end of the galaxy luminosity function and the halting of star formation in the most massive galaxies ( @xcite and references therein ) . this theory is supported by the observation that agn activity lags star formation by @xmath5 years @xcite , and that many agns lie in the ` green valley ' ( e.g. , @xcite) the sparsely populated region between the red sequence and the blue cloud @xcite . however , @xcite find that out to @xmath6 agns are preferentially hosted by star - forming , rather than quiescent or quenching galaxies . additionally , dust attenuation may redden the color of many agn host galaxies ( e.g. , @xcite , @xcite,@xcite ) , implying that star formation quenching by agns may not control the transition between blue and red galaxies . instead , the connection between agn and star formation activity may simply reflect the fact that both processes require a large gas supply , with possible delays between star formation and agn activity caused by inefficient black hole fueling . for example , simulations by , suggest that star formation peaks as a function of gas density , while torques regulating further inflow are relatively inefficient and limit black hole growth , leaving a reservoir of gas long after the gas supply for star formation is exhausted . while it is difficult to disentangle the contributions of star formation and agn activity of high redshift galaxies from optical observations alone , x - ray imaging acts as a better tracer of agn activity , with unobscured agns emitting heavily in both hard and soft bands , and obscured agn dominated by hard x - ray emission ( e.g. @xcite ) . galactic x - ray emission also traces stellar activity , through low mass x - ray binaries ( lmxbs ) , whose primarily soft x - ray emission correlates with stellar mass , as well as high mass x - ray binaries ( hmxbs ) and supernova , whose harder emission is reflective of star formation . soft x - ray emission from hot gas is also apparent in the halos of massive ellipticals ( e.g. , @xcite ) . = 3.0 in = 3.0 in = 3.0 in = 4.2 in the x - ray galaxy connection has been systematically studied at low redshift , providing a rough census of x - ray luminosity with stellar mass , galaxy type , and redshift ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? for example , @xcite find that the agn fraction for massive early type galaxies increases with redshift , while the x - ray emission in less massive galaxies evolves very little out to @xmath7 . late - type galaxies evolve significantly in x - ray luminosity with redshift , and furthermore their x - ray emission shows a strong dependence on stellar mass @xcite . at higher redshifts , x - ray galaxy studies have primarily focused on direct x - ray detections ( e.g. , @xcite ) or on specific galaxy populations , such as lyman break galaxies ( lbgs , steidel et al . 1996 ) , submillimeter bright galaxies ( smgs ) , and distant red galaxies ( drgs , franx et al . for example , @xcite find that the x - ray emission is dominated by an agn in only 15% of the smgs out to @xmath8 in the chandra deep field north ( see also * ? ? ? * ) , while for lbgs and drgs the x - ray emission seems almost completely accounted for by star formation ( e.g. , reddy & steidel 2004 ; rubin et al . however , these different selection techniques target galaxies at different redshifts and stellar masses , and even combined are not representative of the full distant galaxy population . hence , a complete census of the x - ray properties of distant galaxies is still missing . using the high - quality near - infrared photometry from the newfirm medium band survey ( nmbs,@xcite ) in combination with chandra data , it is possible to systemically study the x - ray properties of an observed k - band complete sample of distant galaxies . in this study , we utilize the @xmath1 galaxy from the nmbs and 32 sed templates of @xcite , along with the overlapping c - cosmos field , to better characterize the star formation - agn connection out to @xmath4 . with accurate @xmath9 and @xmath10 measurements from @xcite , indicating instantaneous star formation rates and stellar population ages , along with separate hard and soft band x - ray luminosities , we attempt to correlate agn activity with star formation properties , stellar mass , and redshift . cccccccc 1 & 121 & 1.95 & 14.4 & -11.79 & 0.4 & @xmath11 & @xmath12 + [ 1.5ex ] 2 & 291 & 1.94 & 14.4 & -11.79 & 0.2 & @xmath13 & @xmath14 + [ 1.5ex ] 3 & 436 & 1.81 & 6.2 & -24.93 & 0.5 & @xmath15 & @xmath16 + [ 1.5ex ] 4 & 211 & 1.77 & 2.1 & -18.35 & 0.3 & @xmath17 & @xmath18 + [ 1.5ex ] 5 & 95 & 1.70 & 24.6 & -10.59 & 0.7 & @xmath19 & @xmath20 + [ 1.5ex ] 6 & 22 & 1.67 & 19.1 & -11.67 & 0.1 & @xmath21 & @xmath22 + [ 1.5ex ] 7 & 72 & 1.66 & 20.5 & -11.29 & 0.2 & @xmath23 & @xmath24 + [ 1.5ex ] 8 & 64 & 1.66 & 38.7 & -10.49 & 1.2 & @xmath25 & @xmath26 + [ 1.5ex ] 9 & 94 & 1.65 & 42.7 & -10.59 & 1.1 & @xmath27 & @xmath15 + [ 1.5ex ] 10 & 60 & 1.62 & 33.1 & -10.58 & 0.5 & @xmath28 & @xmath17 + [ 1.5ex ] 11 & 95 & 1.58 & 39.1 & -10.4 & 0.7 & @xmath23 & @xmath29 + [ 1.5ex ] 12 & 46 & 1.58 & 17.7 & -10.58 & 1.4 & @xmath30 & @xmath31 + [ 1.5ex ] 13 & 114 & 1.45 & 63.3 & -10.71 & 2.4 & @xmath32 & @xmath33 + [ 1.5ex ] 14 & 71 & 1.45 & 69.4 & -9.66 & 2.6 & @xmath34 & @xmath35 + [ 1.5ex ] 15 & 64 & 1.40 & 61.7 & -10.34 & 2.8 & @xmath17 & @xmath18 + [ 1.5ex ] 16 & 155 & 1.40 & 67.4 & -9.92 & 2.1 & @xmath34 & @xmath36 + [ 1.5ex ] 17 & 103 & 1.40 & 66.2 & -9.66 & 1.9 & @xmath19 & @xmath37 + [ 1.5ex ] 18 & 95 & 1.38 & 86.0 & -9.34 & 2.3 & @xmath25 & @xmath38 + [ 1.5ex ] 19 & 33 & 1.37 & 86.1 & -9.66 & 2.6 & @xmath25 & @xmath39 + [ 1.5ex ] 20 & 34 & 1.37 & 108.0 & -8.81 & 2.6 & @xmath40 & @xmath41 + [ 1.5ex ] 21 & 139 & 1.35 & 79.5 & -9.15 & 1.9 & @xmath25 & @xmath29 + [ 1.5ex ] 22 & 82 & 1.34 & 66.9 & -10.24 & 1.7 & @xmath42 & @xmath43 + [ 1.5ex ] 23 & 51 & 1.33 & 92.4 & -8.69 & 2.2 & @xmath44 & @xmath45 + [ 1.5ex ] 24 & 26 & 1.30 & 130.7 & -9.25 & 2.1 & @xmath46 & @xmath42 + [ 1.5ex ] 25 & 159 & 1.30 & 94.1 & -10.71 & 1.4 & @xmath47 & @xmath48 + [ 1.5ex ] 26 & 80 & 1.30 & 107.1 & -8.81 & 1.9 & @xmath49 & @xmath50 + [ 1.5ex ] 27 & 108 & 1.27 & 105.8 & -8.78 & 1.6 & @xmath51 & @xmath52 + [ 1.5ex ] 28 & 144 & 1.25 & 106.2 & -9.92 & 1.2 & @xmath32 & @xmath21 + [ 1.5ex ] 29 & 64 & 1.22 & 102.8 & -8.61 & 1.4 & @xmath53 & @xmath54 + [ 1.5ex ] 30 & 142 & 1.21 & 104.0 & -8.65 & 1.2 & @xmath55 & @xmath56 + [ 1.5ex ] 31 & 101 & 1.18 & 109.7 & -8.65 & 1.0 & @xmath57 & @xmath58 + [ 1.5ex ] 32 & 50 & 1.13 & 138.5 & -8.51 & 0.9 & @xmath59 & @xmath60 + [ 1.5ex ] [ tab : sed_props ] in this work , we make use of the photometric catalogs from the nmbs in the cosmos field @xcite . the nmbs uses five custom nir filters ( three j - band and two h - band ) , covers 0.2 deg@xmath61 of the cosmos field @xcite , and is complemented by fuv - mir data sets ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the photometric redshifts and stellar population properties are derived using eazy @xcite and fast @xcite . we utilize the nmbs sample of @xmath62 galaxies at @xmath2 of @xcite ( with mass - redshift distribution displayed in fig . [ fig : mz_dist ] ) , and employ the spectral classification therein . in summary , the galaxies were divided into 32 subsamples based on similarities in their spectral energy distributions ( seds ) . first , the rest - frame photometric seds of all individual galaxies were compared , and based on the similarities in their full sed shape , it was decided whether or not two galaxies were analogs . next , the galaxy with most analogs was identified , which together formed the first subsample . after removal of all galaxies of the first subsample , the next galaxy with most analogs was identified . this procedure was repeated until nearly all galaxies were divided . eventually , some galaxies were re - assigned , if they would have been a better match with a later iteration . thus , our classification method is model independent . = 6.5 in = 3.25 in = 3.0 in = 3.0 in = 3.0 in = 3.0 in a composite sed is constructed for each subsample by combining the rest - frame photometry of all galaxies in the bin . each subsample contains 22 - 436 galaxies for which there is x - ray coverage as well . the final composite sed sample spans almost the entire galaxy population at these redshifts . each composite sed shows detailed spectral features such as the h@xmath63+[nii]+[sii ] and h@xmath64+[oiii ] emission lines , balmer or 4000 breaks , mgii absorption at 2800 , the continuum break at 2640 , and the dust absorption feature at 2175 ( see fig . 4 of @xcite , and three examples in fig . [ fig : stacks ] ) . as the sample is k - band ( and not stellar mass ) selected , it will be slightly biased toward star - forming galaxies with lower @xmath65 ratios . furthermore , as our photometric redshift code only uses stellar population templates , we have a bias against agn - dominated seds . although @xmath10 measures stellar population age ( as stellar opacity increases with age and strengthens absorption lines of ionized metals such as ca@xmath66 h and k ) , rather than ongoing star formation activity , @xcite show a clear correlation between @xmath10 , @xmath9 , and the specific star formation rate ( ssfr ) . galaxies with shallow @xmath67 breaks have high @xmath9 and are fitted by stellar population models with high ssfrs , while galaxies with strong breaks have low values of @xmath9 and low ssfrs . basic properties of each sed type are shown in table [ tab : sed_props ] . in this paper , we study the x - ray properties of the different sed types . we make use of the c - cosmos survey @xcite , which has an effective exposure of 80 - 160ks , and a sampling of 1/pixel . merged images of 0.5 - 2 kev ( soft ) , 2 - 7 kev ( hard ) , and 0.5 - 7 kev ( full ) are publicly available , along with the point source catalog of @xcite . each field of the c - cosmos survey is observed with up to six overlapping pointings , ensuring relatively uniform sensitivity , despite differing point - spread functions ( @xcite ) . we obtain 60x60 c - cosmos cutouts of each galaxy field from @xcite . as a vast majority of the galaxies are not detected in the x - ray , we opt to stack by sed type , and further subdivide by redshift and/or stellar mass . we note that direct detections are included in the stacks and thus the further analysis . we select source counts within a 5 radius of the nir detection , and make background estimates by taking 30 random 5 radius apertures in each frame , excluding 5 apertures around the optical source or known x - ray sources @xcite . we then de - redshift the source according to its photometric redshift ( or spectroscopic redshift when available ) , using a photon index of @xmath68 , to determine a source luminosity , via the equation @xmath69 where @xmath70 is the observed frame emission in the @xmath71 bandpass , and @xmath72 is the luminosity distance . we explore alternate photon indices , as the steepness of galaxy x - ray spectra varies according to the strength of the various types of x - ray sources , and find that the observed trends in x - ray luminosity do not change significantly with differing @xmath73 ( see [ sec:3.2 ] for further discussion ) . we then determine the average luminosity of all sources per bin . to determine the error in each stack , we bootstrap resample the sources , subtracting a random background aperture ( of the selected 30 ) from each frame . the error is the standard deviation of the average stack luminosities of the bootstrap simulations . stacked detections of less than @xmath74 ( as determined by the bootstrapped background estimates ) were deemed insignificant . sample stacks for three sed types of full - band x - ray data are shown in fig . [ fig : stacks ] in this section , we explore x - ray properties of the galaxy population at @xmath2 . to determine the dependence of x - ray luminosity on galaxy type , we seek to disentangle the effects of redshift and stellar mass . we attempt to remove the effects of mass and redshift by splitting galaxies into bins , and eventually normalizing by stellar mass ( [ sec:3.1 ] ) . in [ sec:3.2 ] , we look at x - ray hardness as well as expected xrb contribution to galaxy luminosity in an attempt to distinguish the sources of x - ray flux . as each sed type probes a different mass distribution , we first explore the x - ray luminosity as a function of stellar mass by splitting each sub sample into three mass bins ( @xmath75 , and @xmath76 ) in the left panel of fig . [ fig : luminm ] . the x - ray luminosity is correlated with stellar mass , such that more massive galaxies emit at higher rates regardless of stellar age . we determine a least squares fit to the data via resampling , assuming a normal distribution with standard deviation as calculated by the bootstrap errors for each data point . we calculate the line of best fit for each resampling , weighing individual points by the number of galaxies in each stack , and determine the average and standard deviations of the fits to be @xmath77 . to further examine this relation we stack all galaxies in 10 bins of stellar mass , and show these as the large gray filled circles in the right panel of right panel of fig . [ fig : luminm ] , with errors calculated by resampling the galaxies in each bin . we further divide these 10 mass bins into 3 redshift bins ( colored ) , and find that higher redshift galaxies have higher x - ray luminosities at the same mass , when compared to lower redshift galaxies . at @xmath78 there is a clear correlation between the x - ray luminosity and stellar mass , which is broadly consistent with a one - to - one relation . the x - ray luminosity is increasing with redshift , and the trend with mass seems to flatten . as we probe different mass ranges at different redshifts , the increase of x - ray luminosity with redshift further steepens the correlation between x - ray luminosity and stellar mass . thus , to further examine the trend with redshift and other galaxy properties , we normalize the x - ray luminosity by the stellar mass from here on . as @xmath79 is roughly constant , we may estimate the typical accretion rate of the galaxy sample . assuming a @xmath80 black hole in a galaxy of stellar mass @xmath81 , and a bolometric correction of @xmath82 @xcite , we find the average eddington ratio of our sample is @xmath83 . we will argue below that accretion is the dominant source of emission . we further split the sample by redshift ( @xmath84 and @xmath85 ; for 9 total bins ) and restack using the method described in section [ sec:2 ] , normalizing by stellar mass ( such that we measure total(@xmath86total(stellar mass ) . [ fig : lz ] shows @xmath87(full ) / @xmath88 vs. @xmath89 for all bins combined . we find @xmath90 , suggesting that x - ray luminosity is higher at earlier times for galaxies of similar mass . this trend does not change upon making a minimum mass cut at log@xmath91 , the minimum mass detectable at @xmath92 . to remove the effects of any redshift correlation , we normalize by the above luminosity - redshift relation ( as if all galaxies were at the average redshift , z=0.95 ) after recombining all mass / redshift bins , and divide total stack luminosity by total stellar mass contained in each stack , to explore trends in @xmath10 independent of galaxy mass and redshift . [ fig : ldav ] shows mass and redshift - normalized x - ray luminosity as a function of stellar age ( as measured by the strength of the 4000 break ) , with the colors indicating @xmath93 . we find the following correlation between the normalized x - ray luminosity as a function of stellar population age : @xmath94 , suggesting that there is a moderate correlation between @xmath10 and galaxy x - ray luminosity . this correlation may have caused the flattening or increase of the x - ray luminosity for the lowest mass bins for each redshift interval in figure 3b , as these bins are dominated by star - forming galaxies with slightly higher x - ray luminosities . it may also explain the flattening at the highest masses , as quiescent galaxies with slightly lower x - ray luminosities are dominating these bins . we also explore the x - ray properties of the 32 different galaxy types of fig . [ fig : ldav ] in u - v and v - j color space in fig . [ fig : uvvj ] , in comparison to the color of all detected nmbs galaxies with @xmath95 ( grey scale ) . this diagram is a powerful diagnostic in distinguishing the blue and red galaxy populations , and separates the red passively - evolving galaxies from the dusty red star - formers ; quiescent galaxies are primarily located in the upper left quadrant of this diagram @xcite . the x - ray luminosity decreases along the star forming sequence , and the quiescent galaxies have an average x - ray luminosity per unit mass almost a dex lower than that of the bluest galaxies in the sample . to first order , we observe a linear trend in galaxy x - ray luminosity as a function of stellar mass ( fig . [ fig : luminm ] ) . after normalizing by stellar mass , we observe underlying redshift and stellar population age relations , finding that galaxy x - ray luminosity increases at higher redshifts ( fig . [ fig : lz ] ) , and also is lower for older stellar populations ( fig . [ fig : ldav ] ) . in this section , we attempt to distinguish between the stellar , agn , and gaseous components of the observed x - ray flux . we start by comparing the measured x - ray flux with the expected x - ray flux from xrbs . to determine the expected x - ray flux from xrbs , we utilize the star formation rates and stellar masses derived in @xcite , and x - ray luminosity relations for high - mass x - ray binaries ( hmxbs ) and low - mass x - ray binaries ( lmxbs ) . luminosity , stellar mass , and star formation rate relations are taken from @xcite and @xcite , such that @xmath96 and @xmath97 ( see @xcite , and @xcite for further discussion of these relations ) . we plot the observed x - ray luminosity vs. expected x - ray luminosity from xrbs in the left panel of figure [ fig : lx_expected1 ] . the blue galaxies are well - described by the x - ray luminosity of xrbs alone . however , for galaxies with strong 4000 breaks , the contribution from xrbs can not account for the full x - ray flux . this implies that a large fraction of their x - ray luminosities comes from agn and/or hot gas contributions . in the middle panel of figure [ fig : lx_expected1 ] we show the relative contribution of x - ray flux from non - stellar sources . this plot illustrates that the non - stellar x - ray sources contribute a significant fraction of the x - ray flux , in particular for galaxies with larger 4000 breaks ( luminosities of several galaxies with smaller 4000 breaks are upper limits ) . in the right panel of figure [ fig : lx_expected1 ] we show the measured x - ray flux corrected for xrbs , normalized by stellar mass , as a function of @xmath10 . interestingly , we find that @xmath98 . this suggests that the x - ray luminosity difference between red and blue galaxies is almost entirely driven by non - stellar sources . we note that galaxies with higher values of @xmath99 may be dominated by emission from star formation ( middle and right panels ) ; however , these measurements are not well - constrained . = 3.0 in = 3.0 in = 2.3 in next we explore the hardness ratio , given by @xmath100 , where h is the number of hard band ( 2 - 8 kev ) counts and s is the number of soft band ( 0.5 - 2 kev ) counts ( fig . [ fig : lx_expected ] ) , as a function of @xmath10 and redshift to distinguish probable x - ray sources . we plot lines to determine the hardness ratio that would be observed in a galaxy with the listed obscuring column densities , assuming a photon index of 1.1 ( solid ) and 2.1 ( dashed ) . we similarly plot the photon index as a function of redshift by combining the galaxies into @xmath10 and @xmath89 bins , and measuring the hard and soft counts , assuming a constant hardness ratio ( fig . [ fig : hrz ] ) . we find that the obscuring column density increases as a function of redshift , and may be higher for galaxies with intermediate values of @xmath10 ( green ) . interestingly , these are galaxies with high values of @xmath99 . this suggests a greater agn contribution to the average galaxy x - ray luminosity at higher redshifts , consistent with the observed increase in total x - ray luminosity at the same redshifts . = 3.0 in = 3.0 in = 3.0 in = 3.0 in after correcting for an increase in x - ray luminosity with both stellar mass and redshift , we find a difference of 1 dex between the x - ray luminosities of the highest and lowest sfr galaxies.to separate the effects of black hole accretion and stellar luminosity components , we subtract the expected x - ray binary contribution and normalize by stellar mass , finding that a significant fraction of the x - ray luminosity is from non - stellar sources . this excess is highest for passive red galaxies . we note that both hot gas and black hole accretion may contribute to differences in @xmath101 . while hot gas likely remains an important part of the x - ray luminosity function out to at least @xmath102 @xcite , we note that its contribution to the rest - frame soft band decreases significantly at higher redshifts , as galactic hot gas primarily emits at @xmath103 kev . additionally , x - ray emission from hot gas is unlikely directly correlated with star formation rate for galaxies of similar masses . the x - ray luminosities of local galaxies can vary by orders of magnitude for galaxies of similar stellar or black hole masses @xcite . the cosmological hot gas accretion rate onto halos is predicted to plateau at @xmath78 , however the used models are still highly uncertain ( , ) . we thus attribute the trends in x - ray luminosity to differences in agn accretion rates . previous studies have found similar trends in galaxy mass and accretion rates , though we probe lower black hole accretion rates , averaging @xmath83 ( assuming the @xmath104 relation , @xcite ) . both @xcite and @xcite find an increase in agn luminosity with stellar mass out to @xmath105 and @xmath106 , respectively , for objects several orders of magnitude brighter than those in our sample . @xcite finds that the distribution of eddington ratios is independent of stellar mass , suggesting that the physical mechanism responsible for fueling agn is similar at all scales and times . our results support this picture , with no increase in x - ray luminosity per unit stellar mass from @xmath107 making the plausible assumption that the non - stellar x - ray emission is dominated by agn , our study suggests that black holes have higher accretion rates in star forming galaxies than quiescent galaxies , though there is significant scatter in the relation . this result may seem in contrast to many studies that found a lack of sfr - agn connection up to @xmath108 @xcite . we note that these studies have not corrected for the mass evolution of their sample , and thus probe different galaxy populations at different redshifts . @xcite find that agn host galaxies have ssfrs @xmath109 lower relative to a normal galaxy sample out to @xmath108 , but do not probe the lowest x - ray luminosity regimes , selecting only sources with @xmath110 . this paper explores a different regime than the high - sfr , high accretion rates of the bright agns that are likely associated with mergers ( e.g. @xcite ) . low luminosity agns are most likely fueled by minor mergers and disk instabilities , and may accrete at low rates for periods which long outlast any intense periods of induced star formation ( @xcite and references therein ) . while this study probes a large range of galaxy sed types , masses , and redshifts , we note several limitations of this study . first , by stacking galaxies we are observing the average detection within the stack ; unusually luminous galaxies could dominate the stacked detection . we compare to the c - cosmos point source catalog @xcite , and find that in our stacked galaxy populations , with the exception of a few high - redshift bins in which one lone x - ray source is detected , non - detections typically provide @xmath111 of the stack luminosity . this suggests that luminosities of individual bright agns do not dominate the stacks . this is perhaps unsurprising , as this study automatically eliminates the more unusual seds of type i agns by fitting to galaxy spectra when deriving photometric redshifts ( a vast majority of low - luminosity agns appear obscured , while those at higher luminosity are primarily unobscured ; see , e.g. , @xcite ) . bias from the removal of type i agns may be most prominent for the high mass galaxy sample , as these galaxies are more frequently agn hosts and have a higher unobscured fraction ( e.g. @xcite ) . galaxies with seds that do not resemble any galaxy sed template ( 17% of the sample ) average a full - band luminosity of @xmath112 erg / s , significantly higher than most unnormalized sed types , suggesting that agns may contribute significantly to the luminosities and sed shapes of these galaxies . we explore the x - ray properties of @xmath1 galaxies at @xmath2 by stacking the galaxies as a function of sed type ( categorized by 32 sed templates in @xcite ) , while controlling for mass and redshift . we find that @xmath113 is roughly linearly correlated with galaxy mass , and also increases slightly with increasing redshift . after normalizing by galaxy mass and correcting for the redshift correlation , x - ray luminosity decreases with an increase in stellar population ages ( as measured by @xmath10 ) . we utilize local xrb luminosity functions to determine the stellar contribution to the galaxy x - ray luminosities . we find that there is a significant luminosity excess for most galaxies , particularly those with high values of @xmath10 , suggesting large x - ray contributions from hot gas and/or black hole accretion . consistent with other studies , we find that low luminosity agns show at best a mild preference for galaxies with shallower @xmath114 breaks and higher ssfrs ( e.g. , @xcite ) . first , there is a large scatter in x - ray luminosity from non - stellar sources for similar galaxy types . second , there is no strong correlation between the x - ray emission from non - stellar sources and spectral type . however , we observe that obscuration ( as determined by hardness ratio ) appears to increase with increasing redshift as the photon index decreases . this is consistent with the slight increase in x - ray luminosity per unit stellar mass with redshift , suggesting an increase in agn accretion rates in these galaxies . we note that we do not probe the highest black hole accretion rates , as our selection methods bias against galaxies for which the continuum emission is not dominated by stellar light . while it is difficult to differentiate sources of x - ray emission from c - cosmos data alone , a similar analysis of cosmos radio data ( such as that undertaken by @xcite of lyman break galaxies in the field ) , mid - infrared data , and 6 - 79kev x - ray data from nustar could provide more direct information about the obscured agn activity in these galaxies , helping to differentiate them from heavily star - forming galaxies that are x - ray hard . in addition , nir spectra of @xmath115 galaxies can aid in more accurate measurements of agn demographics using emission line ratios [ nii]/h@xmath63 and [ oiii]/h@xmath64 . an analogous analysis using the deeper chandra deep field south data could also be performed , helping to constrain the amount of variability within individual galaxy sed types . we would like to acknowledge the nmbs and the chandra cosmos teams for their observations and creation of the catalogs . we would like to thank ryan hickox and jane rigby for useful discussions . acknowledges support of the national science foundation graduate research fellowship .
we examine how the total x - ray luminosity correlates with stellar mass , stellar population , and redshift for a @xmath0-band limited sample of @xmath1 galaxies at @xmath2 from the newfirm medium band survey in the cosmos field . the galaxy sample is divided into 32 different galaxy types , based on similarities between the spectral energy distributions . for each galaxy type , we further divide the sample into bins of redshift and stellar mass , and perform an x - ray stacking analysis using the chandra cosmos ( c - cosmos ) data . we find that full band x - ray luminosity is primarily increasing with stellar mass , and at similar mass and spectral type is higher at larger redshifts . when comparing at the same stellar mass , we find that the x - ray luminosity is slightly higher for younger galaxies ( i.e. , weaker 4000 breaks ) , but the scatter in this relation is large . we compare the observed x - ray luminosities to those expected from low and high mass x - ray binaries ( xrbs ) . for blue galaxies , xrbs can almost fully account for the observed emission , while for older galaxies with larger 4000 breaks , active galactic nuclei ( agn ) or hot gas dominate the measured x - ray flux . after correcting for xrbs , the x - ray luminosity is still slightly higher in younger galaxies , although this correlation is not significant . agn appear to be a larger component of galaxy x - ray luminosity at earlier times , as the hardness ratio increases with redshift . together with the slight increase in x - ray luminosity this may indicate more obscured agns or higher accretion rates at earlier times .
modulation classification is the process of choosing the most likely scheme from a set of predefined candidate schemes that a received signal could belong to . various approaches have been proposed to address this problem . there has recently been growing interest in modulation classification for applications such as software defined radio , cognitive radio and interference identification @xcite . existing classification methods can generally be categorized into two main groups : feature based classifiers and likelihood based ( ml ) classifiers . the ml classifiers give the minimum possible classification error of all possible discriminant functions given perfect knowledge of the signal s probability distribution . however , this approach is very sensitive to modeling errors such as imperfect knowledge of the signal to noise ratio ( snr ) or phase offset . further , such approaches have very high computational complexity and are thus impractical in actual hardware implementation . to address these issue , various feature based techniques such as cumulant - based classifiers @xcite and cylostationary - based classifiers have been proposed @xcite . recently , goodness - of - fit ( gof ) tests such as the kolmogorov - smirnov ( ks ) @xcite distribution distance have been proposed to identify the constellation used in qam modulation @xcite . based on the ks classifier , we proposed a new reduced complexity kuiper ( rck ) classifer in @xcite . the rck classifier only finds the empirical cumulative distribution function ( ecdf ) in a small set of predetermined testpoints that have the highest probability of giving the maximum distribution distance , effectively sampling the distribution function . the algorithm offered reduced computational complexity by removing the need to estimate the full ecdf while still providing better performance than the ks classifier . it also increased the robustness of the classifier to imperfect parameter estimates . the idea of improving the classification accuracy of the rck classifier by using more testpoints was proposed in @xcite . the method is referred to as variational distance ( vd ) classifier where testpoints are selected to be the pdf - crossings of two classes being recognized . the sum of the absolute distances is then used as the final discriminating statistic . we refer to methods such as rck and vd , that utilize the value of the ecdf at a small number of testpoints , as sampled distribution distance classifiers . in this work we derive the optimal discriminant functions for classification with the sampled distribution distance given a set of testpoint locations . we also provide a systematic way of finding testpoint locations that provide near optimal performance by maximizing the bhattacharyya distance between classes . finally , we present results that compare the performance of this approach with existing techniques . following @xcite , we assume a sequence of @xmath0 discrete , complex , i.i.d . and sampled baseband symbols , @xmath1 $ ] , drawn from a constellation @xmath2 , transmitted over awgn channel . the received signal , under constellation @xmath3 , is given as @xmath4 $ ] , where @xmath5 , @xmath6 . we further define the snr as @xmath7/\sigma^2 $ ] . the task of the modulation classifier is to find @xmath8 , from which @xmath9 is drawn from . without loss of generality , we consider unit power constellations . let @xmath10=f(\mathbf{r})$ ] where @xmath11 is the chosen mapping from received symbols @xmath9 to the extracted feature vector @xmath12 , where @xmath13 is the length of the feature vector . possible feature maps include @xmath14 ( magnitude , @xmath15 ) , the concatenation of @xmath16 and @xmath17 ( quadrature , @xmath18 ) , the phase information @xmath19 ( angle , @xmath15 ) , among others . the theoretical cdf of @xmath20 given @xmath3 and @xmath21 , denoted as @xmath22 , is assumed to be known _ a priori _ ( methods of obtaining these distributions , both empirically and theoretically , are presented in ( * ? ? ? iii - a ) ) . in this paper we focus on algorithms that use the ecdf defined as @xmath23 as the discriminating feature for classification . here , @xmath24 is the indicator function whose value is 1 if the function argument is true , and 0 otherwise . if the complete ecdf resulting from the entire feature vector , @xmath12 , is used for classification , we get the conventional distribution distance measures such as kuiper , kolmogorov - smirnov , anderson - darling and others . details of these measures are discussed in @xcite . once the ecdf is found and the appropriate distribution distance is calculated , the candidate constellation with minimum distance is chosen . however , prior work in @xcite have shown that improved classification accuracy can be achieved at much lower computational complexity and with increased model robustness by finding the value of the ecdf at a small number of specific testpoints . we describe these methods formally by defining a set of @xmath25 testpoints : @xmath26 $ ] , with @xmath27 . for notational consistency , we also define the following virtual test points , @xmath28 and @xmath29 in addition to @xmath30 . evaluating the ecdf from ( [ eq : ecdf ] ) at @xmath30 gives us @xmath31 $ ] , @xmath32 . we refer to any classifier that utilizes the feature vector @xmath33 as a _ sampled distribution distance - based classifier_. as an example , the variational distance ( vd ) classifier from @xcite proposed forming @xmath30 from ecdf points that give either a local maxima or minima of the difference between two theoretical cdfs of the candidate classes . instead of using the sampled ecdf directly , vd classifier finds the number of samples that fall between two consecutive testpoints , which is equivalent to taking the difference of the ecdf at consecutive testpoints , @xmath34 . in this paper our goal is to optimize the classification accuracy of the sampled distribution distance classification approach defined as @xmath35 intuitively , there are two ways to improve @xmath36 . first , since different testpoints have varying distribution distance , it is expected that different weights should be assigned to each testpoint . second , the choice of the number and location of the points along the ecdf should also be investigated to find the proper balance between complexity and classification accuracy . both of these improvements are addressed in the following subsection . we first assume that @xmath30 has been selected _ a priori _ and our goal is to find the optimal classifier for the resulting feature vector @xmath33 . we want to find a discriminant function @xmath37 for each @xmath38 $ ] , for every candidate constellation @xmath3 . where we follow the rule : @xmath39 it is well established in decision theory that if the performance metric used is average classification error , the optimal classifier is based on the _ bayes decision procedure _ @xcite . this procedure can be stated as : @xmath40 using the prior probabilities @xmath41 , the posterior probabilities @xmath42 could be found from @xmath43 using bayes formula . thus , finding the pdf of the feature vector conditioned on the modulation scheme , @xmath44 , effectively gives us the optimal classifier in the minimum error rate sense . the testpoints partition @xmath12 into @xmath45 regions . an individual sample , @xmath46 , can be in region @xmath47 , such that @xmath48 , with a given probability , completely determined by the cdf , @xmath22 . the number of samples that fall into each of the regions , @xmath49 $ ] , where @xmath50 corresponds to region @xmath51 , @xmath52 , is jointly distributed according to a multinomial probability mass function ( pmf ) given as @xmath53 where @xmath54 $ ] , and @xmath55 is the probability of an individual sample being in region @xmath47 . given that @xmath12 is drawn from @xmath3 , @xmath56 , for @xmath57 . given a particular @xmath33 , the number of samples in each of the @xmath45 regions could be found as @xmath58 where @xmath59 and @xmath60 . this gives a mapping from any given @xmath33 to @xmath61 and therefore to the pmf @xmath62 as defined in ( [ eq : multinomial ] ) . therefore we have the complete class - conditional pdf , @xmath63 with @xmath64 in ( [ eq : multinomial ] ) determined by @xmath22 , the cdf of class @xmath3 . thus we have the optimal classifier . we will refer to @xmath33 and @xmath61 conditioned on class @xmath3 as @xmath65 and @xmath66 . although the multinomial pmf in ( [ eq : multinomial ] ) can be used for minimum error rate classification , its calculation is very computationally intensive . to address this issue we note that asymptotically the multinomial pmf , @xmath62 in ( [ eq : multinomial ] ) , approaches a multivariate gaussian distribution , @xmath67 as @xmath68 . where , @xmath69 since @xmath33 is simply the cumulative sum of @xmath61 ( i.e. @xmath70 ) , which is a linear operation , it follows that @xmath71 where , @xmath72 having shown that the feature vector @xmath33 is asymptotically gaussian distributed , we can proceed to apply the _ bayes decision procedure _ in ( [ eq : bdp ] ) . however , the full multivariate pdfs are not required to perform classification because the optimal discriminant functions for gaussian feature vectors are known to be quadratic with the following form @xcite : @xmath73 where @xmath74 and @xmath75 in the following sections we will simply refer to this classifier as the bayesian approach . similar to rck @xcite and vd @xcite the bayesian approach only needs to store the testpoint locations for a fixed set of snrs since the theoretical cdf is dependent on snr . given a @xmath30 of size @xmath25 , vd and rck require both @xmath30 and @xmath76 for each class @xmath3 . in contrast , the bayesian approach requires the same vector @xmath30 , an @xmath77 matrix @xmath78 , a vector @xmath79 of size @xmath25 , and a scalar @xmath80 for each class @xmath3 . however , there are typically no more than 12 testpoints ( total number of pdf - crossings ) , so this additional storage requirements are negligible . the bayesian approach also requires the calculation of a quadratic form expression ( [ eq : discriminant ] ) . again , due to the fact that only a relatively small number of testpoints is used , the additional complexity is minimal . in this subsection we present a method for choosing testpoint locations , @xmath30 , that provide good classification performance . the method of using the pdf - crossings make intuitive sense , since it tries to find the testpoints that provide the maximum difference in the theoretical cdf while providing some heuristic rule that the testpoints will be spaced apart . tespoints that are too close to each other are not as effective because the ecdf tends to be highly correlated and thus provide minimal additional information . another issue with using the pdf - crossing is that it does not factor in knowledge of the correlation between testpoints . as we have shown in section [ subsec : optimal ] , the distribution @xmath33 follows an approximate multivariate gaussian with statistics given in ( [ eq : mu ] ) and ( [ eq : sigma ] ) . therefore , the class - conditional means @xmath76 and covariance matrices @xmath81 are sufficient to completely describe the distribution of the feature vectors conditioned on @xmath3 . thus , these statistics are also sufficient to find the optimal testpoint locations , @xmath82 . however , since @xmath81 are clearly not equal for all @xmath3 , a closed form expression for the classification accuracy for this problem does not exist . instead , a @xmath83-dimensional integration is required and the limits , determined by the decision boundaries defined by ( [ eq : discriminant ] ) , are non - trivial . as is typically done in this scenario , we replace exact @xmath36 with a sub - optimum distance metric that is easier to evaluate and does not require a @xmath83-dimensional integral . in particular we use the bhattacharyya distance first studied for signal selection in @xcite shown to be a very effective as a `` goodness '' criterion in the process of of selecting effective features to be used in classification . the metric is shown here for reference : @xmath84 note that the bhattacharyya distance is calculated between 2 classes . as a result , the search for testpoints can only be performed for the @xmath85 case . however , this could be done sequentially through all the possible pairs of @xmath3 . as @xmath86 is a function of @xmath76 and @xmath81 which are functions of our testpoint selection , @xmath30 , then we can express it as @xmath87 . we thus find the good candidate testpoint by @xmath88 under the constraint @xmath27 . as this is an @xmath25-dimensional optimization problem , a closed - form solution is beyond the scope of this letter paper . instead , we turn to numerical optimization methods ( gradient descent methods ) to find local maxima . the intial point of these procedures could be chosen to coincide with the pdf - crossings or equally spaced over some interval . for the results section we focus on the quadrature feature which is a concatenation of the i and q component of each symbol . in fig . [ fig : testpoints ] , we show the results of the testpoint selection procedure with @xmath89 , under 0 db snr , for varying number of testpoints with the two class being 4-qam and 16-qam . . the solid line shows the cdf difference between the two classes ( 4-qam and 16-qam , under snr=0 db , @xmath90 ) ] the solid line plot corresponds to the difference of the two theoretical cdfs . we note that in the vd classifier the local maxima and minima of this plot are used as the testpoints . however , we find that the numerical optimization finds `` good '' testpoints to be close , but not exactly at the local maxima and minima . this is due to the additional information provided by the covariance matrices . in contrast to vd classifier that has a fixed number of testpoints ( 4 for this particular problem ) corresponding to the number of local maxima and minima , the optimization procedure allows more flexibility in choosing the number of testpoints . in fig . [ fig : testpoints ] , we show the result of the optimization procedure for a range of 1 to 8 testpoints . it confirms our intuition that `` good '' testpoints tend to be 1 ) spaced apart to avoid high correlation , 2 ) concentrated around locations that have high cdf difference , and 3 ) are not necessarily the same for different values of @xmath25 . this result further confirms the need to jointly optimize the testpoint locations . as mentioned in the previous section , the proposed approach has the flexibility of varying the number of testpoints . this effectively gives more flexibility to trade - off classification accuracy with computational complexity . this idea is illustrated in fig . [ fig : vary_tp ] . for @xmath89 and snr=0 db , we show the classification accuracy of the proposed method as the number of testpoints is increased from 1 to 8 , for all possible pairs of @xmath3 . the dotted lines correspond to the accuracy of the ml classifier which serves as an upperbound to classification accuracy , while the dashed lines correspond to that of the vd classifier . note that both are plotted as horizontal lines because ml does not utilize testpoints , while vd has a fixed number of testpoints corresponding to the pdf - crossings . for all possible pairs of constellations of interest.the classification accuracy of both ml and vd classifiers are also shown for comparison . ( snr=0 db , @xmath0=200 ) ] [ fig : vary_tp ] we see that the proposed method is able to exceed the accuracy of the vd classifier with as low as 3 testpoints . further , the method s accuracy could be improved by adding more testpoints but at the cost of higher complexity . we also note that with additional testpoints , the bayesian classifier reaches classification accuracy close to the ml classifier . finally , in fig . [ fig : vary_snr ] , we compare the performance of the proposed method with the existing techniques under varying snr with @xmath90 symbols used for classification . to have a fair comparison , the same number of testpoints are used for both vd and bayesian . for the entire range of snr the proposed bayesian approach is shown to provide substantial gains over the vd classifier . we emphasize again that asymptotically , the proposed approach is the optimal classifier when using the sampled distribution distance as the discriminating feature . also shown in the plot are the classification accuracy of the ml classifier which acts as the upperbound , and the conventional kuiper classifier . = 200 symbols used for classification . the same number of testpoints are used for both vd and bayesian . ] [ fig : vary_snr ] in this letter we presented the optimal discriminant functions for classifying using the sampled distribution distance . this method was shown to provide substantial gains compared to other existing approaches . the performance of this method is also shown to be close to the ml classifier but at significantly lower computational complexity . although modulation classification is presented in this paper to illustrate the basic concept , the approach is not limited to this application . the same classifier can be generalized to any classification problem where the cdf of each class is available . p. urriza , e. rebeiz , p. pawelczak , and d. cabric , `` computationally efficient modulation level classification based on probability distribution distance functions , '' _ ieee commun . _ , vol . 15 , no . 5 , pp . 476478 , may 2011 .
in this letter , we derive the optimal discriminant functions for modulation classification based on the sampled distribution distance . the proposed method classifies various candidate constellations using a low complexity approach based on the distribution distance at specific testpoints along the cumulative distribution function . this method , based on the bayesian decision criteria , asymptotically provides the minimum classification error possible given a set of testpoints . testpoint locations are also optimized to improve classification performance . the method provides significant gains over existing approaches that also use the distribution of the signal features .
in this paper we study the many server n - system shown in figure [ fig.n ] , with poisson arrivals and exponential service times , under first come first served and assign to longest idle server policy ( fcfs - alis ) , as the number of servers becomes large . before describing the model in detail , we will first discuss our motivation for studying this system . the n - system is one of the simplest special cases of the so called parallel server systems , as defined in @xcite and further studied in @xcite . the general model has customers of types @xmath0 , servers of types @xmath1 , and a bipartite compatibility graph @xmath2 where @xmath3 if customer type @xmath4 can be served by server @xmath5 . arrivals are renewal with rate @xmath6 , where successive customer types are i.i.d . with probabilities @xmath7 , there is a total of @xmath8 servers , of which @xmath9 are of type @xmath5 , and service times are generally distributed with rates @xmath10 . assume the system is operated under the fcfs - alis policy , that is servers take on the longest waiting compatible customer , and arriving customers are assigned to the longest idle compatible server . for this general system necessary and sufficient conditions for stability ( positive harris recurrence for given @xmath6 ) , or for complete resource pooling ( there exists critical @xmath11 such that the system is stable for @xmath12 , and the queues of all customer types diverge for @xmath13 ) can not be determined by 1st moment information alone ( as shown by an example of foss and chernova @xcite ) . in particular , under fcfs - alis calculation of the matching rates @xmath14 , which are long term average fractions of services performed by servers of type @xmath5 on customers of type @xmath4 , is intractable . in the special case that service rates depend only on the server type , and not on customer type , with poisson arrivals and exponential service times , the system has a product form stationary distribution , as given in @xcite . in that case matching rates can be computed from the stationary distribution . the following conjecture was made in @xcite : if the system is stable and has complete resource pooling for given @xmath15 , and we let both become large together , the behavior of the system simplifies : there will exist @xmath16 such that servers of type @xmath5 perform a fraction @xmath16 of the services , and the matching rates @xmath14 will converge to the rates for the fcfs infinite matching model with @xmath17 , as calculated in @xcite ( see also @xcite ) . the conjecture is based on the following heuristic argument : in steady state the times that each server becomes available form a stationary process which is only mildly correlated with the other servers and so servers become available approximately as a superposition of almost independent stationary processes which in the many server limit becomes a poisson process , and server types are then i.i.d . with probabilities @xmath16 , while customer types arrive as an i.i.d . sequence with probabilities @xmath7 , which corresponds exactly to the model of fcfs infinite matching . in our current study of the many server n - system we shall verify the conjectured many server behavior for this simple parallel server system . to do so we start from the known stationary distribution of the n - system with many servers , as derived from @xcite , and study its behavior as @xmath18 . as it turns out , the product form stationary distribution even for this simple case is far from simple , and the derivations of limits , which use summations over server permutations and asymptotic expansions of various expressions are quite laborious . we feel that this emphasizes the difficulty of verifying the conjectured behavior of the general system , which remains intractable at this time . we mention that the n - system with just two servers , has been the subject of several papers , @xcite . in this paper , our focus is on the n - system with many servers under fcfs - alis policy , and its limit property . the rest of the paper is structured as follows : in section [ sec.model ] we describe the model , in section [ sec.fluid ] we use some heuristic arguments to obtain a guess at the limiting behavior . in section [ sec.stationary ] we obtain the stationary behavior under many server scaling . in section [ sec.numerics ] we illustrate our results with some numerical examples . to improve the readability of the paper we have put all the proofs for section [ sec.stationary ] in the appendix . in our n - system customers of types @xmath19 and @xmath20 arrive as independent poisson streams , with rates @xmath21 . there are skill based parallel servers , @xmath22 servers of type @xmath23 which are flexible and can serve both types , and @xmath24 servers of type @xmath25 which can only serve type @xmath19 customers . we assume service times are all independent exponential , with server dependent rates . the service rate of an @xmath23 server is @xmath26 , the service rate of an @xmath25 server is @xmath27 , see fig [ fig.n ] . we let @xmath28 . service policy is fcfs - alis . the system is obviously markovian . in @xcite the following state description for the server dependent poisson exponential system with @xmath29 server types and @xmath30 customer types was used : imagine the customers arranged in a single queue by order of arrivals , and servers are attached to customers which they serve , and the remaining idle servers are arranged by increasing idle time , see figure [ fig : ivostate ] the state is then @xmath31 , where @xmath32 is a permutation of the @xmath8 servers , the first @xmath33 servers are the ordered busy servers , and the last @xmath4 servers are the ordered idle servers , and where @xmath34 are the queue lengths of the customers waiting for one of the servers @xmath35 , and skipped ( can not be served ) by servers @xmath36 . for the special case of the n - system , the following three random quantities are important : @xmath37 the number of idle servers of type @xmath23 , @xmath38 the number of idle servers of type @xmath25 , and @xmath39 the number of servers of type @xmath25 which follow the last server of type @xmath23 in the sequence @xmath32 . we let @xmath40 be the total number of idle servers . because of the structure of the n - system , and the fcfs - alis policy the following properties hold for @xmath41 and @xmath42 : ( i ) : : there are no customers waiting for any server which precedes the last @xmath23 server in the permutation . in other words , for all @xmath43 we have @xmath44 . in particular , if there is an idle type @xmath23 server , in other words if @xmath45 , then there are no waiting customers at all . ( ii ) : : if there are any idle servers , then there are no type @xmath19 customers waiting for service , in other words , if @xmath46 then all the waiting customers are of type @xmath20 . ( iii ) : : if there are no idle servers , then only the last queue can contain type @xmath19 customers , in other words , if @xmath47 then the last queue may contain customers of both types , but all the other waiting customers are of type @xmath20 . denote @xmath48 then a necessary and sufficient condition for stability is @xmath49 we shall require a stronger condition of complete resource pooling , defined by @xmath50 where @xmath51 equals the long run fraction of services performed by @xmath23 servers . the value of @xmath51 will be calculated in the next section . using the results of @xcite we can then write the exact stationary distribution of this system . we wish to show that as the arrival rates and the number of servers increase the system simplifies , and we get very precise many server scaling limits . we will investigate the behavior of the system when @xmath52 are fixed , and @xmath53 . to be precise , we shall then have @xmath54 , @xmath55 , all of which go to @xmath56 . we perform the following heuristic calculation : as long as the system is underloaded ( @xmath57 ) , each server of type @xmath23 will have a cycle of service of mean length @xmath58 , followed by an idle period , and similarly each server of type @xmath25 will have service of mean length @xmath59 followed by an idle period . the key idea now is that when @xmath18 , the idle periods should have the same length for both types , because of alis . let @xmath60 be the average length of the idle time . the average cycle times will be : @xmath61 and @xmath62 . denote by @xmath51 the long run fraction of services performed by @xmath23 servers , and @xmath63 for type @xmath25 . the flow rate out of one type @xmath23 server is @xmath64 , the flow rate out of all type @xmath23 servers should equal @xmath65 . similarly the flow rate out of all type @xmath25 servers should equal @xmath66 . that is , @xmath67 now we solve for @xmath60 and @xmath51 : we rewrite @xmath68 and eliminate @xmath51 : @xmath69 to get a quadratic equation for @xmath60 : @xmath70 here @xmath71 by @xmath72 , so the equation has one positive and one negative root . solving for positive @xmath60 we get : @xmath73 note : for the case of @xmath74 we get @xmath75 . from @xmath60 and little s law we can obtain the average number of idle servers in pool 1 and pool 2 , denoted by @xmath76 and @xmath77 respectively . @xmath78 the values of @xmath51 and @xmath63 are then : @xmath79 note : both are positive , so @xmath80 . also , when @xmath81 we get @xmath82 . the value of @xmath83 does not come into the equation for @xmath60 , or the calculation of @xmath51 . hence , once we solve and obtain @xmath51 , the property of complete resource pooling will consist of checking that @xmath84 . we will show that the following holds for the stationary queue , as @xmath18 : * @xmath85 is distributed as a geometric random variable , taking values @xmath86 with probability of success @xmath87 . it is independent of @xmath88 . * @xmath89 is close to a bivariate normal , with means @xmath90 , variances @xmath91 and correlation @xmath92 * successive idle servers except for the last @xmath93 are i.i.d . of type @xmath23 with probability @xmath51 and of type @xmath25 with probability @xmath63 we first obtain the stationary distribution for each state @xmath94 . we note that the stationary probabilities depend mainly on the values of @xmath95 . let @xmath96 denote the service rate of the server at position @xmath5 . [ thm.markov ] the stationary distribution of the state @xmath94 of the fcfs - alis many server n - system is given by : @xmath97 where @xmath98 is a normalizing constant . this follows for all three parts of ( [ eqn.pistate ] ) by utilizing properties ( i),(ii),(iii ) in section [ sec.model ] and substituting into equation ( 2.1 ) , theorem 2.1 , in @xcite . before we manipulate equation ( [ eqn.pistate ] ) , we introduce a lemma to facilitate the calculation . [ thm.sum ] let @xmath99 denote a permutation of @xmath100 given positive real numbers @xmath101 , we have @xmath102 where @xmath103 denotes the set of all the permutations of @xmath101 . now we can get the joint stationary distribution of @xmath95 . we denote by @xmath104 the stationary probability of @xmath105 , @xmath106 and @xmath107 . [ thm.pii1i2k ] the steady state joint distribution of @xmath95 is given by : @xmath108 where @xmath109 is a normalizing constant . in this section we obtain the asymptotic distribution of @xmath110 conditional on @xmath105 , as @xmath18 . we first show that as @xmath18 , the probability of no idle servers of type @xmath23 goes to zero , and so the probability that customers need not wait goes to 1 . next we condition on @xmath105 and show @xmath111 , where @xmath112 where @xmath60 is given in ( [ eqn.tsolution ] ) . finally , we condition on @xmath105 and show that the scaled and centered values of @xmath110 converge in distribution to a bivariate normal distribution . [ prop.zeroqueue ] when @xmath18 , as long as @xmath57 , @xmath113 , @xmath114 from this theorem we see that when @xmath18 , @xmath115 . therefore , @xmath116 for any @xmath117 . from equation ( [ eqn.piki ] ) , given @xmath105 , the limiting stationary distribution as @xmath18 is @xmath118 [ thm.fluidlimit ] conditional on @xmath105 , @xmath119 converge to @xmath120 in probability for any @xmath121 . that is , for any @xmath122 , when @xmath18 , we have @xmath123 after showing the fluid limit result , we are now ready to show the central limit result . [ thm.idledistribution ] for any @xmath124 , when @xmath18 , we have @xmath125 \right)\ ] ] where @xmath126 @xmath127 @xmath128 [ thm.kdist ] for any @xmath121 , as @xmath18 , @xmath129 theorem [ thm.kdist ] shows that @xmath85 converges in distribution to a geometric distribution , so @xmath130 . therefore , we can extend theorem [ thm.fluidlimit ] and theorem [ thm.idledistribution ] into unconditional versions . [ thm.uncondition ] when @xmath18 , @xmath85 becomes independent of @xmath131 and @xmath132 . @xmath133 converges in distribution to the bivariate normal distribution described in [ eqn.clt ] . consider a special case when @xmath134 , we have @xmath135 . @xmath76 and @xmath77 can be easily solved : @xmath136 when @xmath18 , @xmath137 converges in distribution to a bivariate normal distribution with mean @xmath138 , variance @xmath139 and correlation @xmath140 the total idleness has mean of @xmath141 and variance of @xmath142 in the infinite matching model corresponding to the n - system there is an infinite sequence of customers , of types @xmath143 , where the customer types are i.i.d . , type is @xmath19 with probability @xmath83 and @xmath20 with probability @xmath144 , and an independent sequence of servers , of types @xmath145 , where the server types are i.i.d . , type is @xmath23 with probability @xmath51 and @xmath25 with probability @xmath63 , and compatibility graph with arcs @xmath146 . successive customers and servers are matched according to fcfs : each server is matched to the first compatible customer that was not matched to a previous server , and each customer is matched to the first compatible server that was not matched to a previous customer . after @xmath8 of the customers have been matched , consider the sequence of remaining servers . let @xmath147 be the number of servers of type @xmath25 that are first in this sequence , preceding the first server of type @xmath23 . the @xmath148 is a markov chain . the steady state distribution for this markov chain is that @xmath149 , which is exactly the limiting distribution of @xmath150 in ( [ thm.kdist ] ) . we test our results by investigating an n - system . @xmath151 , @xmath152 , @xmath153 , @xmath154 . from our approximation results , as long as @xmath155 , or @xmath156 , both pools should have similar utilization . so the average number of idle servers in each pool is close to 50 , with variance of @xmath157 . we use exact stationary distribution to verify this . now we can calculate the expectation and variance of idle number in each pool exactly , listed in the following table . .the exact calculation [ cols="^,^,^,^,^",options="header " , ] we can see that even when @xmath158 , the improved approximation is not bad . we are grateful to ivo adan for helpful discussion of this paper . summation over the geometric terms @xmath162 in ( [ eqn.pistate ] ) gives @xmath163 next we see that in this expression , permutations of @xmath32 with the same @xmath164 have a similar structure . we now sum over all the permutations of the appropriate @xmath165 . by lemma [ thm.sum ] we obtain @xmath166 each permutation of the remaining servers , @xmath167 has the same stationary probability . it remains to count the number of permutations . when @xmath168 we have @xmath169 . for each permutation we choose 1 type @xmath23 server and @xmath170 out of @xmath24 type @xmath25 servers to form the last @xmath171 servers . the number of permutations is @xmath172 when @xmath173 , we have @xmath174 . for each permutation , we choose @xmath175 out of @xmath22 type @xmath23 servers and @xmath176 out of @xmath24 type @xmath25 servers . we then choose 1 from the @xmath175 idle servers of type @xmath23 , and @xmath170 from the @xmath176 idle servers of type @xmath25 to obtain the last @xmath171 servers . the number of permutations is @xmath177 multiplying the terms in ( [ eqn.pipermutation ] ) by the appropriate number of permutations and defining @xmath178 gives ( [ eqn.piki ] ) . [ [ proofs-of-theorems-prop.zeroqueue-thm.fluidlimit-and-thm.idledistribution ] ] proofs of theorems [ prop.zeroqueue ] , [ thm.fluidlimit ] and [ thm.idledistribution ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ( i ) : : we show that @xmath179 where @xmath180 . note that by @xmath181 we have @xmath182 . note also that @xmath183 . ( ii ) : : we show that @xmath184 \\ & & \qquad \exp\left[-n_2 \left(\log\left(1-\frac{m_2}{n_2}\right)+\frac{m_2}{n_2 } \right)\right]\end{aligned}\ ] ] where @xmath76 and @xmath77 are defined in ( [ eqn.meanidle ] ) . ( iii ) : : we show that as @xmath18 @xmath185 which proves the proposition . first we calculate @xmath186 we use induction to calculate @xmath187 from @xmath188 to @xmath189 . when @xmath188 , @xmath190 suppose @xmath191 then @xmath192 therefore , the induction is valid and we have @xmath193 @xmath194 next we calculate @xmath195 similar to the induction calculating @xmath196 above , we can obtain @xmath197 therefore , @xmath198 where @xmath199 is a poisson random variable with parameter @xmath200 . using stirling s approximation , @xmath201 recall that @xmath202 and note that @xmath203 . note also that when @xmath18 , @xmath199 can be approximated by a normal distribution with mean @xmath200 and variance @xmath200 . next we analyze @xmath204 in 3 cases depending on @xmath205 . * when @xmath206 , from the normal distribution approximation , when @xmath207 , @xmath208 . therefore , @xmath209 * when @xmath210 , @xmath211 . when @xmath207 , the normal distribution approximation gives @xmath212 . @xmath213 * when @xmath214 , when @xmath18 , the normal distribution approximation gives @xmath215 . we need more care to treat this case . for any @xmath216 , @xmath217 therefore , @xmath218 in fact , for any fixed @xmath5 , when @xmath18 , @xmath219 for any @xmath122 , let @xmath220 . we have @xmath221 . there exists an @xmath222 such that when @xmath223 , for any @xmath224 , @xmath225 therefore , @xmath226 therefore , when @xmath18 , @xmath227 we have @xmath228 since @xmath238 when @xmath239 , we have @xmath240 when @xmath18 , note that @xmath241 is of the order of @xmath242 . therefore , @xmath243 increases exponentially . when @xmath244 , @xmath245 converges to a constant ; when @xmath210 , @xmath245 increases in the order of @xmath246 . therefore , when @xmath18 and @xmath247 , @xmath248 when @xmath249 , @xmath250 we have that @xmath251 therefore , when @xmath18 , @xmath248 this completes the proof that when @xmath18 , @xmath252 first we show that the weak convergence is valid given @xmath253 . then we show that the same holds when @xmath105 , for any fixed @xmath170 . when @xmath253 , we prove the convergence in probability in 2 steps . * we show that for all states @xmath254 , the conditional probability is dominated by a bounded constant multiple of the conditional probability of some point on the boundary of the rectangle @xmath255 @xmath256 @xmath257 . * when @xmath18 , we approximate the conditional probability of the points in the rectangle @xmath255 @xmath256 @xmath257 . we then show that the probability of points on the boundary is negligible compared with the conditional probability at @xmath258 . * when @xmath263 and @xmath264 , we have @xmath265 . therefore , @xmath266 ; * when @xmath267 and @xmath268 , we have @xmath269 . therefore , @xmath270 ; * when @xmath271 , @xmath272 and @xmath273 , we have @xmath274 . therefore , @xmath275 ; * when @xmath271 , @xmath272 and @xmath276 , we have @xmath277 . therefore , @xmath278 ; * when @xmath279 , @xmath280 and @xmath281 . as long as @xmath282 , we have @xmath283 . when @xmath8 is large , this requires @xmath284 as long as @xmath285 , we have @xmath286 . when @xmath8 is large , this requires @xmath287 eventually the movement stops at the boundary which are @xmath290 away from @xmath90 . therefore , the probability of any state @xmath291 satisfying @xmath288 or @xmath289 would be dominated by the probability of some point at the boundary . when @xmath297 $ ] and @xmath298 $ ] , and @xmath8 grows large , we can use stirling s approximation . @xmath299 where @xmath300 . note that @xmath301 and @xmath302 . define @xmath303 and @xmath304 , we have @xmath305 $ ] and @xmath306 $ ] . @xmath307 we define @xmath308 the first order derivatives on @xmath309 and @xmath310 : @xmath311 noting @xmath312 , we can verify that @xmath313 solve the first order conditions . look at the second order derivatives : @xmath314 @xmath315 the hessian matrix is negative definite . therefore , @xmath316 is strictly concave on @xmath317 and reaches its unique global maximum at @xmath318 . since @xmath316 is strictly concave and reaches its unique global maximum at @xmath318 . the maximum of @xmath316 on @xmath319\times [ \delta , 1-\theta-\delta]\backslash ( f_1-\epsilon , f_1+\epsilon)\times(f_2-\epsilon , f_2+\epsilon)$ ] is on the boundary @xmath320 . since the boundary is a compact set , the maximum is attainable , denoted by @xmath321 , where @xmath322 . to obtain the asymptotic distribution of @xmath332 as @xmath207 we need to consider , by theorem [ thm.fluidlimit ] , only values @xmath333 for which @xmath334 and @xmath335 . we write @xmath336 , with @xmath337 , @xmath338 . note that @xmath339 are of the same order of magnitude as @xmath340 , and we only consider @xmath333 of the same order of magnitude . @xmath341 where the use of stirling s approximation is justified for large @xmath8 . here @xmath260 and @xmath300 . we clearly have : @xmath342 so we can treat that part as a constant . consider @xmath343 then from the taylor expansion of the logarithm function , we have @xmath344 therefore , @xmath345 similar expansions are valid for @xmath346 and @xmath347 . we now use the calculations in section [ sec.fluid ] to evaluate all the @xmath246 coefficients . by ( [ eqn.meanidle ] ) we have @xmath350 @xmath351 therefore , @xmath352 we are left with @xmath353 define @xmath354 @xmath355 @xmath356 we have @xmath357 therefore , @xmath358 given @xmath253 converges in distribution as @xmath207 to the bivariate normal distribution as stated in ( [ eqn.clt ] ) . when @xmath327 , and @xmath18 , similarly , @xmath359 we again write @xmath336 , with @xmath337 , @xmath338 . we then have @xmath360 we can now use the same approximation as for @xmath361 to show that @xmath133 converge to the same bivariate normal distribution . take a fixed arbitrary @xmath362 . fix @xmath363 , for any @xmath333 satisfying @xmath364 , @xmath365 and @xmath366 , from ( [ eqn.piki ] ) , noting @xmath367 for any @xmath368 and @xmath369 , we have @xmath370 therefore , @xmath371 for fixed @xmath372 , @xmath373 note the above inequality is valid for any @xmath333 satisfying @xmath364 , @xmath365 , we have @xmath374 from theorem [ thm.fluidlimit ] , there exists an @xmath375 such that when @xmath376 , @xmath377 then we have , @xmath378 this upper bound can be arbitrarily close to 0 when choosing @xmath379 , @xmath376 and @xmath380 . therefore , we have shown the tightness of @xmath85 , that is @xmath381 using @xmath382 for fixed @xmath363 , when @xmath376 , the ratio @xmath383 is lower bounded by @xmath384 and upper bounded by @xmath385 for any @xmath333 satisfying @xmath364 , @xmath365 and @xmath366 , in addition to ( [ eq : upperbound ] ) , we have the lower bound @xmath386 now we have @xmath387.\ ] ] therefore , @xmath388,\ ] ] that is , @xmath389.\ ] ] for fixed @xmath170 , as @xmath18 , the lower bound and the upper bound in ( [ eq : bounds ] ) both converge to @xmath390 . noting @xmath379 can be arbitrarily close to 0 , we have @xmath391 this together with the tightness ( [ eq : tightness ] ) proves ( [ eqn.limk ] ) . bell , s. l. , williams , r. j. ( 2001 ) . dynamic scheduling of a system with two parallel servers in heavy traffic with resource pooling : asymptotic optimality of a threshold policy . _ the annals of applied probability _ , 11(3 ) , 608 - 649 . ghamami , s. , ward , a. r. ( 2013 ) . dynamic scheduling of a two - server parallel server system with complete resource pooling and reneging in heavy traffic : asymptotic optimality of a two - threshold policy . _ mathematics of operations research _ , 38(4 ) , 761 - 824 . pinsky , mark . the normal approximation to the hypergeometric distribution . unpublished manusript , https://www.dartmouth.edu/ chance / teaching_aids / books_articles / probability_book / pinsky - hypergeometric.pdf
the n - system with independent poisson arrivals and exponential server - dependent service times under first come first served and assign to longest idle server policy has explicit steady state distribution . we scale the arrival and the number of servers simultaneously , and obtain the fluid and central limit approximation for the steady state . this is the first step towards exploring the many server scaling limit behavior of general parallel service systems .
the purpose of this section is to derive a rather general but simple model describing a novel class of lasing devices , those are driven by superconducting leads biased at voltage @xmath18 , have the optical resonance frequency at half the josephson generation frequency , @xmath19 , and contain many quantum states capable of emitting light . the purpose of the general approach is to capture the essential properties of this class of devices , without involving specific microscopic details of the used model . in this section we start by briefly repeating the essentials of the model for the hjl at microscopic level and extend it to contain a large number of effective emitters , yet at the microscopic level . we formulate the generic features of the model which are not specific to the details of the microscopic description . finally , we present the general but simple model where all these generic features are incorporated . we formulate the model in terms of a fokker - planck equation for three variables . the hjl is described in ref . as an optical resonator driven by a single quantum emitter . the emitter is a double quantum dot in a _ p - n _ semiconductor nanowire , capable of containing up to two electrons and two heavy holes . the energy of the corresponding discrete quantum states is determined by orbital and coulomb energies and is represented by hamiltonian @xmath20 . owing to the superconducting proximity effect , a gap is induced at the quantum dots for both electrons and holes , represented by hamiltonian @xmath21 where the @xmath22 are the proximity induced pair potentials , with the phases @xmath23 retaining the values of the corresponding superconducting leads . owing to gauge invariance , only the superconducting phase difference , @xmath24 will have a physical significance . interaction of the double quantum dot with the resonator mode occurs according to @xmath25 where @xmath26 is the frequency detuning from resonance , with @xmath27 the resonance frequency of the optical resonator and @xmath18 the bias over the device . the operator @xmath28 , taken in rotating frame , represents the electron hole recombination required for the photon emission . it can be seen as the dipole moment operator that drives the resonator mode . coefficient @xmath29 is the coupling strength of the dipole moment to the resonator mode . it is given by the dipole strength owing to quantum fluctuations of the electric field in the resonator mode . the interaction hamiltonian is assumed to be perturbative as to allow for a large photon number before nonlinearities become manifest , which will induce saturation of the dipole moment . the dynamics of the resonator mode can be treated in a semiclassical fashion , because a large number of photons will occupy the mode . in view of this , the @xmath30 , @xmath31 and @xmath32 will be replaced by their respective expectation values in the photon - dependent part of the full hamiltonian . a scaled quantity is defined , @xmath33 . within this approximation , the expectation value of the dipole moment , @xmath34 , is found by taking the derivative of the spectrum @xmath35 to @xmath36 , with @xmath35 and eigenstates @xmath37 associated with the full hamiltonian , @xmath38 . hence , the value of the dipole moment , @xmath39 , depends on the particular eigenstate of the double quantum dot . on the one hand , the dipole moment depends on the optical field , @xmath40 , but on the other hand , the evolution of @xmath40 is determined by the dipole moment . this leads to a set of self - consistent equations @xmath41 stationary lasing is found when @xmath42 for finite values of @xmath40 . this state must be stable against small perturbations . to get lasing , the rate at which photons are created in the resonator mode , @xmath43 $ ] , needs to exceed the cavity escape rate , @xmath3 . in common lasers , the imaginary part of the susceptibility , im@xmath44 $ ] , results from pumping to excited states . usually , to achieve lasing , the pumping must be sufficiently intense to invert the population . for the hjl , it is the superconductivity that creates the required complex susceptibility so that the population inversion is not necessary . in fact , the ability of the superconductors to absorb or supply cooper pairs undermines the notions of groundstate and excited state in the quantum dots . for instance , the induced pair potentials enable the possibility of transitions from lower to _ higher _ energy states , while _ emitting _ a photon , with the required energy provided by the bias . this feature allows for cycles of emission or absorption of photons , in order to ( de)populate the resonator mode . decoherence in the hjl with a single emitter is caused by spontaneous switchings between quantum dot eigenstates . owing to the finite cavity linewidth , emission of off - resonant photons occurs inducing these switchings they occur at a slow rate , @xmath45 , and therefore introduce extra dynamics in the hjl described by a master equation . this is in addition to the already present dynamics of eq . . it was found in ref . that even for an optimal choice of parameters , not every quantum dot eigenstate couples to the resonator mode strongly enough to support lasing . the ones that do support lasing will yield different stationary values @xmath46 . since only some states support lasing , the optical field may even extinguish after a switch , to turn on again after another one . hence , the switchings change @xmath46 , causing large intensity and phase fluctuations . most notably , for every lasing state @xmath46 there is another one at @xmath47 . when going from some lasing state to a nonlasing state and returning to the same lasing state again , the optical field can go to both @xmath48 with equal probability . hence the switchings are responsible for decoherence . the coherence time is determined by the switching rate , @xmath1 , and is similar to that of the common laser.@xcite even though the hjl , as described above , has a rather specific design , the framework provided by eqs . - is quite general . it applies to a class of devices where a biased josephson junction contains a structure of which the eigenstates couple to both superconducting leads and which can emit light by electron hole recombination . the light is emitted into an optical resonator . when the structure in the josephson junction is large enough it can be divided into @xmath0 blocks . every block represents an effective quantum emitter , which is a quantum system with a number of discrete eigenstates that can emit light . every emitter must couple to both superconducting leads and must emit light into the resonator mode . in principle , the emitters interact with each other , but we may model them to be independent . as an example of such a structure , one might consider an ensemble of _ p - n _ semiconductor nanowires with a double quantum dot , the nanowire that was used in the previous section , or one might consider a two - dimensional electron gas between the superconducting electrodes.@xcite increasing the number of emitters in the josephson junction to enhance the lasing is a method reminiscent of a line of research where superconducting lasers are made based on stacked josephson junctions.@xcite there , the josephson junctions are connected in series so that cascade amplification of the radiation occurs . the setup of the hjl is different though , because the emitters can be regarded as josephson junctions connected in parallel rather than in series . the dipole moment corresponding to the structure in the josephson junction is determined by its energy , which is taken in a rotating frame of reference . when the structure contains @xmath49 emitters the total energy is given by the sum of the energies of the emitters . assuming that emitter @xmath50 is in a state with energy @xmath51 , the total energy is simply @xmath52 . this yields a value of the total dipole moment given by @xmath53 , which should be inserted in eq . to look for solutions that support lasing , @xmath46 . every emitter is subjected to spontaneous switchings that change its eigenstate . therefore , at time scales longer than the switching times it is not the total energy ( dipole moment ) that is relevant but the average total energy ( dipole moment ) . the switching dynamics defines a master equation for each of the emitters , where the probability @xmath54 of emitter @xmath50 to be in state @xmath55 evolves as @xmath56p_m^i \end{aligned}\ ] ] with @xmath57 the transition rate from state @xmath58 in emitter @xmath50 . the quantity in brackets @xmath59 $ ] defines a transition matrix , which necessarily has a null vector to ensure the conservation of probability . the null vector gives the stationary occupation probabilities , @xmath60 . with these , the average total energy is given by @xmath61 , where @xmath62 is the energy of state @xmath55 in emitter @xmath50 . because of the coupling to the resonator mode , the energy depends on @xmath40 and the total average dipole moment is @xmath63 . now this quantity determines the lasing solutions of eq . . let us discuss a complication that might easily arise when considering this model or designing a device . in our approach we consider spontaneous switchings only to be a result of off - resonant photon emission . in the limit of a large number of emitters the associated transition rate matrix , eq . , turns out to be approximately symmetric , with @xmath64 . the transition rates @xmath57 are calculated using fermi s golden rule . they are , up to some constants , given by the product of a matrix element squared , @xmath65 , for a given emitter @xmath50 , and a lorentzian shaped density of states , @xmath66 $ ] , with @xmath67 the energy difference between the eigenstates @xmath68 and @xmath69 . we note that the photon number generically remains the same during a transition ; @xmath40 is constant . only after the transition may relaxation to the new stationary state occur . for a large number of emitters , however , a single switching is not enough to change the electrical field in the mode significantly . the stationary state @xmath46 remains almost the same after the transition and therefore also the eigenstates and energies remain approximately the same . consequently , the transition rates also remain the same . since the energies @xmath67 have a large mismatch with the resonance frequency ( the emitted photons are far off - resonant ) they do not depend significantly on detuning , @xmath70 , and we have @xmath71 , for fixed @xmath40 . the transition rate from state @xmath68 to @xmath69 is therefore equal to the transition rate of the reversed process . hence , because the transition rates remain approximately the same after a transition , the transition rate matrix is approximately symmetric . innocent as it may seem , this feature of the transition matrix poses a problem for our device , because as a result the average value of the dipole moment , @xmath72 , must be approximately zero . the null vector for a symmetric transition matrix must be a vector with all entries equal . therefore , in the stationary solution to the master equations , all the states in an effective emitter @xmath50 have the same probability @xmath73 to be occupied . the total average dipole moment is then given by @xmath74 . however , for the summation over @xmath55 we find @xmath75 = 0 , \end{aligned}\ ] ] where tr@xmath76 is the trace over the space of states of emitter @xmath50 . the equality to zero follows because tr@xmath77 $ ] is independent of @xmath36 . consequently @xmath78 . we note that this is generic for any kind of system governed by our hamiltonian and where the relevant states defining the dipoles , are equally populated . naively one would expect that a large number of emitters would result in a large number of photons in the resonator mode . we see that this is not so . on the contrary , the total dipole moment could be zero and no photons are present . we need to adjust the model used in ref . to get a nonzero average dipole moment . this is done by making the transition rate matrix asymmetric . then the occupation probabilities for the states of an emitter will not be equal anymore , resulting in an imbalance in populations of states . there is no _ a priori _ reason to favor , for instance , the state with higher energy over the state with lower energy , for higher occupation probability . the notion of higher or lower energy of states has become obscured by the superconducting pair potentials . hence , the created imbalance is not the same as a population inversion . importantly , recombination of electron - hole pairs will still be characterized by the phase lock , because they are still coherent with the superconductors . despite introducing incoherent transition rates , the special characteristics of the device are thus maintained . an asymmetry in the transition rate matrix can come about in various ways . the simplest way is to take into account the photon emission into the environment , rather than to the resonant mode . if the rate of this incoherent process is comparable with the emission rate to the resonant mode , the average dipole moment is of the order of its maximum value per emitter . in a realistic system we expect relaxation transitions between the emitter states that do not involve photon emission . these relaxation channels will generally yield asymmetric transition rate matrices . one can also engineer transition rates to get a sufficiently asymmetric transition matrix . a straightforward approach is to change the rates that are already present by changing the electromagnetic environment . a switching to a state with higher ( lower ) energy is accompanied by emission of an off - resonant photon , with energy @xmath79 ( @xmath80 ) , where @xmath81 . by changing the density of states in favor of one of these two , the spontaneous emission rate for the corresponding process increases with respect to the other one , thus creating the desired imbalance in transition rates . this might be done by inserting the device in a low quality factor resonator with resonance frequency about either of the frequencies . the low quality factor will ensure the rapid escape of the photon , before it is reabsorbed again . alternatively , the electromagnetic environment can be changed by shining with a red or blue detuned laser on the device , favoring transitions to lower and higher energy eigenstates respectively , proportional to the laser intensity . this resembles the technique of resolved sideband cooling / heating.@xcite another possibility is to add new transition rates to the system by adding nonsuperconducting leads to the device and directly pump electrons and/or holes into the optically active region . this , however , leads to big challenges in the technical device design . we have discussed the general features of possible microscopic realizations of a hjl with a large number of emitters . let us construct a model that does not rely on microscopic details . the relevant quantity to deal with is the @xmath40-dependent part of the total energy of the device , which determines the dipole moment . we show that the energy has a generic dependence on the phase difference between the superconducting leads . generally , we note that both common lasers and superconductors are systems characterized by a spontaneously broken @xmath82-symmetry , a phase symmetry . as such they are characterized by an order parameter which has a magnitude and a phase . any actual realization of the order parameter bears no apparent connection to the original unbroken symmetry . however , as a remnant of the symmetry , every value of the phase is equally probable and corresponding to states with equal energy . states with different phases are degenerate so that the energy of the system is independent of the phase . our model combines both of these systems , lasers and superconductors , which leads to a phase lock between the optical phase and the superconducting phase difference . indeed , owing to the interaction between these two systems a phase dependence must enter the energy of the system so that not both of the spontaneously broken symmetries can be maintained . the emission of a single photon is caused by the recombination of an electron - hole pair , with the latter carrying the phase of their respective condensates . the phase of the terms in the relevant interaction hamiltonian , eq . , is therefore given by @xmath83 , with @xmath84 the optical phase and @xmath85 the superconducting phase difference . in fact , the full interaction depends on the cosine of this phase combination , leading to a similar dependence in the energy . as a consequence an energy minimum is found for a specific value of @xmath83 , which is a manifestation of the phase lock between optical phase and superconducting phase difference . a further inspection of the phase dependence of the energy shows that there must , in fact , be two equivalent values of the phases . changing @xmath86 leaves the system invariant since the superconducting phase difference has not changed , but the phase in the interaction has changed to @xmath87 , so we conclude that @xmath88 . hence , the original spontaneously broken @xmath82 symmetries of the hamiltonian have been reduced to a single one and a spontaneously broken symmetry under rotations of @xmath6 of the optical phase . the remaining @xmath82 symmetry comprises invariance under a simultaneous change in optical phase and superconducting phase difference such that the quantity @xmath83 is unchanged . we stress that this reasoning applies to any system that combines two order parameters with spontaneously broken @xmath82 symmetries , with an interaction as in eq . . any such system manifests a phase lock . to find lasing in the general model , we need to have an expression for the energy of the device this allows us to extract essential physics from the model . generally , the energy should depend on the value of light field in the resonator , @xmath89 , and the pair potentials , @xmath90 , in the superconductors : @xmath91 . we use parameter limits where the four variables are relatively small , enabling us to use a taylor expansion for the energy . this limit corresponds to a weak drive of the resonator mode . to define `` small '' we go back to the model where the dipole medium consists of many emitters . we can associate an energy scale , @xmath92 , to each of these emitters . the induced pair potentials should be small compared to this scale , @xmath93 . in a realistic device this can be engineered by making suitable ( tunnel ) barriers between the superconductors and emitters . for the second limit , it is reasonable to take the interaction between dipole and resonator mode in the weak coupling limit . hence the coupling constant @xmath29 of the interaction hamiltonian , eq . [ eq : ham_photon ] , is much smaller than @xmath92 . therefore , this interaction perturbs the dipole energy by a value of the order of @xmath92 only when a large number of photons , @xmath94 , occupies the resonator mode . we want to make sure that the stationary value of photon number , @xmath95 , is always much smaller than @xmath96 , so that the use of an expansion of the energy in terms of @xmath97 is always justified . when writing the taylor expansion of the energy , @xmath91 , only a limited number of terms are allowed because of the specific phase dependence that is required . as was discussed in the previous section , the energy should depend on the phase combination , @xmath98 . therefore , the lowest order phase - dependent term in the energy must be proportional to @xmath99complex conjugate . this is a fourth - order term in the taylor expansion . all other terms of equal or lower order are proportional to powers of @xmath100 or @xmath101 . writing only the terms with immediate relevance , the energy is , up to fourth - order in taylor expansion @xmath102 in this expression the phase of the induced pair potentials was absorbed into the optical phase , so it is redefined as @xmath103 . furthermore , we absorbed the induced pair potentials in the prefactor @xmath104 , with @xmath105 a constant . the primes in @xmath106 and @xmath107 refer , respectively , to first and second derivatives at @xmath108 of the function @xmath109 , which is independent of the induced pair potentials . the lowest order term , @xmath110 , is a sum of @xmath111 and some other terms depending on @xmath101 but not on @xmath97 and @xmath112 . these terms will therefore not be relevant for the dipole moment . there are also terms proportional to @xmath113 , but these are small compared to @xmath106 , because they are of higher order in the taylor expansion , and can safely be neglected . it is necessary to keep the @xmath114 term in the expansion , since this nonlinear term stabilizes a possible lasing instability . without this term the light field would grow exponentially under unstable conditions . when calculating the dipole moment later , we will see that @xmath106 adds up to the detuning from resonance . hence , the value of @xmath106 can implicitly be changed by changing the detuning . in fact , to get lasing the detuning should be such that @xmath106 effectively becomes small , of the order of @xmath115 or smaller , or goes to zero . because the detuning @xmath116 , this also sets the scale @xmath117 . we stress that this is an _ effective _ change of @xmath106 . in the expansion of eq . the @xmath106 term is still considered to be larger than the fourth - order terms . with effectively @xmath118 , the @xmath100 term is of the same order as , or smaller than , the @xmath114 term so the latter should be maintained in the expression for the energy . with the above expression for the energy we have managed to separate the dependence of the energy on @xmath97 and @xmath22 from the microscopic details of the driving medium . the latter are all contained in the prefactors of the taylor expansion . substituting eq . into eq . , we find an explicit , general self - consistency equation . noting that @xmath119 and @xmath120 , we have @xmath121 where @xmath122 and @xmath123 is the redefined detuning . this equation describes the deterministic evolution of the electric field , @xmath97 , in the resonator mode . we note that an equivalent complex conjugate equation exists . the energy was considered to be a fixed quantity in the above section , corresponding to the average value of the dipole moment . to fully account for all dynamics in the hjl with many emitters , we also have to consider noise . this can be incorporated in a single fokker - planck equation that encompasses two noise sources . the first noise source is the traditional quantum noise due to photon emission from the mode.@xcite the second noise source describes the fluctuations of parameters in eq . . to understand the latter let us note that eq . is valid not only in average , but also for each specific configuration of the emitter quantum states . the parameters @xmath106 , @xmath107 , and @xmath115 are configuration dependent . since the emitters undergo spontaneous transitions the parameters fluctuate accordingly . explicitly taking into account all the details of the fluctuations would greatly complicate the model . instead , we employ the central limit theorem stating that any ensemble of equal distributions can be described as a gaussian distribution when the number of distributions goes to infinity . therefore , since we have a large number of emitters , as described above , we may regard the parameters as stochastical quantities with gaussian distributions . in principle , the noise in all parameters is expected to be of the same order of magnitude relative to their average values . however , since @xmath106 is of lower order in the taylor expansion of eq . , its fluctuations are more important than that of @xmath107 and @xmath115 . we therefore only consider a single noise source by taking @xmath107 and @xmath115 to be equal to their average values , while @xmath124 , with @xmath125 being the average value and @xmath126 being the deviation . the latter appears in eq . as an addition to the detuning the noise can therefore be seen as fluctuations in detuning or in the resonant frequency . the corresponding variance , @xmath127 , is an important parameter of our general model . we shall take into account that emitters switch on a certain time scale that determines autocorrelations of these detuning fluctuations . the autocorrelation function is thus given by@xcite @xmath128 where @xmath129 is the inverse autocorrelation time of the emitter ensemble it is possible to capture all relevant dynamics compactly in a single nonlinear fokker - planck equation . this is an equation for the probability distribution function @xmath130 of three real values , two of them are represented in the optical field and the third in the detuning deviation . the drift and diffusion terms in the equation respectively describe deterministic and noise driven , indeterministic processes.@xcite the fokker - planck equation for our device encompasses two gaussian noise sources , describing fluctuations in three quantities . first , there is the dynamics of @xmath97 which is subjected to a noise source , @xmath131 , describing the fundamental fluctuations in @xmath97 . this is the quantum noise associated with the discrete changes of photon number in the resonator mode . it satisfies , @xmath132 and @xmath133 . second , as discussed , the noise in @xmath106 is equivalent to detuning fluctuations . this is represented by the quantity @xmath126 , with average @xmath134 and variance @xmath127 . the fokker - planck equation reads @xmath135b p + i a b^ * p \right\ } \right ] + \frac{\gamma}{2}\frac{\partial^2 p}{\partial b\partial b^ * } + \gamma \left\ { \frac{\partial}{\partial \tilde\omega } \tilde\omega p + \langle { \tilde\omega}^2 \rangle \frac{\partial^2 p}{\partial { \tilde\omega}^2 } \right\ } , \end{aligned}\ ] ] the single derivative terms on the right - hand side of eq . are the drift terms . evaluating @xmath136 by using eq . will yield the self - consistency equation for @xmath137 . we remark that in this context the brackets , @xmath138 , denote the average over the classical distribution @xmath139 , and _ not _ the expectation value of the operator @xmath140 . the diffusion is described by the second derivative terms on the right - hand side of eq . . let us note that all the parameters we use have the dimension of frequency . it is convenient to make them dimensionless , measuring them in relative units of @xmath141 . this amounts to setting @xmath142 in eq . . for clarity , we rescale the detuning deviation in eq . to appropriate units , by taking @xmath143 , which is the scale of the frequency domain where lasing is possible , as we show in sec . [ sec : ii ] . additionally we have @xmath144 . as a consequence , the term in eq . proportional to @xmath145 changes to @xmath146 . other terms retain their form . we conclude this section by stressing that this fokker - planck equation applies to a whole class of lasers , where a biased josephson junction contains a structure of which the eigenstates couple to both superconducting leads and which can emit light by electron - hole recombination into the resonator mode . in the previous section we have derived a model for a class of lasers that is driven by superconductivity and that exhibits a phase lock . here we show that lasing with a phase lock is indeed possible and we give expressions for the lasing threshold and the number of photons in the resonator mode . we find three different regimes , with a different number of stationary lasing states . , for the stationary solutions of eq . , as a function of the rescaled detuning @xmath147 , with @xmath148 . thick solid ( dashed ) lines represent stable ( unstable ) solutions , labeled by @xmath149 . three parameter regimes , ( i , ii and iii ) , are found based on the number of stationary solutions . the vertical dotted lines represent boundaries between these regimes . below this plot is a panel that illustrates the existing stationary solutions in the @xmath150-@xmath151 plane , for each regime . solutions with @xmath152 occur twofold with opposite sign . solid ( empty ) circles represent stable ( unstable ) stationary states . the bottom panel shows the dependence of the regimes on the parameters @xmath153 and @xmath147 . the diagonal lines are the lasing thresholds @xmath154 as discussed in the main text . the line between regimes ii and iii is the one where @xmath149 becomes real - valued . the plot of the upper figure is taken along the dashed line . when crossing the boundary along arrow(s ) ( i ) and ( ii ) [ ( iii ) ] , the system undergoes a second [ first ] order phase transition . ] a full description of the dynamics of lasing would require solving fokker - planck equation , but this is difficult , if not impossible , even for a relatively simple system such as ours . instead , we first search for stationary lasing solutions to the deterministic equation of motion , eq . , thus disregarding noise . once these solutions are found we study the effects of noise when near a stable stationary lasing state . this is the topic of the subsequent two sections . to find the desired stationary solutions to eq . , with @xmath155 , we rewrite them as real equations . defining @xmath156 leads to a set of equations , @xmath157 where @xmath158 is the photon number in the resonator mode . these equations can be readily solved to find an expression for @xmath2 and for the phase of @xmath97 , which was defined as @xmath159 , @xmath160 , \\ \tan\left(\frac{\varphi_\pm}{2}\right ) & = -a\mp \sqrt{a^2 - 1 } , \end{split } \end{aligned}\ ] ] where we conveniently rescaled the detuning , @xmath161 . hence , we find stationary lasing solutions . the fixed value of @xmath162 implies that @xmath84 and @xmath85 are locked . note that the equations are invariant under the transformation @xmath163 and @xmath164 , implying the occurrence of two solutions which differ in phase by @xmath6 , as predicted in the previous section . indeed , eqs . are invariant under the transformation @xmath165 . hence @xmath149 and @xmath162 represent four stationary solutions . besides these , there is another stationary solution , being @xmath166 , delivering a total of five possible stationary solutions . we remark that @xmath162 only depends on @xmath115 and not on @xmath147 and @xmath107 . furthermore , a large photon number can be achieved when @xmath167 . studying the stationary solutions of eq . we find three different regimes that differ in the number of physical solutions . since @xmath149 represents the number of photons in the resonator mode , it must be either a real and positive quantity or it must be zero . as a result we note that lasing solutions , with @xmath168 , can only be achieved when @xmath169 . then , depending on @xmath147 , three different regimes exist ( fig . [ fig : lasing ] ) . we choose @xmath170 but similar regimes can be found when @xmath171 . for @xmath172 and @xmath173 , both @xmath174 so that the only physical solution is @xmath175 . we call the parameter regime with a single solution regime i. regime ii is given by the condition that @xmath176 , where @xmath177 and @xmath178 . including @xmath175 there are now three physical solutions . the third regime is when @xmath179 and @xmath173 , where both @xmath168 , corresponding to a total of five physical solutions . even though there can be multiple physical solutions , not all of them will support stationary lasing , because some will be unstable against small perturbations . to find the condition for stability we expand eq . , in terms of @xmath150 and @xmath151 , up to first order about a stationary value . this yields @xmath180 where @xmath150 and @xmath151 are at the stationary values , @xmath181 and @xmath182 are the respective deviations from the stationary values and eq . was used to rewrite some terms . when an eigenvalue of the matrix is negative ( positive ) , a perturbation in the direction of the corresponding eigenvector decays ( grows ) exponentially to ( from ) the stationary value of @xmath150 and @xmath151 . the case where both eigenvalues are negative corresponds to a stable stationary lasing state . after a fluctuation in any direction , the system will relax to the stationary state again . with one positive and one negative eigenvalue there is a stable and an unstable direction against perturbations , thus defining a saddle point . the case where both eigenvalues are positive can not occur , because the trace of the matrix is smaller than zero . the condition of the stability therefore depends on the sign of the determinant of the matrix , positive being the stable stationary state . evaluating the determinant and again using eq . we derive that the stability for a physical solution @xmath168 is determined by sign@xmath183 $ ] , whereas the stability of the solution at @xmath175 is determined by sign@xmath184 $ ] . to illustrate how the system relaxes to a stable stationary lasing state after a small perturbation , we give the eigenvalues of the matrix in eq . . these are , for the case of @xmath185 , @xmath186 and , for the case of @xmath187 , @xmath188 . hence , when @xmath115 is sufficiently large , the system oscillates about the stationary point with a frequency @xmath189 and relaxes to it in @xmath190 rotations . for @xmath170 the stability of the stationary solutions in the various regimes is as depicted in fig . [ fig : lasing ] . regime ii is the most relevant one for lasing as the @xmath175 solution is unstable , whereas the @xmath185 solution is stable . in regime iii the @xmath185 and @xmath175 solutions are all stable , whereas the @xmath187 solutions are unstable . when @xmath171 , the qualitative picture is the same . quantitatively , the location of the regimes changes . in the phase diagram of fig . [ fig : lasing ] , the regimes i and iii are interchanged so that the plots for @xmath171 are mirror images of the plots for @xmath170 . additionally , the @xmath187 ( @xmath185 ) solution is now the stable ( unstable ) one . the boundaries between the regimes can be seen as lasing thresholds because they separate the lasing regimes from the nonlasing regime . the boundaries between regimes i and ii and between regimes ii and iii are lines in parameter space where the determinant of the matrix of eq . is zero at @xmath175 . they are given by @xmath191 . for negative @xmath107 the value @xmath192 ( @xmath193 ) corresponds to the boundary between regimes i and ii ( ii and iii ) . the boundary between regimes i and iii is determined by the condition @xmath194 . at this boundary , the @xmath175 solution is still stable , but two pairs of stable and unstable solutions arise at values @xmath152 . when gradually crossing the boundaries from i to ii or ii to iii , the system undergoes a second order phase transition [ arrows ( i ) and ( ii ) in fig . [ fig : lasing ] ] . thereby the @xmath175 state becomes unstable . in contrast to this , when crossing the boundary from regime i to iii [ arrow ( iii ) in fig . [ fig : lasing ] ] the system will undergo a first - order phase transition , because all physical stationary solutions will arise directly at nonzero values , two stable and two unstable . we note that @xmath115 can not be increased indefinitely as the pairing potentials must remain small enough to satisfy the adopted approximations . to conclude this section , we find that regime ii is the only regime where lasing occurs without an interruption caused by a large fluctuation . this is because only there the @xmath195 solution is unstable . in contrast to this , in regime iii a large fluctuation can bring the device from a lasing state to the stable nonlasing state at @xmath175 thereby interrupting the emission of light . regime ii occurs in a small interval at large detuning , where @xmath196 . the number of photons in the resonator mode , being of the order of @xmath197 , is large when @xmath198 . in the remainder of this work , we assume we have chosen the parameters @xmath115 and @xmath147 such that the system is in regime ii . we take @xmath170 so that the @xmath185 solution is the lasing solution . let us consider fluctuations in the device . it is important to note that because of the phase lock the laser can not lose coherence by a random drift of the optical phase as in common lasers . on the contrary , after a small fluctuation a damping force returns the system to the stable stationary lasing state . the average value of the optical phase is therefore retained . as in the hjl,@xcite the only possibility to lose coherence is when the device switches to another stable stationary lasing state . this requires a sufficiently large fluctuation in photon number or optical phase . we envision these large fluctuations to occur only rarely , on exponentially long time scales , implying that the noise in @xmath97 must be linear , i.e. dominated by small fluctuations about the stationary lasing state . in this section we investigate the hjl in the linear noise approximation . there are two noise sources causing intrinsic quantum fluctuations in the resonator mode and fluctuations in detuning . therefore , the steady state of the hjl can be described by a multivariate gaussian probability distribution , for which we calculate the variances . several parameter limits will be discussed concisely for the photon number variance . we also find an expression for the corresponding noise spectrum . the large fluctuations that cause decoherence of the laser are the topic of the next section . , where we neglect for clarity the detuning noise on the @xmath199 axis . the value of @xmath97 evolves along the lines . the dashed line , indicated by @xmath115 , represents a large fluctuation , where @xmath97 , starting in the vicinity of a stable lasing point , crosses the unstable point to the other stable lasing point . this process shall occur on exponentially long time scales , @xmath200 ( see main text ) . the solid erratic lines , indicated with @xmath201 , represent small fluctuations in the vicinity of the stable points . these fluctuations typically occur on a time scale @xmath202 ( see main text ) . the two kinds of fluctuation contribute independently to the noise spectrum , of which an illustration is shown on the right - hand side . the sharp peak , indicated with @xmath115 , corresponds to the large fluctuations . it is a lorentzian - shaped peak with exponentially small width . the small fluctuations provide a broadband background , indicated by @xmath201 . details on the peaks can be found in the main text . ] we first describe an intuitive picture of the shape of the full noise spectrum , containing the spectrum of both large and small fluctuations . a sketch of this is given in fig . [ fig : noise ] . the large fluctuations occur at exponentially long time scales , @xmath200 , satisfying @xmath203 . here , as we see later in this section , the small fluctuations occur at much shorter time scales , determined by the largest of the response time of the resonator mode , @xmath204 , or the autocorrelation time of the detuning noise , @xmath205 , yielding @xmath206 . because of this separation of time scales , both kinds of fluctuation contribute independently to the noise spectrum . this is given by the fourier transform of the correlator of @xmath207 , @xmath208 at time scales @xmath200 the correlator is expected to decay because of the switching to another lasing state . the switching changes the optical phase but retains the stationary photon number @xmath209 . the correlator decays as @xmath210 the noise spectrum of this correlator is a lorentzian with exponentially small width , @xmath211 and with peak height @xmath212 . at time scales @xmath202 small fluctuations have a dominating contribution to the correlator . writing @xmath213 for a small fluctuation , @xmath214 , we have @xmath215 . the requirement that the fluctuations @xmath216 are small implies that they are gaussian distributed as shown below . one expects the noise spectrum to consist of three peaks . one peak is associated with the fluctuations in detuning , and has the width @xmath217 . two side peaks , shifted with respect to zero detuning , are associated with the intrinsic quantum fluctuations and have width @xmath218 . these peaks are shifted because of the oscillatory behavior of small deviations from equilibrium position , which was described in the previous section in the context of stability analysis . in fact , the shifts should be @xmath219 . the noise spectrum approximately corresponds to a sum of three lorentzians . here the peak heights must be @xmath220 . adding up the two contributions , the total noise spectrum is a high and very narrow peak on a background formed by three low and broad peaks , as depicted in fig . [ fig : noise ] . in this section we concentrate on the small fluctuations that are responsible for the broad background in the noise spectrum of the figure . a full description of the small fluctuations can be found by linearizing the fokker - planck equation of eq . . this amounts to expanding @xmath97 to @xmath213 and only retaining up to linear terms in @xmath221 , @xmath222 , and @xmath223 . the resulting fokker - planck equation can be solved exactly,@xcite but we do not do this , because we are only interested in the steady state probability distribution . the latter is fully characterized by the variances of the variables of the system . instead of directly using the linearized fokker - planck equation of eq . to calculate the variances , we use the equivalent system of langevin equations for a set of appropriate variables . the advantage is that in the course of this procedure , we directly obtain the noise spectrum associated with the linear noise . the linearized form of eq . is fully equivalent@xcite to the following system of langevin equations , @xmath224 a\nonumber\\ & -i \left ( a + \omega''b_s^2 \right ) a^*-i\tilde\chi(t ) \sqrt{a^2 - 1}\ b_s + \xi_b(t ) , \\ \dot{\tilde\chi } = & - \gamma\tilde\chi + \xi_{\tilde\chi}(t ) . \nonumber \end{aligned}\ ] ] here it is noted that the first equation is complex and represents two equations . there are two different langevin forces satisfying @xmath225 with other correlators being zero . the source @xmath226 represents the quantum noise in the resonator mode , while the source @xmath227 represents the noise in the detuning . the current set of equations is not the most convenient one because the variables @xmath221 and @xmath222 are not orthogonal to each other . it will be most natural to work in a system with a variable @xmath228 , representing radial displacements or photon number changes , and a variable @xmath229 , representing azimuthal displacements or phase changes . the relation between these two bases comes about in the following way . one can write the deviation @xmath230 in vector form in a cartesian coordinate system as @xmath231 . alternatively , in a cylindrical coordinate system @xmath232 , where the @xmath233 follows because @xmath221 is taken about @xmath234 , which is separated from the origin with a distance @xmath233 . the variations @xmath229 represent the variations in optical phase , but the variations @xmath235 correspond to the fluctuations of the square root of @xmath2 . hence , @xmath236 . using the transformation relations , @xmath237 and @xmath238 , one can derive expressions for @xmath181 and @xmath182 to find that @xmath239e^{i\varphi } \end{aligned}\ ] ] using this we can rewrite the above set of langevin equations . reminding the reader that we use for the lasing solution @xmath240 and @xmath241 [ eq . ] , we arrive at @xmath242 here we have three langevin forces satisfying @xmath243 the other correlators are zero . with the change of variables , we now need to find the steady - state probability distribution to the fokker - planck equation for the variables @xmath244 , @xmath229 , and @xmath223 . this is given by@xcite @xmath245 where @xmath246 is simply the variance of the distribution of the detuning fluctuations . now using the langevin equations we can find the variance matrix @xmath247 according to the following procedure . first , we take the fourier transform of eqs . . for any variable @xmath248 we have @xmath249 $ ] . the resulting system of equations is solved for @xmath250 , @xmath251 and @xmath252 . to get the variances , we calculate the inverse fourier transform at @xmath253 of the expectation value of products of the fourier transformed variables . for instance , we have @xmath254 according to this procedure , we find the variances and write them in a form that will be convenient later : @xmath255 } \nonumber \\ & + \frac{1}{4n_s } \left[1 + \frac{a^2}{(a^2 - 1)(1+\chi)}\right ] , \nonumber\\ \langle\delta\varphi^2\rangle & = \frac{\gamma(a^2 - 1)\langle\tilde\chi^2\rangle } { 2[4(a^2 - 1)(1+\chi)+ \gamma^2 + 2\gamma ] } + \frac{2+\chi}{16n_s } , \nonumber \\ \frac{\langle\delta n\delta\varphi\rangle}{n_s } & = \frac{1}{4n_s\sqrt{(a^2 - 1 ) } } , \\ \frac{\langle \delta n\tilde\chi\rangle}{n_s } & = \frac{4(a^2 - 1)\langle\tilde\chi^2\rangle } { 4(a^2 - 1)(1+\chi)+ \gamma^2 + 2\gamma } , \nonumber \\ \langle \delta \varphi\tilde\chi\rangle & = \frac{-\gamma \sqrt{a^2 - 1}\langle\tilde\chi^2\rangle } { 4(a^2 - 1)(1+\chi)+ \gamma^2 + 2\gamma } .\nonumber \end{aligned}\ ] ] we remark that some variances diverge when approaching the lasing threshold , which is when @xmath256 or @xmath257 . also , the combination @xmath258 is equal to the determinant of the stability matrix of eq . . a larger value of the determinant implies higher stability , as the variances decrease . with the expressions for the variances , we are able to quantify in which parameter regimes the linear fokker - planck equation and the steady - state distribution of eq . are valid . their validity is ensured when @xmath259 and @xmath260 , so that all expressions in eq . must be much smaller than one . note that under these conditions , there is no _ a priori _ bound on @xmath223 , which is already purely linear . it will be particularly useful to study the expression for @xmath261 in a bit more detail , because it is always of the order of , or larger than the other ones . additionally , this is an interesting quantity , because large fluctuations in @xmath244 lead to switchings and are thus important for the long coherence times . the variance of the fluctuations in photon number is studied in a few limits . for reasons of simplicity and clarity , it is most convenient to be sufficiently far from the lasing threshold , so we take @xmath262 and @xmath263 . keeping this in mind , the first limit to be studied is the one where @xmath264 . then we have @xmath265 we note a few properties . to start with , the term proportional to @xmath266 is associated with the noise in the detuning , whereas the @xmath267 term is associated with the intrinsic quantum noise of the resonator mode . the latter exactly corresponds to the fundamental noise expected in any common laser.@xcite in line with this division , there are two distinct regimes when considering the dependence on @xmath95 . for small enough values of @xmath95 the intrinsic quantum noise dominates so that the variance decreases inversely proportional to @xmath95 . increasing @xmath95 beyond a critical value the variance saturates as the @xmath267 term becomes unimportant and the noise in detuning starts to dominate . we further note that the magnitude of the fluctuations is independent of @xmath115 and independent of @xmath129 in the limit @xmath268 . the criterion to have small fluctuations now leads to the requirements that @xmath10 and @xmath269 . the two previously mentioned limits of @xmath129 in the detuning noise dominated regime can be intuitively understood . if @xmath268 , it is smaller than the decay rate of the resonator mode so that the fluctuations in @xmath223 are slow enough for the optical field to adiabatically follow them . the number fluctuations are then purely determined by the equilibrium dipole distribution . when @xmath270 , the optical field can not follow the dipole fluctuations adiabatically . however , because @xmath264 , the optical field will respond fast to the dipole fluctuation before it fades away . in fact , the optical field will start to rotate about the stationary point with a period @xmath271 and the fluctuation only fades away after @xmath272 rotations about the stationary lasing state . the variance of the number fluctuations is now determined by the instantaneous , out - of - equilibrium value of the fluctuation of @xmath223 , as determined by @xmath273 in eq . , leading to an extra factor of @xmath129 compared to the previous case with @xmath268 . we continue by studying the variance of photon number in the second limit , which is @xmath274 and leads to @xmath275 in this limit , again there is in the dependence on @xmath95 a division in two regimes : one where the variance is dominated by the noise in detuning and one where the intrinsic quantum noise of the resonator mode is most important . in contrast to previous case there is now a dependence on both @xmath115 and @xmath129 . here , the criterion to have small fluctuations leads to the requirements that @xmath10 and @xmath276 . there is also an intuitive way to understand the detuning noise - dominated regime in the limit @xmath274 . here , the dynamics of the detuning fluctuations is on much shorter time scales , @xmath277 , than the dynamics of the resonator mode , @xmath278 , so that the latter has a reduced time window to respond to the detuning fluctuations . compared to the earlier case with @xmath279 , the reduction factor is @xmath280 . as part of the calculation done to find the variances , we can also find an expression for the noise spectrum , which is defined as @xmath281 . defining @xmath282 we have @xmath283[(\nu-\sqrt{d-1})^2 + 1 ] } \left[4a^2 + 2\nu^2 + d(1+\chi)+ \frac{8n_s(a^2 - 1)[4(a^2 - 1)+\nu^2]\gamma \langle\tilde \chi^2\rangle } { \gamma^2+\nu^2 } \right ] \end{aligned}\ ] ] the term proportional to @xmath266 is directly related to the noise in the detuning , whereas the other term is directly related to the quantum noise of the resonator mode . which of the two terms is the most important depends , of course , on the parameter regime , as discussed above . the noise spectrum can be written in three parts , as we show in the appendix and discussed in the intuitive picture above , with each part corresponding to a peak . there is a central lorentzian peak at @xmath284 with width @xmath285 and there are two non - lorentzian sideband peaks at @xmath286 . in the limit @xmath263 and @xmath262 the sideband peaks become lorentzian with width @xmath287 . in this limit , the central peak typically dominates over the sideband peaks when @xmath288 . for @xmath289 , the central peak almost vanishes compared to the sideband peaks . we have now described the linear fluctuations in the hjl and derived conditions on parameters for which fluctuations are small , @xmath290 . the noise in various parameter regimes was studied . for small @xmath95 , but still much larger than @xmath291 , the intrinsic quantum noise of the resonator mode was found to have a dominating contribution to the photon number fluctuations , while for large @xmath95 , the noise in the detuning is the most important . there , the magnitude of number fluctuations depends on @xmath129 and the ratio @xmath292 . the noise spectrum corresponding to the small fluctuations was found and has a central peak with varying width , and two sideband peaks with width @xmath287 . in the previous section we studied small fluctuations in the hjl under the assumption that they dominate the fluctuation spectrum . these small fluctuations do not cause decoherence of the laser . decoherence is caused by large fluctuations that switch the stationary lasing state to one with a phase difference of @xmath6 ; see fig [ fig : lasing ] . in this section we study the large fluctuations thoroughly to find the time scales at which the switchings would occur . this gives the decoherence time of the hjl . the probability to have a large fluctuation can be calculated using the formalism of optimal paths.@xcite this formalism is , for instance , used in classical diffusion driven systems where the transition rate from one stationary state to another , across a high barrier , is calculated . the path that crosses the barrier and has the largest probability is called the optimal path . its shape and probability can be found using the principle of least action . its probability determines the transition rate . in this section , we introduce the formalism of optimal paths and apply it to our system , after which some generic properties of the action are derived . after this we show how the coherence time is related to the optimal paths . we investigate the dependence of the action of the optimal paths on the device parameters . this is done both by using the results of previous section and by simulating the optimal paths . to start with , we outline the dynamics of the device in the absence of noise , describe the effect of noise and explain the relation to kramers escape problem . in sec . [ sec : ii ] we have derived the requirements on stable stationary lasing in the hjl . let us suppose that the optical field in the hjl is not in a point of stable lasing and that @xmath293 . if we disregard the noises we know that the field will evolve to one of the points . this evolution is described by eq . . we visualize the dynamics by making a stream plot , as is done in fig . [ fig : stream ] . to describe the dynamics it is convenient to associate the field components , @xmath294 , with coordinates of a `` particle '' subject to a coordinate - dependent `` force field '' that causes the motion of the particle with the velocity proportional to the force . starting from a given initial condition , the particle will flow towards a stable stationary point . its trajectory corresponds to a streamline in the plot . ( horizontal ) and @xmath151 ( vertical ) , defined according to @xmath295 . the gray streamlines represent a `` force field '' . when a `` particle '' is placed in the force field it will evolve along these lines to one of the attractors , indicated by @xmath115 and @xmath296 , depending on the initial condition . when placed at the thick red ( dashed ) or blue ( solid ) line , the particle will not evolve to one of the attractors but to the saddle point , indicated by @xmath297 . these lines form the separatrix of the system , that separates the domains of attraction of the attractors . ] figure [ fig : stream ] corresponds to regime ii as defined in sec . [ sec : ii ] . in the figure the three stationary points are indicated by @xmath297 , @xmath115 , and @xmath296 . the point @xmath297 , at @xmath298 is the unstable saddle point , whereas the points @xmath115 and @xmath296 are stable points or attractors . for a general initial condition in the vicinity of the saddle point the particle is repelled from this vicinity . there is , however , a particular line called stable direction such that the particle is attracted to the saddle point if the initial condition is chosen to be on this line . the streamlines coming to the saddle point at the stable direction form the separatrix of eq . , which is the boundary separating the domains of attraction of the attractors @xmath115 and @xmath296 . we see that the particle circles around an attractor before reaching it . this signifies that the force field can not be represented as a gradient of a scalar potential . the force field must also have a transversal component that causes the rotation and can be represented as the curl of a vector potential . full equations of motion shall also include a variation of detuning @xmath223 . with this we have three coupled equations of motion that define the force field in the space of three variables , @xmath299y , \\ \dot y & = -y -\left[\sqrt{a^2 - 1}(\chi+\tilde\chi ) + a + \omega '' n\right]x \\ \dot{\tilde\chi } & = -\gamma\tilde\chi . \end{split } \end{aligned}\ ] ] the last equation gives the relaxation of the detuning to its equilibrium position . the specifics of this equation is the presence of a _ saddle line_. the line @xmath298 is a saddle line because for each value of @xmath147 the point @xmath298 is a saddle point with respect to the motion in @xmath294 directions . in sec . [ sec : iii ] noise was introduced . with the langevin noise added to the dynamics , fig . [ fig : stream ] no longer gives a full description of the particle evolution . the particle in the ` force field ' will experience random kicks , causing it to change from one trajectory to another , even when initially at rest at an attractor . effectively , the kicks cause the particle to diffuse away from its starting point , even against the streamlines of the force field . nonetheless , the force field counters the diffusive flow trying to bring the particle back to the attractor . these two kinds of motion have a fundamentally different nature . the evolution along the flow lines is deterministic , in contrast to the diffusive motion which is nondeterministic . importantly , the diffusive motion of the particle lifts up the restriction of the particle to stay in the domain of attraction of an attractor . a series of random kicks may repel the particle from an attractor and eventually pull it over the separatrix into the domain of attraction of the other attractor . after that the particle relaxes naturally along a streamline to the other attractor . this trajectory corresponds to the large fluctuation we are after : a switch between the two stable lasing states has occurred . let us suppose , in contrast to what is the case for eq . , that a particle moves in a potential well , of which the gradient defines the corresponding force field , and that the fluctuation - dissipation theorem is applicable , like in kramers escape problem . in this case the fokker - planck equation can be solved to calculate the escape probability from the potential well.@xcite this escape probability involves the arrhenius factor , @xmath300 , where @xmath301 is the height of the barrier that is crossed by the particle , counted from the potential minimum , and @xmath14 is the effective temperature of the particle . however , because of the transversal component of the force field associated with eq . , we can not use this general approach to solve fokker - planck equation . there , the escape probability does not depend on a potential barrier height , but on the shape and probability of the path that crosses the separatrix . we note that the potential barrier and the separatrix both form the boundary of the domain of attraction where the particle is trapped . with the conceptual framework in mind , let us continue by showing how the probability to have a transition from one stationary state to the other , is related to the concept of the optimal path . we only consider paths that start in stationary point @xmath115 and end in stationary point @xmath296 . the time elapsing during the transition will be infinite , going from @xmath302 to @xmath303 , because the particle can only leave @xmath115 and approach @xmath296 asymptotically . it is possible to express the probability distribution for the fluctuations in terms of an action . we start with a general set of langevin equations , written in vector form , @xmath304 the langevin forces @xmath305 represent gaussian white noise , satisfying @xmath306 and @xmath307 , thus supposing an infinitesimally short autocorrelation time . the noise can be expressed as a single probability distribution that is a product of the probability distributions of the separate langevin forces at each moment in time , @xmath308 \end{aligned}\ ] ] with @xmath105 the normalization constant . one may interpret this expression in another perspective , as the probability distribution for histories of fluctuations , because the integral in the exponent represents the history of fluctuations . each history corresponds to a unique progress in time of @xmath309 . we stress that the values of @xmath310 at each time @xmath311 remain strictly independent from the ones at other times @xmath312 . so how does this concept of histories aid us ? even though the langevin forces might have infinitesimally short autocorrelation times , the vector field @xmath313 will need some finite time to relax to the stationary state . during this time , the history of fluctuations will be relevant . the resulting cumulative effect of fluctuations is best captured by using eq . to rewrite the exponent of @xmath139 , yielding @xmath314 hence , we indeed see that the noise probability distribution can be described using the action , defined as @xmath297 . the probability distribution , @xmath139 , is strongly reminiscent of a path integral . indeed , @xmath139 is a probability density which , upon integration over all paths , gives the probability to go from @xmath115 to @xmath296 . when fluctuations are small , as is described in sec . [ sec : iii ] , the path integral is dominated by the `` classical '' path and by the quadratic fluctuations about it.@xcite the classical path is found by the principle of least action . it is what we have called the optimal path . the optimal path corresponds to the lowest value of the action for the transition , @xmath315 , which relates to the transition probability @xmath316 with @xmath317 the prefactor corresponding to the quadratic fluctuations . hence , the probability of a transition from @xmath115 to @xmath296 depends on the optimal path . without specifying any details about @xmath318 we note two things . first , by inspection we see that @xmath319 , with the equality satisfied for the deterministic path , for which @xmath320 . second , the integrand of @xmath297 is equivalent to a lagrangian and because it only implicitly depends on time , a conserved quantity similar to the hamiltonian exists @xmath321=0 . \end{aligned}\ ] ] the equality of the hamiltonian to zero is not universal , but is appropriate for the path we consider . this path connects the stationary points , where @xmath322 and therefore @xmath323 . being a conserved quantity , @xmath324 must always be zero at the path connecting the stationary points . the action as it is given in eq . is not in the most convenient form . to minimize this action we need to find both the shape of the optimal path and the speed with which the path is traveled . by exploiting the conserved hamiltonian , we can relate the speed to the force field resulting in an action where the optimal path is only determined by its shape . to show this , we choose the diffusion constants , without losing generality , as @xmath325 for all @xmath50 . this is allowed , because we can always rescale @xmath326 and @xmath327 accordingly . applying this to eq . we then find that @xmath328 . with this , the terms @xmath329 and @xmath330 can be rewritten and added to each other to become @xmath331 . because both terms in the lagrangian now only contain a single time derivative , we can parametrize time at will . we parametrize it as a monotonously growing function @xmath332 , such that @xmath333 and @xmath334 . consequently @xmath335 , so that the action becomes @xmath336 minimizing this action will result in the optimal path irrespective of the exact form of parametrization @xmath337 and therefore irrespective of the speed with which the optimal path is traveled . from the expressions for the action , eq . , an intuitive understanding of the dynamics of the system can be obtained . suppose that we are at some point in the force field @xmath313 and move away from this point a random infinitesimal distance @xmath338 . the probability to move in a certain direction is then determined by the increase in action associated with this direction , which is , in turn , determined by the angle between the force field vectors , @xmath318 and @xmath338 . with the two vectors aligned the action is smallest ( zero ) and the probability to move in that direction the highest . again , this corresponds to the deterministic solution . with the vectors counter aligned the action is largest and the probability to move in that direction the lowest . close to stable or unstable stationary points , when @xmath339 is of the order of the diffusion constants , the absolute difference between alignment and counteralignment is small because @xmath340 is close to zero . the probabilities to go in any direction become of the same order of magnitude . this is the regime where the quadratic fluctuations are important . far away from stationary points , however , @xmath340 can be quite large so that the absolute difference between alignment and counteralignment becomes large . then , because the action appears in the exponent , the probability to follow the deterministic path will be very close to unity . going in some other direction will cause a large increase in action . concluding this section we make two remarks . first , the probability distribution of eq . is a solution to the fokker - planck equation in the _ wentzel - kramers - brillouin _ approximation.@xcite the solution is exact in the limit where the diffusion constants are infinitesimally small . second , the coherence time , or escape time , is related to the transition probability,@xcite yielding @xmath341 here , @xmath342 relates to the quadratic fluctuations around the optimal path . we disregard this prefactor and only consider the action of the optimal path , which determines the order of magnitude of the coherence time . let us apply the above described ideas and calculate the action for the hjl , using the fokker - planck equation of eq . . the fokker - planck equation contains three variables . two of them , @xmath294 , associated with the optical field , defined as @xmath343 , and one associated with the fluctuations in detuning , @xmath223 . we redefine the latter as @xmath344 , where @xmath345 is the diffusion constant of @xmath223 . as a result , the diffusion constants associated with fluctuations in @xmath346 are all equal to one . because any fokker - planck equation , with constant diffusion terms , has an equivalent langevin equation,@xcite eq . can be written in the form of eq . , with @xmath347 where @xmath348 is a dimensionless quantity . we remind the reader that @xmath115 , @xmath129 , @xmath70 , @xmath349 , @xmath106 , @xmath107 and @xmath350 [ eq . ] are dimensionless with the corresponding frequencies being measured in units of @xmath141 . the drift terms @xmath318 can be explicitly written in terms of scalar and vector potentials @xmath351 , \end{split } \end{aligned}\ ] ] demonstrating that the vector field corresponding to @xmath318 contains both longitudinal and transversal parts . the action for the hjl is now simply given by eq . with the vector and the diffusion constants given by eq . now that we have discussed the formalism of optimal paths and have established the relation with the escape time , we want to find explicitly the value of the optimal action , depending on the system parameters of the hjl . before that , let us present the estimation of the action for the optimal path , @xmath315 , in different regimes . for this we use the result of sec . [ sec : iii ] . in the following section we simulate numerically the optimal paths and the corresponding action , to validate the findings of this section . in sec . [ sec : iii ] we have discussed the fluctuations in the device in the linear noise approximation . under these assumptions the fluctuations are small and obey a gaussian distribution with variances given by eq . . for the order - of - value estimations we can use this formula for small fluctuations to find the probability of large fluctuations . in particular , large fluctuations in @xmath2 are important , because at @xmath175 , the particle reaches the saddle point enabling the switching to the opposite stable point . large fluctuations in either of the other variables will not drive the particle across the separatrix . the probability for this fluctuation in gaussian approximation is simply proportional to @xmath352 $ ] , with @xmath353 given by eq . @xmath354 } \nonumber \\ & + \frac{1}{2n_s } \left(1 + \frac{a^2}{(a^2 - 1)(1+\chi)}\right ) \end{aligned}\ ] ] we can thus estimate @xmath355 , where the coefficient is different in different parameter regimes . let us list here all possible limiting cases corresponding to different relations between parameters of the device . in the case ( i ) the large fluctuation is caused by quantum photon noise . the fluctuation is estimated by the last term in the expression and the optimal action is given by @xmath356 @xmath357 being a coefficient of the order of one . the probability of the large fluctuation is enhanced near the lasing threshold , @xmath358 , and is exponentially small otherwise , provided @xmath10 . in all other cases the probability of the large fluctuation is mainly caused by the detuning noise . in the limit @xmath268 , when the switching time of the emitters is longer than the damping time of the optical resonator , we can distinguish two cases . the case ( ii ) takes place near the lasing threshold , @xmath359 , and is characterized by @xmath360 . the estimation for @xmath315 reads @xmath361 and is inversely proportional to the variance of the detuning fluctuations . in the opposite case ( iii ) @xmath362 . the estimation is given by @xmath363 and does not depend on @xmath129 . if the switching times are relatively short , @xmath364 , we distinguish two cases depending on the ratio of @xmath129 and @xmath115 . in case ( iv ) @xmath289 and the estimation reads @xmath365 while in the opposite case ( v ) ( @xmath288 ) it reads @xmath366 in all these cases coefficients @xmath367 are supposed to be of the order of one . the boundaries in parameter space separating case ( i ) from cases ( ii)-(v ) are found by comparing the corresponding estimations . now that we have the estimations of the optimal action and the understanding of its dependence on relevant parameters , we quantify numerically the shape of the optimal path and the action . we describe the numerical method and we comment briefly on its limitations . we compare the numerical results and the estimations and discuss the peculiarities of the path shapes . we use the action of eq . . a path @xmath368 starts at the attractor at positive @xmath150 , negative @xmath151 and @xmath369 ( @xmath115 in fig . [ fig : stream ] ) and ends at the point @xmath296 . the path can be separated into two parts . during the first part the path proceeds from @xmath115 untill a point at the separatrix that divides the domains of attraction . during the second part the path follows the natural motion of the particle satisfying equations of motion eq . . this part does not contribute to the action , since @xmath370 , and thus does not have to be evaluated numerically . in all cases investigated numerically , we have found that the first part of the path ends at the saddle line @xmath298 at some yet to be optimized value of @xmath199 . to find the optimal path we could first formulate a boundary value problem , using the differential euler - lagrange equation from the action eq .. the solution to this is the optimal path , which can only be found using a shooting method . with this method , one changes initial conditions and integrates the euler - lagrange equation trying to reach the predefined final condition . this method is not efficient to find the trajectories we are looking for because they have a relatively low probability . we implement a more efficient method . we directly minimize the action of a predefined path , that depends on a finite number of parameters , as function of the path parameters . for the path parametrization , we implement a bezier curve of some sufficiently high order @xmath2 ( from 12 to 16 depending on the region in the parameter space ) , @xmath371 the order plus one equals to the number of points @xmath372 in @xmath373 space that define the curve . the starting point of the bezier curve , @xmath374 , is fixed to the attractor while the last point , @xmath375 is fixed to the saddle line , @xmath376 being the subject of the optimization procedure , as well as the coordinates of all other points . the accuracy of this method depends on the order of the bezier curve used . the higher the order , the better the accuracy . the order required to have a sufficiently accurate path will depend on the complexity of the shape of the path . we have calculated the optimal paths for several parameter regimes , to validate the estimations of eqs . - and find numerical values of the prefactors . in fig . [ fig : action](a ) we show the dependence of the value of the optimal action on the number of photons in the resonator mode , @xmath2 , for three different values of @xmath129 and @xmath377 . for sufficiently small values of @xmath2 , @xmath297 is linear in @xmath2 and case ( i ) applies , yielding @xmath378 for all three values of @xmath129 . for large values of @xmath2 the action @xmath297 saturates to a value depending on @xmath129 . for @xmath379 case ( iii ) applies , with @xmath380 , while case ( iv ) applies for @xmath381 with @xmath382 . the curve for @xmath383 at large @xmath2 corresponds to a crossover regime between cases ( iv ) and ( v ) . in fig . [ fig : action](b ) we show the dependence of the value of the action on @xmath129 for two values of @xmath384 . in the limit @xmath385 case ( iii ) applies , with also here @xmath380 . in the limit @xmath289 case ( iv ) applies with @xmath386 depending on @xmath266 . at @xmath387 @xmath382 , which is consistent with its value found previously , whereas at @xmath388 @xmath389 . figure [ fig : action](c ) shows the dependence of the value of the action on the detuning @xmath147 for three values of @xmath384 . for @xmath390 case ( i ) applies with @xmath378 , which is consistent with the value of @xmath357 for fig . [ fig : action](a ) . for @xmath391 case ( iii ) applies with @xmath380 , which is also consistent with the value previously found . the curve with @xmath392 corresponds to the crossover regime between cases ( i ) and ( iii ) so that no value can be extracted for any of @xmath393 . we study the shapes of the optimal paths for the cases listed , with representative examples given in fig . [ fig : optimalpaths ] . rather surprisingly , we find that the shapes of the optimal paths are qualitatively similar to the shapes of the deterministic paths , that are defined by eqs . . both the optimal path and the deterministic path circle counter clockwise around the stable point @xmath394 while moving respectively from / to the point the three optimal paths in the right column of fig . [ fig : optimalpaths ] , with @xmath395 , correspond to case ( i ) . in this case , the intrinsic photon number fluctuations dominate . as a signature of this , the path circles around the stationary point @xmath394 in the @xmath396 plane , without any significant displacement in the @xmath199 direction . the optimal paths in the column with @xmath397 correspond to the regime where the fluctuations in detuning dominate . here the displacements in the @xmath199 direction are two orders of magnitude larger than for the optimal paths of the previous case . the upper left path corresponds to case ( iii ) , where the photon number fluctuations are sufficiently fast that the optical field adiabatically follows the fluctuations in the detuning . the path has a shape resembling a piece of parabola and reaches the saddle line close to the value of @xmath199 corresponding to the lasing threshold ( fig . [ fig : lasing ] ) . for the path with @xmath381 [ case ( iv ) ] , detuning fluctuations occur on shorter time scales so that the photon number fluctuations become relatively more important , leading to an increased number of circulations of the path around the stationary point @xmath394 . the other optimal paths correspond to crossovers between the cases . the general trend is however clear : with increasing @xmath129 ( @xmath398 ) the displacement of the optimal path in the @xmath199 direction ( @xmath396 plane ) becomes smaller . we briefly comment on the accuracy of the method used . the order of the bezier curve that is required to have an accurate optimal path is different for the different cases . the simple optimal path of case ( iii ) is found with high accuracy ( up to 1% ) with a bezier curve of the order @xmath399 , while the optimal path with @xmath381 and @xmath400 is a bezier curve of order @xmath401 and is less accurate ( up to 10% ) . we do not use a higher order because it would greatly increase the computation time without providing new insight . the shape of the optimal path can , however , be inferred , especially for case ( i ) . since the optimal path resembles the deterministic path and the deterministic path in case ( i ) circles @xmath272 around the stable point ( sec.[sec : ii ] ) , we expect that also the optimal path circles @xmath272 times around @xmath394 . to calculate this optimal path requires a bezier curve of at least the order of @xmath402 . the method is thus inefficient for very big @xmath115 . for the value @xmath403 that was used , we are far away from the lasing threshold , while being able to find the optimal path with sufficient accuracy . for the cases other than the case ( i ) the number of times the path circles around @xmath394 decreases , resulting in more accurate optimal paths . in this article we have shown that a device that combines stimulated emission of light with superconductivity can produce coherent radiation with exponentially long coherence times . we have introduced a general model for such hjl , where a big number of quantum states are coupled to two superconducting leads and are capable of emitting light by electron - hole recombination into a resonant optical mode . we argued that there should also be incoherent emission of photons in the environment that creates an imbalance in the population of emitter states required for lasing . the stochastic dynamics of the hjl is compactly described by a single fokker - planck equation , eq . , for three real variables , incorporating two noise sources . one noise source is the intrinsic quantum noise associated with the discrete changes of photon number in the resonator mode . the other noise source is related to the random switching between the eigenstates of the quantum emitters and causes the fluctuations of detuning . we have found a lasing regime where the nonlasing state is unstable . the optical phase is locked to the superconducting phase difference in two possible stable lasing states with a phase difference of @xmath6 . in this case , small fluctuations do not lead to decoherence , which is caused by a large fluctuation whereby the device switches between the two stable lasing states . to evaluate the coherence time of the hjl , we study these large fluctuations estimating their probability by the optimal path method . our numerical results show agreement with simple estimations deduced from the distribution of small fluctuations . we have found the decoherence times to be exponentially long @xmath404 $ ] provided @xmath405 . the optimal action is inversely proportional either to the number of photons in the laser or to the variance of the detuning . the latter is inversely proportional to the number @xmath49 of emitters in the laser . thereby we prove the feasibility of the exponentially long times for a big number of emitters . @xmath315 was also shown to depend on @xmath129 , the ratio of emitter switching rates to damping rate from the resonant mode , and the distance to the lasing threshold . to give an example we can express these conditions for a simple microscopical model formulated in terms of the parameters of the interaction hamiltonians [ eqs . and ] in sec . [ sec : i ] . first , we estimate the system parameters @xmath115 and @xmath107 [ eq . ] assuming that the energy of the states of the emitters is only perturbed by these two interactions . taking @xmath92 as the typical energy scale of the eigenstate of a single emitter and using perturbation theory , we can therefore estimate the order of magnitude of @xmath115 and @xmath107 @xmath406 here it is assumed that the population of emitter states is asymmetric as discussed in sec.[sec : i ] . to have lasing in the hjl requires @xmath407 . additionally , a large number of photons , @xmath2 , in the resonator mode requires @xmath408 , implying that @xmath409 . second , we estimate the variance of the detuning fluctuations , @xmath223 , which correspond to the fluctuations of @xmath106 [ eq . ] scaled with @xmath115 . using perturbation theory , we estimate @xmath410 . its variance is of the order of @xmath411 , with @xmath412 being a constant depending on the transition rates between the emitter eigenstates , where it is of the order of one when all rates are of the same order . this leads to @xmath413 where the estimation for @xmath115 was used . typically , this quantity should be much smaller than one . we list several ideas for possible applications of the hjl . in the hjl model , we neglected voltage fluctuations of the bias . when present , these fluctuations will transfer to fluctuations in the optical phase of the laser . by measuring these and providing negative feedback the voltage fluctuations , and hence the optical fluctuations , can be reduced and stabilized . another idea is similar to what was proposed in ref . . one can embed two hjls in a superconducting interference device ( squid ) and let them interact optically , so that an aharonov - bohm effect manifests in a shift of the optical phases that is flux - dependent . as an alternative to the latter idea , one can also let two distant squids interact via optically coupled hjls . this may allow for coherent interactions between widely separated superconducting qubits . finally , the exponentially long coherence time of the hjl can make it very suitable for metrology purposes , where the measurement of optical frequencies in the next - generation atomic clocks require ultrastable optical lasers.@xcite we acknowledge fruitful discussions with f. hassler , c. padurariu and c.j.o . verzijl and financial support from the dutch science foundation nwo / fom . the noise spectrum of eq . can be written in three parts , each part representing a peak . it is symmetric in @xmath414 and is given by @xmath415 with @xmath282 . defining @xmath416 , the coefficients are given by @xmath417 w + 32n_s\gamma\langle\tilde\chi^2\rangle ( a^2 - 1)^2\left [ 4(d-1)+\chi(d-\gamma^2 ) \right ] \big\ } , \\ c_2 & = \frac{1}{16dw}\big\ { 2w\left [ 4a^2+d(1+\chi ) \right ] + 64n_s\gamma\langle\tilde\chi^2\rangle ( a^2 - 1)^2\left [ \gamma^2 -4 + d(3+\chi ) \right ] \big\ } , \\ c_3 & = \frac{1}{4w}8n_s\langle\tilde\chi^2\rangle ( a^2 - 1)\left[4(a^2 - 1)-\gamma^2\right ] , \end{aligned}\ ] ] y.j . doh , j.a . van dam , a.l . roest , e.p.a.m . bakkers , l.p . kouwenhoven and s. de franceschi , science * 309 * , 272 ( 2005 ) ; j.a . van dam , yu.v . nazarov , e.p.a.m . bakkers , s. de franceschi and l.p . kouwenhoven , nature * 442 * , 667 ( 2006 ) . y. hayashi , k. tanaka , t. akazaki , m. jo , h. kumano and i. suemune , appl . express * 1 * , 011701 ( 2008 ) ; h. sasakura , s. kuramitsu , y. hayashi , k. tanaka , t. akazaki , e. hanamura , r. inoue , h. takayanagi , y. asano , c. hermannstdter , h. kumano and i. suemune , phys . rev . lett . * 107 * , 157403 ( 2011 ) ; i. suemune , h. sasakura , y. hayashi , k. tanaka , t. akazaki , y. asano , r. inoue , h. takayanagi , e. hanamura , j .- h . huh , c. hermannstdter , s. odashima and h. kumano , japanese journal of applied physics * 51 * , 010114 ( 2012 ) . krasnov , phys . 97 * , 257003 ( 2006 ) ; l. ozyuzer , a. e. koshelev , c. kurter , n. gopalsami , q. li , m. tachiki , k. kadowaki , t. yamamoto , h. minami , h. yamaguchi , t. tachiki , k. e. gray , w .- k . kwok and u. welp , science * 318 * , 1291 ( 2007 ) ; t. m. benseman , a. e. koshelev , k. e. gray , w .- k . kwok , u. welp , k. kadowaki , m. tachiki and t. yamamoto , phys . b * 84 * , 064523 ( 2011 ) .
we describe a superconducting device capable of producing laser light in the visible range at half the josephson generation frequency , with the optical phase of the light locked to the superconducting phase difference . an earlier proposed device , the so called `` half - josephson laser '' [ phys . rev . lett . 107 , 073901 ( 2011 ) ] , can not provide long coherence times , because of spontaneous switchings between the emitter states . to circumvent this we consider @xmath0 emitters driving an optical resonator mode . we derive a general model that captures essential physics of such devices while not depending on specific microscopic details . we find the conditions under which the coherence times are exponentially long , thus surpassing the fundamental limitation on the coherence times of common lasers . for this we study the noise in the device . in particular , we are interested in the rate of large fluctuations of the light field in the limit where the typical fluctuations are small . the large fluctuations are responsible for switching of the laser between stable states of radiation and therefore determine the coherence time . all numerous degrees of freedom of a bulk superconductor are frozen at low temperature . so its state is characterized by a single variable , the quantum phase of the superconducting order parameter.@xcite in principle , this phase remains constant , that is coherent , for infinitely long time . another example of a highly coherent system is a laser . there a macroscopic number of photons form a coherent state characterized by an optical phase and an amplitude.@xcite in contrast to superconductors the coherence time is not infinite , but limited by the intrinsic noise in the device , caused by spontaneous photon emission . this leads to a phase drift and thus to coherence loss at typical time scales of@xcite @xmath1 , with @xmath2 the number of photons in the mode and @xmath3 the cavity escape rate . coupling these two very different systems may open up novel ways of controlling either of them , allowing for manipulation of one with the other . recently , a device has been proposed@xcite that combines the coherence of laser light with the coherence of the superconducting order parameter , dubbed the `` half - josephson laser '' ( hjl ) . this device was based on yet another recently proposed device , the `` josephson light emitting diode''@xcite ( joled ) , which , it was shown , can be used for quantum manipulation purposes.@xcite the light in the joled is generated by a biased semiconductor _ p - n _ junction realized on a nanowire . to acquire a narrow emission spectrum of the junction , quantum dots are defined at the _ p - n _ interface . using nanowires an efficient coupling can be achieved between semi- and superconductors . by connecting both ends of the described _ p - n _ semiconductor nanowire to superconductors , superconductivity is induced at the quantum dot regions that alter the quantum dot states . electrons and holes jump from the leads to the quantum dots and recombine there , emitting a photon . even though the joled is biased with a voltage much larger than the superconducting gap , the superconducting correlations are retained because any direct , nonradiative charge transfer is blocked by the potential barriers in the _ p - n _ junction . the electron - hole recombination can happen in two ways . a `` blue '' photon at about the josephson frequency , @xmath4 , is emitted when cooper pairs from either side of the junction recombine . the phase of this blue photon is locked to the superconducting phase difference . a process resembling spontaneous parametric down - conversion@xcite allows for the emission of two `` red '' photons in stead of a single blue one , at about half the josephson frequency , @xmath5 . the joled is the basis for the setup of the hjl . the latter exploits the emission of red photons by the joled and amplifies it by stimulated emission in an optical resonator . in contrast to the case of the joled , where red photons are emitted spontaneously , the red photons in the resonant mode of the hjl form a coherent state . the phase of this coherent state is locked to the superconducting phase difference . lasing was found for the red emission with the optical phase locked to the superconducting phase difference , where only two values of the optical phase are allowed , having a phase difference of @xmath6 . the importance of the phase locking in this device is that it can result in exponentially long laser coherence times . the idea to prolong coherence times by phase locking has been thoroughly explored in the field of lasers.@xcite using , for instance , the technique of injection locking@xcite one could pump a laser with another one with longer coherence time , so that the pumped laser inherits this longer coherence time . however , in that case the coherence time is still limited by the fundamental noise processes in the pump laser , with some coherence time @xmath7 . the essential difference with the hjl is that the laser light inherits the coherence time of the superconductors , which is infinitely long . it will not become infinitely long , though , because the fundamental noise in the laser is still present and causes switchings to the lasing state with opposite phase , with exponentially small probability . hence , the coherence times can be exponentially long . unfortunately , despite this promising prospect , the hjl does not exhibit these exponentially long coherence times because switchings between quantum dot states result in a shorter coherence time . the magnitude and phase of the laser light depends on the actual quantum state of the joled in the hjl . not every state supports lasing however , and the ones that do give rise to various magnitudes and phases of the laser light . switchings between the quantum dot states therefore cause decoherence of the hjl . the switching rates are suppressed because they occur in combination with the emission of off - resonant photons , which is allowed because of the finite linewidth of the optical resonator . for an optimal choice of parameters it is found that , despite this very different decoherence mechanism , the coherence time is , @xmath7 , similar to that of the common laser and much shorter than the envisioned exponentially long coherence times . the design for the hjl is based on a single quantum emitter , putting rather high demands on the quality factor of the optical resonator . it was proposed@xcite to relax these demands by considering a large number , @xmath0 , emitters in a single resonator , increasing proportionally the upper bound on the damping rate at which lasing is achieved . as an added benefit , intensity fluctuations are expected to reduce as @xmath8 , thus realizing a less fluctuating light intensity . in this article , we thoroughly investigate the alternative idea , where , instead of a single joled , we consider a large number of emitters forming a dipole moment to drive the optical resonator mode . the model that we introduce will be quite general as we will only consider macroscopic quantities and leave out the microscopic details . the crucial aspect of the design , the coupling of the quantum states to both superconducting leads , will of course remain , since it both drives the laser and induces the phase lock between optical phase and superconducting phase difference . the high demands on the quality factor of the resonator can be relaxed because it is driven by a big number of emitters . we find that stationary lasing is possible with two equivalent values of the optical phase , as in the hjl , and with exponentially long coherence times owing to the phase lock . rather surprisingly , the sole increase in number of emitters does not directly guarantee the increased number of photons in the mode . the average dipole moment of all emitters will turn out to be zero when the emitters are coupled to the resonator mode only . incoherent emission to other modes is required to create a population imbalance,@xcite resulting in a nonzero average dipole moment . owing to the large number of emitters , single switchings will not cause decoherence as in the hjl . rather , they induce minor fluctuations of the dipole moment , transferring to detuning fluctuations in the resonator mode . these again transfer to fluctuations of magnitude and phase of the lasing state . the detuning fluctuations compete with the intrinsic fluctuations of the resonator mode . when both are sufficiently small , the superconducting phase difference will remain locked to a single value of the optical phase . the major decoherence mechanism of the laser is then provided by large fluctuations , represented by the tails of the noise distributions . these bring the optical field to the state with opposite phase . large switchings occur rarely , on exponentially long time scales . the importance of such a feature is hard to underestimate . indeed , examining these processes we have found coherence times ( surpassing the fundamental limit for common lasers ) , scaling as @xmath9 , up to some critical value @xmath10 , where it saturates . while the linewidth of the laser is directly related to this coherence time , there are large tails in the line shape associated with the noise in the dipole moment as well as to the intrinsic quantum noise . in the model we study , we neglect the noise source causing decoherence of the superconducting phase difference : voltage fluctuations . the superconducting phase difference is fixed only for ideal voltage bias . in case of nonideal bias , the voltage fluctuations will drive the phase evolution and cause the drift of the superconducting phase difference . the corresponding coherence time can be estimated using the voltage correlation function for a junction with impedance @xmath11 @xmath12 with @xmath13 the thermal energy for temperature @xmath14 . with this the average value of the phase is given by @xmath15 where the second josephson relation and the voltage correlation function were used for the final expression . from this we find a coherence time of @xmath16 . the coherence time can thus be prolonged by reducing temperature or impedance of the nanostructure . alternatively one can increase the number of josephson junctions to reduce the noise . the latter approach is used in metrology to realize the josephson voltage standard . a voltage drift of a few nv per hour can be realized@xcite in that way , leading to coherence times on the order of @xmath17 s. hence the voltage fluctuations may be disregarded when studying the decoherence in the device . this article is organized as follows . in sec . [ sec : i ] we derive the governing equations for a broader class of devices , characterized by a large number of emitters . we start by briefly treating the model of the hjl , as described in ref . , and subsequently argue that the resulting equations can be generalized to represent a wide range of devices . in sec . [ sec : ii ] it is shown that lasing with a phase lock occurs for these general equations in certain parameter regimes . there are distinguished regimes , separated by first and second order phase transitions . in sec . [ sec : iii ] conditions are derived to determine the parameter regime where the noise can be considered in linear approximation . the frequency dependence of the noise , when in a stationary lasing state , is calculated . the noise in linear approximation can not cause the decoherence of the laser owing to the phase lock . however , occasional large fluctuations will cause decoherence . these are treated in sec . [ sec : iv ] , for which the formalism of optimal paths is used . finally , we conclude in sec . [ sec : v ] .
it is widely accepted that the electronic properties of the cuprates are well described by the three - band hubbard model @xmath10 @xcite . zhang and rice @xcite suggested that this model can be reduced to an effective one - band @xmath11 model under certain simplifying hypothesis : small o - cu hopping @xmath12 and zero on - site o coulomb repulsion @xmath13 , cu - o coulomb nearest - neighbor interatomic repulsion @xmath0 and o - o hopping @xmath14 . while the one - band effective models can explain the magnetic properties of high-@xmath15 materials and can provide magnetic mechanisms for superconductivity @xcite , these models , with fixed parameters , do not take into account the effects of the cu - o repulsion @xmath0 . the density of carriers in cuprate superconductors is very low . the average distance between two holes is larger than 7 and the cu - o distance is @xmath16 1.9 and therefore @xmath0 is expected to be poorly screened . neglecting screening @xmath17 ev , while direct calculations give @xmath18 ev @xcite and constrained - density - functional results predict @xmath19 ev @xcite . thus , as pointed out early after the discovery of high-@xmath15 superconductivity , the effects of @xmath0 can be important @xcite . these effects include a charge - transfer mechanism for superconductivity @xcite , and marginal -fermi - liquid behavior @xcite . it is known that a sufficiently large @xmath0 induces a charge - transfer instability ( cti ) and a valence instability ( vi ) related with cu - o charge transfer @xcite . these instabilities are associated with phase separation because of the coupling between the fluctuations of the valence and the total density @xcite . the three - band hubbard model including @xmath0 has been studied by exact diagonalization of small clusters @xcite , random - phase approximation @xcite , gutzwiller variational wave function @xcite , and slave - bosons with 1/n expansion , for the particular case @xmath20 and with @xmath13 and @xmath0 treated in the hartree approximation @xcite . the results of these works are qualitatively similar in general . increasing @xmath0 the system reaches the cti , vi and the related phase separation . near the phase - separation boundary , the effective interaction in the cooper _ s _ and _ d _ channels becomes attractive . bang _ et al . _ in weak coupling @xcite and raimondi _ et al_. in strong coupling @xcite have shown that even though the energy @xmath21 of the charge - transfer collective excitonic mode at zero wave vector , decreases with increasing @xmath0 , it remains finite at the cti . however , this mode is coupled to the zero - sound mode , leading to a charge - transfer - mode mediated attraction which allows the violation of the landau stability criterion @xmath22 . this leads to the simultaneous divergence of the compressibility and charge - transfer static susceptibility indicating that phase separation and the cti take place . in spite of the qualitative agreement on the facts mentioned above among the different techniques , the regions of phase separation differ in the strong and weak coupling approaches , and the critical value of @xmath0 where the instabilities take place depends strongly on the approximation used also , the effect of the variation of the parameters has not been investigated in detail and realistic values of @xmath9 have not been studied so far . thus , further research on these subjects seems necessary . on the other hand , it is important to address the question of to what extent the above mentioned properties of the three - band hubbard model can be accounted for using a one - band effective model derived from the former through a sufficiently accurate mapping procedure . generalized @xmath11 and one - band hubbard models have been obtained performing systematic low - energy reductions of the three - band hubbard @xcite . these reductions either are based on a cell method @xcite or use an effective spin - fermion model with renormalized parameters as an intermediate step in the derivation @xcite . in contrast to the original derivation of zhang and rice , these methods allow to extend the mapping procedure to realistic and large values of cu - o hopping @xmath23 , and show that the zhang - rice singlets are stable for large cu - o covalency . there have been an important amount of research devoted to the study of the validity of the low - energy reduction and the stability of the zhang - rice singlets . in particular , the roles of local triplets states @xcite and apical o ions @xcite have been recently investigated . exact diagonalization of small clusters have shown that the low - energy spectrum @xcite and magnetic properties @xcite of the three - band model can be well reproduced by a @xmath24 model , where @xmath25 is the next - nearest - neighbor hopping and @xmath26 is a next - nearest - neighbor hopping combined with a nearest - neighbor spin flip . this three - body term , with the sign corresponding to large o - o hopping @xmath27 in the original three - band model , can stabilize a resonance - valence - bond superconducting ground state @xcite . in this paper , starting from the three - band hubbard model , we derive an effective generalized one - band hubbard model , using the cell method generalized to take into account properly the intercell part of the o intratomic repulsion @xmath13 and the cu - o interatomic repulsion @xmath0 . as a consequence of these interactions , the parameters of the effective model become dependent on the particle density . solving the effective one - band model exactly in a @xmath28 cluster , and also in the slave - boson mean - field theory of kotliar and ruckenstein @xcite we study the above mentioned valence instability ( vi ) , the regions of phase separation ( ps ) , and also the dependence of the effective one - band parameters with @xmath0 and doping . the advantage of our approach in comparison with other methods discussed above for the study of the vi , cti and ps , is that the intracell correlations , including cu on - site repulsion @xmath9 and a large part of the @xmath0 and @xmath13 terms , are taken into account exactly at each cell . this allows us in particular to take realistic values of @xmath9 ( @xmath29 ev ) , while in previous treatments @xmath9 was either small or infinite . in section ii we describe the different hamiltonians and the method for low - energy reduction . section iii contains the results and section iv is a short summary and discussion . we start from the three - band extended hubbard model : @xmath30 the sum over @xmath31 ( @xmath32 ) runs over all cu ( o ) ions . the vector @xmath33 ( @xmath34 ) connects a cu ( o ) site with one of its four nearest o atoms . the operator @xmath35 ( @xmath36 ) creates a hole with symmetry @xmath37 ( @xmath38 ) at site @xmath31 ( @xmath32 ) with spin @xmath39 . the phases of half of the orbitals have been changed in such a way that for all directions , the hoppings are positive ( @xmath23 , @xmath40 ) . the parameters of the model are known approximately from constrained - density - functional calculations @xcite . the first step in the cell method @xcite is to change the basis of the o orbitals to linear combinations which hybridize ( @xmath41 ) and do not hybridize ( @xmath42 ) with @xmath43 orbitals , due to the term in @xmath23 at each point * k * of the reciprocal space . we denote the wannier functions of these orbitals as @xmath44 and @xmath45 respectively . the @xmath46 ( _ non - bonding _ states ) lie very high in energy and are neglected . after the change of basis , the original hamiltonian can be separated in two parts , one containing the intracell ( and generally larger ) terms and another containing the intercell terms : @xmath47 with @xmath48 \nonumber \\ & & + u_{pd}\sum_{i\neq jl\sigma \sigma ^{\prime } } [ f(i - j , i - l)n_{i\sigma ^{\prime } } ^d\alpha _ { j\sigma } ^{\dagger } \alpha _ { l\sigma } + h.c . ] \nonumber \\ & & + u_p\sum_{ijlm}^{\prime } h(i - j , i - l , i - m)\alpha _ { i\uparrow } ^{\dagger } \alpha _ { m\uparrow } \alpha _ { j\downarrow } ^{\dagger } \alpha _ { l\downarrow } \label{inter}\end{aligned}\ ] ] the functions of the lattice vectors @xmath49 , @xmath50 , @xmath51 , @xmath52 and @xmath53 are given in ref . they decay rapidly with increasing argument and as a consequence , most of the original hoppings and interactions are contained in @xmath54 . the @xmath55 indicates that in the last sum the term with @xmath56 is excluded . the ordinary cell method consists in solving @xmath57 exactly in the subspaces of 0 , 1 and 2 particles and retaining the ground state in each subspace . the non trivial retained eigenstates have the form : @xmath58 @xmath59 the energies of these states will be denoted @xmath60 and @xmath61 respectively . the first term of @xmath62 corresponds to the zhang - rice singlet in the original derivation @xcite . our modification consists in considering the coefficients as variational parameters which are determined minimizing the total energy . we treat the terms in @xmath0 of @xmath63 with @xmath64 replacing @xmath65 by its expectation value . using the fact that @xmath66 , this is equivalent to the treatment of ref . similarly in the terms in @xmath13 of @xmath63 with @xmath67 and @xmath68 we replace @xmath69 by its expectation value . the same treatment is done for @xmath70 and @xmath71 . the remaining terms except those with @xmath67 and @xmath71 were neglected . the intercell hamiltonian is written as a function of the retained eigenstates of @xmath57 , and mapping the latter into those of the one - band hubbard model at site @xmath31 , one obtains the following effective model : @xmath72+t_{bb}n_{i,-\sigma } n_{j,-\sigma } + h.c . \nonumber \\ & & \sum_{<ij>\sigma \sigma ^{\prime } } v_{11}(1-n_{i,-\sigma } ) ( 1-n_{j,-\sigma ^{\prime } } ) n_{i\sigma } n_{j\sigma ^{\prime } } \nonumber \\ & & \sum_{<ij>\sigma \sigma ^{\prime } } v_{12}((1-n_{i,-\sigma } ) n_{j,-\sigma ^{\prime } } + n_{i,-\sigma } ( 1-n_{j,-\sigma ^{\prime } } ) ) n_{i\sigma } n_{j\sigma ^{\prime } } \nonumber \\ & & \sum_{<ij>\sigma \sigma ^{\prime } } v_{22}n_{i,-\sigma } n_{j,-\sigma ^{\prime } } n_{i\sigma } n_{j\sigma ^{\prime } } \label{1b}\end{aligned}\ ] ] we obtain the ground state energy of @xmath73 as a function of the three free variational parameters ( eqs . [ para1 ] and [ para2 ] ) using two different approaches : 1 . the slave - boson approximation ( sb ) of kotliar and ruckenstein @xcite after treating the nearest - neighbor repulsions in the hartree - fock approximation . exact diagonalization ( ed ) of a square cluster with 10 unit cells . we begin this section by studying the parameters of the one - band effective model and its dependence with the original parameters and doping . as a basis for our study we take the parameters of the original three - band hubbard model determined from constrained - density - functional approximation @xcite . we take the unit of energy as @xmath74 ev . however we also study the effect of @xmath0 and @xmath9 within a wider range . in the ( unrealistic ) limit @xmath13 @xmath75 the intercell interactions vanish ( @xmath76 ) and the variational parameters @xmath77 , @xmath78 ( eqs . [ para1 ] and [ para2 ] ) which minimize the total energy correspond to the eigenstates of the cell hamiltonian @xmath57 ( eq . [ cell ] ) for all dopings . thus the effective one - band parameters are independent of doping . increasing @xmath0 and @xmath13 , the intercell interactions appear and the variational parameters start to differ from those corresponding to the low - energy eigenstates of @xmath57 to take into account better the intercell interactions . the variational parameters as well as the one - band effective parameters become doping dependent , although this dependence is very weak for small @xmath0 and @xmath79 in figs . 1 - 4 we show the one - band parameters as functions of doping for different values of @xmath0 using ed . the effective on - site repulsion @xmath80 increases with @xmath0 except for small values of @xmath81 and high doping where the amount of o states in the local singlet ( eq . [ para1 ] ) increases and the mixed cu - o part ( which pays @xmath82 ) decreases . related with this fact , @xmath80 decreases with doping for small values of @xmath83 this reflects a _ metallization _ of the system with a larger amount of cu - o covalency which should be reflected in an increase of the conductivity . this might be related to the sudden increase in the hall conductivity observed in la@xmath84sr@xmath85cuo@xmath86 for @xmath87 @xcite . as a consequence of the larger amount of o holes in the doubly occupied cells , and the change in the variational parameters @xmath78 ( eq . [ para2 ] ) to minimize the loss of energy due to the cu - o repulsion @xmath88 the energy @xmath61 and the amount of o holes in the singly - occupied cells increase with doping for small @xmath81 and large @xmath0 . for others values of @xmath81 and @xmath0 , @xmath61 is rather insensitive to doping ( see fig . 2 ) . for small values of @xmath0 the three correlated hoppings of the effective one - band model are rather insensitive to doping and similar in magnitude . for large values of @xmath0 the last two terms in eq . ( [ inter ] ) can not be neglected and affect somewhat the magnitude of these hoppings . as for the case of the correlated hoppings , if @xmath89 the different effective nearest - neighbor repulsions are similar between them and rather independent of hopping . however in contrast to the former , the @xmath90 increase significantly with @xmath0 . for small @xmath81 and large @xmath0 , in spite of the fact that there is a significant charge transfer from cu to o with doping , the effective repulsions do not change very much because the decrease ( increase ) in cu - o repulsion is approximately compensated by the increase ( decrease ) in o - o repulsion . in figs . 5 and 6 we show the dependence of the expectation value @xmath91 @xmath92 on @xmath81 . as expected , @xmath93 increases with @xmath81 . for small @xmath0 the results obtained by exact diagonalization ( ed ) and slave bosons ( sb ) are qualitatively similar , although there are quantitative differences of the order of 20% for larger dopings . for larger values of @xmath0 there are some important qualitative differences between both methods which are discussed below , however both methods coincide in that for intermediate and small values of @xmath81 , @xmath93 _ decreases _ with doping as a consequence of cu - o charge transfer which overcompensates the effect of doping ( see figs . 6 and 7 ) . this is related with the increase in cu - o covalency and the _ metallization _ mentioned before . in contrast , for large values of @xmath94 @xmath93 _ increases _ faster with doping than for small values of @xmath0 . as a consequence of these opposite effects of @xmath0 for small and large @xmath81 , the dependence of @xmath93 with @xmath81 becomes more abrupt for large @xmath88 eventually driving the valence instability ( vi ) . the effect of reducing @xmath9 induces a larger cu occupancy @xmath93 at large values of @xmath81 favoring the vi . the static valence susceptibility is defined by : @xmath95 where @xmath96 is the total occupation per cell and @xmath97 is the o occupation per cell . @xmath98 differs from the charge - transfer susceptibility @xmath99 because the latter is defined at constant chemical potential instead of at constant @xmath96 . the valence susceptibility @xmath100 is shown in fig . 8 . for each value of @xmath101 , there is a critical value of @xmath0 ( @xmath102 ) for which @xmath98 diverges at a critical value of @xmath81 ( @xmath103 ) , indicating the presence of a vi . for @xmath104 there is a discontinuous transition as shown for example in fig . 6 for @xmath105 . the minimum possible value of @xmath106 occurs for maximum doping ( @xmath107 ) . the critical value @xmath108 increases with decreasing @xmath109 . note that @xmath102 found with sb is always lower than that found with ed . also , at @xmath110 the sb approximation gives an artificially large @xmath98 at the metal - insulator transition due to the vanishing of the double occupancy ( @xmath111 ) in the insulating phase and the exaggerated large increase of @xmath111 with @xmath112 in the metallic phase near the metal - insulator boundary . within the ed , @xmath111 grows smoothly with @xmath112 in absence of the vi as one expects . in fig . 9 we show the phase - separation diagram calculated with the sb approximation . contrary to previous strong - coupling mean - field approximations , where always one of the phases between which phase separation takes place has @xmath110 @xcite , we obtain , in qualitative agreement with weak - coupling approximations @xcite that phase separation takes place between a phase with large doping @xmath107 and another one with @xmath113depending of the parameters . as expected , the region of phase separation grows with increasing @xmath0 . at least for @xmath114 ( _ i.e. _ for parameters near those obtained by constrained - density - functional approximation @xcite ) there is no phase separation . the effect of decreasing @xmath9 on phase separation is to suppress it for large values of @xmath81 . this is related with the change of the character of the local singlet @xmath62 ( eq . [ para1 ] ) from mainly cu - o or o - o to a doubly occupied cu orbital . for low values of @xmath115 , the phase diagram is rather independent of @xmath116 also , the effects of @xmath0 and @xmath9 on the phase diagram on a qualitative level can be understood on general physical grounds : increasing these repulsions favors localization , inhibits the kinetic energy terms , and as a result the dependence of the energy on occupation is more flat , favoring phase separation . although using ed we have only a few possible different densities , due to the small size of the cluster , the results for the compressibility at these densities agree qualitatively with the sb phase separation diagram . however there are some important differences in a quantitative level : a ) there is no phase separation for @xmath117 . b ) for @xmath118 there is a small island inside which the system phase separates in two phases , both with compositions inside the interval @xmath119 . some of these features can be seen in fig . 10 , where we show the chemical potential @xmath50 as a function of doping . in table i we list the densities for which the compressibility is negative ( a sufficient but not necessary condition for phase separation ) according to our ed results for different parameters . in this work , we have studied the effects of cu - o coulomb repulsion @xmath0 on valence instabilities and phase separation . using a variational form of the cell method , we derive a one - band effective model with doping dependent parameters . this model has the form of a generalized hubbard with correlated hoppings and nearest - neighbor repulsions . this approach allows us to study realistic values of the on - site cu repulsion @xmath120 ev and is accurate enough for moderate values of @xmath0 and on - site o repulsion @xmath13 . we obtain that for parameters near those derived using constrained - density - functional approximation @xcite , in particular @xmath121 ev , the low- energy physics of the three - band hubbard model can be well represented by the effective one - band model , with doping independent parameters . the effect of @xmath0 is merely to renormalize the effective parameters and there are neither valence instabilities nor phase separation . for larger @xmath0 the effective parameters become strongly doping dependent and for @xmath122 ev valence instabilities and phase separation appear . in agreement with previous weak - coupling results @xcite , we obtain that phase separation starts to occur for large doping values ( @xmath123 according to the exact - diagonalization results or @xmath124 according to the slave - boson ones ) . the valence instability begins at @xmath107 in both treatments . this facts are in contrast to results of alternative previous strong - coupling approaches @xcite . this is probably due to the facts that on the one hand , the slave - boson approximation produces an artificial increase of @xmath98 at the metal - insulator transition ( @xmath110 ) , as discussed in section iii , favoring ps at low @xmath109 . on the other hand in refs . @xcite , the @xmath8 and @xmath13 terms were taken in a mean - field approximation , essentially equivalent to hartree - fock , ( and since @xmath9 is large , this induces an artificially large increase of the energy with doping ) , while in our present treatment most of the @xmath0 term and a large part of the @xmath13 one are treated exactly . in fact , we have verified that treating both terms in hartree - fock before performing the low - energy reduction to the one - band model ( as was done in previous studies of the metal - insulator transition @xcite ) , the results become qualitatively similar to those of refs . we are indebted to e. r. gagliano for providing us the program for the exact diagonalization . one of us ( m.e.s . ) would like to thank m. balia for very helpful discussions . m. e. s. is supported by the consejo nacional de investigaciones cientficas y tcnicas ( conicet ) , argentina . a. a. a. is partially supported by conicet . v. j. emery , phys . lett . * 58 * , 2794 ( 1987 ) . c. m. varma , s. schmitt - rink , and e. abrahams , solid state commun . * 62 * , 681 ( 1987 ) . p. b. littlewood , c. varma , and e. abrahams , phys . * 63 * , 2602 ( 1989 ) . f. c. zhang and t. m. rice , phys . b * 37 * , 3759 ( 1988 ) . e. dagotto and j. riera , phys lett . * 70 * , 682 ( 1993 ) ; e. s. heeb and t. m. rice , europhys . lett . * 27 * , 673 ( 1994 ) . c. d. batista and a. a. aligia , lt5375b . han rushan , c. chew , k. phua , and z. gan , phys . b * 39 * , 9200 ( 1989 ) . a. k. mc mahan , j. annett , and r. martin , phys . rev b * 42 * , 6268 ( 1990 ) ; m. hybertsen , m. schlter , and n. christensen , phys . rev b * 39 * , 9028 ( 1989 ) ; m. hybertsen , e. stechel , m. schlter , and d. jennison , phys . rev . b * 41 * , 11068 ( 1990 ) . ; j. grant and a. mc mahan , phys . rev . lett . * 66 * , 488 ( 1991 ) . e. r. gagliano , a. g rojo , c. a. balseiro , and b. alascio , solid state commun . * 64 * , 901 ( 1987 ) ; c. a. balseiro , a. g. rojo , e. r. gagliano , and b. alascio , phys . * b * 38 , 9315 ( 1988 ) ; j. e. hirsch , s. tang , e loh jr . , and d. j. scalapino , phys . lett .*60 * , 1668 ( 1988 ) . a. sudb , c. m. varma , t. giamarchi , e. b. stechel , and r. t. scalettar , phys . lett . * 70 * , 978 ( 1993 ) . i. e. perakis , c. m. varma , and a. e. ruckestein , phys . lett . * 70 * , 3867 ( 1993 ) ; c. m. varma , phys . lett . * 75 * , 898 ( 1995 ) . j. hicks , a. ruckenstein , and s. schmitt - rink , phys . b * 45 * , 8185 ( 1992 ) y. bang , g. kotliar , r. raimondi , c. castellani , and m. grilli , phys . b * 47 * , 3323 ( 1993 ) . r. raimondi , c. castellani , m. grilli , y. bang , and g. kotliar , phys . b * 47 * , 3331 ( 1993 ) ; m. grilli , r. raimondi , c. castellani , c. di castro , and g. kotliar , phys . rev . lett . * 67 * , 259 ( 1991 ) . p. b. littlewood , phys . b * 42*,10075 ( 1990 ) . n. kothekar , k. quader , and d. allender , phys . b * 51 * , 5899 ( 1995 ) . s. n. coppersmith and p. b. littlewood , phys . rev . b * 41 * , 2646 ( 1990 ) . d. hirashima , y. ono , t. matsuura , and y. kuroda , j. phys . soc . jpn . * 61 * , 649 ( 1992 ) . s. caprara and m. grilli , phys . b * 49 * , 10805 ( 1994 ) . h. b. schttler and a. j. fedro , phys . rev . b * 45 * , 7588 ( 1992 ) . j. h. jefferson , h. eskes , and l. f. feiner , phys . b * 45 * , 7959 ( 1992 ) l. f. feiner , j. h. jefferson , and r. raimondi , preprint . r. hayn , v. yushankhai , and s. lovstov , phys . b * 47 * , 5253 ( 1993 ) , and references therein . m. e. simn , m. balia , and a. a. aligia , physica c * 206 * , 297 ( 1993 ) . m. e. simn and a. a. aligia , phys . b * 48 * , 7471 ( 1993 ) . v. i. belinicher and a. l. chernyshev , phys . b * 49 * , 9746 ( 1994 ) ; v. belinicher , a. chernyshev , and l. popovich , phys . rev . b * 50 * , 13768 ( 1994 ) . a. a. aligia , m. e. simn , and c. d batista , phys . rev . b * 49 * , 13061 ( 1994 ) . c. d. batista and a. a. aligia , phys . rev . b * 47 * , 8929 ( 1993 ) . m. e. simn and a. a. aligia , phys . rev b * 52 * , 7701 ( 1995 ) . r. raimondi , l. feiner , and j. jefferson , physica c * 235 - 240 * , 2203 ( 1994 ) . c. d. batista and a. a. aligia , phys . b * 48 * , 4212 ( 1993 ) ; e * 49 * , 6436 ( 1994 ) . j. m. eroles , c. d. batista , and a. a. aligia , bd5722 . g. kotliar and a. e. ruckenstein , phys . * 57 * , 1362 ( 1986 ) . n. p. ong , z. z. wang , j. clayhold , j. m. tarascon , l. h. greene , and w. r. mckinnon , phys . b * 35 * , 8807 ( 1987 ) ; k. sreedhar and p. ganguly , phys . rev . b * 41 * , 371 ( 1990 ) * fig . 1 : * on - site repulsion of the effective one - band model ( eq . [ 1b ] ) obtained by exact diagonalization ( ed ) ( see text ) as a function of hole doping @xmath109 for different values of the cu - o charge transfer energy ( @xmath81 ) and cu - o repulsion : @xmath0=1 ( solid ) , 2 ( dashed ) , 3 ( dotted ) , 4 ( dotted - dashed ) . other parameters of the original three - band model ( eq . [ 3b ] ) are : @xmath125 , @xmath126 , @xmath127 , @xmath128 . * fig . 2 : * one particle on - site energy of the effective one - band model ( eq . [ 1b ] ) as a function of doping for different values of @xmath129 and @xmath130 the three lower curves correspond to @xmath131 and the other three to @xmath132 . other parameters as in fig fig . 3 : * nearest - neighbor hoppings of the effective one - band model ( eq . [ 1b ] ) as a function of doping for different values of @xmath129 ( ( a ) @xmath131 , ( b ) @xmath132 ) and @xmath0 . other parameters as in fig fig . 4 : * nearest - neighbor coulomb repulsions of the effective one - band model ( eq . [ 1b ] ) as a function of doping for different values of @xmath81 and @xmath0=1 ( solid ) , and 3 ( dotted ) . other parameters as in fig . 1 . * fig . 5 : * cu occupancy as a function of cu - o charge transfer energy ( @xmath81 ) for different dopings ( indicated at the right of the corresponding curve ) obtained by ( a ) exact diagonalization and ( b ) slave bosons . other parameters are @xmath133 , @xmath125 , @xmath126 , @xmath127 , @xmath134 * fig . 6 : * same as fig . 5 for @xmath135 @xmath136 . * fig . 7 : * cu occupancy as a function of doping for different values of @xmath81 ( indicated at the right of the corresponding curve ) and @xmath137 , obtained by slave bosons ( sb ) . other parameters are @xmath118 , @xmath126 , @xmath127 , @xmath134 * fig . 8 : * valence susceptibility as a function of @xmath81 for the different positive dopings of fig . 6 obtained by ( a ) exact diagonalization and ( b ) slave bosons . other parameters as in fig . fig . 9 : * phase - separation diagram @xmath81 vs. @xmath109 obtained using sb for different values of @xmath9 and @xmath0 . the dotted line indicates the first - order valence transition for @xmath138 ( for @xmath139 there are no first order vi transition ) . the solid squares represent the same obtained with ed . other parameters as in figs . 1,5 . * fig . 10 : * chemical potential as a function of doping for several values of @xmath81 calculated with ( a ) sb and ( b ) ed . other parameters are @xmath118 , @xmath125 , @xmath126 , @xmath127 , @xmath134 * table i : * values of the density @xmath96 for which the compressibility is negative according to ed for different values of @xmath9 , @xmath8 and @xmath81 . other parameters are @xmath126 , @xmath127 , @xmath140
we study the cu - o valence instability ( vi ) and the related phase separation ( ps ) driven by cu - o nearest - neighbor repulsion @xmath0 , using an effective extended one - band hubbard model ( @xmath1 ) obtained from the extended three - band hubbard model , through an appropriate low - energy reduction . @xmath1 is solved by exact diagonalization of a square cluster with 10 unit cells and also within a slave - boson mean - field theory . its parameters depend on doping for @xmath2 or on - site o repulsion @xmath3 . the results using both techniques coincide in that there is neither vi nor ps for doping levels @xmath4 if @xmath5 ev . the ps region begins for @xmath6 ev at large doping @xmath7 and increases with increasing @xmath8 . the ps also increases with increasing on - site cu repulsion @xmath9 . pacs numbers : 74.72.-h , 74.25.jb
the entanglement dynamics of open quantum systems has attracted considerable interests over recent years @xcite . by investigating a system of two qubits interacting independently with their corresponding vacuum reservoirs , ref . @xcite has pointed out the sudden termination of the entanglement initially owned by the two qubits at a finite interval , which is called entanglement sudden death ( esd ) . this intriguing phenomenon has recently been demonstrated experimentally by linear optics systems exp - lo1,exp - lo2 and by atomic ensembles @xcite . a great deal of theoretical investigations of esd have so far been reported yu1,lopez , ikram , asma , davidovich , ficek1,ficek2,bellomo1,bellomo2,zeno , bellomo3,paris , cao , wang . for example , based on the same model as in ref . @xcite , it was found that esd regarding the qubits would always occur under the thermal and squeezing reservoirs @xcite . meanwhile , studies of esd have been extended to multi - particle systems @xcite , which indicated the delay of esd time with more qubits involved . in addition , some studies have also showed that entanglement of qubits will revive in the case of a commonly shared reservoir @xcite or of independent non - markovian reservoirs @xcite . a question naturally arises : where has the lost entanglement gone when esd occurs ? an answer to the question has been given recently by @xcite . considering the systems and the reservoirs as a whole , ref . @xcite presented the lost entanglement to be transferred to reservoir degrees of freedom , which is called `` entanglement sudden birth '' ( esb ) of the reservoirs . intriguingly , esd of the systems and esb of the reservoirs are of some symmetry and esb might be occurring before , simultaneously with or even after esd , depending on different initial states . however , the discussions in @xcite are restricted to the weak coupling regime under the markovian approximation , which only works when the reservoir correlation time is small compared to the relaxation time of the system @xcite . if the qubit is strongly coupled to its environment or the characteristic time of the system is shorter than the environmental correlation time open - book , solid , such as in recent experiments with cavity qed system exp - qed or solid - state systems @xcite , the situation would be much complicated and we have to give up the markovian approximation in treatment . the present paper is focused on the study of entanglement dynamics without the markovian approximation for two qubits interacting , respectively , with two independent reservoirs . for the initial condition with the two qubits entangled , we will study esd and esb , and show their different variation from the markovian treatment . we consider a two - qubit system coupling to two uncorrelated vacuum reservoirs at zero temperature . since there is no interaction between the two pairs of qubit - reservoir , the dynamics of the whole system can be obtained simply from the evolution of the individual pairs lopez , bellomo1 . the hamiltonian of the individual qubit - reservoir pair under the rotating wave approximation is given by ( @xmath0@xmath1where @xmath2 is the resonant transition frequency of the qubit between levels @xmath3 and @xmath4 with @xmath5 and @xmath6 , @xmath7 @xmath8 @xmath9 @xmath10 are the frequency , creation ( annihilation ) operator and the coupling constant for the _ _ k__th mode of the reservoir . for simplicity in our treatment , only one excitation of the reservoir will be considered . consequently , the initial state @xmath11 will evolve to @xmath12 where @xmath13 and @xmath14 is the reservoir state with one excitation in the _ _ k__th mode and other states in vacuum . assuming the initial conditions @xmath15 and @xmath16 we could obtain the following equation by solving the schrdinger equation of eq . ( 2)@xmath17@xmath18 where the correlation function @xmath19 in the limit of @xmath20 is of the form @xmath21 with @xmath22 the spectral density of the reservoir . in what follows , we will consider in our calculation the lorentzian spectral distribution @xmath23 open - book@xmath17 in which @xmath24 and @xmath25 with @xmath26 and @xmath27 the qubit relaxation time and the reservoir correlation time , respectively open - book . in a non - markovian regime , @xmath28 i.e. , the reservoir correlation time is longer than the relaxation time of the qubit , and @xmath29 means a markovian regime @xcite@xmath30 as a result , in the non - markovian regime , the probability amplitude can be easily solved as@xmath31 , \label{4}\ ] ] with @xmath32 if we set @xmath33 , eq . ( 2 ) can be rewritten as@xmath34 with @xmath35 /\tilde{c}(t)$ ] ( @xmath36 ) . we will employ concurrence @xcite to assess entanglement . in the case of the density matrix with the `` x '' type@xmath37 the concurrence could be easily calculated by @xcite@xmath38 we consider that the two qubits are initially entangled and the two reservoirs are in zero - temperature vacuum states , which is described as yu - esd,@xmath39using eq . ( 5 ) , we could straightforwardly reach the reduced density matrices for the two qubits,@xmath40for the two reservoirs,@xmath41for the qubit-1 and reservoir-1,@xmath42and for the qubit-1 and reservoir-2,@xmath43 as eqs . ( 9)-(12 ) are all of the `` x '' types , the concurrence of the partitions @xmath44 @xmath45 , @xmath46 and @xmath47 can be calculated directly using eq . in addition , eq . ( 7 ) also implies that the esd time of the qubits or the esb time of the reservoirs can be obtained from @xmath48 and @xmath49 however , as we could not solve them in an explicitly analytical way , numerical calculations are given below instead . 1 shows the entanglement dynamics of @xmath50 , @xmath45 , @xmath51 and @xmath47 in the non - markovian regime with @xmath52 in fig . 1(a ) , there are three regimes for entanglement variance regarding the two qubits : for @xmath53 no esd will occur ; for @xmath54 there will be entanglement revival after esd takes place ; as for @xmath55 entanglement will vanish forever at a finite interval @xcite . in contrast to the behavior regarding the two qubits , there are only two @xmath56dependent regimes for entanglement variance regarding the two reservoirs : esb or no esb , as shown in fig . 1 ( b ) . but no matter in which regime , the two reservoirs will finally be entangled in a time - dependently oscillating manner . the entanglement transfer is done step by step from the qubits to the reservoirs , which is reflected in figs . 1(c ) and 1(d ) with the entanglement variance regarding the partitions @xmath46 , @xmath47 etc . . the partitions @xmath46 , @xmath47 behave as something like relays , which get and release entanglement in an oscillating way . meanwhile the @xmath45 entanglement is enhanced also in the oscillating way . to see more clearly and also to make comparison with the markovian treatment , we have plotted fig . 2 for the entanglement dynamics with @xmath57 @xmath58 in the non - markovian regime , e.g. , @xmath59 when esd occurs at @xmath60 , the entanglement of the qubits has been transferred to other partitions , e.g. , @xmath61 , @xmath62 @xmath47 etc . . in more details , we may first visualize that within an interval @xmath63 in which the entanglement transfer from qubit - qubit to qubit - reservoir is greater than that from qubit - reservoir to reservoir - reservoir degrees of freedom , and the entanglement of qubit - reservoir and reservoir - reservoir are both increasing . then from the critical point @xmath64 the qubit - reservoir entanglement begins to be decreasing until @xmath65 at which the entanglement of qubit - reservoir has been totally transferred to the reservoir degrees of freedom . however , in contrast to the case of markovian reservoirs with the entanglement fully transferred to the reservoirs at one time , due to the memory effect of the non - markovian reservoirs , entanglement will partially be recoiled to the qubit - reservoir degrees of freedom , or even back to the qubit - qubit degrees of freedom . as a result , revival of entanglement takes place bellomo1,bellomo2 . this back and forth process will repeat until the entanglement of the qubits have been completely moved to their corresponding reservoirs . finally , the reservoirs are entangled . moreover , fig . 2 also presents that , the stronger the qubits coupling to their reservoirs , the more evident the non - markovian memory effects . in summary , * * * * we have investigated two - qubit entanglement affected by the decoherence from non - markovian reservoirs . in contrast to the smooth decay of the qubit entanglement in the case of markovian reservoirs , the entanglement of the qubits will revive after esd occurs in the non - markovian case . meanwhile , by means of the qubit - reservoir partitions as the relays , the reservoirs will be asymptotically entangled in an oscillating manner , which is a good interpretation of entanglement revival of qubits in the non - markovian reservoirs . we argue that our present study would be useful for quantum information science . for example , the solid - state qubits in systems with strong correlation always experience strong decoherence solid , exp - solid , exp - spin . so quantum information processing on any qubit in such systems should be affected by detrimental influence from the non - markovian reservoir . moreover , fast logic gating is necessary in quantum information processing . as the characteristic time of the system during the fast gating is very short , we have to consider the non - markovian effect from environment in assessing the operational fidelity . we have noticed recent studies on non - markovian reservoir from the microscopic way zhang1,zhang2 . our present work from the phenomenological viewpoint , however , could explain some experimental observations of entanglement loss regarding qubits more straightforwardly . on the other hand , entanglement is the unique resource in quantum information science . as no system could be completely isolated from external noise , it is of interest to understand what happens in the entanglement transfer from the system to the environment , although whether the reservoirs are entangled or not is actually of no practical application with current technology . the work is supported by nnsf of china under grants no . 10774163 and no . 10774042
we investigate the entanglement dynamics of two initially entangled qubits interacting independently with two uncorrelated reservoirs beyond the markovian approximation . quite different from the markovian reservoirs [ c. e. lpez _ et al_. , phys . rev . lett . 101 ( 2008 ) 080503 ] , we find that entanglement sudden birth ( esb ) of the two reservoirs occurs without certain symmetry with respect to the entanglement sudden death ( esd ) of the two qubits . a phenomenological interpretation of entanglement revival is also given . entanglement dynamics , non - markovian reservoir , entanglement sudden death ( esd ) , entanglement sudden birth ( esb ) 03.67.mn , 03.65.ud , 03.65.yz
in general relativity , gravitational energy has no proper local description due to the equivalence principle . although there is a well known total energy defined by arnowitt , deser and misner ( adm ) , we need some definitions for the practical applications which involve a finite region rather than the whole space . this leads to the concept of quasi - local energy . there are many different proposals for defining quasi - local quantities , here we follow the covariant hamiltonian approach by chen , nester and tung ( cnt ) @xcite , and use the method in @xcite to determine the reference . the physical significance of the choice of reference is the choice of the _ zero - point _ value of energy ( * ? ? ? * , a. substruction term ) . for a survey of the covariant hamiltonian approach one may refer to @xcite . a major difficulty for the hamiltonian approach is to find a suitable way to identify the reference . basically there are two ways to determine the reference : the first one is the `` analytic matching '' in which the reference variables are directly restricted from the physical variables , for example , in schwarzschild spacetime , one can calculate the metric and connection as the physical variables and then take the @xmath14 limit to be the reference ones @xcite . although it is sometimes convenient to do this calculation , it is not so clear as to what is the meaning of the reference one gets ; the second is the `` 4d isometric matching '' which is based on the 2-surface @xmath15 isometric embedding into a reference spacetime and the embedded surface @xmath16 have the same orientation and also the spacetime @xmath17 and the reference spacetime @xmath18 have the same time orientation and space orientation . ] and then making an extension in the `` tubular '' neighborhood of @xmath15 in the normal plane . one can identify the local 4-frame of the physical and the reference spacetime _ just on @xmath15_. for an asymptotic flat spacetime , a resonable choice of reference is minkowski spacetime , and for measuring the quasi - local _ energy _ , the corresponding displacement vector field on @xmath15 is the timelike vector field @xmath19 identical to the timelike spacetime killing vector field @xmath20 on @xmath16 of the reference spacetime through the 4d isometric matching @xcite . in @xcite we found that for a specific decomposition of @xmath19 , the result is the same as the wang - yau energy @xcite . one may refer to the surveys on the wang - yau energy @xcite . the application to kerr spacetime using this method was considered in @xcite . the results are ( i ) for @xmath21 , the critical value of the qle agrees with martinez s result @xcite for the slow rotation approximation . ( ii ) the numerical results implies that qle is decreasing corresponding to increasing @xmath11 , and monotonically decreasing to the value @xmath9 w.r.t . the spatial radius @xmath10 . ( iii ) the value of the quasi - local angular momentum is the _ constant _ @xmath22 . later , tam and one of the authors found that this critical value of the qle is actually the brown - york mass ( * ? ? ? * theorem 2.1 ) ( for @xmath23 ) , and also the critical point is the unique solution in the region @xmath24 for the slow rotation approximation ( the global uniqueness is still not clear ) . note that @xmath25 is the outer horizon of the kerr spacetime . in this paper , we analyse the qle of kerr spacetime for unrestricted rotation and inside the ergosphere ( * ? ? ? * box 33.2 ) . ] . the situation is different from the case @xmath26 and slow rotation . for @xmath10 less than @xmath27 , there are limits on the isometric embedding ; for non - slow rotation , there is no known theorem for positivity in some region . as mentioned before , the qle we considered depends on the 2-surface isometric embedding into @xmath8 . there are limits on @xmath11 for the surface s isometric embedding . it is known that in the kerr spacetime , there is no event horizon for @xmath28 , and hence the singularity is naked . for @xmath23 but not small , smarr found that if @xmath29 , the gauss curvature is negative near the poles , which implies that the outer horizon can not be embeded into @xmath8 _ isometrically _ @xcite . one can imagine that when @xmath11 grows up over that limit , the horizon `` warps '' out of @xmath8 from the poles ( * ? ? ? * fig.4 ) until the extreme limit ( @xmath30 ) , and no horizon for @xmath28 . something like tearing a sticker off a table , part of it is out but still some part stays on the table . we are interested in the limit not just for the horizon . the main results of this paper are : ( i ) the gauss curvature is positive if and only if @xmath1 ( see ) , where @xmath31 is the unique real root of @xmath32 . this also gives the limit for the isometric embedding ; ( ii ) the _ integrand _ of the critical value of the qle ( i.e. @xmath33 ) is positive if and only if @xmath3 ( therefore the qle is positive ) . based on these results , we conclude that in a certain region @xmath13 ( see ) , the critical value of kerr qle is a minimum as the corollary from chen - wang - yau s result @xcite . note that the three curves : the outer horizon @xmath34 , @xmath31 and @xmath6 intersect at the point @xmath7 . in the _ triangle - like _ region ( see figure 4 ) , @xmath35 for @xmath36 , the integrand @xmath33 is not always positive . that means we can not use the theorem in @xcite to prove the minimized property in that region , and also it is not mathematically proved that the qle is positive in that region . however , the numerical results imply the positivity and a monotonic decreasing property . this extends the results of @xcite into the region @xmath37 , and also generalizes the result of martinez @xcite to non - slow rotation . about the situation that the qle is decreasing corresponding to the increasing @xmath11 , we found that if we replace @xmath11 by the irreducible mass @xmath12 , the qle is monotonically increasing in @xmath12 . let @xmath38 and @xmath39 be two oriented and time - oriented spacetimes which are considered as the physical spacetime and the reference spacetime , respectively . let @xmath40 be a closed space - like surface in @xmath17 . a 4d isometric matching reference of @xmath15 is defined as a smooth embedding @xmath41 preserving orientations and time - orientations on some open neighborhood @xmath42 of @xmath15 such that @xmath43 on @xmath15 . let @xmath19 be a future directed time - like vector field on @xmath17 . then , the cnt quasi - local energy ( see @xcite ) of @xmath15 with respect to @xmath19 and a 4d isometric matching reference @xmath44 is defined as @xmath45.\ ] ] here , @xmath46 and @xmath47 are the connection forms of the levi - civita connections for @xmath48 and @xmath49 respectively , @xmath50 is the natural inclusion map , and @xmath51 although the 4d isometric matching reference @xmath44 is by definition smooth on a neighborhood of @xmath15 , the value of @xmath52 depends only on the 1-jet of @xmath44 ( see @xcite ) . that is to say , @xmath53 for any two references @xmath54 and @xmath55 with @xmath56 on @xmath15 . consider the axisymetric kerr - like spacetime @xmath57 in the boyer - lindquest coordinate @xcite : @xmath58 where the components @xmath59 are functions of @xmath10 , @xmath60 . for kerr spacetime , they are @xmath61 here , we suppose @xmath62 . then the 4d isometric matching equation for an axially symmetric embedding : @xmath63 into @xmath64 of a constant radius surface @xmath65 is indeed explicitly solvable ( see @xcite ) . set @xmath66 and @xmath67 then , @xmath68 with ( see appendix of @xcite ) @xmath69 where @xmath70 the euler - lagrange equation of @xmath71 is @xmath72 which has an obvious solution @xmath73 . it was shown in @xcite that @xmath74 here @xmath75 is the brown - york mass of @xmath15 @xcite , where @xmath76 and @xmath77 are mean curvatures of @xmath15 and @xmath78 respectively . note that for the brown - york definition , @xmath78 is the isometric embedding of @xmath15 into @xmath8 . here if the constant @xmath79 , @xmath80 surface can be embedded into the constant @xmath81 hyperplane of @xmath82 , then the critical value of the qle is equal to the brown - york mass , as shown in @xcite . let @xmath83 be the gaussian curvature , @xmath84 be the mean curvature of the constant radius surface with @xmath85 , and @xmath86 be the mean curvature of @xmath78 . then , we obtain the following results : [ thm - main ] * when @xmath87 , @xmath88 for any @xmath60 if and only if @xmath89 here , @xmath90 is the unique real root of the cubic polynomial equation : @xmath91 , @xmath923\sqrt[3]{27a^2m+\sqrt 3\sqrt{243a^4m^2+a^6}}}\\ & + \frac{\sqrt[3]{27a^2m+\sqrt 3\sqrt{243a^4m^2+a^6}}}{\sqrt[3 ] 9 } ; \end{split}\ ] ] * when @xmath93 , @xmath94 for any @xmath60 if and only if @xmath95 note that @xmath90 is an increasing function of @xmath96 $ ] with @xmath97 and @xmath98 . so that ( 1 ) of theorem [ thm - main ] improves a result of ( * theorem 3.1 ) . moreover , note that the graphs of @xmath99 , @xmath100 and @xmath101 intersect at the point @xmath102 ( see figure [ fig:4 ] ) . the triangle - like region @xmath103 is defined by @xmath104\}.\ ] ] when @xmath105 is in this region , the constant radius surface @xmath106 is outside the outer horizon with positive gaussian curvature but @xmath107 for some @xmath60 . therefore there is no known theorem to guarantee the positivity of qle . later we show the numerical results which imply positivity and the monotonically decreasing property . by setting @xmath108 , @xmath109 , @xmath110 and preserving @xmath60 and @xmath111 we have @xmath112 , where @xmath113 ^ 2+\frac{\sin^2\theta}{\tilde{\sigma}^2}[(\tilde{r}^2+\tilde{a}^2)d\phi-\tilde{a}d\tilde{t}]^2\\ & + \frac{\tilde{\sigma}^2}{\tilde{\delta}}d\tilde{r}^2+\tilde{\sigma}^2d\theta^2,\nonumber \end{split}\ ] ] @xmath114 , @xmath115 . so without loss of generality , we will simply assume that @xmath116 in the proofs in the following sections . in the previous section , we introduced the qle of kerr spacetime of the surface with constant @xmath2 , and the critical value of the qle is the brown - york mass , which involves the 2-surface isometric embedding into @xmath8 . it can be solved explicitly in the kerr case for positive gaussian curvature . from @xcite , the gaussian curvature @xmath117 of the constant radius surface @xmath106 for the kerr - like spacetime is @xmath118 it can be simplified as @xmath119 let @xmath120 . then from we have @xmath121 by direct computation , we have @xmath122 and @xmath123 we are now ready to prove ( 1 ) of theorem [ thm - main ] . note that @xmath124 @xmath125 and by @xmath126 substituting these into the second equality of , we have @xmath127 ^ 2 } , \end{split}\ ] ] where @xmath128 - 2h_{\sigma\sigma}[(a^2+r^2)^2-\delta a^2\sigma](1-\sigma).\ ] ] then , by , @xmath129\\ & -2h_{\sigma\sigma\sigma}[(a^2+r^2)^2-\delta a^2\sigma](1-\sigma)\\ = & \frac{4r(a^2+r^2)^2a^2 u(\sigma)}{\left((a^2+r^2)-a^2\sigma\right)^4 } , \end{split}\ ] ] where @xmath130\\ & -6a^2[(a^2+r^2)^2-\delta a^2\sigma](1-\sigma ) . \end{split}\ ] ] note that @xmath131 , @xmath132 since @xmath133 , and @xmath134 so , @xmath135 for any @xmath136 $ ] . then , by , @xmath137 is increasing in @xmath138 $ ] . moreover , note that @xmath139 therefore by , we have @xmath140 when @xmath141 and @xmath142 conversely @xmath143 so that if @xmath144 , we must have @xmath145 . this completes the proof of ( 1 ) in theorem [ thm - main ] . in this section , we prove ( 2 ) of theorem [ thm - main ] . the integrand of the critical quasi - local energy is @xmath33 . the mean curvature @xmath77 @xcite of the constant radius surface @xmath106 for kerr - like spacetime is @xmath146 and the mean curvature @xmath76 of @xmath106 isometrically embedded into @xmath147 ( if such an embedding exists ) is @xmath148 let @xmath149 , from we obtain @xmath150 substituting this into , we have @xmath151 from the last equality of for @xmath116 and @xmath120 we have @xmath152 from , when @xmath141 , by direct computation we have @xmath153 to prove the necessary condition of theorem [ thm - main ] ( 2 ) , it is sufficient to consider the north pole @xmath154 . by we have @xmath155 so that @xmath156 . from the mean curvature is @xmath157 consequently , when @xmath141 and @xmath145 , we have @xmath158 therefore @xmath159 is necessary for @xmath160 . conversely , note that the three curves @xmath100 , @xmath99 and @xmath161 intersect at the point @xmath162 . it is clear that @xmath163 when @xmath164 , and @xmath165 for @xmath166 . so , by ( 1 ) of theorem [ thm - main ] , when @xmath141 and @xmath159 , we obtain @xmath144 for @xmath167 . from it is not difficult to see that @xmath168 implies @xmath169 for @xmath170 . hence , by @xmath171 ^ 2}.\ ] ] where @xmath137 is given in and @xmath172 let @xmath173 by direct computation , @xmath174}{(r^2+a^2-a^2\sigma)^2}\\ \leq&\frac{\delta a^2[(r+1)^2(r^2+a^2)^2-(r-1)^2r^4]}{(r^2+a^2-a^2\sigma)^2}\\ \leq & \frac{\delta a^2[(r+1)^2(r^2 + 1)^2-(r-1)^2r^4]}{(r^2+a^2-a^2\sigma)^2}\\ = & \frac{\delta a^2(4r^5 + 2r^4 + 4r^3 + 3r^2 + 2r+1)}{(r^2+a^2-a^2\sigma)^2}\\ \leq & \frac{16\delta a^2r^5}{(r^2+a^2-a^2\sigma)^2}\\ \leq&\frac{16\delta a^2r^5(r^2+a^2)^2}{(r^2+a^2-a^2\sigma)^4}\\ \end{split}\ ] ] since @xmath175 . then , by , @xmath176 where @xmath177 with @xmath42 given in . note that @xmath178 moreover , by @xmath179 since @xmath180 , and by @xmath181 hence @xmath182 for any @xmath136 $ ] . by , @xmath183 is increasing on @xmath138 $ ] . by , we know that @xmath184 when @xmath185 and @xmath186 . so @xmath187 on @xmath138 $ ] and hence @xmath160 for any @xmath60 . in @xcite , we found that the cnt quasi - local energy is closely related to wang - yau s energy @xmath188 @xcite . more precisely , let @xmath189 be an isometric embedding and @xmath190 be the time component of the embedding , and suppose the mean curvature vector @xmath191 of @xmath15 in @xmath17 is spacelike . then , @xmath192 here @xmath193 , @xmath191 and @xmath194 are the mean curvature vector of @xmath15 and @xmath78 respectively , @xmath195 is a future - directed time - like vector such that @xmath196 @xmath197 is orthogonal to @xmath195 and @xmath15 , and pointed outside if @xmath15 encloses a domain @xmath198 , @xmath199 is pointing outside and orthogonal to @xmath78 and @xmath200 , @xmath201 is a future - directed time - like vector which is orthogonal to @xmath78 and @xmath199 , @xmath202 and @xmath44 is a 4d isometric matching extension of @xmath203 @xcite . @xmath204 and @xmath205 here mean the gradient and laplacian of @xmath190 with respect to the induced metric on @xmath15 . by direct computation ( * appendix ) , the first equation of corresponds to @xmath206 let @xmath207 with @xmath208 decided by @xmath209 from the first equation of . then , by the uniqueness of the isometric embedding into @xmath147 and the relation , we obtain @xmath210 where @xmath190 depends only on @xmath60 , and @xmath211 . furthermore , by combining theorem [ thm - main ] and theorem 3 of chen - wang - yau @xcite , and ( other minimizing properties of the wang - yau qle can be found in @xcite ) , we have the following corollary : [ cor - main ] 1 . when @xmath212 , @xmath213 where @xmath190 is only a function of @xmath60 such that @xmath214 has positive gaussian curvature ; 2 . when @xmath212 , @xmath215 where @xmath209 is a function of @xmath60 such that @xmath216 has positive gaussian curvature . here @xmath217 this implies that the critical value of the kerr qle is the global minimum . an interesting question is whether corollary [ cor - main ] is satisfied in the triangle - like region @xmath218 . in order to use chen - wang - yau s result @xcite , one may check that @xmath219 is a critical point of @xmath220 . according to @xcite , @xmath219 is a critical point of @xmath220 if and only if @xmath221 , where @xmath222 is the vector field dual to @xmath223 which is the connection one form defined by @xmath224 . here @xmath225 is the spacelike unit vector normal to @xmath15 and @xmath226 is the future directed timelike vector normal to the hypersurface @xmath198 . by definition @xmath227 , we have @xmath228 which implies @xmath229 , where @xmath230 is the connection coefficient corresponding to the orthonormal frame @xmath231 , @xmath232 , @xmath233 , @xmath234 . here @xmath226 is the locally nonrotating observer ( see the next section ) with @xmath235 and angular velocity @xmath236 . the divergence of @xmath222 is @xmath237,\ ] ] where @xmath238 is the determinant of the metric induced on @xmath15 and @xmath239 , @xmath240 . note that the metric components of kerr spacetime in boyer - lindquest coordinate depend only on @xmath241 so that the connection coefficients are independent of @xmath111 . this implies that @xmath242 vanishes . it is not difficult to find that @xmath243 and consequently @xmath221 , which implies @xmath219 is a critical point of @xmath220 . by the 4d isometric matching , the displacement vector @xmath19 is identical to the reference timelike killing vector : @xmath244 , and @xmath245 on the 2-surface @xmath15 . in the physical spacetime , although the timelike vector @xmath246 becomes spacelike inside the ergosphere , @xmath19 is still timelike , with components ( * ? ? ? * eq(47 ) ) @xmath247 it is the _ locally nonrotating observer _ @xcite , or the so - called zamo ( zero angular momentum observer ) at the critical point @xmath248 : @xmath249 note that this timelike @xmath19 has the form @xmath250 @xcite , where @xmath251 with the angular velocity @xmath236 and @xmath252 is timelike in the regions outside the outer horizon and inside the inner horizon . from the result of @xcite , it follows that for constant @xmath11 , the qle is monotonic decreasing from @xmath253 to large @xmath10 . here we analyze the qle for the region @xmath37 , which has some part of the surface inside the ergosphere . we also recheck the results for @xmath21 . these results imply that : ( i ) for increasing @xmath11 , the qle is _ decreasing _ and ( ii ) it is monotonically decreasing w.r.t . @xmath10 . about ( i ) , if we consider the irreducible mass @xmath12 @xcite @xmath254 then the qle is increasing w.r.t . @xmath12 , where @xmath255 and the equality holds for the schwarzschild limit ( @xmath256 ) and @xmath257 for @xmath30 . it is not difficult to see that @xmath12 is decreasing w.r.t . @xmath11 . so that @xmath258 respectively corresponds to @xmath259 . for the region @xmath37 , it should be considered into two parts : @xmath260 note that for @xmath256 , the schwarzschild case , the irreducible mass @xmath261 and its quasi - local energy at the horizon @xmath253 is @xmath262 . the following figures are the plots by the approximation of the boundary integration at @xmath248 : @xmath263 , where we pick @xmath264 , and @xmath265 starting from @xmath266 to the last step which @xmath267 . each step with the interval @xmath264 . note that @xmath268 . 0.5 , @xmath116 . + the curves are for constant @xmath11 and constant @xmath12 from the lower to the upper bound with the interval @xmath269 . they show that ( i ) qle is decreasing w.r.t . @xmath11 and increasing w.r.t . ( ii ) qle is monotonically decreasing w.r.t . @xmath10.,title="fig : " ] 0.5 , @xmath116 . + the curves are for constant @xmath11 and constant @xmath12 from the lower to the upper bound with the interval @xmath269 . they show that ( i ) qle is decreasing w.r.t . @xmath11 and increasing w.r.t . ( ii ) qle is monotonically decreasing w.r.t . @xmath10.,title="fig : " ] 0.5 , @xmath116 . + the curves are for constant @xmath11 and the irreducible mass @xmath12 from the lower to the upper bound with the interval @xmath269 . they show that ( i ) qle is decreasing w.r.t . @xmath11 and increasing w.r.t . ( ii ) qle is monotonic decreasing w.r.t . note that at @xmath256 ( corresponding to @xmath261 , the spherical symmetric case ) , the outer horizon @xmath270 is only one point , which is @xmath271.,title="fig : " ] 0.5 , @xmath116 . + the curves are for constant @xmath11 and the irreducible mass @xmath12 from the lower to the upper bound with the interval @xmath269 . they show that ( i ) qle is decreasing w.r.t . @xmath11 and increasing w.r.t . ( ii ) qle is monotonic decreasing w.r.t . note that at @xmath256 ( corresponding to @xmath261 , the spherical symmetric case ) , the outer horizon @xmath270 is only one point , which is @xmath271.,title="fig : " ] 0.5 , @xmath116 . + these figures show the numerical approximation of qle outside @xmath253 . figure ( a ) agrees with the result in @xcite ; figure ( b ) is the result for the irreducible mass @xmath12.,title="fig : " ] 0.5 , @xmath116 . + these figures show the numerical approximation of qle outside @xmath253 . figure ( a ) agrees with the result in @xcite ; figure ( b ) is the result for the irreducible mass @xmath12.,title="fig : " ] 0.5 : @xmath272 : dashed , @xmath6 : dotted , @xmath273 : thickness . , title="fig : " ] 0.5 : @xmath272 : dashed , @xmath6 : dotted , @xmath273 : thickness . , title="fig : " ] we analyzed the critical value of the kerr qle ( which is the brown - york mass ) , which depends on the the choice of constant @xmath2 surface @xmath106 in the boyer - lindquest coordinate system . the surface @xmath106 can not be embedded isometrically into @xmath274 if @xmath11 is too large . we found @xmath275 is the limit of the isometric embedding , if @xmath276 ( see ) then the gauss curvature @xmath117 is negative at the poles whereas @xmath144 implies the existence of an isometric embedding . we consider the region outside the outer horizon , i.e. @xmath277 , which is outside the embedding limit @xmath31 if @xmath278 . however , if @xmath29 , the isometric embedding is limited to @xmath1 . on the other hand , we are also interested in the positivity of the quasi - local energy , and the critical value is the brown - york mass for which the integrand is proportional to @xmath33 . it is obvious that if @xmath279 then the positivity is satisfied , which is no problem for small rotation . we found the condition that if @xmath3 , then @xmath279 . it is interesting that the three curves , @xmath34 , @xmath31 and @xmath6 intersect at the point @xmath7 . so that the condition for @xmath279 is the region @xmath13 ( see ) . for non - slow rotation , in the region @xmath280 , the integrand @xmath33 is not positive everywhere on the surface , but the numerical results imply that over the integration the qle is positive inside the triangle - like region . furthermore , it is monotonically decreasing with respect to @xmath10 . regarding the qle decreasing corresponding to an increasing @xmath11 , from _ the second law _ of black hole dynamics , the black hole area _ can never decrease_. the area of a kerr black hole is @xmath281 ( * ? ? ? * box 33.4 ) . suppose there are two black holes with the same mass @xmath9 but different angular momentum per unit mass @xmath282 , @xmath283 , it is not difficult to see that if @xmath284 then @xmath285 . the larger black hole has larger irreducible mass . it is reasonable that a larger black hole would carry larger quasi - local energy . so that if we replace @xmath11 by the irreducible mass @xmath12 , the qle is increasing w.r.t . @xmath12 . by the results of chen - wang - yau , the critical value of kerr qle is a global minimum for the region @xmath13 , and an interesting question is whether this is also true in the triangle - like region for non - slow rotation . we would like to thank professor james m. nester and professor naqing xie for helpful suggestions and discussions . c. yu is partially supported by the yangfan project of guangdong province and nsfc 11571215 . j. l. liu is supported by the china postdoctoral science foundation 2016m602497 and partially supported by nsfc 61601275 . chen c .- m . , nester j. m. and tung r .- s . , `` _ gravitational energy for gr and poincar gauge theories : a covariant hamiltonian approach , _ '' , int . d * 24 * , 1530026 ( 2015 ) [ arxiv:1507.07300v1 ] . nester j. m. , chen c .- m . , liu j .- l . and sun gang , `` _ a reference for the covariant hamiltonian boundary term , _ '' spring proceedings in physics 157 , relativity and gravitation : 100 years after einstein in prague , page 177184 ( 2014 ) [ arxiv:1210.6148v1 [ gr - qc ] ] . sun gang , chen c .- m . , liu j .- l . and nester j. m. , `` _ an optimal choice of reference for the quasi - local gravitational energy and angular momentum , _ '' chin . j. phys . * 52 * , 111 ( 2014 ) [ arxiv : 1307.1309v1 [ gr - qc ] ] .
in this paper , we study the quasi - local energy ( qle ) and the surface geometry for kerr spacetime in the boyer - lindquist coordinate without taking the slow rotation approximation . we also consider in the region @xmath0 , which is inside the ergosphere . for a certain region , @xmath1 , the gaussian curvature is positive of the surface with constant @xmath2 , and for @xmath3 the critical value of the qle is positive . we found that the three curves : the outer horizon @xmath4 , @xmath5 and @xmath6 intersect at the point @xmath7 , which is the limit for the horizon to be isometrically embedded into @xmath8 @xcite . the numerical result indicates that the kerr qle is monotonically decreasing to adm @xmath9 from the region inside the ergosphere to large @xmath10 . for increasing @xmath11 , the qle is decreasing but for increasing irreducible mass @xmath12 , qle is increasing . from the results of chen - wang - yau @xcite , we conclude that in a certain region @xmath13 , the critical value of the kerr qle is a global minimum .